Volume 144 Number 3 February 4, 2011
www.cell.com
Sugar Trans Transferas ferase e Unve Un veils ils Pr Protease tease Bla Blade de Enha En hancers ncers—Discovery —Discovery a and d Mechanism echanisms
See what you’ve been missing
Cell-Cell Interaction / Immune Synapse
Co-localization
Nuclear Localization
Phagocytosis / Internalization
The ImageStreamX imaging flow cytometer. The revolutionary ImageStreamX imaging flow cytometer combines the power of digital fluorescence microscopy with the speed and sensitivity of flow cytometry to enable appli-cations that would be impossible with either technique alone. The ImageStream X boasts an impressive array of features: • 1,000 Cells per Second • 12 Images per Cell
Cell Death / Autophagy
• Five Lasers • 60X/40X/20X Magnification • Extended Depth of Field • Multiwell AutoSampler
Shape Change
ImageStream users have published over 150 peer-reviewed papers to-date. Isn’t it time to see what you’ve been missing?
Discover the power of imaging flow cytometry at www.amnis.com/Cell
www.amnis.com/Cell
! W E N
Intelligent design
inCuSaFe™ copper enriched stainless steel interior
+
+
+
+
Single-beam, dual capture infrared CO2 sensor
+
Zirconia O2 Sensor for multigas model
SafeCell UV protection in situ
Hydrogen peroxide vapor sterilization in situ
1
Integrated Cooling Coil (optional)
The industry’s first in situ H2O2 sterilization with the fastest turnaround. For maximum productivity in clinical, general purpose or the most highly compliant GMP applications, the new SANYO Sterisonic™ GxP CO2 incubator offers an impressive return on investment. With multiple contamination control safeguards, exclusive on-board H2O2 sterilization, FDA-21CFR data capture and graphical LCD display, the Sterisonic™ GxP rewards good laboratory technique with performance you can trust. Learn more, visit www.sterisonic.com or call 800-858-8442. pictured above: Sterisonic™ GxP MCO-19AICUVH with rapid H2O2 vapor sterilization system.
FREEE R OFF [
!
E S O N L IN D E TA IL
]
FREE! FREE!
H2O2 sterilization system accessories ($571 Value), plus BD labware consumables ($250 Value). Limited time upgrade offer with purchase of the Sterisonic™ GxP MCO-19AIC(UVH). BD stem cell starter kit with Sterisonic™ GxP quote. No purchase necessary! Act now. Supplies are limited. ($150 Value)
www.sterisonic.com
©2011 Sanyo Biomedical
BE THE FIRST
to read the latest issue of any Cell Press journal.
Register for Cell Press Email Alerts and get the complete table of contents as soon as the issue publishes online — FREE! Cell Press Email Alerts deliver the news, research, and commentaries featured in each journal’s latest issue, including the full title of every article, direct links to the articles, and the complete author list. Plus, to save you time, each research article has a brief summary highlighting its significant findings. You don’t have to be a subscriber to sign up for Cell Press Email Alerts. While subscribers have instant access to the full text of all articles listed in the Email Alerts, non-subscribers can read the abstracts of all articles as well as the full text of the issue’s Featured Article.
www.cellpress.com
Editor Emilie Marcus Senior Deputy Editor Elena Porro Deputy Editor Robert Kruger Scientific Editors Karen Carniol Kara Cerveny Michaeleen Doucleff Fabiola Rivas Niki Scaplehorn Lara Szewczak Senior Managing Editor Meredith Adinolfi Deputy Managing Editor Andy Smith Lead Illustrator Andrew A. Tang Illustrators Yvonne Blanco Kate Mahan Production Staff Reyna Clancy Editorial Assistant Mary Beth O’Leary
Editorial Board Abul Abbas C. David Allis Genevie`ve Almouzni Uri Alon Angelika Amon Johan Auwerx Richard Axel Cori Bargmann Konrad Basler Bonnie Bassler David Baulcombe Jeffrey Benovic Carolyn Bertozzi Wendy Bickmore Elizabeth Blackburn Joan Brugge Lewis Cantley Joanne Chory David Clapham Andrew Clark Hans Clevers Stephen Cohen Pascale Cossart Jeff Dangl Ted Dawson Pier Paolo di Fiore Marileen Dogterom Julian Downward Bruce Edgar Steve Elledge Anne Ephrussi Ronald Evans Witold Filipowicz Marco Foiani Elaine Fuchs Yukiko Goda Stephen Goff Joe Goldstein
Douglas Green Leonard Guarente Taekjip Ha Daniel Haber Ulrike Heberlein Nobutaka Hirokawa Mark Hochstrasser Arthur Horwich Tony Hunter James Hurley Richard Hynes Thomas Jessell Narry Kim Mary-Claire King David Kingsley Frank Kirchhoff Richard Kolodner John Kuriyan Robert Lamb Mark Lemmon Beth Levine Wendell Lim Jennifer Lippincott-Schwartz Dan Littman Richard Losick Scott Lowe Tom Maniatis Matthias Mann Kelsey Martin Joan Massague´ Iain Mattaj Satyajit Mayor Ruslan Medzhitov Craig Mello Tom Misteli Tim Mitchison Alex Mogilner Paul Nurse Roy Parker
Dana Pe’er Kathrin Plath Carol Prives Klaus Rajewsky Venki Ramakrishnan Rama Ranganathan Anne Ridley James Roberts Alexander Rudensky Helen Saibil Joshua Sanes Randy Schekman Ueli Schibler Joseph Schlessinger Hans Scho¨ler Trina Schroer Geraldine Seydoux Kevan Shokat Pamela Sklar Nahum Sonenberg James Spudich Paul Sternberg Bruce Stillman Azim Surani Keiji Tanaka Craig Thompson Robert Tjian Ju¨rg Tschopp Ulrich von Andrian Gerhard Wagner Detlef Weigel Alan Weiner Jonathan Weissman Matthew Welch Tian Xu Shinya Yamanaka Marino Zerial Xiaowei Zhuang Huda Zoghbi
Cell Office Cell, Cell Press, 600 Technology Square, 5th Floor, Cambridge, Massachusetts 02139 Phone: (+1) 617 661 7057, Fax: (+1) 617 661 7061, E-mail:
[email protected] Online Publication: http://www.cell.com Cell (ISSN 0092-8674) is published biweekly by Cell Press, 600 Technology Square, 5th Floor, Cambridge, Massachusetts 02139. The institutional subscription rate for 2011 is $1,605 (US and Canada) or $1,847 (elsewhere). The individual subscription rate is $320 (US and Canada) or $363 (elsewhere). The individual copy price is $50. Periodicals postage paid at Boston, Massachusetts and additional mailing offices. Postmaster: send address changes to Elsevier Customer Service Americas, Cell Press Journals, 11830 Westline Industrial Drive, St. Louis, MO 63146, USA. The paper used in this publication meets the requirments of ANSI/NISO Z39.48-1992 (Permanence of Paper). Printed by Dartmouth Printing Company, Hanover, NH.
Expand your stem cell library and save today on the latest books on stem cells and regenerative medicine Stem Cells
Stem Cell Anthology
Scientific Facts and Fiction
From Stem Cell Biology, Tissue Engineering, Regenerative Medicine, Cloning and Stem Cell Methods
Christine Mummery, Ian Wilmut, Anja Van de Stolpe and Bernard Roelen November 2010 | 400 pages | Paperback | $79.95 | €57.95 | £48.99 | ISBN: 9780123815354
Principles of Regenerative Medicine, 2nd Edition
Bruce M. Carlson October 2009 | 450 pp. | Hardback | $150.00 | €100.00 | £95.00 |AU$222.00 | ISBN: 9780123756824
Essential Stem Cell Methods
Anthony Atala and Robert Lanza November 2010 | 1400 pages | Hardback | $199.95 | €143.00 | £125.00 | ISBN: 9780123814227
A Volume in the Reliable Lab Solutions Series
Heart Development and Regeneration, 2-Volume Set
Tissue Engineering
Robert Lanza and Irina Klimanskaya April 2009 | 628 pp. | Paperback | $75.00 | €50.95 | £45.99 |AU$111.00 | ISBN: 9780123750617
Nadia Rosenthal and Richard P. Harvey June 2010 | 1072 pp. | Hardback | $199.95 | €143.00 | £125.00 | AU$296.00 | ISBN: 9780123813329
Clemens van Blitterswijk, Peter Thomsen, Jeffrey Hubbell, Ranieri Cancedda, Anders Lindahl Sahlgrenska, Jerome Sohier and David F. Williams March 2008 | 760 pp. | Hardback | $115.00 | €76.95 | £69.99 |AU$170.00 | ISBN: 9780123708694
Essentials of Stem Cell Biology, 2nd Edition
Human Stem Cell Manual
Robert Lanza, Roger Pedersen, John Gearhart, E. Donnall Thomas, Brigid Hogan, James Thomson, Douglas Melton and Sir Ian Wilmut June 2009 | 600 pp. | Hardback | $199.95 | €134.00 | £125.00 | AU$302.00 | ISBN: 9780123747297
Jeanne F. Loring, Robin L. Wesselschmidt and Philip H. Schwartz June 2007 | 488 pp. | Spiral bound | $88.95 | €59.95 | £53.99 |AU$132.00 | ISBN: 9780123704658
Foundations of Regenerative Medicine Clinical and Therapeutic Applications Anthony Atala, Robert Lanza, James Thomson and Robert Nerem September 2009 | 750 pp. | Hardback | $99.95 | €66.95 | £60.99|AU$148.00 | ISBN: 9780123750853
A Laboratory Guide
Handbook of Stem Cells 2-Volume Set with CD-ROM Vol. 1–2 Vol. 1 – Embryonic Stem Cells Vol. 2 – Adult & Fetal Stem Cells Robert Lanza, Roger Pedersen, Helen Blau, E. Donnall Thomas, John Gearhart, James Thomson, Brigid Hogan, Catherine Verfaillie, Douglas Melton, Irving Weissman, Malcolm Moore and Michael West September 2004 | 1,760 pp. | Hardback | $566.00 | €380.00 | £345.00 | AU$817.00 | ISBN: 9780124366435
Cell Stem Cell subscribers save 25% on their book order Secure ordering online at elsevierdirect.com Enter promo code 28024 at check out Prices and publication dates subject to change without notice.
Cell Press President & CEO Lynne Herndon Editor in Chief, Vice President of Content Development Emilie Marcus Vice President of Business Development Joanne Tracy Vice President of Web Development and Operations Keith Wollman Senior Product Manager Mark Van Hussen Director of Marketing Jonathan Atkinson Production Manager Meredith Adinolfi
Display Advertising Northeast/Mid-Atlantic: Victoria Macomber, ph: 508 928 1255; fax: 508 928 1256; e-mail:
[email protected] Midwest/Southeast/Eastern Canada: Inez Herrero-Redman, ph: 585 678 4395; fax: 585 678 4722; e-mail: i.herrero@elsevier. com Northwest/Southwest/Western Canada: Lori Young, ph: 646 370 6312; fax: 212 462 1915; e-mail:
[email protected] California: Elizabeth Loennborn, ph: 714 655 1877; fax: 214 452 9627; e-mail:
[email protected] UK/Europe: James Kenney, ph: +44 20 7424 4216; fax: +44 18 6585 3136; e-mail:
[email protected] Asia: Wendy Xie, ph: +86 10 8520 8827; e-mail: w.xie@ elsevier.com Classified Advertising United States and Canada: Gordon Sheffield, Key Account Manager, ph: 617 386 2189; fax: 617 397 2805; e-mail: g.sheffi
[email protected] Press Officer Cathleen Genova
UK, Europe, and Asia: Sabrina Dodge, Key Account Manager, ph: +44 20 7424 4997; fax: +44 18 6585 3136; e-mail:
[email protected] ª2011 Elsevier Inc. All rights reserved. This journal and the individual contributions contained in it are protected under copyright by Elsevier Inc., and the following terms and conditions apply to their use:
advances in the medical sciences, in particular, independent verification of diagnoses and drug dosages should be made. Although all advertising material is expected to conform to ethical (medical) standards, inclusion in this publication does not constitute a guarantee or endorsement of the quality or value of such product or of the claims made of it by its manufacturer.
Photocopying: Single photocopies of single articles may be made for personal use as allowed by national copyright laws. Permission of the Publisher and payment of a fee are required for all other photocopying, including multiple or systematic copying, copying for advertising or promotional purposes, resale, and all forms of document delivery. Special rates are available for educational institutions that wish to make photocopies for nonprofit educational classroom use. For information on how to seek permission, visit www.elsevier. com/permissions or call (+44) 1865 843830 (UK) / (+1) 215 239 3804 (US). Permissions: For information on how to seek permission, visit www.elsevier.com/ permissions or call (+44) 1865 843830 (UK) / (+1) 215 239 3804 (US). Derivative Works: Subscribers may reproduce tables of contents or prepare lists of articles including summaries for internal circulation within their institutions. Permission of the Publisher is required for resale or distribution outside the institution. Permission of the Publisher is required for all other derivative works, including compilations and translations (please consult www.elsevier.com/permissions). Electronic Storage or Usage: Permission of the Publisher is required to store or use electronically any material contained in this journal, including any article or part of an article (please consult www.elsevier.com/permissions). Except as outlined above, no part of this publication may be reproduced, stored in a retrieval system, or transmitted in any form or by any means, electronic, mechanical, photocopying, recording, or otherwise, without prior written permission of the Publisher. Notice: No responsibility is assumed by the Publisher for any injury and/or damage to persons or property as a matter of products liability, negligence, or otherwise, or from any use or operation of any methods, products, instructions, or ideas contained in the material herein. Because of rapid
Reprints: Article reprints are available through Cell’s reprint service; for information, contact Nicholas Pavlow (e-mail:
[email protected]; ph: (+1) 212 633 3960). Subscription Orders and Inquiries: Mail, fax, or e-mail address changes to Elsevier Customer Service Americas, allowing 4–6 weeks for processing. Lost or damaged issues will be replaced, subject to availability, if Cell Press is notified within the claim period (US and airmail delivery: 3 months from issue date; surface delivery: 4 months from issue date). Periodical delivery in the US can take up to 3 weeks. Airmail delivery can take 2–4 weeks. The price of a single copy of Cell is $50 (excluding special issues). All orders must be prepaid and in writing. Please include the volume and issue number, payment (check or credit card, MasterCard, Visa, or American Express only), and a delivery address. Allow 4–6 weeks for delivery. Mailing address: Elsevier Customer Service Americas, Cell Press Journals, 11830 Westline Industrial Drive, St. Louis, MO 63146, USA. Toll-free phone within USA/Canada: 866 314 2355; phone for outside US/Canada: (+1) 314 453 7038; fax: (+1) 314 523 5170; e-mail:
[email protected]; internet: www.cellpress.com or <www.cell.com>. Funding Body Agreements and Policies: Elsevier has established agreements and developed policies to allow authors whose articles appear in journals published by Elsevier to comply with potential manuscript archiving requirements as specified as conditions of their grant awards. To learn more about existing agreements and policies, visit http://www.cell.com/cellpress/FundingBodyAgreements. Guide for Authors: For a full and complete guide for authors, please go to www.cell.com/ authors.
MASTERING CHANGE
Breakthrough in 5-hmC quantitation for epigenetics Interested in simplifying the study of DNA methylation, particularly 5-hydroxymethylcytosine? Try the EpiMark™ 5-hmC and 5-mC Analysis Kit, a robust enzymatic method for the locus-specific detection of methylated (5-mC) and hydroxymethylated (5-hmC) cytosine. As the first commercially available PCR-based kit to reproducibly identify and quantitate the presence of 5-hmC, this simple 3-step protocol will expand your potential for epigenetics research and biomarker discovery. Identify and quantitate methylation states with the EpiMark™ 5-hmC and 5-mC Analysis Kit
Advantages: s Reproducible quantitation of 5-hmC and 5-mC
% 5-hmC
s Easy-to-use protocols
% 5-mC % Unmethylated C
s Compatible with existing techniques (PCR)
100
s Amenable to high throughput
80 60
Visit neb.com/epigenetics to learn more, and explore the complete listing of EpiMark validated products from NEB.
40 20 0 Brain
Liver
Heart
Spleen
Analysis of the different methylation states in Balb/C mouse tissue samples shows a variation in the amount of 5-hmC present at locus 12.
CLONING & MAPPING
DNA AMPLIFICATION & PCR
RNA ANALYSIS
PROTEIN EXPRESSION & ANALYSIS
GENE EXPRESSION & CELLULAR ANALYSIS
www.neb.com
Leading Edge Cell Volume 144 Number 3, February 4, 2011 IN THIS ISSUE SELECT 315
Molecular Pathology of Alzheimer’s Disease
PREVIEWS 319
Cell Migration: GSK3b Steers the Cytoskeleton’s Tip
G. Yucel and A.E. Oro
321
A Versatile Sugar Transferase Makes the Cut
J.A. Hanover
323
Need Tension Relief Fast? Try Caveolae
S. Mayor
325
Getting Cells and Tissues into Shape
D. Odde
REVIEW 327
Functional and Mechanistic Diversity of Distal Transcription Enhancers
M. Bulger and M. Groudine
SNAPSHOT 454
Chromatin Remodeling: ISWI
A.N. Yadon and T. Tsukiyama
Antibodies and Related Reagents for Signal Transduction Research
™
XP
Monoclonal Antibodies, eXceptional Performance
™
Unparalleled product quality, validation, and technical support.
XP monoclonal antibodies are generated using XMT™ technology, a proprietary monoclonal method developed at Cell Signaling Technology. This technology provides access to a broad range of antibody-producing B cells unattainable with traditional monoclonal technologies, allowing more comprehensive screening and the identification of XP monoclonal antibodies.
eXceptional specificity As with all of our antibodies, the antibody is specific to your target of interest, saving you valuable time and resources.
+ eXceptional sensitivity The antibody will provide a stronger signal for your target protein in cells and tissues, allowing you to monitor expression of low levels of endogenous proteins, saving you valuable materials.
+ eXceptional stability and reproducibility XMT technology combined with our stringent quality control ensures maximum lot-to-lot consistency and the most reproducible results.
= eXceptional Performance™ XMT Technology coupled with our extensive antibody validation and stringent quality control delivers XP monoclonal antibodies with eXceptional Performance in the widest range of applications. Above: Confocal IF analysis of rat cerebellum using β3-Tubulin (D71G9) XP™ Rabbit mAb #5568 (green) and Neurofilament-L (DA2) Mouse mAb #2835 (red). Blue pseudocolor = DRAQ5® #4084 (fluorescent DNA dye).
For additional information and a complete list of available XP™ Monoclonal Antibodies visit…
www.cellsignal.com Orders (toll-free) 1-877-616-2355
| Technical support (toll-free) 1-877-678-8324
[email protected] | Inquiries
[email protected] | Environmental Commitment eco.cellsignal.com
© 2011 Cell Signaling Technology, Inc. XMT™, XP™ , eXceptional Performance™, CST™, and Cell Signaling Technology® are trademarks of Cell Signaling Technology, Inc. / DRAQ5® is a registered trademark of Biostatus Limited
XP™ monoclonal antibodies are a line of high quality rabbit monoclonal antibodies exclusively available from Cell Signaling Technology. Any product labeled with XP has been carefully selected based on superior performance in all approved applications.
Articles Cell Volume 144 Number 3, February 4, 2011 341
Skin Stem Cells Orchestrate Directional Migration by Regulating Microtubule-ACF7 Connections through GSK3b
X. Wu, Q.-T. Shen, D.S. Oristian, C.P. Lu, Q. Zheng, H.-W. Wang, and E. Fuchs
353
The RNA Exosome Targets the AID Cytidine Deaminase to Both Strands of Transcribed Duplex DNA Substrates
U. Basu, F.-L. Meng, C. Keim, V. Grinstein, E. Pefanis, J. Eccleston, T. Zhang, D. Myers, C.R. Wasserman, D.R. Wesemann, K. Januszyk, R.I. Gregory, H. Deng, C.D. Lima, and F.W. Alt
364
Structural Basis of the 9-Fold Symmetry of Centrioles
D. Kitagawa, I. Vakonakis, N. Olieric, M. Hilbert, D. Keller, V. Olieric, M. Bortfeld, M.C. Erat, I. Flu¨ckiger, P. Go¨nczy, and M.O. Steinmetz
376
O-GlcNAc Transferase Catalyzes Site-Specific Proteolysis of HCF-1
F. Capotosti, S. Guernier, F. Lammers, P. Waridel, Y. Cai, J. Jin, J.W. Conaway, R.C. Conaway, and W. Herr
389
Osh Proteins Regulate Phosphoinositide Metabolism at ER-Plasma Membrane Contact Sites
C.J. Stefan, A.G. Manford, D. Baird, J. Yamada-Hanff, Y. Mao, and S.D. Emr
402
Cells Respond to Mechanical Stress by Rapid Disassembly of Caveolae
B. Sinha, D. Ko¨ster, R. Ruez, P. Gonnord, M. Bastiani, D. Abankwa, R.V. Stan, G. Butler-Browne, B. Vedie, L. Johannes, N. Morone, R.G. Parton, G. Raposo, P. Sens, C. Lamaze, and P. Nassoy
THEORIES 414
Influence of Cell Geometry on Division-Plane Positioning
N. Minc, D. Burgess, and F. Chang
(continued)
'O8TREME
2OCHE.EW8 TREME'%.% 4RANSFECTION2EAGENTS 5SEOURNEWPOWERFULTRANSFECTIONREAGENTSTO
2,5X LUCIFERASE
%FFICIENTLYTRANSFECTABROADSPECTRUMOFCELLS INCLUDING DIFFICULT TO TRANSFECTCELLLINES
!VOIDTIME CONSUMINGOPTIMIZATIONSTEPSUSINGAFAST RELIABLEPROTOCOL
%S P M #O
, 48
0
FE
P
8 M #O
P M #O
M
P
!T
%F P M
P
,
#O
#O
+
0 ( M
&ORMOREINFORMATION VISIT WWWX TREMEGENEROCHECOM
8
TRE
M
#O
% %. E'
E' M TRE 8
#E
LLS
A
%.
LO
%
NE
'ENERATEPHYSIOLOGICALLYRELEVANTRESULTSUSINGLOW CYTOTOXICTRANSFECTIONREAGENTS
8 TREME'%.%AND(0$.!4RANSFECTION2EAGENTSCONSISTENTLY OUTPERFORMCOMPETITORREAGENTS(E,ACELLSWERETRANSFECTEDWITHA LUCIFERASEEXPRESSIONPLASMIDUSING8 TREME'%.%4RANSFECTION2EAGENTS ANDSIXDIFFERENTCOMPETITORREAGENTS#OMP,+n%S
&ORLIFESCIENCERESEARCHONLY .OTFORUSEINDIAGNOSTICPROCEDURES 8 42%-%'%.%ISATRADEMARKOF2OCHE ¹2OCHE$IAGNOSTICS!LLRIGHTSRESERVED
2OCHE$IAGNOSTICS'MB( 3ANDHOFER3TRAE -ANNHEIM 'ERMANY
427
Control of the Mitotic Cleavage Plane by Local Epithelial Topology
W.T. Gibson, J.H. Veldhuis, B. Rubinstein, H.N. Cartwright, N. Perrimon, G.W. Brodland, R. Nagpal, and M.C. Gibson
RESOURCE 439
Reference Maps of Human ES and iPS Cell Variation Enable High-Throughput Characterization of Pluripotent Cell Lines
C. Bock, E. Kiskinis, G. Verstappen, H. Gu, G. Boulting, Z.D. Smith, M. Ziller, G.F. Croft, M.W. Amoroso, D.H. Oakley, A. Gnirke, K. Eggan, and A. Meissner
POSITIONS AVAILABLE
On the cover: Human OGT is responsible for the O-GlcNAcylation of many nuclear and cytoplasmic proteins. Here, Capotosti et al. find that OGT also possesses proteolytic activity as it activates the epigenetic cell-cycle regulator HCF-1 by catalyzing its proteolytic maturation. The cover image evokes these results as a cutting blade added to the screwdriver function of a multipurpose pocketknife. The illustration is by David Kendall, a.k.a. ‘‘Pixelmechanic,’’ the winner of an online contest at 99designs.com.
Announcing an innovative new textbook from Academic Cell Primer to The Immune Response, Academic Cell Update Edition By Tak W. Mak and Mary Saunders
Primer to The Immune Response, Academic Cell Update Edition, is an invaluable resource for students who need a concise but complete and understandable introduction to immunology. Academic Cell textbooks contain premium journal content from Cell Press and are part of a new cutting-edge textbook/journal collaboration designed to help today’s instructors teach students to “think like a scientist.”
academiccell.com
Primer to The Immune Response Academic Cell Update Edition Tak W. Mak The Campbell Family Institute for Breast Cancer Research, Ontario, Canada Mary Saunders The Campbell Family Institute for Breast Cancer Research, Ontario, Canada Paperback/456 pages ISBN: 9780123847430 $79.95/£54.99/€64.95
Academic Cell is a dynamic textbook publishing partnership between Academic Press and Cell Press, two market-leading publishers bringing scientific advances from the world of life science research into the classroom. Order online now from: elsevierdirect.com/9780123847430 Request and examination copy from textbooks.elsevier.com
Twitter.com/academiccell
Facebook.com/academiccell
Leading Edge
In This Issue Calling up Hair Follicle Stem Cells PAGE 341
Wound healing in skin relies on hair follicle stem cells whose migration is governed by Wnt signaling. Wu et al. now reveal how the Wnt pathway integrates with downstream cytoskeletal dynamics to direct stem cell mobilization. They show that the kinase GSK3b phosphorylates the microtubule/F-actin crosslinking protein ACF7 to uncouple it from microtubules. Wnt signaling inhibits GSK3b activity to enable polarization and migration of the follicle stem cells.
At the Core of the Centriole PAGE 364
The structure of the centriole is characterized by nine-fold radial symmetry. Kitagawa et al. now provide insight into how this architecture is achieved. They show that the centriolar protein SAS-6 self-assembles into rod-shaped homodimers. Oligomerization of these homodimers is essential for centriole formation in C. elegans and human cells. Structural modeling of the related Chlamydomonas protein Bld12p reveals that nine homodimers form a ring that resembles a centriolar structure. Strikingly, recombinant Bld12p can self-assemble into these structures.
The Exosome AIDs Antibody Diversity PAGE 353
In B cells, the AID protein initiates immunoglobulin class switch recombination (CSR) by deaminating both template and nontemplate DNA strands of immunoglobulin genes. Targeting both strands is necessary to generate the doublestranded breaks that initiate CSR. How AID, which only functions on single-stranded DNA, can deaminate both DNA strands has been an enigma. Basu et al. now show that the RNA exosome targets AID to both strands, revealing a role for the noncoding RNA surveillance machinery in generating antibody diversity.
Sweetening up Proteolysis PAGE 376
The cell-cycle regulator HCF-1 undergoes proteolytic cleavage to create two peptides with differing functions in mitotic progression. Capotosti et al. now report that this cleavage event is unusual in that it is carried out by a glycosylation enzyme, OGT. OGT glycosylates and cleaves HCF-1, and both events are required for the function of each resulting peptide. The findings reveal an unexpected enzymatic activity for OGT and a nexus between glycosylation and proteolysis in cell-cycle regulation.
Caveolae, First Responders to Membrane Stress PAGE 402
Small membrane invaginations, caveolae, have been associated with endocytosis, signaling, and lipid metabolism. Sinha et al. now report an important role for caveolae in absorbing stress on the cell membrane. They show that caveolae flatten and disassemble upon osmotic swelling or stretching and thus buffer surges in membrane tension that could lead to rupture. These findings support a new role of caveolae as a membrane reservoir that allows cells to quickly accommodate sudden and acute mechanical stresses prior to any other cellular response. Cell 144, February 4, 2011 ª2011 Elsevier Inc. 311
Scopus is the largest abstract and citation database of peer-reviewed literature and quality web sources with smart tools to track, analyze and visualize research.
enrich your experience
www.scopus.com
Membranes Keep in Touch with Their PI PAGE 389
Phosphoinositide (PI) lipids are essential signaling molecules in cell growth, polarity, and membrane-trafficking pathways. Here Stefan et al. demonstrate that the PI phosphatase Sac1, located in the ER membrane, is regulated by the ORP family of lipid transfer proteins at sites where the ER contacts the plasma membrane. The coordination revealed here may also regulate PI status at additional membrane contact sites to control signaling between the ER and the various compartments along the secretory and endocytic systems.
Shape-Shifting Cells Find Middle Ground PAGE 414
How do cells sense their own shapes? Minc et al. study the effects of cellular geometry on mitosis by coaxing sea urchin embryos into unnaturally shaped wells. They find that microtubules sense cell shape by probing the cellular space and pulling the nucleus into its central position. Based on these observations, they develop a simple and possibly universal rule that predicts how cells of any shape will divide.
Spindle Conforms to Peer Pressure PAGE 427
For almost 150 years it has been appreciated that the long axis of a dividing cell often correlates with the eventual direction of its mitotic cleavage plane. Gibson et al. now report that the geometries of a mitotic cell’s neighbors strongly influence the orientation of the mitotic cleavage plane in epithelial tissue in both plants and animals. Using a mechanical model, they show how simple packing constraints can explain how growing epithelia maintain an ordered structure.
A Scorecard for iPSCs PAGE 439
Bock et al. report extensive molecular profiling of several iPS and ES cell lines. They find that some but not all iPS cell lines are indistinguishable from ES cells. They go on to develop a highly efficient method for identifying the most useful cell lines for a given application, for example, identifying which cell lines will differentiate most efficiently to motor neurons. The molecular characterizations and assays reported should facilitate the use of ES and iPS cell lines in biology and medicine.
Cell 144, February 4, 2011 ª2011 Elsevier Inc. 313
3TRUGGLINGåTOåKEEPåUPåWITHåå THEåLATESTåLIFEåSCIENCEåNEWS
#ELLå$AILY¬.EWS¬!GGREGATORåHASåTHEåSOLUTION 3UBSCRIBEåTOå&2%%å$AILYå.EWSå!LERTSåATåNEWSCELLCOMåANDå GETåTHEåLATESTåLIFEåSCIENCEåHEADLINESåDELIVEREDåTOåYOURåINåBOX
NEWSCELLCOM
Leading Edge
Select: Molecular Pathology of Alzheimer’s Disease b-amyloid accumulation is a hallmark of Alzheimer’s disease pathology, but how does it cause cognitive impairment? New findings described in this issue’s Select examine how b-amyloid impacts signaling events at the synapse at early stages of disease progression. And at later stages, one study suggests that Alzheimer’s patients exhibit a pronounced deficiency in their capacity to metabolize b-amyloid.
Cause and Eph-ect in Cognitive Dysfunction Although oligomeric b-amyloid is often viewed as enemy number one in Alzheimer’s disease, it doesn’t act alone, and unmasking its accomplices may point the way to new therapeutic strategies. An advance of this kind is made by Cisse´ et al. (2011), whose work offers significant insight into how b-amyloid triggers pathological changes at synapses. The authors examine trangenic mice overexpressing human amyloid precursor protein and provide evidence that b-amyloid binds to the receptor tyrosine kinase EphB2. Binding promotes EphB2’s proteolytic degradation, providing an explanation for previous reports that EphB2 is depleted early in the course of the disease. Prior work has also shown that postsynaptic EphB2 regulates NMDA receptor trafficking and function. In the current work, Cisse´ et al. show that restoring EphB2 expression in the granule cells of the dentate gyrus ameliorates phenotypes associated with impaired NMDA receptor function, including the normalization of long-term potentiation and impressive improvements on multiple cognitive tasks. What aspects of EphB2 function should be studied further by future investigators who have their eyes on potential therapeutics? For starters, the interaction between EphB2 and oligomeric b-amyloid should be characterized in detail, here shown to involve EphB2’s fibronectin domains. In addition, EphB2 is reported to directly interact with the NMDA receptor. Hence, affecting this interaction could impact NMDA receptor surface expression. Additional targets to consider for possible inolvement in pathways elicited by b-amyloid could include other known elements of EphB2 downstream signaling, including the kinase Src. Cisse´ et al. (2011). Nature 469, 47–52.
Caspase-3 Gets an Early Start In Alzheimer’s disease, impaired performance on cognitive tasks precedes the wholesale loss of neurons, suggesting that synaptic dysfunction is to blame for early symptoms. Using a mouse model that expresses a mutant allele of human amyloid precursor protein found in familial Alzheimer’s disease (Tg2576 mice), D’Amelio et al. (2011) examine the onset of memory decline and correlate its timing with molecular changes at the synapse. They reveal that caspase-3 activation is elevated in the transgenic mice at the onset of synaptic dysfunction in the hippocampus. Although the caspase-3 activation is above baseline, it is not high enough to trigger apoptosis and is localized to the postsynaptic compartment. Caspase-3 activity is a known activator of the protein phosphatase calcineurin, which in turn dephosphorylates the AMPA receptor subunit GluR1 to regulate its subcellular distribu- b-amyloid induces AMPA receptor tion. Thus, in the current work, the authors provide evidence linking the (AMPAr) dephosphorylation via a pathelevation in caspase-3 activity to a cascade of events that culminate in way involving caspase-3 (Casp-3) and a reduction in the level of AMPA receptors at postsynaptic densities. As calcineurin (CaN). Image courtesy of further evidence in support of the model and looking forward to future ther- F. Cecconi. apeutic possibilities, the authors show that both the molecular and cognitive phenotypes of Tg2576 mice are improved by the administration of caspase-3 inhibitors. A question left open for future work is to establish what processes downstream of b-amyloid production contribute to elevated caspase-3 activation. D’Amelio et al. (2011). Nat. Neurosci. 14, 69–76.
Cell 144, February 4, 2011 ª2011 Elsevier Inc. 315
.%7
Find Your Ideal Job! så3EARCHåJOBSåBYåKEYWORD å LOCATIONååTYPEå så0OSTåYOURåRESUMEåå ANONYMOUSLYå så#REATEåAå*OBå!LERTåå ANDåLETåYOURåIDEALåå JOBålNDåYOU
careers.cell.com
Where and When Does Tau Go Wrong? Tau is a core component of neurofibrillary tangles, one of the hallmarks of Alzheimer’s disease and other dementias, but when and where does altered tau function have its detrimental effects? According to new findings by Hoover et al. (2010), it is early, at synapses, and is independent of neurodegeneration. The authors characterize mice expressing a human tau variant with a mutation associated with frontotemporal dementia (rTgP301L mice) and show that mutant tau is mislocalized to dendritic spines in the first few months of life. This is noteworthy, as the mislocalization occurs prior to the loss of synapses and neurons (which occur at later stages in this mouse model) yet coincides with the onset of impairment in Tau proteins do not normally go to cognitive tasks and synaptic plasticity. Probing the molecular basis for the dendritic spines (red spines, dysfunction, the authors link the mislocalization phenotype to a reduction in triangles) but are mislocalized AMPA and NMDA receptors in the postsynaptic membrane (see also the recent under disease conditions (yellow work of Ittner et al. Cell 142, 387–397). They further show that the mislocalization spines, arrows). Image courtesy of D. Liao. is dependent on tau phosphorylation. When 14 serines and threonines are mutated to mimic hyperphosphorylation, tau is mistargeted to dendritic spines, and when the same residues are mutated to alanines to block phosphorylation, the mislocalization is alleviated. These findings suggest that efforts to impede tau mislocalization at early stages of tauopathies might prevent some of their most debilitating outcomes, thus raising the question: what molecular mechanisms make phosphorylated tau prone to accumulation in dendritic spines? Hoover et al. (2010). Neuron 68, 1067–1081.
Work Slowdown by the b-Amyloid Clean-Up Crew Given that much of the emphasis in the Alzheimer’s research community is placed on understanding the abberrant production of toxic oligomeric species of b-amyloid, the recent report by Mawuenyaga et al. (2010) casts the molecular pathology of the disease in an interesting and provocative light. By collecting cerebral spinal fluid from individuals infused with 13C6-leucine to label newly synthesized proteins, the authors probed the rates of production and clearance of A theoretical graph shows the effects of increased production (red) or decreased clearance (blue) as soluble b-amyloid. A comparison of control individuals and Alz- possibilities for altered b-amyloid metabolism. heimer’s patients surprisingly reveals that the rates of production Image courtesy of R. Bateman. of b-amyloid (both Ab40 and Ab42) are similar between the two groups. Instead, stark differences are observed in their respective rates of b-amyloid clearance, with late-onset Alzheimer’s patients exhibiting a 30% decrease. The authors calculate that this diminishment of b-amyloid metabolism is consistent with a 10 year timeframe for the buildup of b-amyloid in the disease. An intriguing question now is to determine what might account for the decreased capacity for soluble b-amyloid clearance in Alzheimer’s patients. Also, how early in the disease process is altered b-amyloid metabolism a factor? Does this point the way to new therapeutic approaches? If so, what existing soluble b-amyloid clean-up pathways in the central nervous systems might be ramped up to help counter its toxic accumulation? Mawuenyega et al. (2010). Science 330, 1774. Robert P. Kruger
Cell 144, February 4, 2011 ª2011 Elsevier Inc. 317
Stores up to 57,200 samples
l
–86 ºC for long-term storage
l
3.6 ft3 under-thecounter ULT
l
Durable, energy saving design
130.A1.0102.A © 2011 Eppendorf AG
l
NEW!
Something really cool from Eppendorf Eppendorf is now offering New Brunswick Ultra-Low Temperature (ULT) –86 ºC Freezers. Choose from 11 models to satisfy your space requirements.
l
New Brunswick equipment is legendary for design innovation, quality construction and long-term durability. 304 L stainless steel interior, insulated and gasketed inner doors, and voltage stabilizer are only a few of the many standard features found on the New Brunswick freezer line-up.
l
l
l l
Innova® Freezers—utilizes vacuum insulation technology providing 30% more storage capacity Premium Freezers—a cost saving alternative to the Innova line Energy-efficient—consumes less power per Kilowatt-Hour than comparable ULT freezers CFC and HCFC Free—environmentally friendly and non-ozone depleting 5-year limited warranty and 12 years on vacuum insulation
For more information visit www.eppendorfna.com/freezers
www.eppendorf.com Þ Email:
[email protected] In the U.S.: Eppendorf North America, Inc. 800-645-3050 Þ In Canada: Eppendorf Canada Ltd. 800-263-8715
Leading Edge
Previews Cell Migration: GSK3b Steers the Cytoskeleton’s Tip Gozde Yucel1 and Anthony E. Oro1,* 1Program in Epithelial Biology and Stanford University, School of Medicine, CCSR 2145c, 269 Campus Drive, Stanford, CA 94305, USA *Correspondence:
[email protected] DOI 10.1016/j.cell.2011.01.023
Directed cell migration polarizes the cytoskeleton, allowing the cell to move toward migratory cues. In this issue, Wu et al. (2011) demonstrate that the glycogen synthase kinase 3b (GSK3b) controls microtubule architecture and polarized movement of skin stem cells during wound healing in mammals by regulating the microtubule crosslinking protein ACF7. Development, inflammation, and wound healing all critically depend upon directed cell migration when effector cells move toward particular extracellular cues. For example, when skin is injured, stem cells travel from the hair follicle to the wound to rebuild the skin epithelium. Directed cell migration requires changes in the cell’s polarity and remodeling of the cytoskeleton (Ridley et al., 2003), including the regulation of microtubule, actin, and intermediate filaments at the front of the cell (Figure 1). These dynamic structures mediate changes in cellular shape, but they also form a platform for receiving and generating key morphogenic signals. What cytosolic factors transmit migratory cue information from the cellular membrane to these signaling centers on the cytoskeleton? Although this complex process probably depends on many genetically redundant factors, several kinases and phosphatases are emerging as key players. Initial reports have solidified roles for phosphatidylinositol 3-kinase (PI3K) and the lipid phosphatase PTEN in defining the front and backsides of migrating cells, respectively. Now Wu et al. (2011) find that another workhorse kinase, glycogen synthase kinase 3b (GSK3b), orchestrates polarization and reorganization of microtubules in hair follicle stem cells during wound healing. This study, along with another recent report by Zaoui et al. (2010), demonstrates that GSK3b governs microtubule elongation through its control of the microtubule-actin crosslinking protein ACF7 (actin crosslinking factor-7/microtubule and actin crosslinking factor-1).
GSK3b is a serine-threonine kinase that participates in numerous growth factor and morphogen signaling pathways, including the canonical Wnt pathway. In addition, GSK3b has been implicated in a variety of diseases, such as diabetes, Alzheimer’s disease, bipolar disorder, and cancer (Rayasam et al., 2009). Recent studies have found that GSK3b is a central regulator of cell polarity and cytoskeleton dynamics. The kinase regulates the polymerization of filamentous actin (F-actin) at the front and periphery of the cell by activating the Rho GTPase Rac and ADP-ribosylation factor 6 (Arf6), and by repressing p190a-RhoGAP and APC (adenomatous polyposis coli). Further, GSK3b modulates cell-matrix adhesions by activating paxillin and inhibiting focal adhesion kinase (Sun et al., 2009). A fascinating link has also recently emerged between GSK3b and proteins that associate with the plus (+) ends of microtubules, called ‘‘+tip proteins.’’ The +tip proteins regulate the growth of the noncapped, (+) ends of microtubule filaments at the front of migrating cells (Figure 1). The +tip proteins also interact with membrane proteins, such as Rho GTPases, to relay extracellular migratory signals to microtubules (Sun et al., 2009). Recent studies indicate that GSK3b phosphorylates and inactivates certain +tip proteins, such as CLASP2, providing a link between GSK3b kinase and microtubule elongation (Kumar et al., 2009). Another key class of +tip proteins is spectraplakins. The two mammalian spectraplakins, BPAG1/dystonin and ACF7/ MACF, are large proteins (>500 kDa) that orient the cytoskeleton by binding and
crosslinking actin and microtubule filaments. These spectraplakins also form membrane subdomains and scaffolds for signaling complexes. Mice lacking BPAG1 display skin blistering due to the epidermis splitting from the dermis (Roper et al., 2002). In contrast, loss of ACF7 in the epidermis causes no skin or hair phenotype but delays wound repair. Keratinocytes with mutations in ACF display 80% less movement than wild-type control cells in scratch wound assay in vitro. This phenotype was previously ascribed to defective coordination of microtubule growth along F-actin filaments and stabilization of focal adhesions (Wu et al., 2008). Now Wu et al. and Zaoui et al. elegantly link GSK3b activity with the regulation of ACF7-associated microtubules and directed cell migration. Wu and colleagues investigate the wounding phenotype of mice with mutations in the ACF gene and find a significant delay in wound closure. They found that phosphorylation of the carboxyl terminus of ACF7 by GSK3b diminishes ACF7’s binding to microtubules. In addition, using a phospho-specific antibody, the authors show that phospho-ACF7 localizes to the cytoplasm but not to microtubules, confirming that phosphorylation of ACF7 uncouples the +tip protein from microtubules (Figure 1). Next, the authors generate a mutant version of ACF7 that is refractory to phosphorylation by GSK3b, and indeed, this mutant rescues portions of the ACF7 mutant phenotype. Similarly, Zaoui and colleagues investigate GSK3b’s role in the migration of breast cancer cells responding to the epidermal growth factor (EGF) ligand
Cell 144, February 4, 2011 ª2011 Elsevier Inc. 319
It is well-known that PI3K Heregulin through stimulation regulates GSK3b (Cross of the tyrosine kinase reet al., 1995), but it is still ceptor ErbB2. The group unknown whether migratory had previously shown that cues necessarily regulate Heregulin induces directed PI3K, GSK3b, or both kinases cell protrusions by simultaneously. Having cues triggering the Memo (mediaffect different kinases could ator of ErbB2 motility) allow for greater complexity membrane complex at the of cell shapes and motility leading edge. In addition, outcomes. they found that +tip proteins Finally, identifying endogeAPC and CLASP2 mediate microtubule formation at cell nous migratory cues that protrusions during these control directional migration cells’ migration. In their most through GSK3b may lead to recent study, Zaoui and the discovery of new theracolleagues demonstrate that peutic agents. Heregulin and Memo inactivates GSK3b, other EGF ligands are welland this inhibition, in turn, known migratory cues in targets APC and CLASP2 to wound healing and developthe membrane. In addition, ment. They appear, at least Heregulin activity localizes in breast cancer cells, to ACF7 to the plasma membe a migratory cue that alters brane and microtubules, and the cytoskeleton by acting this localization depends on through GSK3b (Zaoui et al., GSK3b and APC. 2010). Together, the data preIn addition, Wnt ligands are Figure 1. GSK3b Is a Central Regulator of the Cytoskeleton during sented by Wu et al. and Zaoui certainly present in wounded Cell Migration In nonmigrating cells, GSK3b is normally active and phosphorylates proteins et al. strongly link GSK3b skin, and noncanonical Wnt that associate with the plus (+) ends of microtubule filaments, called ‘‘+tip activity and the phosphorylasignaling to the cytoskeleton proteins.’’ Phosphorylation uncouples +tip proteins from microtubules (Wu tion status of ACF7 with cell through Rho kinase plays a et al., 2011), inhibiting growth of the filaments. Migratory cues, such as migration and ACF7 associawell-established role in morHeregulin, inactivate GSK3b by phosphorylation (Zaoui et al., 2010), allowing the local extension of microtubules by +tip proteins. This helps to polarize the phogenic processes, such as tion with microtubules. Howcytoskeleton and direct cellular movement toward the migratory cues. convergent extension during ever, the biochemical mechagastrulation (i.e., when tissue nism of ACF7 regulation by GSK3b remains mysterious. Wu and These two studies also generate many restructures to extend along a perpendiccolleagues report that neither the kinase- additional questions about how migratory ular axis). However, a clear role for canonrefractile nor the phosphomimetic cues sculpt the cytoskeleton through key ical Wnt signaling through GSK3b is mutants of ACF7 could rescue the polarity kinases, such as GSK3b. One central currently lacking for wound healing in and directional movement defect in ACF7 question is the nature of the interactions skin (Ito et al., 2007). Nevertheless, given mutant hair follicle stem cells. This between +tip proteins and how their local- the central role of GSK3b and +tip surprising result portends that regulation ization on microtubules regulates actin proteins in regulating the cytoskeleton of ACF7 by GSK3b is more complex and microtubule architecture, cell migra- during directed cell migration, additional than simply inhibition by phosphorylation tion, and polarity. The complex of ACF7 roles in establishing the cytoskeleton’s and suggests that subsets of the multiple and another +tip protein, EB1 (end- global positional device system are sure phosporylation sites on ACF7 may have binding protein 1), appears necessary to appear. distinct functions. Indeed, such behavior and sufficient for microtubule elongation has been observed for the +tip protein in some cells in culture, but how the other CLASP2 (Kumar et al., 2009). Alterna- +tip proteins function in different physio- REFERENCES tively, cycling of ACF7 phosphorylation logical contexts remains unexplored. Cross, D.A., Alessi, D.R., Cohen, P., Andjelkovich, status may be needed for prolonged Mice with mutations in ACF7 develop M., and Hemmings, B.A. (1995). Nature 378, microtubule elongation. Experiments normal skin morphology, suggesting that 785–789. examining the phosphorylation and func- other spectraplakins or +tip proteins Ito, M., Yang, Z., Andl, T., Cui, C., Kim, N., Millar, tion of ACF7 (and other +tip proteins) in must control the polarized growth of the S.E., and Cotsarelis, G. (2007). Nature 447, single-molecule, dynamic microtubule hair follicle or developing skin epithelium. 316–320. Another open question is how the Kumar, P., Lyle, K.S., Gierke, S., Matov, A., Danassays will be necessary to understand precisely how GSK3b operates at the signaling pathways of PI3K intersect with user, G., and Wittmann, T. (2009). J. Cell Biol. leading edge. those of GSK3b in generating cell polarity. 184, 895–908. 320 Cell 144, February 4, 2011 ª2011 Elsevier Inc.
Rayasam, G.V., Tulasi, V.K., Sodhi, R., Davis, J.A., and Ray, A. (2009). Br. J. Pharmacol. 156, 885–898. Ridley, A.J., Schwartz, M.A., Burridge, K., Firtel, R.A., Ginsberg, M.H., Borisy, G., Parsons, J.T., and Horwitz, A.R. (2003). Science 302, 1704–1709.
Roper, K., Gregory, S.L., and Brown, N.H. (2002). J. Cell Sci. 115, 4215–4225. Sun, T., Rodriguez, M., and Kim, L. (2009). Dev. Growth Differ. 51, 735–742. Wu, X., Kodama, A., and Fuchs, E. (2008). Cell 135, 137–148.
Wu, X., Shen, Q., Oristian, D., Lu, C., Zheng, Q., Wang, H., and Fuchs, E. (2011). Cell 144, this issue, 341–352. Zaoui, K., Benseddik, K., Daou, P., Salaun, D., and Badache, A. (2010). Proc. Natl. Acad. Sci. USA 107, 18517–18522.
A Versatile Sugar Transferase Makes the Cut John A. Hanover1,* 1Laboratory
of Cellular and Molecular Biology, NIDDK, National Institutes of Health, Bethesda, MD 20892-0851, USA *Correspondence:
[email protected] DOI 10.1016/j.cell.2011.01.025
The nutrient sensor O-GlcNAc transferase modifies proteins with the O-GlcNAc moiety. In this issue, Capotosti et al. (2011) reveal that O-GlcNAc transferase not only glycosylates the cell-cycle regulator host cell factor 1 but activates it through proteolytic cleavage, providing a surprising link between metabolism and epigenetic regulation of the cell cycle. Host cell factor 1 (HCF-1) is an evolutionarily conserved epigenetic regulator that undergoes an unusual proteolytic processing event to generate stably associated fragments that regulate different stages of the cell cycle (Kristie et al., 2010). In vertebrates, the molecular identity of the factor responsible for HCF-1 cleavage has been cloaked in mystery. With a plot twist worthy of a best-selling detective novel, Capotosti et al. now unmask an unexpected culprit—the nutrient-sensing epigenetic regulator O-linked GlcNAc transferase (OGT) (Figure 1). Their findings provide compelling evidence that OGT has a dual enzymatic role, catalyzing both O-GlcNAcylation and cleavage of HCF-1. This partnering of two epigenetic regulators to control cell-cycle progression has important implications for fields as diverse as viral pathogenesis, stem cell biology, and diseases of aging including cancer, diabetes, and neurodegeneration. Although the HCF-1 protein was originally identified as a target for the Herpes simplex virus VP16 transcriptional activator, subsequent work has shown that
HCF-1 normally coordinates passage through the cell cycle (Goto et al., 1997). The N-terminal subunit, HCF-1N, promotes progression through G1, whereas the C-terminal subunit, HCF-1C, regulates mitosis and cytokinesis. In vertebrates, these domains are separated by six conserved HCF-1PRO repeats, which contain many proline and threonine residues and an invariant glutamate that marks the site of proteolysis. Cleavage at this site is required for HCF-1 regulation of M phase events. Although HCF-1 activity in all animals is regulated by proteolysis, the HCF-1 sequence differs greatly between invertebrates and vertebrates such that different animal lineages employ distinct mechanisms for HCF-1 processing. Lower metazoans employ a threonine-directed endopeptidase (taspase) to sever the two HCF-1 subunits, but vertebrates use a different strategy (Capotosti et al., 2011). Multiple pieces of evidence hint that posttranslational modification of HCF-1 might be important for cleavage. Capotosti et al. show that inhibitors of phosphorylation do not alter processing, but
a presumptive OGT inhibitor and depletion of OGT by RNA interference prevent HCF-1 maturation. Although O-GlcNAcylation of HCF-1 depends on the presence of the threonine-rich HCF-1PRO repeats, it is several threonines in the HCF-1N N-terminal subunit that are glycosylated. OGT does more than simply glycosylate HCF-1 before it is cleaved. Capotosti et al. show that OGT glycosylates and then remains bound to a cleavage-resistant HCF-1 mutant. Furthermore, they find that purified recombinant OGT cleaves the HCF-1PRO repeat in vitro. The authors suggest that O-GlcNAc-catalyzed autoproteolysis of HCF-1 does not efficiently cleave HCF-1PRO repeats in vitro. Providing an in vivo validation of this conclusion, the authors replace the vertebrate HCF-1PRO domain with a taspase cleavage site. Although this variant is efficiently processed by taspase, it is neither recognized nor cleaved by OGT, and this heterologous HCF-1 does not rescue HCF-1 function. Together, these data strongly suggest that OGT-dependent glycosylation and cleavage are critical for the M phase functions of
Cell 144, February 4, 2011 ª2011 Elsevier Inc. 321
HCF-1, building a compelling thus has profound implicacase that OGT is the longtions for fields ranging from sought HCF-1 protease viral pathogenesis and stem (Figure 1). cell biology to diseases of OGT contains no obvious aging. Viruses, like Herpes protease-like domains, so simplex, notoriously subvert how does it catalyze HCF-1 the host cell-cycle machinery proteolysis? The authors for maintaining dormancy examine two possibilities. and viral production (Kristie OGT is part of an enzyme et al., 2010). Integrating both superfamily with known stress and nutrient availhydrolases as members, and ability, host O-GlcNAcylation as such, it could retain cryptic is a likely regulator of viral protease activity. Based on latency. Stem cell pluriposimilarities between the tency and self-renewal may OGT-mediated cleavage of also be regulated by OGT HCF-1PRO and the proteoland HCF-1 as both are ysis of E. coli LexA by its copcomponents of the Oct4rotease, RexA, the authors centered transcription factor favor a model in which OGT network implicated in stem Figure 1. OGT GlcNAcylates and Cleaves HCF-1 for Cell-Cycle promotes a conformational cell pluripotency (Love et al., Progression change in the HCF-1PRO 2010b). In adulthood, obesity O-GlcNAc transferase (OGT) couples the nutrient-dependent synthesis of UDP-GlcNAc to the O-GlcNAcylation of numerous targets including the celland nutrient excess are assodomain, repositioning the cycle regulator host cell factor 1 (HCF-1). OGT is also necessary and sufficient ciated with an increased risk cleavage site and the Nfor proteolysis of the HCF-1PRO domain; together O-GlcNAcylation and terminal O-GlcNAc moieties for diabetes, cardiovascular cleavage of HCF-1 promote changes in chromatin organization that are required for progression through G1-S and M phases of the cell cycle. OGT disease, and cancer; to stimulate hydrolysis (Caand HCF-1 are tethered together in a tight complex and share a number of genome-wide association potosti et al., 2011). known binding partners and targets, participating in chromatin remodeling and studies have linked the OGT The recent finding that cell-cycle regulation to influence stem cell fate, development, and diseaseand MGEA5 loci to these OGT modulates cell-cycle related processes including immunity, cancer, diabetes, and neurodegeneration. same diseases (Hanover dynamics through its modifiet al., 2010; Butkinaree et al., cation and processing of HCF-1 provides definitive support for the enzymes of O-GlcNAc cycling 2010). The recently uncovered conneca growing body of evidence linking O- balance the activity of numerous kinases, tion between OGT and HCF-1 provides GlcNAc to epigenetics and disease (Han- influencing the robustness of cell prolifer- an important clue for understanding how over et al., 2010 and Love et al., 2010b). ation and tissue homeostasis decisions the environmentally responsive hexosO-GlcNAc is added to and removed (Butkinaree et al., 2010; Hanover et al., amine-signaling pathway intersects with from proteins by separate enzymes in 2010). Both OGT and MGEA5 have been human development and disease, higha manner reminiscent of dynamic phos- previously implicated in cell-cycle pro- lighting the importance of the O-GlcNAc phorylation. OGT couples the nutrient- gression (Wang et al., 2010) and epige- modification in epigenetic regulation of dependent synthesis of UDP-GlcNAc to netic regulation (Hanover et al., 2010). In cell proliferation and signaling during the O-GlcNAcylation of Ser/Thr residues Drosophila, OGT is encoded by the home- development, adulthood, and senesof a variety of targets. This link between otic gene ‘‘Super sex combs,’’ a gene cence. nutrient availability and O-GlcNAcylation previously linked to the epigenetic regulaof proteins through the hexosamine tion of the polycomb and trithorax groups signaling pathway provides cells with (Love et al., 2010b). In C. elegans, muta- REFERENCES a mechanism to sense and respond to tions that affect O-GlcNAc cycling a variety of environmental conditions. provide insight into the signaling Butkinaree, C., Park, K., and Hart, G.W. (2010). BiSome of the known O-GlcNAcylated cascades triggered by stress, infection, ochim. Biophys. Acta 1800, 96–106. proteins include signaling kinases, and aging (Love et al., 2010a). In verte- Capotosti, F., Guernier, S., Lammers, F., Waridel, nuclear pores components, various chro- brates, OGT and HCF-1 coexist in deace- P., Cai, Y., Jin, J., Conaway, J.W., and Herr, W. (2011). Cell 144, this issue, 376–388. matin-remodeling enzymes, and RNA tylase and methyltransferase chromatinpolymerase II. In addition, one isoform of remodeling complexes, providing yet Davis, D.B., Lavine, J.A., Suhonen, J.I., Krautkramer, K.A., Rabaglia, M.E., Sperger, J.M., FernanOGT is targeted to mitochondria where it another link to epigenetic regulatory dez, L.A., Yandell, B.S., Keller, M.P., Wang, I.M., coordinates apoptotic and metabolic events (Wysocka et al., 2003). et al. (2010). Mol. Endocrinol. 24, 1822–1834. The pairing of OGT with HCF-1 regufunctions by adding O-GlcNAc to key Goto, H., Motomura, S., Wilson, A.C., Freiman, mitochondrial proteins (Hanover et al., lates cell-cycle progression by placing it R.N., Nakabeppu, Y., Fukushima, K., Fujishima, 2010). The O-GlcNAc moiety is removed under the control of stress-triggered, M., Herr, W., and Nishimoto, T. (1997). Genes by the O-GlcNAcase, MGEA5. Together nutrient-responsive O-GlcNAcylation and Dev. 11, 726–737. 322 Cell 144, February 4, 2011 ª2011 Elsevier Inc.
Hanover, J.A., Krause, M.W., and Love, D.C. (2010). Biochim. Biophys. Acta 1800, 80–95. Kristie, T.M., Liang, Y., and Vogel, J.L. (2010). Biochim. Biophys. Acta 1799, 257–265. Love, D.C., Ghosh, S., Mondoux, M.A., Fukushige, T., Wang, P., Wilson, M.A., Iser, W.B., Wol-
kow, C.A., Krause, M.W., and Hanover, J.A. (2010a). Proc. Natl. Acad. Sci. USA 107, 7413– 7418.
Wang, Z., Udeshi, N.D., Slawson, C., Compton, P.D., Sakabe, K., Cheung, W.D., Shabanowitz, J., Hunt, D.F., and Hart, G.W. (2010). Sci. Signal. 3, ra2.
Love, D.C., Krause, M.W., and Hanover, J.A. (2010b). Semin. Cell Dev. Biol. 21, 646–654.
Wysocka, J., Myers, M.P., Laherty, C.D., Eisenman, R.N., and Herr, W. (2003). Genes Dev. 17, 896–911.
Need Tension Relief Fast? Try Caveolae Satyajit Mayor1,* 1National Centre for Biological Sciences (TIFR), Bellary Road, Bangalore 560065, India *Correspondence:
[email protected] DOI 10.1016/j.cell.2011.01.018
Caveolae are protein-driven membrane invaginations that regulate both the physical and chemical composition of the plasma membrane. Sinha et al. (2011) now show that caveolae are membrane reservoirs that are used to rapidly buffer against changes in membrane tension. The cell membrane exhibits fluid-like mechanical properties and is subject to a resting surface pressure that creates a well-defined membrane tension (Evans and Skalak, 1980). Although cellular membrane tension primarily reflects the adhesion of the membrane to the cytoskeleton (Sheetz, 2001), it is also affected by tension in the lipid bilayer. The magnitude of bilayer tension affects the packing of proteins and lipids in the membrane, influencing their properties at the molecular and supramolecular scale—for instance, many mechanosensitive channels are gated by membrane tension (Hua et al., 2010). It is therefore not surprising that cells have evolved many ways of regulating tension to withstand changes in osmotic pressure, fluid shear, or mechanical deformation. Most of the mechanisms of compensation that have been invoked to date involve adding or taking away membrane through exo- or endocytosis, which occur over relatively slow timescales (Sheetz and Dai, 1996). Flask-shaped membrane invaginations known as caveolae have also been implicated in the buffering against changes in plasma membrane composition and physical properties (Parton and Simons, 2007). In this issue, Sinha et al. (2011) now provide evidence that caveolae allow
cells to rapidly compensate for changes in membrane tension. These findings build on the theoretical work of Sens and Turner (2006), who proposed that membrane budding driven by protein coats (such as observed in stable caveolae) could constitute a membrane reservoir that is controlled by membrane tension (Figure 1). This, in turn, would impose an equilibrium tension regulated by the mechanical properties of these budded domains. Several simple predictions emanating from this analysis have provided the motivation behind the work of Sinha et al., who have directly investigated the role of budded caveolar domains as buffers of membrane tension. The authors measured membrane tension using a membrane tether-pulling assay (Sheetz, 2001) while subjecting cells to a variety of means of altering membrane area, from osmotic swelling to elegant micromechanical devices specially designed for this study. Their findings show that it is indeed the budded profile of caveolar domains that act as reservoirs of membrane area, buffering against instantaneous changes in membrane tension. Sinha et al. further find that the role of preformed caveolae in buffering tension is a passive one; buffering of membrane tension occurs in
ATP-depleted cells as well as in plasma membrane spheres detached from other intramembranous organelles. Whereas Sens and Turner suggested that the difference in chemical composition could drive the formation of budded domains due to passive budding mechanisms, Sinha et al. find that this is not altogether true in living cells. They make an important but unexpected discovery that the regeneration of caveolae during recovery from osmotic swelling or from the imposed physical perturbation requires ATP and is regulated by actin-based processes. To establish a potential pathophysiological consequence of disrupting caveolae, the authors show that a mutation in caveolin 3 (P28L) associated with a familial form of muscular dystrophy called hyperCKaemia (FHCK) (Woodman et al., 2004) prevents the reformation of caveolae and the buffering of membrane tension. Although it is not known whether this form of muscular dystrophy is associated with increased membrane damage, such a mechanism for buffering against changes in membrane tension could be important for maintaining the integrity of muscles subject to continuous cycles of stretching and relaxation. In addition to muscle fibers, this newly discovered role
Cell 144, February 4, 2011 ª2011 Elsevier Inc. 323
lack caveolae, in addition to being unable to buffer changes in tension, exhibit very low values of membrane tension, at least in some cell types. This has important implications for understanding the role for caveolae in mechanosignaling and tensional homeostasis. Thus, further progress in this area hinges on ascertaining a detailed molecular mechanism behind the mechanical destabilization of caveolae upon the application of force and their active reconstruction following the cessation of the force. ACKNOWLEDGMENTS S.M. would like to thank Madan Rao (Raman Research Institute and NCBS) for his comments and a JC Bose Fellowship (Department of Science and Technology, India) for generous research support. REFERENCES
Figure 1. Caveolae Buffer against Rapid Changes in Membrane Tension Caveolae (flask-shaped invaginations with striated coats containing the protein caveolin) act as passive membrane reservoirs buffering membrane tension by flattering out upon the application of force. This releases some caveolar coat components, the cavins (green circles), into the cytoplasm. Upon the removal of force, caveolae are reformed by ATP-dependent processes. It is tempting to speculate (gray arrows) that the released cavins trigger cellular processes to induce formation of additional caveolae, thereby increasing the number of caveolae and the size of the membrane reservoir. (Images of invaginated and flattened caveolae shown on the right are from Rothberg et al. [1992]).
of caveolae could function in a range of physiological contexts, including osmotic swelling of epithelial cells or mechanical stretching or shear stresses as experienced by endothelial cells. A thorough investigation of all of the known diseases of caveolin mutations, termed ‘‘cavelinopathies’’ (Woodman et al., 2004), in the context of buffering against changes in membrane tension would likely help to clarify some of the reasons for the pleiotrophic phenotypes for these diseases. These studies also raise intriguing questions about how cells maintain surface area homeostasis. It may seem paradoxical that several studies have shown that chronic membrane stresses lead to an increase in caveolar domains (Parton and Simons, 2007), whereas Sinha and coworkers show that there is
an instantaneous loss of caveolar domains accompanying an increase in membrane stresses. A potential explanation for this could involve the release of cavin, a key structural component of caveolae (Hill et al., 2008) that was originally characterized as a transcription factor called PTRF (Jansa et al., 2001). In this scenario, cavin might activate transcriptional circuits that are necessary for the synthesis of new caveolae, thereby providing a bigger protective buffer against chronic stresses. Further, how is a set point for membrane tension established? The ability to actively construct caveolar domains should allow the cell to establish a set point for the regulation of membrane tension (Figure 1). Consistent with this notion, Sinha et al. show that cells that
324 Cell 144, February 4, 2011 ª2011 Elsevier Inc.
Evans, E.A., and Skalak, R. (1980). Mechanics and Thermodynamics of Biomembranes, First Edition (Boca Raton, FL: CRC Press). Hill, M.M., Bastiani, M., Luetterforst, R., Kirkham, M., Kirkham, A., Nixon, S.J., Walser, P., Abankwa, D., Oorschot, V.M., Martin, S., et al. (2008). Cell 132, 113–124. Hua, S.Z., Gottlieb, P.A., Heo, J., and Sachs, F. (2010). Am. J. Physiol. Cell Physiol. 298, C1424– C1430. Jansa, P., Burek, C., Sander, E.E., and Grummt, I. (2001). Nucleic Acids Res. 29, 423–429. Parton, R.G., and Simons, K. (2007). Nat. Rev. Mol. Cell Biol. 8, 185–194. Rothberg, K.G., Heuser, J.E., Donzell, W.C., Ying, Y.S., Glenney, J.R., and Anderson, R.G. (1992). Cell 68, 673–682. Sens, P., and Turner, M.S. (2006). Phys. Rev. E Stat. Nonlin. Soft Matter Phys. 73, 031918. Sheetz, M.P. (2001). Nat. Rev. Mol. Cell Biol. 2, 392–396. Sheetz, M.P., and Dai, J. (1996). Trends Cell Biol. 6, 85–89. Sinha, B., Ko¨ster, D., Ruez, R., Gonnord, P., Bastiani, M., Abankwa, D., Stan, R.V., Butler-Browne, G., Vedie, B., Johannes, L., et al. (2011). Cell 144, this issue, 402–413. Woodman, S.E., Sotgia, F., Galbiati, F., Minetti, C., and Lisanti, M.P. (2004). Neurology 62, 538–543.
Leading Edge
Previews Getting Cells and Tissues into Shape David Odde1,* 1Department of Biomedical Engineering, University of Minnesota, 7-132 Nils Hasselmo Hall, 312 Church Street SE, Minneapolis, MN 55455, USA *Correspondence:
[email protected] DOI 10.1016/j.cell.2011.01.022
Cells of all shapes and sizes are able to calculate the location of their middles in order to divide into two during mitosis. Minc et al. (2011) and Gibson et al. (2011) now show that simple mechanical models accurately predict cleavage-plane positioning, and that geometrical interactions between neighboring cells are sufficient to generate ordered patterns of mitosis in growing epithelia. Biological tissues are remarkable in their ability to maintain ordered structures through development, growth, and repair. Cells appear to ‘‘know’’ what constitutes proper organization and form appropriate shapes in response to external and internal cues. How do the molecular components of the cell work together as dynamic ensembles to establish proper tissue organization? In this issue of Cell, two studies (Minc et al., 2011 and Gibson et al., 2011) together give us a better understanding of how ordered tissue structures can naturally and robustly emerge from the self-assembly and mechanical properties of the constituent molecules and cells. Both studies elegantly demonstrate how cell shape, most likely sensed by the mitotic spindle, serves as a major determinant of future cell and tissue development. The tendency of cells to divide along their longest axis was first noted almost 150 years ago (Hofmeister, 1863). How does a mitotic spindle know where this axis lies? Minc et al. address this question by observing the initial mitosis of sea urchin embryos that have been forced into microfabricated chambers of precisely defined size and shape. They find that mitosis is robust to a wide variety of imposed shapes ranging from simple circles, ellipses, squares, and rectangles to more exotic stars, L shapes, and teardrop shapes. Importantly, they find good agreement with the general observation that spindles tend to align with the longest axis. However, in weakly rectangular cells, the cleavage plane forms not along the diagonal but along the longest axis of symmetry. This axis is predicted by the
nucleus, which becomes located at the geometric center of the cell and elongates perpendicular to the future cleavage plane. Nuclear positioning and elongation in this system depend on intact microtubules, which appear to exert a tensional force on the elastic nucleus presumably via their interactions with the cell cortex (Figure 1A). Inspired by these observations and previous modeling efforts (Grill et al., 2003), Minc et al. construct a mechanical model for cleavage-plane positioning based on microtubule length-dependent nuclear pulling forces (Hays et al., 1982). Their model accurately predicts cleavage-plane positioning in cells of different shapes and types, even dramatically reproducing the spindle orientations of cells in a 1900 drawing of pigeon spermatocytes. The results suggest that cellshape-dependent interactions between the cortex and the nuclear envelope are major determinants of cleavage-plane positioning, although it remains to be understood exactly how these forces are generated in a microtubule length-dependent manner. Which forces determine cell shape and division orientation in a multicellular tissue such as a growing epithelium? One important factor to consider must be the shape of neighboring cells. Gibson et al. begin with the observation that in multiple tissue types, from the Drosophila wing disc to cucumber epidermis, there is a strong bias toward hexagonal shapes, with other polygons occurring less frequently (Gibson et al., 2006; AegerterWilmsen et al., 2010; Patel et al., 2009; Farhadifar et al., 2007). To investigate the forces underlying the maintenance of
this ordered arrangement of hexagonal cells, Gibson et al. construct and test models in which the shapes of cells surrounding a mitotic cell influence the orientation of the cleavage plane. In the case of a quadrilateral cell lying adjacent to a mitotic cell, a simple mechanical model of outward pressure resisted by elastic spring-like edges predicts that the mitotic cell will deform so that its short axis points toward the adjacent quadrilateral (Figure 1B, left panel). In contrast, an adjacent octagon will also induce deformation in the mitotic cell, but in this case the mitotic cell long axis will point to the adjacent octagonal cell (Figure 1B, right panel). Assuming that cells tend to divide along their long axes, cleavage-plane orientation should depend upon the number of sides in the neighboring polygon. In this way, quadrilaterals are biased toward gaining edges, whereas octagons tend not to gain edges. The shape-dependent division rule, in contrast to a random division rule, achieves a narrower distribution of cell shapes and therefore yields more ordered epithelial tissues. Gibson et al. test their assumption that cells divide their longest axis by observing cell division live in this tissue. They find that although the long-axis division rule appears to hold in the majority of cases, the division plane in a proportion of cells fails to form as predicted. It would be interesting to test whether a combination of the microtubule length-dependent mechanism put forward by Minc et al. and the neighboring cells model might have enhanced predictive power in situations in which neither on its own is effective. For example, cleavage-plane
Cell 144, February 4, 2011 ª2011 Elsevier Inc. 325
Figure 1. Cell Shape Regulates Positioning of the Nucleus and the Mitotic Cleavage Plane (A) Microtubules sense cell shape, pulling the nucleus into the cell center and promoting nuclear elongation and cleavage-plane positioning. Efficient centering depends on long microtubules exerting a stronger tensional force than short microtubules. (B) The neighbors of a mitotic cell influence its shape and positioning of the cleavage plane. Four-sided neighbors tend to elongate the central cell such that their short axes point toward the quadrilateral, whereas eight-sided neighbors tend to attract the long axis. Because it is most often the long axis that is divided during mitosis, quadrilaterals tend to gain sides whereas octagonal cells tend not to.
orientation appears to be specified during interphase or prophase and persists through mitosis. However, Minc et al. show that if metaphase-arrested single sea urchin cells are forced to adopt a different shape, the cleavage plane can reposition within 30 min in a microtubule-dependent manner. How then is the cleavage-plane position normally preserved through mitosis, even though cell shape becomes relatively rounded during the division process? Perhaps the geometry of neighboring nonmitotic cells plays a role here. In many disease processes including cancer, cells may lose their sense of order giving rise to aberrant geometric structures. Which aspects of these models become altered to drive pathologic changes in cell shape? Further work in this direction will not only provide insights into the mechanisms of cancer progression but may also reveal more clues as
to how intracellular signaling and mechanical constraints interact (Meyers et al., 2006). Taken together, these models both demonstrate that surprisingly simple mechanical properties can underlie the apparently complex topological calculations performed by dividing cells. They suggest that membrane-dependent interactions linking neighboring cell membranes, and microtubule-dependent interactions connecting these membranes to individual nuclei, form an integrated mechanical network that is responsible for maintaining geometrical order in a growing tissue.
Farhadifar, R., Ro¨per, J.C., Aigouy, B., Eaton, S., and Ju¨licher, F. (2007). Curr. Biol. 17, 2095–2104. Gibson, M.C., Patel, A.B., Nagpal, R., and Perrimon, N. (2006). Nature 442, 1038–1041. Gibson, W.T., Veldhuis, J.H., Rubinstein, B., Cartwright, H.N., Perrimon, N., Brodland, G.W., Nagpal, R., and Gibson, M.C. (2011). Cell 144, this issue, 427–438. Grill, S.W., Howard, J., Schaffer, E., Stelzer, E.H., and Hyman, A.A. (2003). Science 301, 518–521. Hays, T.S., Wise, D., and Salmon, E.D. (1982). J. Cell Biol. 93, 374–389. Hofmeister, W. (1863). Jahrbucher fu¨r Wissenschaft und Botanik 3, 259–293. Meyers, J., Craig, J., and Odde, D.J. (2006). Curr. Biol. 16, 1685–1693.
REFERENCES
Minc, N., Burgess, D., and Chang, F. (2011). Cell 144, this issue, 414–426.
Aegerter-Wilmsen, T., Smith, A.C., Christen, A.J., Aegerter, C.M., Hafen, E., and Basler, K. (2010). Development 137, 499–506.
Patel, A.B., Gibson, W.T., Gibson, M.C., and Nagpal, R. (2009). PLoS Comput. Biol. 5, e1000412.
326 Cell 144, February 4, 2011 ª2011 Elsevier Inc.
Leading Edge
Review Functional and Mechanistic Diversity of Distal Transcription Enhancers Michael Bulger1,* and Mark Groudine2,3 1Center
for Pediatric Biomedical Research, Department of Pediatrics, University of Rochester, Rochester, NY 14627, USA of Basic Sciences, Fred Hutchinson Cancer Research Center, Seattle, WA 98109, USA 3Department of Radiation Oncology, University of Washington, Seattle, WA 98105, USA *Correspondence:
[email protected] DOI 10.1016/j.cell.2011.01.024 2Division
Biological differences among metazoans and between cell types in a given organism arise in large part due to differences in gene expression patterns. Gene-distal enhancers are key contributors to these expression patterns, exhibiting both sequence diversity and cell type specificity. Studies of long-range interactions indicate that enhancers are often important determinants of nuclear organization, contributing to a general model for enhancer function that involves direct enhancer-promoter contact. However, mechanisms for enhancer function are emerging that do not fit solely within such a model, suggesting that enhancers as a class of DNA regulatory element may be functionally and mechanistically diverse. Introduction Genomic DNA acts as a carrier of information in two fundamental ways. First, in transcribed genes, it specifies the sequences of protein-coding mRNAs and functional RNAs, as well as information encoded in RNA that affects its processing and stability. Second, in regulatory sequences, it provides sites for transcription factors to bind, establishing the appropriate levels and expression patterns of those genes. A large proportion of the regulatory information that is necessary for gene expression is confined to the promoter region immediately upstream of transcription start sites. In single-celled organisms, this information can serve to specify absolute levels of transcription and in many cases can mediate alternate responses (up- or downregulation) to external stimuli. Metazoans present a challenge in this regard. A single genome specifies many morphologically distinct cell types and also directs the ordered processes of development and differentiation that lead to the varied structures present in an adult multicellular organism. Sequencing of metazoan genomes has not revealed a simple correlation between genome size, as measured by the number of genes, and relative complexity, as measured by the number of cells and cell types and diversity of behavior. The 959–1031 cells of the nematode Caenorhabditis elegans and the trillions of cells in a typical human are both specified by 20,000–25,000 protein-coding genes. Thus, morphological and developmental complexity is not a function of increased numbers of genes but of alternative mechanisms. For example, a higher proportion of vertebrate genes are subject to alternative splicing, as compared to invertebrates (Kim et al., 2007a). Notably, however, complexity can be generated by diversification of the patterns in which genes are expressed, both spatially and temporally, within an organism. Such diversification is enabled by a correspondingly more complex set of regulatory information in the genomes of metazoans. In particular, in metazoans transcriptional regulation is
often decoupled from the confines of the promoter-proximal region and distributed among distal sequence elements, termed enhancers, which can be located far from the transcription start site. This distribution of regulatory sequences evades the limitations that are inherent in systems in which transcription is a function solely of the few hundred base pairs immediately upstream of a gene promoter. Recent developments in genomics, coupled with studies of covalent histone modifications, structural features of genomic DNA, functional assays for regulatory elements, methods to investigate nuclear organization, and cross-species sequence comparisons, have revealed enhancers, as a class of regulatory element, to be generally and fundamentally important for the normal regulation of genes and thus for the generation of the morphological and behavioral diversity that characterizes multicellular organisms. Finding Enhancers Structure and Sequence Enhancers were first characterized using transient reporter gene assays in cultured cell lines. The activity associated with such elements—first described for viral sequences (Banerji et al., 1981; Moreau et al., 1981) and subsequently for sequences originating from metazoan gene loci (Banerji et al., 1983; Gillies et al., 1983) —is the activation of transcription regardless of the element’s location or orientation relative to the promoter within a plasmid construct. This flexibility is the defining hallmark of enhancers and remains part of their functional definition. They are commonly found within the introns of the genes that they regulate (or, in fact, within the introns of neighboring genes) and often at prodigious distances from the promoter. One of the most extreme examples known is a limb bud enhancer for the mouse Sonic hedgehog (Shh) gene, which is located within the intron of another gene more than 1 Mb from the Shh gene promoter (Lettice et al., 2003; Sagai et al., 2005). Cell 144, February 4, 2011 ª2011 Elsevier Inc. 327
This flexibility, however, impedes attempts to comprehensively identify and catalog the full population of enhancers within the genome or, indeed, the full complement of enhancers that act upon a single selected gene. Whereas the promoter of a gene can be located simply by sequencing the 50 end of its mRNA, no similarly clear-cut criterion exists that can pinpoint the location of an enhancer or the target gene for its activity. Enhancer detection therefore relies on a number of imperfect measures of chromatin structure and sequence functionality. Enhancers are typically found to colocalize with disruptions in chromatin structure revealed by hypersensitivity to digestion by DNaseI. DNaseI hypersensitivity was first discovered at the promoter region of the Drosophila hsp70 gene (Wu, 1980) and is usually thought to result from short (100–300 bp) regions of genomic DNA from which nucleosomes are excluded due to the binding of transcription factors (Elgin, 1988; Gross and Garrard, 1988), although DNA bending by transcription factors has also been implicated (Stamatoyannopoulos et al., 1995; Leach et al., 2001). Enhancers consist of clusters of cognate binding sites for transcription factors that can both exclude nucleosomes and bend DNA and are therefore marked by such nuclease hypersensitivity. For several decades, scans of gene loci for putative enhancer elements involved the slow and laborious indirect end-labeling assay, but recent advances have allowed the mapping of nuclease hypersensitivity genome-wide using microarrays or high-throughput sequencing (Crawford et al., 2006). Another criterion employed for predicting putative enhancer sequences is noncoding sequence conservation. The full utility of this approach has only been realized in the past decade, when multiple fully sequenced genomes have been available for comparison. The principle, however, is well established, resting on the assumption that conservation of DNA sequence across evolution in regions that do not encode proteins implies regulatory function. Nonfunctional sequences, in contrast, will accumulate mutations over evolutionary timescales and eventually diverge. High-level expression of the mammalian b-globin genes in erythroid cells, for example, requires a set of sequence elements termed the locus control region (LCR) that is distributed across a region of 20–30 kb located upstream of the gene cluster (Bender et al., 2000; Moon and Ley, 1990; Grosveld et al., 1987; Hardison et al., 1997). Each of these sequence elements shows significant sequence conservation among all mammalian genomes sequenced thus far. More generalized studies have similarly indicated that evolutionary DNA sequence conservation can predict enhancer activity (Visel et al., 2009b; Noonan and McCallion, 2010). On the other hand, several well-documented examples exist of genes that are expressed in identical patterns in different species but that require the activities of enhancers that look nothing alike. Both Drosophila and sepsid flies, for example, express orthologous even-skipped genes with identical patterns in the developing embryo, and this expression pattern is governed by multiple enhancers. Sequence comparisons, however, fail to reveal any significant similarities between the DNA sequences of these enhancers (Hare et al., 2008). Similarly, expression of the Pax2 gene in the Drosophila melanogaster eye is regulated, in part, by a cone-specific enhancer within which multiple 328 Cell 144, February 4, 2011 ª2011 Elsevier Inc.
transcription factors bind in a specific pattern that can be disrupted by small deviations in sequence of, or spacing between, binding sites. Surprisingly, however, the cone-specific Pax2 enhancers in other Drosophila species exhibit very little conservation of sequence or binding site spacing even though they function identically when transferred into D. melanogaster (Swanson et al., 2010). A systematic evaluation of a 40 kb region encompassing the zebrafish phox2b gene showed that a large proportion of sequences with demonstrable regulatory activity were not identified by measures of evolutionary sequence constraint (McGaughey et al., 2008). In fact, comparisons of genome-wide transcription factor binding patterns across species indicate that a large proportion of enhancers are species specific. In an analysis of mouse and human hepatocytes, for example, 41%–89% of binding sites for four transcription factors were found to be species specific (Odom et al., 2007). A subsequent study examined genomewide binding of the liver-specific transcription factors CEBPA and HNF4a in multiple species (Schmidt et al., 2010). Only 10%–22% of binding sites were shared between placental mammals (human, mouse, or dog), and this number was even lower in comparisons between placental mammals and a nonplacental mammal (opossum) or chicken. Moreover, variation in transcription factor binding patterns appears to have a predominantly genetic origin. In mouse hepatocytes harboring an intact human chromosome 21, binding patterns of the transcription factors HNF1a, HNF4a, and HNF6 are nearly identical to those observed on chromosome 21 in human hepatocytes (Wilson et al., 2008). Combined with other indications that a large proportion of functional enhancers is not subject to evolutionary constraint (ENCODE Project Consortium, 2007; McGaughey et al., 2008), such findings indicate that enhancers can ‘‘turn over’’ rapidly over evolutionary timescales, even when associated gene expression patterns are conserved. Thus, conservation of regulatory function is not always reflected in conservation of DNA sequence. An approach to enhancer prediction that relies on sequence conservation alone will fail to identify a large proportion of enhancers. Chromatin and Transcription Factor Signatures Recently, genome-wide studies have suggested that enhancers exhibit a characteristic chromatin ‘‘signature’’ (Figure 1A). This signature consists of monomethylation of histone H3 lysine 4 (H3K4Me1) in the absence of significant trimethylation (H3K4Me3) (ENCODE Project Consortium, 2007; Heintzman et al., 2007; Koch et al., 2007). Notably, H3K4Me3 is associated with active gene promoters, which in turn exhibit low levels of H3K4Me1 at the transcription start site. In other studies, however, a sharp divide between H3K4Me3 present at promoters, but not enhancers, or H3K4Me1 present at enhancers, but not transcription start sites, has not been as obvious (Barski et al., 2007; Wang et al., 2008). The basis for the discrepancy is not clear, although each study utilized different sources of chromatin (transformed cell lines versus primary T lymphocytes), protocols for isolation of chromatin (formaldehyde crosslinking and sonication versus micrococcoal digestion of native chromatin), and antibodies. It is nevertheless now possible to catalog enhancers by identifying histone methylation signatures that are not associated with other functional elements.
Figure 1. Chromatin Signatures Enhancers and Promoters
at
(A) A histone H3K4 methylation signature marks many enhancers and promoters. Red circles indicate methyl groups on histone H3K4. H3K4 monomethylation is enriched at enhancers and is generally low at transcription start sites. By contrast, H3K4 trimethylation is largely absent from enhancers and appears to predominate at active promoters. (B) A more complicated picture of histone modifications at enhancers. At the left, nucleosomes near a ‘‘poised’’ enhancer in a stem cell are marked by H3K4 monomethylation, but also by H3K27 trimethylation, so the enhancer is not active. Upon differentiation, the H3K27 trimethyl mark is lost, while enhancer-binding factors recruit p300 or other histone acetyltransferases, resulting in H3K27 acetylation (blue circles). This results in enhancer activation and active transcription of target genes.
The H3K4 methylation signature has been correlated with enhancer activity in gain-of-function assays (ENCODE Project Consortium, 2007; Heintzman et al., 2007) but imperfectly. This result is perhaps unsurprising because the transient reporter gene assay is often a poor proxy for the activities of regulatory elements integrated in the genome, which can be active in narrow windows of development and/or cellular differentiation. Enhancer predictions become more significant, however, when the histone methylation signature is combined with other indicators of enhancer activity in a specific cell type, such as transcription cofactor binding—most often, the acetyltransferase p300 (Visel et al., 2009a; Blow et al., 2010; Ghisletti et al., 2010). In addition, studies utilizing embryonic stem (ES) cells and multiple primary cell types suggest that acetylation of histone H3K27 in combination with H3K4Me1 is correlated with enhancers near active genes, whereas H3K4Me1, in the absence of H3K27 acetylation, appears to mark inactive or ‘‘poised’’ enhancers (Creyghton et al., 2010; Rada-Iglesias et al., 2010). Notably, p300 can acetylate H3K27. Moreover, many poised enhancers in ES cells were instead associated with H3K27Me3, and at a subset of these enhancers, the H3K27Me3 mark was replaced with H3K27Ac upon differentiation of ES cells along a neuronal pathway (Figure 1B; Rada-Iglesias et al., 2010). The available evidence suggests that H3K4Me1 represents a generalized, although perhaps not all-inclusive, mark for distal enhancers in a given cell type. Additional modifications can then distinguish between enhancers that are active and those that are potentiated for activity in response to growth conditions or cell fate decisions. The most accurate predictions for enhancer activity to date are derived from combinatorial analysis of the binding of multiple transcription factors across the genome in Drosophila (Zinzen et al., 2009). In this study, binding of five transcription factors known to be involved in the differentiation of different muscle cell types from mesoderm was mapped genome-wide at different stages of development. A machine learning method was then applied to this data set in order to derive predictions for enhancers that are active in different cell types. Using this approach, the authors were able to identify 77% of all previously characterized muscle-specific enhancers in five different cell
types in Drosophila and had a roughly equivalent success rate when predicted sequences were tested for enhancer activity. If one accepts the description of a chromatin ‘‘signature’’ for enhancers, current evidence then suggests that mammalian genomes harbor an abundance of such elements and that they are the major determinant of cell type specificity in gene expression. Examination of genome-wide H3K4 methylation patterns in two cell lines—K562, a human erythroleukemia, and HeLa, a human cervical carcinoma—resulted in an estimate of 24,000– 36,000 enhancers in each line (Heintzman et al., 2009). When the locations of these enhancers were compared between the two lines, only 5000 were found to be present in both. Analysis of histone modification patterns across a region comprising 1% of the human genome, compared among five different cell lines, revealed a frequency of H3K4Me1 distal from promoters that agrees roughly with the genome-wide K562/HeLa study and also indicated that a much higher number of enhancers defined by this criterion exhibited cell type specificity, as compared to promoters (Koch et al., 2007). Genome-wide mapping of nuclease hypersensitive sites (HSs) also suggests that enhancers are the primary determinant of cell type specificity. A survey of HSs across 1% of the human genome in six cell lines (including HeLa and K562) established a strong correlation between hypersensitive sites that were cell type specific and enhancer elements, as defined by the H3K4Me signature (Xi et al., 2007). Based on such observations, it has been suggested that the human genome might harbor as many as 1 3 106 enhancers (Heintzman et al., 2009). At present, however, this is little more than the roughest of estimates because it is not clear how extensive variability in the H3K4Me signature actually is in vivo. For example, as yet, no studies have addressed the degree of overlap between cells within a specific lineage at different stages of differentiation or similar cells at different stages of embryonic development. There is also no a priori reason to expect that enhancers as a general class of functional genomic element should necessarily exhibit the same histone methylation pattern. A study of human CD4+ T cells that investigated 39 distinct chromatin-associated marks identified several different histone modifications that correlated with putative enhancers, but no Cell 144, February 4, 2011 ª2011 Elsevier Inc. 329
Figure 2. Enhancer Looping and Variant Models (A) Simplified schematic of how enhancers might interact directly with promoters. Nucleosomes are shown as yellow circles, and the default state for chromatin is shown as a 30 nm fiber. Factors are bound to both the enhancer (E, orange bar) and promoter (P, blue bar). These factors can potentially interact with each other, and when they do, transcription is activated. The simplest mechanism by which such interaction might be accomplished is via free diffusion in the nucleus (horizontal arrow). Enhancer-promoter interactions might be facilitated by additional factors (blue circles) that bind to the intervening sequences and organize them to bring the enhancer and promoter into proximity (vertical arrows). Alternatively, both enhancer and promoter can interact with RNA polymerase II, which then serves to bring the elements into proximity via association with a common RNA polymerase II transcription ‘‘factory’’ (diagonal arrow). (B) Tracking. The enhancer-bound complex (red oval) actively scans along the chromatin fiber until it encounters the promoter complex (pink oval) and activates transcription.
single modification was completely predictive (Wang et al., 2008). Finally, despite studies such as this, dozens of histone modifications remain untested. Thus, a true enhancer census remains a subject of speculation. Mechanisms of Enhancer Function Enhancer-Promoter Interactions Since the discovery of enhancers, the dominant model for their mechanism of action on promoters has invoked direct interactions (Figure 2A). This model is commonly termed ‘‘looping,’’ as it requires that the intervening DNA be looped out or otherwise organized in order to permit the enhancer-promoter interaction (Bulger and Groudine, 1999; Blackwood and Kadonaga, 1998; de Laat et al., 2008). Alternative models for enhancer function have chiefly differed from the basic ‘‘looping’’ premise only in how the enhancer-promoter interaction is established—whether by free or facilitated diffusion within the nucleus or by an active ‘‘scanning’’ or ‘‘tracking’’ mechanism (Figure 2B) in which the enhancer diffuses one-dimensionally along the chromatin fiber in search of a promoter (Blackwood and Kadonaga, 1998). In addition, more indirect models have been proposed, including ‘‘oozing’’ or ‘‘linking,’’ in which a complex is nucleated at the enhancer and then polymerizes along the chromatin fiber bidirectionally until it reaches a promoter (Ptashne, 1986; Dorsett, 1999; Bulger and Groudine, 1999). In one variation of this model, RNA polymerase II or other complexes are loaded at the enhancer and then actively move along the DNA until reaching a promoter. The ‘‘looping’’ model in particular has received abundant support from studies of nuclear architecture utilizing ‘‘chromosome conformation capture’’ (3C) and its high-throughput derivatives. In this assay, interactions between two genomic regions are identified by crosslinking, restriction endonuclease digestion, intermolecular ligation, and PCR analysis of the resulting ligated products (Cullen et al., 1993; Dekker et al., 2002; Miele and Dekker, 2009). Using this procedure, specific genomic 330 Cell 144, February 4, 2011 ª2011 Elsevier Inc.
restriction fragments located far from each other on the linear genome can be found to interact with a greater frequency than more proximal fragments if the sequences colocalize within the three-dimensional space of the nucleus. This approach has been adapted to produce genome-wide maps of the threedimensional associations that a specific locus makes among all of the other sequences in the genome (‘‘4C’’) (Zhao et al., 2006; Simonis et al., 2006) or even associations among multiple sequences located throughout the genome (‘‘5C’’ and ‘‘Hi-C’’) (Dostie et al., 2006; Lieberman-Aiden et al., 2009). 3C has been employed to reveal interactions between distal sequence elements—primarily enhancers and promoters— within multiple loci in mammalian genomes (Miele and Dekker, 2008; de Laat et al., 2008). It is now common to find distal enhancers that colocalize with the promoters they regulate, which has uniformly been interpreted to be the result of direct enhancer-promoter interactions that are necessary for gene activation. Moreover, at multiple gene loci, strong correlations have been made between active transcription and the ability to reveal such associations. Within the mammalian b-globin locus, for example, transcription factor knockouts that eliminate b-globin gene transcription uniformly result in the loss of colocalization of the gene with the b-globin LCR, as revealed by 3C (Drissen et al., 2004; Vakoc et al., 2005). Additional support for direct communication between enhancers and promoters is provided by indications that enhancer-promoter interactions can be specific. As a general rule, enhancers are capable of activating transcription from heterologous promoters, and in fact in the majority of gain-offunction assays, reporter gene expression follows the pattern governed by the enhancer, not the promoter. Some notable exceptions to this principle have been demonstrated, however. In the Drosophila Antennapedia gene complex, for example, an enhancer specifically mediates activation of the Sex combs reduced (Scr) gene despite the presence of another active gene, fushi tarazu (ftz), located between it and the Scr promoter
(Calhoun et al., 2002; Calhoun and Levine, 2003). This interaction requires a ‘‘tethering’’ element near the Scr promoter, which can direct the enhancer to activate the ftz gene if it is moved to the ftz promoter instead. A similar element within the promoter of the Drosophila yellow gene renders heterologous promoters capable of being activated by the otherwise highly specific yellow gene enhancers (Melnikova et al., 2008). In addition, several studies have indicated that some enhancers can exhibit a preference for specific classes of gene promoters, for example, between promoters that harbor a canonical TATA box versus promoters that contain a DPE (Ohtsuki et al., 1998; Butler and Kadonaga, 2001). Transcription of the ftz gene is mediated by an enhancer bound by the transcription factor Caudal, which specifically activates genes with DPE-containing promoters (Juven-Gershon et al., 2008). Nuclear Organization Evidence supporting enhancer-promoter interactions is part of a large and growing body of studies that have suggested that nuclear architecture is a major determinant of gene expression. Several generalized properties of genomic organization within the nucleus have been established, and enhancers have been shown to influence many of them. First, each chromosome occupies a distinct ‘‘territory’’ within the nucleus, although up to 20% of the volume of the nucleus may be comprised of intermingling of neighboring chromosomes (Cremer and Cremer, 2010; Branco and Pombo, 2007). In addition, specific sequences on a given chromosome have been observed to extrude or ‘‘loop’’ out from the main body and can even be found in other chromosomal territories (CTs). A study of the b-globin locus, for example, found that the region extruded from its CT specifically in erythroid cells (Ragoczy et al., 2003). This extrusion or looping from the CT occurred prior to high-level b-globin gene expression and was dependent upon the presence of the b-globin LCR. Furthermore, ectopic integration of a b-globin LCR into a gene-dense region of the mouse genome resulted in more frequent extrusion/looping of the region away from its CT (Noordermeer et al., 2008). The limb bud enhancer of the Shh gene is similarly required for extrusion of the gene locus from its CT (Amano et al., 2009). These results suggest that some enhancers mediate a change in nuclear localization for genes in their vicinity and that this represents a step distinct from transcriptional activation. Second, in many metazoan cell types, active genes appear to be localized in the interior of the nucleus, whereas silent genes are found at the periphery. This correlation is not absolute. The vicinity of the nuclear pore in yeast, for example, has actually been associated with active genes (Taddei, 2007), and thus there can be distinct compartments at the nuclear periphery. However, the nuclear periphery in general appears to exert a repressive influence on genes that localize there. For example, artificial tethering of reporter genes to the nuclear lamina or nuclear periphery has been shown to result in downregulation of the reporter and of neighboring genes, although not all genes are affected (Andrulis et al., 1998; Finlan et al., 2008; Reddy et al., 2008). A link between this pattern and enhancer function is again provided by the b-globin locus. A study of b-globin locus positioning within the nucleus during erythroid differentiation demonstrated that, at early maturational stages, the locus is found near
the nuclear periphery (Ragoczy et al., 2006). As differentiation proceeds and b-globin transcription is activated, the locus moves more toward the interior of the nucleus. This process requires the presence of the b-globin LCR, although here it is not clear whether relocalization is a function of the LCR directly or whether it represents an indirect consequence of LCR-dependent b-globin gene activation. Separate studies of transgenes under the control of an enhancer derived from the b-globin LCR, however, indicated that, at ectopic integration sites, the enhancer was required for localization of the transgene far from regions of centromeric heterochromatin, and this was in turn associated with a higher, stochastically determined probability of the gene being active at all (Francastel et al., 1999). Third, the nucleus harbors a number of self-organized substructures, proximity to which can affect gene expression (Ferrai et al., 2010). In addition to the nuclear lamina, such substructures include nucleoli, Cajal bodies, PML bodies, splicing speckles, and other features that represent concentrations of factors that can influence transcription—in the case of splicing speckles, for example, of the splicing machinery and other mRNA processing factors. Such substructures can serve as the basis for colocalization of genes that are otherwise located far apart on a linear chromosome or even on different chromosomes. For example, erythroid-specific genes have been found to be positioned near common splicing speckles in erythroid cells (Brown et al., 2008), whereas several muscle-specific genes have been shown to localize to shared speckles in differentiated muscle cells (Moen et al., 2004). Finally, gene loci can colocalize in the nucleus on the basis of shared associations with specific factors. Perhaps the most notable of these is with RNA polymerase II itself. Visualization using antibodies against RNA polymerase II or labeling of primary mRNA transcripts have suggested that transcription is localized to a limited number of RNA polymerase II ‘‘factories’’ (Iborra et al., 1996; Sutherland and Bickmore, 2009). Transcription factories have been proposed to underlie the observation that active gene loci distributed across single chromosomes or even located on separate chromosomes tend to colocalize in the nucleus (Osborne et al., 2004; Simonis et al., 2006). In fact, some studies have suggested that transcription factories occur in different varieties, corresponding to transcription mediated by different factors (Xu and Cook, 2008; Schoenfelder et al., 2010). In addition to RNA polymerase II, other factors involved in transcriptional regulation appear to organize into discrete foci in the nucleus and can either directly or indirectly bring distal gene loci into proximity with each other. These include, for example, special AT-rich sequence binding protein 1 (SATB1), a protein expressed in thymocytes and several other cell types that has been implicated in anchoring disparate genomic loci via longrange interactions (Cai et al., 2003; Cai et al., 2006). In thymocytes SATB1 is observed in a ‘‘cage-like’’ distribution around, but not coincident with, concentrations of centromeric heterochromatin. SATB1, in turn, binds to promoter-distal regulatory elements in multiple gene loci, and loss of the factor results in disruptions of normal gene expression patterns and locus-wide chromatin structure. A special case of this category of interaction is presented by CTCF, a zinc finger transcription factor that can have many roles Cell 144, February 4, 2011 ª2011 Elsevier Inc. 331
in gene regulation. Notably, CTCF binding to insulator elements can block enhancer-mediated gene activation when such elements are located between an enhancer and promoter (Phillips and Corces, 2009). CTCF is capable of self-association (Pant et al., 2004; Yusufzai et al., 2004) and has been implicated in the formation of chromatin loops in vivo by 3C-based approaches (Kurukuti et al., 2006; Splinter et al., 2006; Hou et al., 2010). Interestingly, CTCF associates with cohesins, protein complexes that physically connect sister chromatids during mitosis and meiosis, and this association appears to be necessary for enhancer-blocking activity (Rubio et al., 2008; Wendt et al., 2008). A recent study of enhancers in embryonic stem cells demonstrated that enhancers and promoters were also associated with cohesin (Kagey et al., 2010), and the cohesin-loading factor Nipbl (Nipped-B in Drosophila) was one of two factors that emerged from a genetic screen in Drosophila for factors involved in enhancer-promoter communication (Rollins et al., 1999). Common association with cohesins provides a potential mechanistic link between long-range associations observed between CTCF-binding sites and between enhancers and their cognate promoters. Alternative Mechanisms and Remaining Issues 3C and its variants are inherently descriptive assays, and as with most studies of nuclear organization, they do not provide obvious ways to distinguish between correlation and causation. Thus, whereas the spatial colocalization of active enhancer and promoter regions revealed by 3C is suggestive of direct interactions that mediate gene activation, it is formally just as likely that it represents a consequence of a distinct activating mechanism. RNA polymerase II transcription factories, for example, provide an alternate mechanism by which enhancer-promoter colocalization might take place (Figure 2A). RNA polymerase II is recruited to many enhancers (Heintzman et al., 2007; Koch et al., 2008; Kim et al., 2010). Although the function of this recruitment, if any, is unknown, the likelihood that RNA polymerase II-bound enhancers and promoters might colocalize by virtue of coincidental association in such transcription factories has rarely elicited much commentary. In some cases, it has been speculated that part of the role of the enhancer is to transfer RNA polymerase II to the promoter directly (Zhu et al., 2007; Leach et al., 2001), but this has not been demonstrated, and in the case of the b-globin locus, RNA polymerase II still associates with the gene promoter in the absence of the LCR (Sawado et al., 2003). The function of RNA polymerase II recruitment to enhancers, if it does not involve transfer to the promoter, is not clear, but recent studies have shown that enhancers themselves are transcribed. In one study, a population of 12,000 enhancers was defined by the combination of H3K4Me1 and binding of the transcription cofactor CBP (Kim et al., 2010). Roughly 25% of the enhancers defined in this fashion were found to be associated with RNA polymerase II, and this subset was, in turn, transcribed into RNA. The product RNAs, termed eRNAs, were short, bidirectional, and not polyadenylated, and in at least one case, eRNA transcription required the presence of the target promoter. In another study, 70% of extragenic RNA polymerase II binding in primary macrophages was found to map to enhancers, as defined by the H3K4Me1 signature (De Santa et al., 2010). 332 Cell 144, February 4, 2011 ª2011 Elsevier Inc.
Such findings complement earlier studies that have shown transcription originating from specific enhancers, for example at HS2 of the b-globin LCR (Kim et al., 2007b; Zhu et al., 2007). In a few cases, nongenic transcription originating from enhancers has been shown to be necessary for normal gene activation. For example, at transgenic human growth hormone loci in mice, a large region located upstream of the growth hormone gene is transcribed in an enhancer-dependent fashion (Ho et al., 2006). Insertion of a transcriptional terminator in this region partially eliminates this nongenic transcription pattern and, in turn, results in a decrease in growth hormone gene expression. Another study has suggested that some noncoding RNAs (ncRNAs) act as enhancers on neighboring genes (Ørom et al., 2010). The authors identified a set of several thousand unique, long ncRNAs and then focused on a subset that exhibited differential expression during keratinocyte differentiation. RNAi-mediated knockdown of several of these ncRNAs resulted in a decrease in expression of neighboring genes. Further analysis of one ncRNA, located near the gene for the transcription factor Snai1, revealed that it behaved as a classical enhancer in transient reporter gene assays and furthermore that it was the RNA itself that was required for the effect. The ncRNAs characterized in this study appear to be distinct from eRNAs; the former, for example, are polyadenylated, whereas the latter are not. The significance of eRNA transcription is not known, and the mechanism by which ncRNAs might mediate activation of neighboring (but distal) genes is similarly unclear. Such studies, however, reveal an unforeseen complexity to the roles of RNA polymerase II association with enhancers and of the association of enhancers with transcription factories and suggest functions for noncoding RNA for which conventional models of enhancermediated gene activation currently do not account (Figure 3C). Several studies have provided evidence for a mechanism of enhancer function distinct from, or perhaps complementary to, the nuclear colocalization implied by 3C-based studies. Within the b-globin locus, some chromatin-modifying enzymes— notably, the histone H3 lysine 4 methyltransferase MLL2 and the histone H3 lysine 9 methyltransferase G9a—have been shown to be recruited by association with the enhancer-binding factor NF-E2 and then to spread locus-wide (Demers et al., 2007; Chaturvedi et al., 2009). In addition, a class of enhancer activity is suggested by studies of enhancers within the endogenous mouse b-globin locus and within a transgenic human growth hormone locus in mice. In both cases, the active genes are embedded within a larger ‘‘domain,’’ extending for 15–30 kb and defined by high levels of histone hyperacetylation. Deletion of binding sites for the transcription factor Pit-1 within the transgenic human growth hormone locus (Ho et al., 2002) or of an evolutionarily conserved enhancer located between the embryonic b-globin genes (G. Fromm and M.B., unpublished data) results in complete loss or significant decreases in expression, respectively, coupled with complete loss of the hyperacetylated domain. Thus, enhancers can also control chromatin modifications that are distributed continuously over large regions in a manner analogous to silencers that control heterochromatic domains (Figure 3B; see also below). Another function for enhancers is emerging from studies of epigenetic marks in pluripotent embryonic stem (ES) cells
Figure 3. Alternative Enhancer Function
Mechanisms
of
(A) Placeholding. In ES cells, the enhancer (E, orange bar) is occupied by ES cell-specific transcription factors, but the gene promoter (P, blue bar) is not active. Upon differentiation along the lineage in which the gene is normally expressed, the ES cell-specific factors are downregulated, but new cell-specific factors occupy the enhancer in their place and mediate gene activation (upperright). Upon differentiation into another lineage, the ES cell-specific factors are not replaced, and the locus is inactivated (lower-right). (B) Spreading. The enhancer is bound by a factor or factors that recruit chromatin-modifying or other activities (light and dark blue ovals), which then spread along the chromatin fiber bidirectionally until they reach a promoter. Histones may be modified throughout the region (nucleosomes shaded blue), leading to the formation of a distinct chromatin domain. (C) Noncoding RNA. The enhancer is bound by RNA polymerase II and transcribed. By as yet undefined mechanisms, the noncoding RNA then mediates transcriptional activation of a neighboring but still distal gene promoter.
(Figure 3A). In several cases, ‘‘pioneer’’ transcription factors have been found to associate with distal enhancers in ES cells, although they do not mediate gene activation. These factors are then lost upon ES cell differentiation. If differentiation proceeds along a lineage within which the enhancer is normally active, the pioneer factor is replaced with another factor that mediates enhancerdependent gene activation. If differentiation proceeds along another lineage, loss of the pioneer factor results in inactivation of the enhancer and thus the entire gene locus. For example, an enhancer within the liver-specific Alb1 gene locus is bound by the transcription factor FoxD3 in ES cells; binding of FoxD3 prevents DNA methylation at this enhancer. Upon differentiation into endoderm, FoxD3 is replaced by FoxA1, the activating factor, and the Alb1 gene is transcribed (Xu et al., 2009). Differentiation into another lineage is accompanied by DNA methylation at the enhancer. Similar models have been presented for the macrophage/dendritic cell gene IL12b, the thymocyte-specific Ptcra gene (Xu et al., 2009), and for an enhancer within the pre-B cellspecific l5/VpreB1 gene locus (Liber et al., 2010). These examples suggest a general model in which developmental and lineage maturation decisions to activate or to silence a given gene locus are postponed until specific time points via the binding of pioneer transcription factors at enhancer elements. Factor binding to the enhancer thus acts as a ‘‘placeholder’’ for a later step in gene regulation. Although this mechanism for enhancer function does not involve direct activation of gene promoters, it is no less necessary for the proper regulation of transcription and represents an indirect method by which enhancers can affect gene activation. Thus, whereas most studies of enhancer function have focused on models involving direct enhancer-promoter interactions and the assays that can reveal them, other studies indicate that regulation at a distance can involve indirect mechanisms as well. Transcription of enhancers, noncoding RNAs, enhancer placeholding, and spreading mechanisms are likely to be important for the function of at least some enhancers, and observa-
tions of such phenomena raise the larger possibility that simple models of enhancer-promoter looping are not sufficient to account for enhancer activity. Enhancers, Disease, and Evolution While initially discovered through the direct activation of gene transcription, enhancers influence a variety of fundamental cellular phenomena, such as stem cell multipotency and chromosomal and nuclear organization. Enhancer function is also emerging as an important component in human disease and metazoan evolution. Assuming that an enhancer is marked by a nucleosome-free region that can extend for 200–300 bp, estimates of the abundance of enhancers suggest that they could comprise as much as 10% of the human genome. For comparison, the total extent of protein-coding sequences in the human genome is estimated at 2%–3%. Mutations within proteincoding sequences are well established as a basis for human disease, with many thousands of known examples, so it is unsurprising to find that mutations within enhancers can similarly lead to heritable disorders. For example, some instances of X-linked deafness type 3 are associated with loss of an enhancer region located 900 kb upstream of the POU3F4 gene (de Kok et al., 1996). An enhancer located within the first intron of the RET proto-oncogene occurs in a variant that confers a 20-fold increased risk for Hirschsprung’s disease (Emison et al., 2005). Numerous other examples have been presented, indicating that noncoding distal regulatory sequences represent a target for mutations that can lead to disease (Kleinjan and van Heyningen, 2005; Visel et al., 2009b; Noonan and McCallion, 2010). Thus far, however, the proportion of known mutations that localizes to enhancer sequences does not align with the apparent complexity of the enhancer population in the genome. A very small percentage of mutations documented in the Human Gene Mutation Database map to noncoding DNA, and the majority of these correspond to promoter-proximal regions (Noonan and McCallion, 2010). In part, this may stem from the historical Cell 144, February 4, 2011 ª2011 Elsevier Inc. 333
difficulty in mapping and characterizing gene-distal enhancers and thus reflect an artificial bias for protein-coding regions. Another contributing factor, however, may be that enhancers are not as sensitive to single base pair alterations as proteincoding sequences. Though such changes have the potential to fundamentally alter transcription factor association and render an enhancer nonfunctional, they are more likely to alter binding affinities or developmental patterns of association in ways that can change function more subtly without eliminating it. While a mutation in a protein-coding region will affect function in every cell in which the protein is expressed, a mutation in an enhancer may affect only part of the expression pattern. Enhancer mutations are therefore a potentially powerful engine for intraspecies variation and, insofar as such changes affect selectable traits, for evolutionary divergence (Rebeiz et al., 2009; Visel et al., 2009b; Levine, 2010; Noonan and McCallion, 2010). Illustrations of the principle include freshwater stickleback fish populations, which lack pelvic fins that are present in saltwater stickleback populations and in ancestral fish. Freshwater sticklebacks have lost an enhancer that regulates the gene for the homeobox transcription factor Pitx1. In stickleback fish that have pelvic fins, this enhancer specifically activates Pitx1 gene expression in the pelvic fins during development (Chan et al., 2010; Levine, 2010). Another study defined a series of mutations in an enhancer for the ebony gene in Drosophila (Rebeiz et al., 2009). Populations of Drosophila in Uganda exhibit a correlation between their abdominal pigmentation and the elevation of their primary habitat. Abdominal pigmentation is influenced by the product of the ebony gene, in the absence of which the abdomen exhibits a dark, melanic phenotype. An analysis of dark versus light-colored lines resulted in the identification of a distal enhancer for the ebony gene, mutations in which give rise to phenotypic variation. Several studies have systematically attempted to define human-specific mutations in noncoding genomic DNA sequences that are otherwise highly conserved among mammalian species. Such analyses are able to define isolated sequences, termed ‘‘human-accelerated conserved noncoding elements’’ (HACNSs) or ‘‘human-accelerated regions’’ (HARs), that in some cases appear to be enhancers (Pollard et al., 2006; Prabhakar et al., 2006; Noonan, 2009). Notably, one element, HACNS1/HAR2, consists of 81 bp, within which 16 humanspecific substitutions have accumulated during evolution. When tested in reporter constructs in transgenic mice, the human version acts as an enhancer to drive gene expression in the developing anterior limb and a few other locations (Prabhakar et al., 2008). Versions of this element that lack the humanspecific substitutions fail to direct gene expression to the developing limb. Although the biological significance of this expression pattern, including the gene(s) influenced by HACNS1/HAR2, is unknown, such behavior provides a concrete demonstration that changes in enhancer function can alter expression patterns to generate variation. Action at a Distance in Single-Celled Organisms Abundant evidence exists to assign a wide-ranging and crucial role for enhancers in accomplishing complex patterns of gene expression that underlie cell type specificity in metazoans. More334 Cell 144, February 4, 2011 ª2011 Elsevier Inc.
over, such function has a significant impact on considerations of evolution and disease in multicellular organisms. In prokaryotes and unicellular eukaryotes, however, DNA sequences that mediate gene regulation over large genomic distances represent exceptions rather than the rule. Such exceptions, however, may be informative when compared and contrasted to the activity of metazoan enhancers. In the budding yeast S. cerevisiae, for example, regulatory sequences are usually limited to upstream activation sequences (UASs) that are located within a few hundred base pairs of the promoter. Relocation of a UAS at greater distances results in loss of function (Dobi and Winston, 2007). Artificial model systems in yeast, however, have shown that bridging of distal sequences can be accomplished. Integration of expression cassettes within yeast telomeres, which naturally fold and can bring distal elements within close proximity of each other, or use of the self-associating Drosophila GAGA factor to mediate looping between two sequences have demonstrated that gene activation can be accomplished by simple looping interactions (de Bruin et al., 2001; Petrascheck et al., 2005). In addition, regulatory elements that appear to act like enhancers by activation at a distance and/or from positions downstream of the promoter have been found at select gene loci in S. cerevisiae. The most notable example in this regard is the gene for the HO endonuclease, an enzyme involved in mating-type switching that is expressed in a brief window during late G1 phase of the cell cycle. The process of HO gene activation requires binding of the transcription factor Swi5 to two sites located more than 1 kb upstream of the promoter (Figure 3A). Swi5, in turn, is required for recruitment of chromatin remodeling factors and results in histone acetylation across the entire region between the binding sites and the promoter. Interestingly, binding of Swi5 is a transient event, and HO gene activation occurs well after Swi5 is no longer present, suggesting that these distal regulatory elements initiate an activation mechanism that is maintained by chromatin remodeling at the promoter in an indirect fashion (Cosma et al., 1999; Krebs et al., 1999). Studies in yeast also provide a cautionary note that may be applied to enhancer studies in metazoans. For example, early investigations of the S. cerevisiase rRNA genes identified an apparent transcriptional enhancer located downstream of the 35S rRNA gene. In transgene constructs, this enhancer region mediated 10- to 30-fold increases in 35S rRNA transcription and was also selective, having no effect on the 5S rRNA gene located between the enhancer region and the 35S rRNA promoter (Neigeborn and Warner, 1990; Morrow et al., 1993). The results suggested a simple model in which the enhancer interacted with the 35S rRNA gene promoter by a looping mechanism, with the intervening 5S gene isolated within the loop. Subsequent studies, however, demonstrated that the enhancer was unnecessary for endogenous 35S rRNA expression (Wai et al., 2001) and instead functioned in ectopically integrated reporter constructs to recruit RNA polymerase I, which is otherwise primarily localized within the nucleolus. Such findings may be instructive for any study in which the potential function of a distal enhancer is investigated outside of its normal context. Finally, yeast heterochromatin provides a well-established model for action at a distance in single-celled eukaryotes.
Notably, the simplest version of this model, exemplified by silencing at the yeast mating-type gene loci, is entirely indirect: silencer elements within the locus serve to nucleate a complex of Sir proteins (Rusche et al., 2003). The complex includes the histone deacetylase Sir2, which deacetylates the histone tails of nearby nucleosomes. Other components of the complex bind with high affinity to deacetylated histone tails and then serve to recruit more of the Sir2 deacetylase, resulting in a progressive spreading of histone deacetylation and silencing factors along the chromatin fiber (Figure 4A). In this way, DNA sequences that in themselves do not encode information sufficient for silent regulation are packaged in a repressive chromatin structure. It has been shown that, once established, silencing is maintained in yeast cells even after the silencer itself is excised by an in vivo recombination strategy, although the silencer is still required for re-establishment of the silent state after cell division (Holmes and Broach, 1996). Silent domains in yeast, however, can be established and maintained discontinuously. For example, silent domains at yeast telomeres or within the HMR locus can encompass artificially inserted active genes. Such behavior involves the activity of ‘‘proto-silencer’’ elements located elsewhere in the domain that have no activity by themselves but can augment the activity of canonical silencers (Talbert and Henikoff, 2006). Recently, it has been demonstrated that silencers within the HMR locus in yeast colocalize, as determined by 3C (Valenzuela et al., 2008). Thus far, however, it is not clear whether discontinuous silencing requires such direct interactions or whether it occurs indirectly by localization to nuclear subcompartments enriched in silencing factors. Such issues directly parallel corresponding aspects of enhancer function, as described above. Studies of enhancer elements in bacteria have provided elegant demonstrations of both the looping and tracking models (reviewed by Xu and Hoover, 2001). A number of bacterial genes regulated by the s54 holoenzyme are also controlled by factors that bind to enhancer sequences that can be located up to 3 kb upstream and 1.5 kb downstream of the promoter. These enhancer binding proteins then interact directly with the s54 holoenzyme via DNA looping, and notably, in some cases, the looping mechanism is facilitated by additional factors that bind between the enhancer and promoter and bend DNA (Figure 4B). In contrast, a distal enhancer that is required for activation of phage T4 late genes functions by the loading of a ring-shaped trimer of a factor termed gp45, followed by one-dimensional diffusion of the trimeric protein along the DNA until it reaches the promoter (Figure 4C).
Figure 4. Mechanisms of Distal Regulatory Element Function in Yeast and Bacteria (A) Mating-type silencing in S. cerevisiae (spreading). In this simplified representation, a discrete silencer element is bound by several factors (ORC, Abf1p, and Rap1p), which in turn recruit a complex of SIR proteins that includes a histone deacetylase. Deacetylated nucleosomes (in pink) are, in turn, recognized by the SIR complex, which can then deacetylate additional nucleosomes, resulting in a progressive spread of histone deacetylation and associated SIR complexes. Genes within the region are silenced. (B) Enhancer-binding protein interactions with s54-dependent promoters in bacteria (looping). An enhancer sequence is recognized by an enhancer-
binding protein (EBP), and in turn, the promoter is recognized by a s54 holoenzyme that includes RNA polymerase. This closed complex is only activated upon direct interaction between the EBP and s54 (bottom), which can be facilitated by DNA bending mediated by integration host factor (IHF) binding between the enhancer and promoter. (C) Activation of phage T4 late genes (tracking). A sliding clamp, consisting of a trimer of gp45 polypeptides, is loaded at a promoter-distal site by the gp44gp62 complex. The gp45 trimer then tracks or slides along the DNA until it reaches the promoter, where it mediates gene activation. In this representation, RNA polymerase and additional factors (gp33 and gp55) are present at the promoter in a closed complex prior to the arrival of gp45, but this is not known; in other models, the polymerase can track with gp45, for example.
Cell 144, February 4, 2011 ª2011 Elsevier Inc. 335
In summary, studies in organisms that typically lack enhancer activity indicate that distal sequences can affect gene expression by widely different mechanisms and imply that many enhancers could have evolved independently. In addition, there is no reason to exclude the possibility that a given enhancer might work by a combination of mechanisms. Thus, such studies argue against simple, unified models for enhancer function in metazoans. Conclusions Enhancers were discovered nearly 30 years ago, and for most of the intervening period, they existed in the experimental literature largely as an odd feature of certain tissue-specific loci, with a peculiar ability to activate transcription from promoters over large genomic distances. Only recently have advances in technology and understanding provided a fuller delineation of the wide role that distal enhancers play in gene regulation in metazoans. Their function is not only likely to constitute a primary basis for differential gene expression that underlies cell type specificity, but also to have crucial functions in stem cell pluripotency, human disease, and metazoan evolution. Mechanistically, enhancers lie at the nexus of transcription, nuclear organization, chromatin structure, epigenetics, and noncoding RNA. In accordance with such a complex spectrum of biological functions, it seems unlikely that enhancers constitute a monolithic class of regulatory element that works via a single, unified mechanism. ACKNOWLEDGMENTS We are grateful to M. Bender, M. Conerly, R. Kamakaka, J. Palis, T. Ragoczy, H. Rincon-Arano, J. Ritlund, and D. Strongin for critical reading of the manuscript and helpful suggestions.
REFERENCES Amano, T., Sagai, T., Tanabe, H., Mizushina, Y., Nakazawa, H., and Shiroishi, T. (2009). Chromosomal dynamics at the Shh locus: limb bud-specific differential regulation of competence and active transcription. Dev. Cell 16, 47–57. Andrulis, E.D., Neiman, A.M., Zappulla, D.C., and Sternglanz, R. (1998). Perinuclear localization of chromatin facilitates transcriptional silencing. Nature 394, 592–595. Banerji, J., Rusconi, S., and Schaffner, W. (1981). Expression of a beta-globin gene is enhanced by remote SV40 DNA sequences. Cell 27, 299–308. Banerji, J., Olson, L., and Schaffner, W. (1983). A lymphocyte-specific cellular enhancer is located downstream of the joining region in immunoglobulin heavy chain genes. Cell 33, 729–740.
Brown, J.M., Green, J., das Neves, R.P., Wallace, H.A., Smith, A.J., Hughes, J., Gray, N., Taylor, S., Wood, W.G., Higgs, D.R., et al. (2008). Association between active genes occurs at nuclear speckles and is modulated by chromatin environment. J. Cell Biol. 182, 1083–1097. Bulger, M., and Groudine, M. (1999). Looping versus linking: toward a model for long-distance gene activation. Genes Dev. 13, 2465–2477. Butler, J.E., and Kadonaga, J.T. (2001). Enhancer-promoter specificity mediated by DPE or TATA core promoter motifs. Genes Dev. 15, 2515–2519. Cai, S., Han, H.J., and Kohwi-Shigematsu, T. (2003). Tissue-specific nuclear architecture and gene expression regulated by SATB1. Nat. Genet. 34, 42–51. Cai, S., Lee, C.C., and Kohwi-Shigematsu, T. (2006). SATB1 packages densely looped, transcriptionally active chromatin for coordinated expression of cytokine genes. Nat. Genet. 38, 1278–1288. Calhoun, V.C., and Levine, M. (2003). Long-range enhancer-promoter interactions in the Scr-Antp interval of the Drosophila Antennapedia complex. Proc. Natl. Acad. Sci. USA 100, 9878–9883. Calhoun, V.C., Stathopoulos, A., and Levine, M. (2002). Promoter-proximal tethering elements regulate enhancer-promoter specificity in the Drosophila Antennapedia complex. Proc. Natl. Acad. Sci. USA 99, 9243–9247. Chan, Y.F., Marks, M.E., Jones, F.C., Villarreal, G., Jr., Shapiro, M.D., Brady, S.D., Southwick, A.M., Absher, D.M., Grimwood, J., Schmutz, J., et al. (2010). Adaptive evolution of pelvic reduction in sticklebacks by recurrent deletion of a Pitx1 enhancer. Science 327, 302–305. Chaturvedi, C.P., Hosey, A.M., Palii, C., Perez-Iratxeta, C., Nakatani, Y., Ranish, J.A., Dilworth, F.J., and Brand, M. (2009). Dual role for the methyltransferase G9a in the maintenance of beta-globin gene transcription in adult erythroid cells. Proc. Natl. Acad. Sci. USA 106, 18303–18308. Cosma, M.P., Tanaka, T., and Nasmyth, K. (1999). Ordered recruitment of transcription and chromatin remodeling factors to a cell cycle- and developmentally regulated promoter. Cell 97, 299–311. Cremer, T., and Cremer, M. (2010). Chromosome territories. Cold Spring Harb. Perspect. Biol. 2, a003889. Crawford, G.E., Davis, S., Scacheri, P.C., Renaud, G., Halawi, M.J., Erdos, M.R., Green, R., Meltzer, P.S., Wolfsberg, T.G., and Collins, F.S. (2006). DNase-chip: a high-resolution method to identify DNase I hypersensitive sites using tiled microarrays. Nat. Methods 3, 503–509. Creyghton, M.P., Cheng, A.W., Welstead, G.G., Kooistra, T., Carey, B.W., Steine, E.J., Hanna, J., Lodato, M.A., Frampton, G.M., Sharp, P.A., et al. (2010). Histone H3K27ac separates active from poised enhancers and predicts developmental state. Proc. Natl. Acad. Sci. USA 107, 21931–21936. Cullen, K.E., Kladde, M.P., and Seyfred, M.A. (1993). Interaction between transcription regulatory regions of prolactin chromatin. Science 261, 203–206. de Bruin, D., Zaman, Z., Liberatore, R.A., and Ptashne, M. (2001). Telomere looping permits gene activation by a downstream UAS in yeast. Nature 409, 109–113. Dekker, J., Rippe, K., Dekker, M., and Kleckner, N. (2002). Capturing chromosome conformation. Science 295, 1306–1311.
Barski, A., Cuddapah, S., Cui, K., Roh, T.Y., Schones, D.E., Wang, Z., Wei, G., Chepelev, I., and Zhao, K. (2007). High-resolution profiling of histone methylations in the human genome. Cell 129, 823–837.
de Kok, Y.J., Vossenaar, E.R., Cremers, C.W., Dahl, N., Laporte, J., Hu, L.J., Lacombe, D., Fischel-Ghodsian, N., Friedman, R.A., Parnes, L.S., et al. (1996). Identification of a hot spot for microdeletions in patients with X-linked deafness type 3 (DFN3) 900 kb proximal to the DFN3 gene POU3F4. Hum. Mol. Genet. 5, 1229–1235.
Bender, M.A., Bulger, M., Close, J., and Groudine, M. (2000). Beta-globin gene switching and DNase I sensitivity of the endogenous beta-globin locus in mice do not require the locus control region. Mol. Cell 5, 387–393.
de Laat, W., Klous, P., Kooren, J., Noordermeer, D., Palstra, R.J., Simonis, M., Splinter, E., and Grosveld, F. (2008). Three-dimensional organization of gene expression in erythroid cells. Curr. Top. Dev. Biol. 82, 117–139.
Blackwood, E.M., and Kadonaga, J.T. (1998). Going the distance: a current view of enhancer action. Science 281, 60–63.
Demers, C., Chaturvedi, C.P., Ranish, J.A., Juban, G., Lai, P., Morle, F., Aebersold, R., Dilworth, F.J., Groudine, M., and Brand, M. (2007). Activatormediated recruitment of the MLL2 methyltransferase complex to the betaglobin locus. Mol. Cell 27, 573–584.
Blow, M.J., McCulley, D.J., Li, Z., Zhang, T., Akiyama, J.A., Holt, A., PlajzerFrick, I., Shoukry, M., Wright, C., Chen, F., et al. (2010). ChIP-Seq identification of weakly conserved heart enhancers. Nat. Genet. 42, 806–810. Branco, M.R., and Pombo, A. (2007). Chromosome organization: new facts, new models. Trends Cell Biol. 17, 127–134.
336 Cell 144, February 4, 2011 ª2011 Elsevier Inc.
De Santa, F., Barozzi, I., Mietton, F., Ghisletti, S., Polletti, S., Tusi, B.K., Muller, H., Ragoussis, J., Wei, C.L., and Natoli, G. (2010). A large fraction of extragenic RNA pol II transcription sites overlap enhancers. PLoS Biol. 8, e1000384.
Dobi, K.C., and Winston, F. (2007). Analysis of transcriptional activation at a distance in Saccharomyces cerevisiae. Mol. Cell. Biol. 27, 5575–5586.
Holmes, S.G., and Broach, J.R. (1996). Silencers are required for inheritance of the repressed state in yeast. Genes Dev. 10, 1021–1032.
Dorsett, D. (1999). Distant liaisons: long-range enhancer-promoter interactions in Drosophila. Curr. Opin. Genet. Dev. 9, 505–514.
Hou, C., Dale, R., and Dean, A. (2010). Cell type specificity of chromatin organization mediated by CTCF and cohesin. Proc. Natl. Acad. Sci. USA 107, 3651–3656.
Dostie, J., Richmond, T.A., Arnaout, R.A., Selzer, R.R., Lee, W.L., Honan, T.A., Rubio, E.D., Krumm, A., Lamb, J., Nusbaum, C., et al. (2006). Chromosome Conformation Capture Carbon Copy (5C): a massively parallel solution for mapping interactions between genomic elements. Genome Res. 16, 1299– 1309. Drissen, R., Palstra, R.J., Gillemans, N., Splinter, E., Grosveld, F., Philipsen, S., and de Laat, W. (2004). The active spatial organization of the beta-globin locus requires the transcription factor EKLF. Genes Dev. 18, 2485–2490. Elgin, S.C. (1988). The formation and function of DNase I hypersensitive sites in the process of gene activation. J. Biol. Chem. 263, 19259–19262. Emison, E.S., McCallion, A.S., Kashuk, C.S., Bush, R.T., Grice, E., Lin, S., Portnoy, M.E., Cutler, D.J., Green, E.D., and Chakravarti, A. (2005). A common sex-dependent mutation in a RET enhancer underlies Hirschsprung disease risk. Nature 434, 857–863. ENCODE Project Consortium. (2007). Identification and analysis of functional elements in 1% of the human genome by the ENCODE pilot project. Nature 447, 799–816. Ferrai, C., de Castro, I.J., Lavitas, L., Chotalia, M., and Pombo, A. (2010). Gene positioning. Cold Spring Harbor Perspect. Biol. 2, a000588. Finlan, L.E., Sproul, D., Thomson, I., Boyle, S., Kerr, E., Perry, P., Ylstra, B., Chubb, J.R., and Bickmore, W.A. (2008). Recruitment to the nuclear periphery can alter expression of genes in human cells. PLoS Genet. 4, e1000039. Francastel, C., Walters, M.C., Groudine, M., and Martin, D.I. (1999). A functional enhancer suppresses silencing of a transgene and prevents its localization close to centrometric heterochromatin. Cell 99, 259–269. Ghisletti, S., Barozzi, I., Mietton, F., Polletti, S., De Santa, F., Venturini, E., Gregory, L., Lonie, L., Chew, A., Wei, C.L., et al. (2010). Identification and characterization of enhancers controlling the inflammatory gene expression program in macrophages. Immunity 32, 317–328. Gillies, S.D., Morrison, S.L., Oi, V.T., and Tonegawa, S. (1983). A tissuespecific transcription enhancer element is located in the major intron of a rearranged immunoglobulin heavy chain gene. Cell 33, 717–728. Gross, D.S., and Garrard, W.T. (1988). Nuclease hypersensitive sites in chromatin. Annu. Rev. Biochem. 57, 159–197. Grosveld, F., van Assendelft, G.B., Greaves, D.R., and Kollias, G. (1987). Position-independent, high-level expression of the human beta-globin gene in transgenic mice. Cell 51, 975–985. Hardison, R., Slightom, J.L., Gumucio, D.L., Goodman, M., Stojanovic, N., and Miller, W. (1997). Locus control regions of mammalian beta-globin gene clusters: combining phylogenetic analyses and experimental results to gain functional insights. Gene 205, 73–94. Hare, E.E., Peterson, B.K., Iyer, V.N., Meier, R., and Eisen, M.B. (2008). Sepsid even-skipped enhancers are functionally conserved in Drosophila despite lack of sequence conservation. PLoS Genet. 4, e1000106. Heintzman, N.D., Stuart, R.K., Hon, G., Fu, Y., Ching, C.W., Hawkins, R.D., Barrera, L.O., Van Calcar, S., Qu, C., Ching, K.A., et al. (2007). Distinct and predictive chromatin signatures of transcriptional promoters and enhancers in the human genome. Nat. Genet. 39, 311–318. Heintzman, N.D., Hon, G.C., Hawkins, R.D., Kheradpour, P., Stark, A., Harp, L.F., Ye, Z., Lee, L.K., Stuart, R.K., Ching, C.W., et al. (2009). Histone modifications at human enhancers reflect global cell-type-specific gene expression. Nature 459, 108–112. Ho, Y., Elefant, F., Cooke, N., and Liebhaber, S. (2002). A defined locus control region determinant links chromatin domain acetylation with long-range gene activation. Mol. Cell 9, 291–302. Ho, Y., Elefant, F., Liebhaber, S.A., and Cooke, N.E. (2006). Locus control region transcription plays an active role in long-range gene activation. Mol. Cell 23, 365–375.
Iborra, F.J., Pombo, A., Jackson, D.A., and Cook, P.R. (1996). Active RNA polymerases are localized within discrete transcription ‘‘factories’’ in human nuclei. J. Cell Sci. 109, 1427–1436. Juven-Gershon, T., Hsu, J.Y., and Kadonaga, J.T. (2008). Caudal, a key developmental regulator, is a DPE-specific transcriptional factor. Genes Dev. 22, 2823–2830. Kagey, M.H., Newman, J.J., Bilodeau, S., Zhan, Y., Orlando, D.A., van Berkum, N.L., Ebmeier, C.C., Goossens, J., Rahl, P.B., Levine, S.S., et al. (2010). Mediator and cohesin connect gene expression and chromatin architecture. Nature 467, 430–435. Kim, E., Magen, A., and Ast, G. (2007a). Different levels of alternative splicing among eukaryotes. Nucleic Acids Res. 35, 125–131. Kim, A., Zhao, H., Ifrim, I., and Dean, A. (2007b). Beta-globin intergenic transcription and histone acetylation dependent on an enhancer. Mol. Cell. Biol. 27, 2980–2986. Kim, T.K., Hemberg, M., Gray, J.M., Costa, A.M., Bear, D.M., Wu, J., Harmin, D.A., Laptewicz, M., Barbara-Haley, K., Kuersten, S., et al. (2010). Widespread transcription at neuronal activity-regulated enhancers. Nature 465, 182–187. Kleinjan, D.A., and van Heyningen, V. (2005). Long-range control of gene expression: emerging mechanisms and disruption in disease. Am. J. Hum. Genet. 76, 8–32. Koch, C.M., Andrews, R.M., Flicek, P., Dillon, S.C., Karao¨z, U., Clelland, G.K., Wilcox, S., Beare, D.M., Fowler, J.C., Couttet, P., et al. (2007). The landscape of histone modifications across 1% of the human genome in five human cell lines. Genome Res. 17, 691–707. Koch, F., Jourquin, F., Ferrier, P., and Andrau, J.C. (2008). Genome-wide RNA polymerase II: not genes only!. Trends Biochem. Sci. 33, 265–273. Krebs, J.E., Kuo, M.H., Allis, C.D., and Peterson, C.L. (1999). Cell cycleregulated histone acetylation required for expression of the yeast HO gene. Genes Dev. 13, 1412–1421. Kurukuti, S., Tiwari, V.K., Tavoosidana, G., Pugacheva, E., Murrell, A., Zhao, Z., Lobanenkov, V., Reik, W., and Ohlsson, R. (2006). CTCF binding at the H19 imprinting control region mediates maternally inherited higher-order chromatin conformation to restrict enhancer access to Igf2. Proc. Natl. Acad. Sci. USA 103, 10684–10689. Leach, K.M., Nightingale, K., Igarashi, K., Levings, P.P., Engel, J.D., Becker, P.B., and Bungert, J. (2001). Reconstitution of human beta-globin locus control region hypersensitive sites in the absence of chromatin assembly. Mol. Cell. Biol. 21, 2629–2640. Lettice, L.A., Heaney, S.J., Purdie, L.A., Li, L., de Beer, P., Oostra, B.A., Goode, D., Elgar, G., Hill, R.E., and de Graaff, E.A. (2003). A long-range Shh enhancer regulates expression in the developing limb and fin and is associated with preaxial polydactyly. Hum. Mol. Genet. 12, 1725–1735. Levine, M. (2010). Transcriptional enhancers in animal development and evolution. Curr. Biol. 20, R754–R763. Liber, D., Domaschenz, R., Holmqvist, P.H., Mazzarella, L., Georgiou, A., Leleu, M., Fisher, A.G., Labosky, P.A., and Dillon, N. (2010). Epigenetic priming of a pre-B cell-specific enhancer through binding of Sox2 and Foxd3 at the ESC stage. Cell Stem Cell 7, 114–126. Lieberman-Aiden, E., van Berkum, N.L., Williams, L., Imakaev, M., Ragoczy, T., Telling, A., Amit, I., Lajoie, B.R., Sabo, P.J., Dorschner, M.O., et al. (2009). Comprehensive mapping of long-range interactions reveals folding principles of the human genome. Science 326, 289–293. Melnikova, L., Kostuchenko, M., Silicheva, M., and Georgiev, P. (2008). Drosophila gypsy insulator and yellow enhancers regulate activity of yellow promoter through the same regulatory element. Chromosoma 117, 137–145. Miele, A., and Dekker, J. (2008). Long-range chromosomal interactions and gene regulation. Mol. Biosyst. 4, 1046–1057.
Cell 144, February 4, 2011 ª2011 Elsevier Inc. 337
Miele, A., and Dekker, J. (2009). Mapping cis- and trans- chromatin interaction networks using chromosome conformation capture (3C). Methods Mol. Biol. 464, 105–121. Moon, A.M., and Ley, T.J. (1990). Conservation of the primary structure, organization, and function of the human and mouse beta-globin locus-activating regions. Proc. Natl. Acad. Sci. USA 87, 7693–7697. McGaughey, D.M., Vinton, R.M., Huynh, J., Al-Saif, A., Beer, M.A., and McCallion, A.S. (2008). Metrics of sequence constraint overlook regulatory sequences in an exhaustive analysis at phox2b. Genome Res. 18, 252–260. Moen, P.T., Jr., Johnson, C.V., Byron, M., Shopland, L.S., de la Serna, I.L., Imbalzano, A.N., and Lawrence, J.B. (2004). Repositioning of muscle-specific genes relative to the periphery of SC-35 domains during skeletal myogenesis. Mol. Biol. Cell 15, 197–206. Moreau, P., Hen, R., Wasylyk, B., Everett, R., Gaub, M.P., and Chambon, P. (1981). The SV40 72 base repair repeat has a striking effect on gene expression both in SV40 and other chimeric recombinants. Nucleic Acids Res. 9, 6047– 6068. Morrow, B.E., Johnson, S.P., and Warner, J.R. (1993). The rRNA enhancer regulates rRNA transcription in Saccharomyces cerevisiae. Mol. Cell. Biol. 13, 1283–1289. Neigeborn, L., and Warner, J.R. (1990). Expression of yeast 5S RNA is independent of the rDNA enhancer region. Nucleic Acids Res. 18, 4179–4184. Noonan, J.P. (2009). Regulatory DNAs and the evolution of human development. Curr. Opin. Genet. Dev. 19, 557–564. Noonan, J.P., and McCallion, A.S. (2010). Genomics of long-range regulatory elements. Annu. Rev. Genomics Hum. Genet. 11, 1–23. Noordermeer, D., Branco, M.R., Splinter, E., Klous, P., van Ijcken, W., Swagemakers, S., Koutsourakis, M., van der Spek, P., Pombo, A., and de Laat, W. (2008). Transcription and chromatin organization of a housekeeping gene cluster containing an integrated beta-globin locus control region. PLoS Genet. 4, e1000016. Odom, D.T., Dowell, R.D., Jacobsen, E.S., Gordon, W., Danford, T.W., MacIsaac, K.D., Rolfe, P.A., Conboy, C.M., Gifford, D.K., and Fraenkel, E. (2007). Tissue-specific transcriptional regulation has diverged significantly between human and mouse. Nat. Genet. 39, 730–732. Ohtsuki, S., Levine, M., and Cai, H.N. (1998). Different core promoters possess distinct regulatory activities in the Drosophila embryo. Genes Dev. 12, 547–556.
Human-specific gain of function in a developmental enhancer. Science 321, 1346–1350. Ptashne, M. (1986). Gene regulation by proteins acting nearby and at a distance. Nature 322, 697–701. Rada-Iglesias, A., Bajpai, R., Swigut, T., Brugmann, S.A., Flynn, R.A., and Wysocka, J. (2010). A unique chromatin signature uncovers early developmental enhancers in humans. Nature, in press. Published online December 15, 2010. 10.1038/nature09692. Ragoczy, T., Telling, A., Sawado, T., Groudine, M., and Kosak, S.T. (2003). A genetic analysis of chromosome territory looping: diverse roles for distal regulatory elements. Chromosome Res. 11, 513–525. Ragoczy, T., Bender, M.A., Telling, A., Byron, R., and Groudine, M. (2006). The locus control region is required for association of the murine beta-globin locus with engaged transcription factories during erythroid maturation. Genes Dev. 20, 1447–1457. Rebeiz, M., Pool, J.E., Kassner, V.A., Aquadro, C.F., and Carroll, S.B. (2009). Stepwise modification of a modular enhancer underlies adaptation in a Drosophila population. Science 326, 1663–1667. Reddy, K.L., Zullo, J.M., Bertolino, E., and Singh, H. (2008). Transcriptional repression mediated by repositioning of genes to the nuclear lamina. Nature 452, 243–247. Rollins, R.A., Morcillo, P., and Dorsett, D. (1999). Nipped-B, a Drosophila homologue of chromosomal adherins, participates in activation by remote enhancers in the cut and Ultrabithorax genes. Genetics 152, 577–593. Rubio, E.D., Reiss, D.J., Welcsh, P.L., Disteche, C.M., Filippova, G.N., Baliga, N.S., Aebersold, R., Ranish, J.A., and Krumm, A. (2008). CTCF physically links cohesin to chromatin. Proc. Natl. Acad. Sci. USA 105, 8309–8314. Rusche, L.N., Kirchmaier, A.L., and Rine, J. (2003). The establishment, inheritance, and function of silenced chromatin in Saccharomyces cerevisiae. Annu. Rev. Biochem. 72, 481–516. Sagai, T., Hosoya, M., Mizushina, Y., Tamura, M., and Shiroishi, T. (2005). Elimination of a long-range cis-regulatory module causes complete loss of limb-specific Shh expression and truncation of the mouse limb. Development 132, 797–803. Sawado, T., Halow, J., Bender, M.A., and Groudine, M. (2003). The beta -globin locus control region (LCR) functions primarily by enhancing the transition from transcription initiation to elongation. Genes Dev. 17, 1009–1018.
Ørom, U.A., Derrien, T., Beringer, M., Gumireddy, K., Gardini, A., Bussotti, G., Lai, F., Zytnicki, M., Notredame, C., Huang, Q., et al. (2010). Long noncoding RNAs with enhancer-like function in human cells. Cell 143, 46–58.
Schmidt, D., Wilson, M.D., Ballester, B., Schwalie, P.C., Brown, G.D., Marshall, A., Kutter, C., Watt, S., Martinez-Jimenez, C.P., Mackay, S., et al. (2010). Five-vertebrate ChIP-seq reveals the evolutionary dynamics of transcription factor binding. Science 328, 1036–1040.
Osborne, C.S., Chakalova, L., Brown, K.E., Carter, D., Horton, A., Debrand, E., Goyenechea, B., Mitchell, J.A., Lopes, S., Reik, W., and Fraser, P. (2004). Active genes dynamically colocalize to shared sites of ongoing transcription. Nat. Genet. 36, 1065–1071.
Schoenfelder, S., Sexton, T., Chakalova, L., Cope, N.F., Horton, A., Andrews, S., Kurukuti, S., Mitchell, J.A., Umlauf, D., Dimitrova, D.S., et al. (2010). Preferential associations between co-regulated genes reveal a transcriptional interactome in erythroid cells. Nat. Genet. 42, 53–61.
Pant, V., Kurukuti, S., Pugacheva, E., Shamsuddin, S., Mariano, P., Renkawitz, R., Klenova, E., Lobanenkov, V., and Ohlsson, R. (2004). Mutation of a single CTCF target site within the H19 imprinting control region leads to loss of Igf2 imprinting and complex patterns of de novo methylation upon maternal inheritance. Mol. Cell. Biol. 24, 3497–3504.
Simonis, M., Klous, P., Splinter, E., Moshkin, Y., Willemsen, R., de Wit, E., van Steensel, B., and de Laat, W. (2006). Nuclear organization of active and inactive chromatin domains uncovered by chromosome conformation capture-on-chip (4C). Nat. Genet. 38, 1348–1354.
Petrascheck, M., Escher, D., Mahmoudi, T., Verrijzer, C.P., Schaffner, W., and Barberis, A. (2005). DNA looping induced by a transcriptional enhancer in vivo. Nucleic Acids Res. 33, 3743–3750. Phillips, J.E., and Corces, V.G. (2009). CTCF: master weaver of the genome. Cell 137, 1194–1211. Pollard, K.S., Salama, S.R., Lambert, N., Lambot, M.A., Coppens, S., Pedersen, J.S., Katzman, S., King, B., Onodera, C., Siepel, A., et al. (2006). An RNA gene expressed during cortical development evolved rapidly in humans. Nature 443, 167–172.
Splinter, E., Heath, H., Kooren, J., Palstra, R.J., Klous, P., Grosveld, F., Galjart, N., and de Laat, W. (2006). CTCF mediates long-range chromatin looping and local histone modification in the beta-globin locus. Genes Dev. 20, 2349–2354. Stamatoyannopoulos, J.A., Goodwin, A., Joyce, T., and Lowrey, C.H. (1995). NF-E2 and GATA binding motifs are required for the formation of DNase I hypersensitive site 4 of the human beta-globin locus control region. EMBO J. 14, 106–116. Sutherland, H., and Bickmore, W.A. (2009). Transcription factories: gene expression in unions? Nat. Rev. Genet. 10, 457–466.
Prabhakar, S., Noonan, J.P., Pa¨a¨bo, S., and Rubin, E.M. (2006). Accelerated evolution of conserved noncoding sequences in humans. Science 314, 786.
Swanson, C.I., Evans, N.C., and Barolo, S. (2010). Structural rules and complex regulatory circuitry constrain expression of a Notch- and EGFRregulated eye enhancer. Dev. Cell 18, 359–370.
Prabhakar, S., Visel, A., Akiyama, J.A., Shoukry, M., Lewis, K.D., Holt, A., Plajzer-Frick, I., Morrison, H., Fitzpatrick, D.R., Afzal, V., et al. (2008).
Taddei, A. (2007). Active genes at the nuclear pore complex. Curr. Opin. Cell Biol. 19, 305–310.
338 Cell 144, February 4, 2011 ª2011 Elsevier Inc.
Talbert, P.B., and Henikoff, S. (2006). Spreading of silent chromatin: inaction at a distance. Nat. Rev. Genet. 7, 793–803.
Wu, C. (1980). The 50 ends of Drosophila heat shock genes in chromatin are hypersensitive to DNase I. Nature 286, 854–860.
Vakoc, C.R., Letting, D.L., Gheldof, N., Sawado, T., Bender, M.A., Groudine, M., Weiss, M.J., Dekker, J., and Blobel, G.A. (2005). Proximity among distant regulatory elements at the beta-globin locus requires GATA-1 and FOG-1. Mol. Cell 17, 453–462.
Xi, H., Shulha, H.P., Lin, J.M., Vales, T.R., Fu, Y., Bodine, D.M., McKay, R.D., Chenoweth, J.G., Tesar, P.J., Furey, T.S., et al. (2007). Identification and characterization of cell type-specific and ubiquitous chromatin regulatory structures in the human genome. PLoS Genet. 3, e136.
Valenzuela, L., Dhillon, N., Dubey, R.N., Gartenberg, M.R., and Kamakaka, R.T. (2008). Long-range communication between the silencers of HMR. Mol. Cell. Biol. 28, 1924–1935.
Xu, H., and Hoover, T.R. (2001). Transcriptional regulation at a distance in bacteria. Curr. Opin. Microbiol. 4, 138–144.
Visel, A., Blow, M.J., Li, Z., Zhang, T., Akiyama, J.A., Holt, A., Plajzer-Frick, I., Shoukry, M., Wright, C., Chen, F., et al. (2009a). ChIP-seq accurately predicts tissue-specific activity of enhancers. Nature 457, 854–858. Visel, A., Rubin, E.M., and Pennacchio, L.A. (2009b). Genomic views of distantacting enhancers. Nature 461, 199–205. Wai, H., Johzuka, K., Vu, L., Eliason, K., Kobayashi, T., Horiuchi, T., and Nomura, M. (2001). Yeast RNA polymerase I enhancer is dispensable for transcription of the chromosomal rRNA gene and cell growth, and its apparent transcription enhancement from ectopic promoters requires Fob1 protein. Mol. Cell. Biol. 21, 5541–5553. Wang, Z., Zang, C., Rosenfeld, J.A., Schones, D.E., Barski, A., Cuddapah, S., Cui, K., Roh, T.Y., Peng, W., Zhang, M.Q., and Zhao, K. (2008). Combinatorial patterns of histone acetylations and methylations in the human genome. Nat. Genet. 40, 897–903. Wendt, K.S., Yoshida, K., Itoh, T., Bando, M., Koch, B., Schirghuber, E., Tsutsumi, S., Nagae, G., Ishihara, K., Mishiro, T., et al. (2008). Cohesin mediates transcriptional insulation by CCCTC-binding factor. Nature 451, 796–801. Wilson, M.D., Barbosa-Morais, N.L., Schmidt, D., Conboy, C.M., Vanes, L., Tybulewicz, V.L.J., Fisher, E.M.C., Tavare´, S., and Odom, D.T. (2008). Species-specific transcription in mice carrying human chromosome 21. Science 322, 434–438.
Xu, M., and Cook, P.R. (2008). Similar active genes cluster in specialized transcription factories. J. Cell Biol. 181, 615–623. Xu, J., Watts, J.A., Pope, S.D., Gadue, P., Kamps, M., Plath, K., Zaret, K.S., and Smale, S.T. (2009). Transcriptional competence and the active marking of tissue-specific enhancers by defined transcription factors in embryonic and induced pluripotent stem cells. Genes Dev. 23, 2824–2838. Yusufzai, T.M., Tagami, H., Nakatani, Y., and Felsenfeld, G. (2004). CTCF tethers an insulator to subnuclear sites, suggesting shared insulator mechanisms across species. Mol. Cell 13, 291–298. Zhao, Z., Tavoosidana, G., Sjo¨linder, M., Go¨ndo¨r, A., Mariano, P., Wang, S., Kanduri, C., Lezcano, M., Sandhu, K.S., Singh, U., et al. (2006). Circular chromosome conformation capture (4C) uncovers extensive networks of epigenetically regulated intra- and interchromosomal interactions. Nat. Genet. 38, 1341–1347. Zhu, X., Ling, J., Zhang, L., Pi, W., Wu, M., and Tuan, D. (2007). A facilitated tracking and transcription mechanism of long-range enhancer function. Nucleic Acids Res. 35, 5532–5544. Zinzen, R.P., Girardot, C., Gagneur, J., Braun, M., and Furlong, E.E. (2009). Combinatorial binding predicts spatio-temporal cis-regulatory activity. Nature 462, 65–70.
Cell 144, February 4, 2011 ª2011 Elsevier Inc. 339
Skin Stem Cells Orchestrate Directional Migration by Regulating Microtubule-ACF7 Connections through GSK3b Xiaoyang Wu,1 Qing-Tao Shen,2 Daniel S. Oristian,1 Catherine P. Lu,1 Qinsi Zheng,1,3 Hong-Wei Wang,2 and Elaine Fuchs1,* 1The Howard Hughes Medical Institute and Laboratory of Mammalian Cell Biology and Development, Rockefeller University, New York, NY 10065, USA 2Department of Molecular Biophysics and Biochemistry, Yale University, 333 Cedar Street, New Haven, CT 06520, USA 3Tri-institutional Training Program in Chemical Biology, Weill Cornell Medical College, New York, NY 10065, USA *Correspondence:
[email protected] DOI 10.1016/j.cell.2010.12.033
SUMMARY
Homeostasis and wound healing rely on stem cells (SCs) whose activity and directed migration are often governed by Wnt signaling. In dissecting how this pathway integrates with the necessary downstream cytoskeletal dynamics, we discovered that GSK3b, a kinase inhibited by Wnt signaling, directly phosphorylates ACF7, a > 500 kDa microtubule-actin crosslinking protein abundant in hair follicle stem cells (HF-SCs). We map ACF7’s GSK3b sites to the microtubule-binding domain and show that phosphorylation uncouples ACF7 from microtubules. Phosphorylation-refractile ACF7 rescues overall microtubule architecture, but phosphorylationconstitutive mutants do not. Neither mutant rescues polarized movement, revealing that phospho-regulation must be dynamic. This circuitry is physiologically relevant and depends upon polarized GSK3b inhibition at the migrating front of SCs/progeny streaming from HFs during wound repair. Moreover, only ACF7 and not GSKb-refractile-ACF7 restore polarized microtubule-growth and SC-migration to ACF7 null skin. Our findings provide insights into how this conserved spectraplakin integrates signaling, cytoskeletal dynamics, and polarized locomotion of somatic SCs. INTRODUCTION Directional cell movement is essential for developmental morphogenesis, tumor metastasis, and wound repair. A typical migrating cell adopts front-rear polarity with asymmetrical distribution of signaling molecules and cytoskeletal components. During establishment of polarity, temporal capture and stabilization of microtubules (MTs) occur near filamentous actin (F-actin)-enriched leading edges, which enable reorientation of the MT-organizing center and Golgi complex to ensure biased vesicular transport for directional migration (Etienne-Manneville, 2004; Siegrist and
Doe, 2007). Among many regulatory molecules required for cell migration, the ubiquitously-expressed serine and threonine kinase glycogen synthase kinase 3b (GSK3b) is particularly important in transmitting upstream signaling necessary for establishing cell polarity and guiding directional movement (Sun et al., 2009). In contrast to most other protein kinases, GSK3b activity is high in nonstimulated cells, but it is dampened at the leading edge of scratch-wounded astrocytes and endodermal cells in vitro and in developing neurons of brain slices ex vivo (Etienne-Manneville and Hall, 2003; Kodama et al., 2003; Yoshimura et al., 2006). In vitro, GSK3b inhibition and cell polarity are known to be provoked by CDC42 (Etienne-Manneville and Hall, 2003). In vivo, GSK3b activity is inhibited by Wnt signaling, which in turn leads to b-catenin stabilization, which plays a critical and near universal role in stem cell (SC) activation and migration (Fuchs, 2009; Nusse et al., 2008). Despite this tantalizing connection, the possible physiological relevance among Wnt signaling, SC activation, and cytoskeletal remodeling remains unclear. Another key unaddressed question is how GSK3b regulates the dynamic changes in MT organization and stabilization that transpire during polarized movements of somatic SCs. Hair follicles (HFs) provide an ideal paradigm to address these issues. Adult HFs undergo homeostasis through cyclical bouts of active growth (anagen), regression (catagen), and rest (telogen) (Figure S1A available online). They also participate in epidermal re-epithelialization during wound healing (Blanpain and Fuchs, 2009; Ito et al., 2005; Tumbar et al., 2004). Both processes rely upon a resident population of SCs, which reside in a specific niche called the bulge, located at the base of the noncycling portion of the HF (Figure S1A). Wnt signaling emanating from polarized epithelial-mesenchymal crosstalk in the SC niche is critical for hair follicle stem cell (HF-SC) activation during tissue homeostasis (Greco et al., 2009). In addition, localized Wnt signaling/b-catenin at a wound site initiates the migration of SCs after injury (Ito et al., 2005, 2007). Based upon these points, the potential exists for using skin as a model to dissect how signaling cues to SCs might regulate the cytoskeleton to orchestrate cell polarization and directional cell movement. In searching for potential cytoskeletal regulators in this process, we noticed that HF-SCs display significantly more actin crosslinking Cell 144, 341–352, February 4, 2011 ª2011 Elsevier Inc. 341
factor-7 (ACF7; also called MACF1, MT and actin crosslinking factor-1) than other skin epithelial cells (Blanpain et al., 2004; Morris et al., 2004; Tumbar et al., 2004) (Figure S1B). Unique to multicellular organisms, spectraplakins such as ACF7 can bind both MT and actin networks (Jefferson et al., 2004; Ro¨per et al., 2002). Although broadly expressed, their essential functions are most clearly revealed in muscle, neurons and skin epithelial cells, which maintain elaborate yet dynamic cytoskeletal networks. Mutations in the single Drosophila spectraplakin gene (Kakapo/short-stop/shot) cause a wide variety of cellular and tissue defects that include perturbations in actin-MT organization, cell-cell adhesion and integrin-mediated epidermal attachment to muscle. There are two evolutionarily conserved mammalian counterparts. Mice lacking BPAG1/dystonin display sensory neuron and muscle degeneration and have gross defects in cytoskeletal organization and function. By contrast, mice lacking ACF7 exhibit early embryonic lethality (Chen et al., 2006; Kodama et al., 2003). Recent studies show that mice conditionally lacking ACF7 display defects in cell migration. This was true for both K14-Cre cKO animals, impaired in skin wound healing (Wu et al., 2008) and for Nestin-Cre cKO mice, defective in brain development (Goryunov et al., 2010). While the loss of function data underscore the importance of spectraplakins in coordinating the cytoskeletal dynamics necessary for cells to polarize and move in a directed fashion (Rodriguez et al., 2003), the mechanisms underlying the regulation of spectraplakin-mediated actin-MT connections remain unknown. Similarly, lacking are the molecular details of the circuitry that must link upstream signaling pathways to cytoskeletal remodeling in order for SCs to migrate from their niche. In the present report, we make major inroads into understanding this process. We identify a link between GSK3b and ACF7 and further reveal the in vivo relevance of this connection in polarized locomotion of skin stem cells upon injury. RESULTS C-Terminal Tail Interactions Govern Binding between ACF7 and Microtubules ACF7’s carboxy-terminal tail (CT) contains a Gas2-related (GAR) domain and a (GSR-repeat domain (GSR) domain. Previous studies suggest, and we have confirmed, that both domains are involved in the interaction with MTs (Sun et al., 2001; Wu et al., 2008). To obtain structural information on this interaction, we incubated ACF7(CT) with polymerized MTs and conducted ultrastructural analyses. Under saturating conditions [4:1 molar ACF7 (CT):tubulin heterodimer], ACF7(CT) markedly enhanced the electron density along the MT surface (Figure 1A). When compared with naked MTs, ACF7(CT)-coated MTs were 10 nm thicker in diameter (projection profile in Figure 1A). Fourier transform analyses further indicated that whether decorated with ACF7(CT) or not, assembled MTs displayed a 40A˚ layer line corresponding to the packing of tubulin dimers. However, only ACF7(CT)-decorated MTs displayed discrete 80A˚ layer lines, suggesting that ACF7 might associate with the MT lattice with a rather weak distinction between a- and b- tubulins (Figure 1A). Interaction with the MT lattice usually involves the acidic C-terminal tails of tubulin subunits that protrude from the MT 342 Cell 144, 341–352, February 4, 2011 ª2011 Elsevier Inc.
surface. To test this hypothesis, we performed binding assays between ACF7(CT) and increasing concentrations of taxol-stabilized MTs. Just prior to adding ACF7(CT), we exposed half the polymerized MTs to subtilisin to shave protruding tubulin tails (MTDC-tail). Following ultracentrifugation, pellets were then analyzed by SDS-polyacrylamide gel electrophoresis (PAGE) and Coomassie Blue (CB) staining. Subtilisin-treated MTs still pelleted after ultracentrifugation, confirming that MTs remained assembled after treatment. Only a slight increase was noted in tubulin’s electrophoretic mobility, consistent with tail removal (compare asterisked lanes in Figure 1B). However, this modification markedly diminished ACF7(CT)’s binding to MTs (Figure 1B). When ACF7(CT)’s concentration was progressively increased, its association with MTs became saturating, reflected by an increase in the soluble pool of free ACF7(CT). A Scatchard plot of the data are shown in Figure 1C. ACF7(CT) binding to untreated MTs gave a KD of 1.4 3 107 M, which was comparable to published results on GAR and GSR domains of ACF7 (Sun et al., 2001). By contrast, the affinity of ACF7(CT) for subtilisin-treated MTs was >103 weaker (KD 1.5 3 106 M). Phosphorylation of ACF7’s C-Terminal Domain by GSK3b Association with tubulin tails usually depends on electrostatic interactions between acidic tubulin C termini and positively charged surfaces of MT-binding proteins. Consistent with this notion, ACF7(CT) contains many lysine (K) and arginine (R) residues. Particularly, the GSR domain (202 amino acids) harbors 36 strongly basic residues, with a calculated isoelectric point at 11.8. Additionally, 32% of residues in ACF7’s GSR domain are serine or threonine, suggestive of the potential to regulate ACF7-MT’s electrostatic interactions through protein phosphorylation. To address this possibility, we transfected primary keratinocytes with ACF7(CT) and labeled them with [32P]-orthophosphate. Our results showed that ACF7(CT) is efficiently phosphorylated in vivo (Figure 1D). We next analyzed phospho-ACF7(CT) by mass spectrometry. We circumvented the difficulties posed by ACF7(CT)’s high percentage of basic residues by choosing protease AspN, which cleaves peptide bonds N-terminal to aspartate. Liquid chromatography-tandem mass spectrometry (LC-MS/MS) identified multiple (up to six) phosphorylation sites within a peptide encompassing the GSR repeats (Figures 1E and 1F). Interestingly, the characteristic GSR repeats within this sequence have six serines that match consensus GSK3b phosphorylation sites [phosphorylation cluster-1 (P1)]. While in silico analysis with different phosphorylation prediction algorithms (Obenauer et al., 2003; Schiller, 2007) revealed additional potential GSK3b sites [phosphorylation cluster 2 (P2), Figure 1F], their incompatible sequence context precluded efficient peptide retrieval for MS/MS analysis. To determine whether ACF7 is specifically phosphorylated by GSK3, we first tested for an endogenous association between ACF7 and GSK3b proteins in HF-SCs. Immunoblot analyses revealed GSK3b in anti-ACF7 immunoprecipitates of wild-type (WT) but not ACF7 null cell lysates (Figure 2A). In vitro kinase (IVK) assays further showed that ACF7 is a substrate for active GSK3b and that GSK3b phosphorylates ACF7(CT) but not ACF7-NT (N-terminal domain of ACF7 serving as a control)
Figure 1. Structural Evidence that Electrostatic Interactions and Phosphorylation May Regulate Associations between ACF7 and MT (A) Ultrastructural analyses of negatively stained MTs either decorated with ACF7(CT) or naked. Data were normalized and are shown superimposed in the upper right as 1D projection profiles along the MT axis. The scale bars represent 20 nm. A fourier transform of an ACF7decorated MT is shown with the 40 A˚ and 80 A˚ layer lines denoted by white arrows. (B) Cosedimentation assay with ACF7(CT) and with decreasing amounts (20, 10, 5, and 2.5 mg) of MTs ± subtilisin. Note small downshift of tubulin bands (asterisks) and also decreased binding to ACF7(CT) that occurs following enzymatic treatment of polymerized MTs. MTDC-tail indicates MT with C-terminal tail removed by subtilisin. (C) Dissociation constants (Kd) of ACF7’s interactions with MT ± subtilisin. Binding assays were performed as in (B) but with a range of ACF7(CT) concentrations. Scatchard plot represents average of three data sets. (D) GST-tagged ACF7(CT) was purified from [32P] orthophosphate-labeled keratinocyte lysates ± calf intestine alkaline phosphatase (CIP) and subjected to autoradiography and immunoblotting (IB). (E) Tandem mass spectrometry peptide mass map of AspN-generated peptides from purified ACF7 (CT) protein. Stars (+) denote multiple signals corresponding to a serine-rich phospho-peptide, whose deduced sequence is shown. (F) Diagram of ACF7 with its many domains. The following abbreviations are used: CH, calponin homology F-actin binding; EF: EF hand motif mediating potential Ca2+ binding; GAR-GSR, Gas2-related and GSR repeats for MT binding. GSR repeats contain two potential GSK3b phosphorylation clusters, shown here, aligned with consensus GSK3b site (phospho-Ser in red, requisite Arg in blue). Figure 1 is associated with Figure S1.
(Figure 2B). Moreover, coexpression of GSK3b with ACF7(CT) in cultured cells resulted in phosphorylation of ACF7 that was sensitive to treatment of phosphatase (Figure 2C). To assess whether our identification of GSK3b phosphorylation sites in ACF7(CT) was correct, we replaced the predicted GSK3b-targeted serines with alanines and repeated our phosphorylation assays in vitro and in vivo. Individually, mutations in P1 and P2 each reduced overall phosphorylation. Combinatorial mutations of both clusters abolished ACF7(CT) GSK3b phosphorylation (Figures 2D and 2E). Phosphorylation of ACF7 by GSK3b Inhibits Microtubule Binding Our finding of functional GSK3b phosphorylation sites in ACF7’s GSR domain hinted a potential role of this signaling event in tempering ACF7’s MT connection by reducing their electrostatic affinity. To test this possibility, we had to first overcome the technical hurdles of ACF7’s enormous size (5380 amino acid residues) and engineer mammalian expression vectors encoding human influenza hemagglutinin epitope (HA)-tagged full-length ACF7 as well as point mutants that converted GSK3b phosphorylation sites at P1 and P2 to either a kinase-refractile version
harboring Ser/Ala mutations (S:A mutant) or a phosphomimetic version, containing Ser/Asp mutations (S:D mutant). We purified the proteins by affinity chromatography (Wu et al., 2008) and carried out in vitro cosedimentation assays with polymerized MTs (Figure 3A). Both HA-tagged ACF7 and its S:A mutant counterpart maintained a strong affinity for MT binding similar to WT ACF7, while ACF7(S:D) exhibited significantly reduced affinity for MTs. Similar results were obtained when the binding assays were repeated with freshly prepared lysates from cells expressing ACF7 or its mutants (Figure 3B). In contrast to the effects of S:A and S:D mutations on MT-binding affinity, ACF7’s F-actin-binding affinity and adenosine triphosphate (ATP) hydrolysis activity were unaffected (Figures S2A and S2B). To directly assess the effects of GSK3b phosphorylation, we next coexpressed constitutively active GSK3b (caGSK3b, S9A mutant) with either WT or kinase-refractile ACF7. As predicted, the MT-binding affinity of WT-ACF7 was reduced when GSK3b was superactivated (Figure 3C). This effect was specific for ACF7’s C-terminal GSR domain, since GSK3b did not alter interactions between ACF7(S:A) and MTs (Figure 3C). Consistent with these results, inhibition of endogenous GSK3b activity by Cell 144, 341–352, February 4, 2011 ª2011 Elsevier Inc. 343
Figure 2. GSK3b Associates with and Phosphorylates ACF7 (A) Bulge SC lysates were analyzed by SDS-PAGE and IB before and after immunoprecipitation (IP) with ACF7 and control IgG Abs. Blots were probed with GSK3b or ACF7 Abs as indicated. (B) In vitro GSK3b kinase (IVK) assays were performed on C- or N-terminal ACF7. Phosphorylation was analyzed by SDS-PAGE and autoradiography. (C) Lysates are from cultured cells expressing HA-tagged ACF7(CT) ± constitutively active (ca) GFP-tagged GSK3b (or Ctrl vector) ± CIP phosphatase. Lysates were analyzed ± anti-HA IP by SDS-PAGE and anti-phosphoserine/threonine (Pi-Ser/Thr) immunoblotting. (D) Cells were transfected with plasmids encoding WT or GSK3-site mutants (see Figure 1F) of HA-tagged ACF7 (CT) (or HA-tagged GST control) and caGSK3b or control vector. After labeling with [32P]-orthophosphate, proteins were subjected to anti-HA IP, SDS-PAGE, staining, and autoradiography. HC is an abbreviation for IgG heavy chain. (E) Recombinant kinase was used to test GSK3b phosphorylation of ACF7(CT) and its different mutants in vitro. Phosphorylation of ACF7(CT) was detected by IB with PiSer/Thr Ab.
exposing HF-SCs to different GSK3-specific inhibitors significantly increased the MT-binding affinity of endogenous ACF7 (Figure 3D). Together, these findings provided compelling support for GSK3b as an important regulator of ACF7’s association with MTs and showed that GSK3b-mediated regulation was exerted exclusively at P1 and P2 of the ACF7’s GSR domain. The Wound-Healing Delay in ACF7-Deficient Skin Arises From Defective HF-SC Migration Previously, we showed that mice targeted for loss of ACF7 in skin are defective in wound repair (Wu et al., 2008), a process known to involve HF-SCs (Ito et al., 2005; Tumbar et al., 2004). When coupled with the appearance of ACF7 on the list of genes upregulated in HF-SCs (Figure S1B) and an emerging role for Wnt signaling in wound repair (Fathke et al., 2006; Ito et al., 2007; Stoick-Cooper et al., 2007), our finding of GSK3b as a potential regulator of ACF7 took on newfound importance and merited further investigation. To begin to evaluate how ACF7 functions in this pathway, we first purified HF-SCs (CD34hia6-integrinhi) by fluorescence activated cell sorting (FACS) versus other basal cells (CD34nega6-integrinhi) (Blanpain et al., 2004). RT-PCR and immunoblot on these purified cell populations verified ACF7’s enrichment in HF-SCs at both mRNA and protein levels (Figure 4A). Immunofluorescence further documented elevated ACF7 in this niche throughout the hair cycle (Figure 4B). Loss of ACF7 in HF-SCs did not alter bulge architecture (data not shown) nor did it affect expression of key bulge markers (Blanpain and Fuchs, 2009) (Figure 4C). In addition, no significant changes were found in proliferation or apoptosis of homeostatic HF-SCs (more details below) or in hair growth or cycling (Wu 344 Cell 144, 341–352, February 4, 2011 ª2011 Elsevier Inc.
et al., 2008). We therefore focused on the hypothesis that the associated defects in wound repair originate from perturbations in HF-SC migration. To test this, we bred our ACF7fl/fl animals with mice expressing a progesterone-regulatable recombinase (K15-Cre-PGR) specifically in bulge SCs (Ito et al., 2005). To monitor HF-SC progeny in a wound response, we further bred these mice to Rosa26-lox-Stop-lox-LacZ reporter animals. As expected, treatment of adult mice with RU486 (a progesterone antagonist) activated Cre and selectively marked HF-SCs. When challenged to a wound, activated LacZ+ bulge cells exited the niche and migrated upward to re-epithelialize wounded epidermis (Figure 4D). This was readily observed by whole-mount staining, which displayed trails of HF-SC-derived blue (LacZ+) cells emanating from perilesional follicles of wounded tissue (dashed arrows). By contrast, ACF7 cKO bulge cells were delayed in this process by 40% compared with WT controls over 4–6 days after injury (Figures 4D and 4E). Importantly, since targeting was specific to bulge cells, the delay was directly attributable to an SC defect. Moreover, ACF7-deficiency did not affect proliferation or apoptosis of bulge SCs, indicating that the defect was rooted in cell migration (Figures S3A and S3B). To examine the contribution of GSK3b in this process, we took a pharmacological approach to manipulate GSK3b activity in vivo. Wounds on WT skin were treated with either lithium chloride (LiCl), which directly inhibits GSK3b, or wortmannin, which activates GSK3b by inhibiting an upstream regulator, phosphoinositol-3 kinase (PI3K). Interestingly, both treatments impaired wound-induced cell migration out of the adult HF-SC niche (Figure 4F), suggesting that spatiotemporal regulation of GSK3b’s activity is required to achieve efficient bulge SC migration in vivo.
Figure 3. GSK3b Phosphorylation of ACF7’s GSR Domain Inhibits Microtubule Binding (A) In vitro binding assays on purified ACF7 proteins. After incubating with 0–20 mg taxol-stabilized MTs, purified WT, or mutant versions of full-length, HA-tagged ACF7 were cosedimented and subjected to IB analyses. Note that phosphomimetic S:D but not phosphorylation-refractile S:A ACF7 shows significantly reduced binding affinity. (B) In vivo MT-binding assays. Lysates were prepared from cells expressing HA-tagged ACF7 or ACF7 mutants, and cosedimentation assays were performed as in (A). (C) Lysates were prepared from cells coexpressing caGSK3 with HA-tagged ACF7 or ACF7(S:A). Cosedimentation assays and IBs were as in (A). (D) Lysates were prepared from untreated cells or cells treated with GSK3b inhibitors (1: LiCl, 2: AR-A01448). Cosedimentation assays and IBs were as in (A). Figure 3 is associated with Figure S2.
The Polarization of MTs along Actin-Focal Adhesion Networks is Abrogated When ACF7 Is Phosphorylated by GSK3b Previously, we showed that when ACF7 is ablated, cultured epidermal keratinocytes cannot coordinate microtubule growth along F-actin filaments, a feature that in turn leads to overstabilization of focal adhesions (FAs) and defective cell movement (Wu et al., 2008). Based on our results thus far, we posited that in vivo, HF-SCs might respond to migratory stimulations such as Wnt signaling by spatiotemporally regulating ACF7-MT connections and promoting the cytoskeletal remodeling needed for polarized migration. To test this hypothesis, we first generated phosphospecific ACF7 Abs against two synthetic phospho-peptides corresponding to the ACF7 GSR sequences encompassing P1 and P2, respectively. Each Ab was specific for the phosphorylated state of its GSK3b target sequence: when ACF7 was not phosphorylated or when the sites were selectively mutated, the Abs failed to recognize the ACF7 protein (Figures S4A and S4B). Isolated HF-SCs can sustain long-term culture without losing stemness (Blanpain et al., 2004), allowing us to investigate the role of ACF7 phosphorylation in vitro. Both ACF7 Abs detected the expected sized band in immunoblots of cultured bulge SC but not ACF7 cKO lysates (Figure 5A). Importantly, these signals were sensitive not only to chemical inhibitors of GSK3b but also to recombinant Wnt3a. These data confirmed the specificity of our phospho-specific Abs and further demonstrated the ability of Wnt signaling to repress ACF7 phosphorylation in its C-terminal tail. We next examined the GSK3b phosphorylation status of endogenous ACF7. In vitro, ACF7 decorated the ends of MTs that are coaligned with underlying F-actin cables (Figure 5B).
By contrast, phospho-ACF7 was diffuse and/ or punctate throughout the cytoplasm and showed no association with these MT cables (Figure 5C). Nevertheless this cytoplasmic staining was specific for ACF7, as it was abolished in ACF7 KO bulge cells (Figure 5C). These findings were consistent with the phosphodependent decreased affinity of ACF7 for MTs that we observed in vitro and revealed a marked correlation between GSK3b phosphorylation of ACF7 and a severing of the ACF7-MT connection. To directly evaluate the effect of GSKb phosphorylation on this process, we overexpressed caGSK3b in WT cultured bulge SCs. In contrast to empty vector alone (Figure 5D, left), caGSK3b dramatically reduced ACF7 localization along MTs (Figure 5D, center). Moreover, and quite remarkably, expression of constitutively active GSK3b transformed the straight and radial MTs of WT cells into a network of bent and curly MTs, reminiscent of the aberrant MT network typifying ACF7 KO cells (Figure 5D, compare with data in Figure 5C). If GSK3 activation and phosphorylation of ACF7 is responsible for severing the polarization of MTs at the migrating front, then inhibiting endogenous GSK3b activity might be expected to stabilize these connections. Indeed, when we treated HF-SCs with LiCl, under conditions that potently inhibited GSK3b activity and ACF7 phosphorylation, ACF7 clustered at the ends of MTs (Figures 5A and 5E). Additionally, polarized sites of converging ACF7, MT, and F-actin usually colabeled with Abs against FA proteins (Figure S4C). ACF7-deficiency stabilizes FA through inhibiting MT targeting to FA (Wu et al., 2008). However, when DsRed-Zyxin-expressing HF-SCs were subjected to videomicroscopy and quantified, the effects of LiCl on FA turnover were modest (Figure S4D, Movie S1). This was also the case for the average size of FAs and the level of focal adhesion kinase (FAK) activity, which influence FA dynamics as well (Figures S4E and S4F). Overall, these results suggest that constitutive association between ACF7 and MTs may not elicit the alterations in FA stability that are seen when ACF7 is missing altogether. Cell 144, 341–352, February 4, 2011 ª2011 Elsevier Inc. 345
Figure 4. Loss of ACF7 Impairs Migration of Bulge Stem Cells In Vivo and In Vitro (A) RT-PCR (upper left), IB (upper right) and quantifications (lower) of ACF7 mRNA (real-time RT-PCR) and protein from isolated bulge HF-SCs and non-SC extracts. Error bars denote standard deviation (SD). (B) Immunofluorescence of skin sections at different hair cycle stages. ACF7 is enriched in bulge (arrows), and although CD34 is present in dermis, its epithelial expression is specific to bulge. Color coding is according to secondary Abs used in detection. Nuclei were counterstained with DAPI (blue). Dashed lines denote basement membrane. The following abbreviations are used: HF, hair follicle; der, dermis; epi, epidermis; DP, dermal papilla. The scale bars indicate 50 mm. (C) Immunofluorescence of sections of HF bulges (arrows) for SC markers indicated. (D and E) Whole-mount LacZ imaging and quantifications of bulge SCs and their progeny from perilesional follicles of WT and ACF7 cKO mice. Dashed lines denote wound boundary; dashed arrows denote trails of bulge SCs and progeny that migrated into wound. The inset is of unwounded K15-Cre-activated skin where LacZ expression is confined to the bulge, not visible on the body surface. Blue cells migrated upward only after wounding. The length of the blue cell trails 4 or 6 days after wounding (P4 and P6) was quantified in (E) by box-whisker plots. (F) Wounded animals were topically treated with LiCl or wortmannin (Wort). Bulge SC migration was quantified as in (E). Figure 4 is associated with Figures S1 and S3.
GSK3b has many targets, which complicates the interpretation of LiCl experiments (Sun et al., 2009). To distinguish the specific effects of GSK3b on ACF7-mediated cytoskeletal dynamics, we performed rescue experiments with our ACF7 phosphorylation mutants. To more precisely control concentration and ensure comparable expression of encoded proteins, we microinjected our expression constructs into primary cultured HF-SCs null for ACF7. Introducing GFP-tagged versions of either full-length or S:A mutant ACF7 restored ACF7’s localization to the MT ends residing near or at the cortex (Figure 5F). Consistent with their MT-binding capability, kinase-refractile ACF7(S:A) and fulllength ACF7 also generated arrays of radial MT trajectories in individual ACF7 null cells (Figure 5F). By contrast, the phosphomimetic ACF7(S:D) was diffusely localized, and its MT organization appeared no different than in uninjected or GFP-injected ACF7 null cells (Figure 5F). Taken together, these experiments provide compelling evidence that GSK3b plays a critical role in regulating ACF7’s dissociation from MTs and that phosphorylation is sufficient to dramatically alter polarized organization of MTs along actin stress fibers converging at FAs. 346 Cell 144, 341–352, February 4, 2011 ª2011 Elsevier Inc.
GSK3b Phosphorylation of ACF7 Must Be Dynamic to Govern Polarization and Directed Movement in Cultured HF-SCs Cultured HF-SCs further allowed us to examine their directional movement in vitro. When subjected to a modified Boyden chamber assay with conditioned feeder fibroblast medium as a chemo-attractant, WT HF-SCs showed a marked, dosedependent migratory response, which was greatly diminished in HF-SCs lacking ACF7 (Figure 6A). Moreover, stimulation of WT HF-SCs’ migration was achieved only when conditioned medium was administered in a positive concentration gradient. These data further documented the chemotactic nature of the response, and confirmed ACF7’s role in sustaining directional cell movement. Consistent with our in vivo observations, treatment of HF-SCs with GSK3b inhibitors or ectopic expression of caGSK3b inhibited the response (Figure 6B). Directionality of movement relies heavily on cell polarity, and prior studies indicated that embryonic endodermal cells lacking ACF7 cannot sustain cell polarity after scratch-wounding in vitro (Kodama et al., 2003). To evaluate how regulated ACF7-MT connections might contribute to cell polarity, we devised
Figure 5. ACF7 Is Required to Polarize MTs along F-Actin-Focal Adhesion Networks, which in Turn Are Inhibited by GSK3b Activity (A) ACF7 protein and its phosphorylation status were evaluated by immunoblot of lysates from KO or WT bulge SCs treated with GSK3b inhibitors or Wnt3a, as indicated. (B and C) Cultured bulge SCs immunolocalized for a-tubulin (MT), ACF7, ACF7 phosphorylation (ACF7-Pi, with P2-specific Abs), or F-actin (phalloidin). Notes: (1) ACF7 but not phospho-ACF7 is enriched at polarized MT tips and (2) the diffuse cytoplasmic ACF7-Pi pattern of WT is absent in KO cells. (D) Immunofluorescence of WT bulge SCs expressing caGSK3b or vector control. Transfected cells are marked with arrows. Note: caGSK3b inhibits ACF7 localization at MT tips and leads to disorganized MT networks. (E) WT bulge cells were treated with LiCl to impair GSK3b activity and subjected to immunofluorescence as indicated. Note clusters of ACF7 at cell periphery. (F) Cultured ACF7 KO bulge SCs were microinjected with expression vectors encoding GFP alone, GFP-ACF7 (rescue), GFP-ACF7(S:A) (GSK3-refractile), or GFP-ACF7 (S:D) (phosphomimetic). Cells were immunolabeled for MTs. Note that only WT and S:A mutant, and not GFP or S:D mutant, localized and rescued MT organization defects that occur in KO bulge SCs in low-Ca2+ medium. Wherever the field includes uninjected cells, injected ones are denoted by an asterisk. For (B–F), boxed areas are magnified in insets. The scale bars represent 20 mm. Figure 5 is associated with Figure S4.
a method to polarize cultured bulge SCs by seeding them at low density on fibronectin-coated dishes and then elevating Ca2+ levels to induce cell-cell adhesion. Under these conditions, the perimeter of isolated WT colonies exhibited significant polarity as determined by immunolocalization of phosphorylated (inactive) GSK3b, Par proteins, and aPKC (Figure 6C, and Figure S4G). Polarization of perinuclear Golgi was particularly prominent, enabling quantification by measuring its preferential localization around the axis bisecting nucleus and colony edge. Interestingly, ACF7-deficient bulge SCs displayed polarized GSK3b phosphorylation but not Golgi (Figure 6C). Moreover, manipulating GSK3b activity in WT HF-SCs disrupted Golgi polarization (Figure 6D). These data placed ACF7 midstream in the pathway
that links polarized GSK3b inhibition at the HFSC front and Golgi polarization in the perinuclear region. Our findings were intriguing in light of prior data showing that MTs are essential for polarizing Golgi assembly (Miller et al., 2009; Siegrist and Doe, 2007). We therefore wondered whether rescuing the ability of MTs to polarize along actin cables might also rescue defective Golgi polarization. To test this possibility, we repeated the polarization assays, this time with bulge SCs microinjected with our GFP-tagged versions of WT and phosphorylation-altered ACF7. WT-ACF7 rescued polarization of HFSCs (Figure 6D). As expected from its failure to bind MTs (Figure 5F), ACF7(S:D) also failed to effectively polarize Golgi (Figure 6D). Interestingly, however, even though ACF7(S:A) efficiently rescued MT organization (Figure 5F), it failed to rescue Golgi orientation (Figure 6D). Finally, we tested the ability of our mutants to rescue the chemotactic migration defects seen in ACF7 null cells. As shown in Figure 6E, only WT-ACF7 rescued the ability of bulge SCs to migrate efficiently. Together, these data showed that the ability of MTs to track along actin cables is not sufficient to achieve either polarization or effective migration of bulge SCs. Moreover, both of these processes require in addition the dynamic regulation of ACF7 phosphorylation since neither the phosphomimetic mutant, the phosphorylation refractile mutant, nor the two Cell 144, 341–352, February 4, 2011 ª2011 Elsevier Inc. 347
Figure 6. GSK3b Phosphorylation Governs ACF7 Functionality in Cell Polarity and Directional Cell Movement in Vitro (A) Isolated HF-SCs were subjected to modified Boyden chamber assays, and a checkerboard analysis was performed to distinguish chemokinesis versus chemotaxis effects. Feeder-conditioned medium was used as chemoattractant and was mixed with serum-free medium to load into upper and lower chambers. Migration was quantified as in Experimental Procedures, and presented as bar graphs. Color coding corresponds to the concentration of chemoattractants in the lower chamber. Concentrations of chemoattractants in upper chamber are indicated in bar graph legends. Note that only WT cells showed significant chemotactic behavior when exposed to a positive chemoattractant gradient. (B) Migration of treated or transfected WT HF-SCs was examined as in (A), with only 0 or 100% of chemoattractants used in lower chamber and no chemoattractants in upper chamber. Note that both global inhibition and artificial activation of GSK3b activity inhibit migration of HF-SCs. (C) Immunofluorescence detects Ser-9 GSK3b phosphorylation (inactivation), GM-130 (Golgi), E-cadherin (E-Cad), and chromatin (DAPI). Arrows denote enriched phospho-GSK3b at the leading edge. Dashed lines denote colony perimeters. q represents angle between solid arrow, denoting outward direction of HF-SC colony and dashed arrow, denoting Golgi orientation. Note polarization of inactivated GSK3b at migrating edge of both WT and ACF7 null HF-SCs, and loss of Golgi polarization in ACF7 null HF-SCs. The scale bar represents 20 mm. (D) Windrose plots of Golgi orientation (quantification of angle q) in treated or untreated WT and microinjected or uninjected ACF7-deficient cells. Note random distribution of q in KO cells; only WT ACF7 restores proper cell polarity. (E) Quantifications of chemotactic behaviors show that only WT ACF7 rescues migration defects in ACF7 null bulge SCs. Error bars represent SD.
mutants combined (Figure 6E) were able to rescue these defects in KO cells. GSK3b Phosphorylation of ACF7 Regulates Bulge SC Migration and Skin Wound Repair In Vivo We next focused on whether dynamic regulation of ACF7 phosphorylation coordinates polarized bulge SC migration during wound repair in vivo. We began by engineering transgenic mice expressing N-terminally GFP-tagged, full-length versions of ACF7 and ACF7 S:A under the control of K14 promoter and enhancer. Mice genotypic for K14-ACF7 or K14-ACF7(S:A) alleles were born in the expected Mendelian numbers and grew normally (Figure S5A). Transgenic ACF7 and ACF7(S:A) GFP-tagged proteins of the correct size were expressed comparably and exhibited the expected differential GSK3b phosphorylation states in vivo (Figure S5A). Immunofluorescence confirmed skin-specific transgene expression (Figure S5B). To determine the ability of these transgenic proteins to compensate for loss of ACF7 in HF-SCs in vivo, we bred our 348 Cell 144, 341–352, February 4, 2011 ª2011 Elsevier Inc.
transgenics to ACF7fl/fl:K15-Cre-PGR:Rosa26-LacZ mice and then induced ablation of endogenous ACF7. Mice expressing these transgenes and not ACF7 in their HF-SCs were visibly normal, and no gross differences were noted in hair cycles (Figure S5C; data not shown). Immunofluorescence for a variety of differentiation markers showed normal morphology and localization patterns (Figures S5D and S5E). Given the normal tissue architecture and homeostasis, we next turned to investigating how ACF7’s GSK3b-phophorylation status affects the ability of HF-SCs to respond and migrate outward to repair epidermis upon skin wounding. In response to punch wounds, only ACF7 cKO mice expressing GFP-ACF7, and not GFP-ACF7(S:A), showed significant rescue of bulge SC migration defects, as measured by LacZ whole-mount staining (Figure 7A, left). This difference appeared to reflect a differential ability to restore directional cell movement, since bulge SC proliferation and apoptosis assays showed no change (Figures S3A and S3B). Although most data shown are for K15-Cre-PGR, similar results were obtained when the wounding challenge was
Figure 7. GSK3b Phosphorylation of ACF7 Plays a Critical Role in Bulge Stem Cell Migration and Skin Wound Repair In Vivo (A) Box and whisker plots (left) and bar graphs (right) show that in vivo migration of HF-SCs and overall skin healing (length of hyperproliferative epithelium in wound) were restored by WT-ACF7 transgene but not ACF7(S:A). Stars (+) represent p value less than 0.05, and error bars represent SE. (B) Outline of methodology used to activate and induce migration of quiescent HF-SCs in skin biopsy (1) by ex vivo wounding. A typical telogen HF is shown in (2) with arrows marking potential direction of HF-SC migration upon wounding. HFSCs are identified by C12-FDG (a fluorogenic substrate of LacZ), and outward migration is visualized by phase contrast (PhC) or fluorescence microscopy. Dashed lines denote basement membrane in (2), or migrating front in (3–4). Stars denote C12FDG(+) cell in (4). (C) Representative examples of SC progeny after exiting and migrating out of the HF bulge. Immunofluorescence was used to visualize ACF7, MT, and F-actin. Boxed areas are magnified as insets. (D–E) Cells migrating out of the bulge SC niche were immunostained for phospho-GSK3b and phospho-ACF7. Note enrichment of inactive (phosphor) GSK3b at migrating front and corresponding lack of GSK3b-phosphorylation of ACF7 at plus ends of MTs localized there. (F) Live imaging of progeny that migrated out of the bulge SC niche during wound response. Cells were microinjected to express GFP-EB1, and MT +tip movements were then monitored by videomicroscopy and tracked automatically (upper). Corresponding MT behavior is colorcoded and described in text and Experimental Procedures. Data for directionalities of MT growth were collected from movies and then quantified and presented as Windrose plots (lower). Arrow denotes direction toward wound front. Dotted lines denote migrating front of cells streaming from the wound-induced skin explant. The scale bar represents 20 mm. Figure 7 is associated with Figures S5 and S6.
broadened by using transgenic mice mated to ACF7fl/fl: K14-Cre mice. Once again, the area of hyperproliferative epithelium that typically migrates into the wound site was only significantly rescued by GFP-ACF7 and not GFP-ACF7(S:A) (Figure 7A, right). Finally we addressed whether changes in GSK3b activity alter ACF7’s ability to coordinate actin- MT dynamics during woundinduced directed migration of bulge SCs out of HFs. For this purpose, we induced a rapid wound response in HFs by cutting the skin at the resting (telogen) phase and placing it into rich medium. In a process analogous to wound healing, the outward migration of activated HF-SC progeny could then be imaged by videomicroscopy and immunofluorescence (Figures 7B–7F and Figure S6). Targeted bulge cell progeny were identified by their ability to cleave a fluorogenic substrate for LacZ (Figure 7B). Consistent with our in vitro results, Golgi complex as well as Par6 and aPKC localized in a polarized fashion in WT bulge SCs at the
leading front (Figure S6A). These cells also contained FAs (Figure S6A) and well-polarized MT bundles that coaligned with F-actin, and ACF7 localized at the interface between these MT +tips and F-actin bundles (Figure 7C). Loss of ACF7 resulted in disorganized MT architecture (Figure 7C). Additionally, ACF7 deficiency or change in GSK3b activity led to perturbations in cell polarity and migration of the marked bulge SCs/progeny that were streaming from the HFs. Interestingly, while both WT and S:A mutant ACF7 rescued the disorganized MT network, only WT ACF7 rescued alterations in cell polarity and migration (Figure 7C and Figures S6B and S6C). Similar to our in vitro results (see Figure 6C) and those obtained from other model systems (Sun et al., 2009), Ser9-phosphorylated (inactive) GSK3b was enriched at the leading edge of migrating HF-SCs (Figure 7D). Consistent with these data, the majority of ACF7 phosphorylated by GSK3b was localized in the cell body and not at the leading edge of HF-SCs (Figure 7E). Cell 144, 341–352, February 4, 2011 ª2011 Elsevier Inc. 349
As a +tip protein, ACF7 can guide movement of MT plus ends. To monitor potential polarity of MT dynamics during woundinduced migration, we microinjected migrating bulge SCs/ progeny with a GFP-tagged EB1 expression vector. Videomicroscopy combined with automated tracking of this plus end (+tip) MT-binding protein revealed polarized MT growth toward the migrating front of WT HF-SCs (Movies 2 and Figure 7F). Directionality of MT growth was largely randomized in KO cells, and only WT-ACF7, and not ACF7(S:A), rescued it (Figure 7F). Taken together, these findings underscore an essential physiological role for GSK3b-mediated phosphorylation of ACF7 in sustaining polarized MT growth and directed HF-SC migration during wound healing. DISCUSSION Aberrant mobilization of SCs in response to injury can delay wound repair and have dire consequences to animal survival (Fuchs, 2009). Exposed to frequent mechanical stresses, skin SCs have developed a unique and elaborate cytoskeletal system. However little is known about how cytoskeletal dynamics are coordinated in these or other SCs. In this report, we’ve demonstrated the role of ACF7 and ACF7-mediated cytoskeletal dynamics in SCs in vivo. Our studies show that ACF7 is required for efficient upward migration of bulge cells in response to wounding and that this function is primarily rooted in ACF7’s ability to coordinate MT dynamics and polarize HF-SCs. Through a comprehensive approach encompassing biochemistry and molecular and cell biology, we further unveiled a hitherto unrecognized regulatory role for GSK3b-mediated phosphorylation of ACF7’s major MT-binding domain. We show that the consequence of this phosphorylation is the attenuation of interactions between ACF7’s basic GSR domain and the acidic C-terminal tubulin tails. Polarized cell movement is an essential component of diverse biological processes such as cancer metastasis, tissue development, and wound healing (Lauffenburger and Horwitz, 1996). All these processes share significant similarities regarding their mechanisms for sensing directional cues and translating them into directional locomotion. Cells under the influence of various chemotactic molecules exhibit polarized activation of signaling proteins, such as the CDC42 Rho GTPase and PI3K, that guide differential remodeling of cytoskeletons at the leading edge versus the back of cells (Etienne-Manneville, 2004; Siegrist and Doe, 2007). MTs play a particularly important role in this process, delivering positional information to establish the proper site of cortical polarity (Siegrist and Doe, 2007). Once MTs and their associated proteins determine the polarity site, a positive feedback loop initiates interactions between the actin-rich cortex and growing (plus) ends of MTs, resulting in reinforcement and maintenance of polarity. Such a mechanism not only provides cells with the ability to sense and amplify small asymmetries in their field but also buffers and maintains the polarity axis after it is established. Common components of MT-based polarity pathways are plusend-directed kinesin motors and MT plus-end-stabilizing proteins, including CLIP-170, EB1, Clasps, and adenomatous polyposis coli (APC) (Akhmanova and Steinmetz, 2008). Our prior 350 Cell 144, 341–352, February 4, 2011 ª2011 Elsevier Inc.
studies added ACF7 to this list (Karakesisoglou et al., 2000; Kodama et al., 2003; Wu et al., 2008). Our current study lends physiological relevance for ACF7’s ability to polarize MTs and identifies a specific role for this connection in enabling HF-SCs to polarize and migrate into a wound site. In addition, our results elucidate a hitherto unappreciated signaling pathway whereby ACF7’s function is regulated dynamically in HF-SCs by GSK3b in order to control the directionality of MT growth, cell polarity, and migration. GSK3b modifies various regulatory components of MT cytoskeletons, including APC and some neuronal MT-associated proteins (Sun et al., 2009). Intriguingly, a recent study showed that ErbB2-induced repression of GSK3 is required for MT capture and targeting of ACF7 to plasma membrane in breast carcinoma cells (Zaoui et al., 2010). Our results identify ACF7 as a GSK3b substrate in skin somatic SCs and raise the possibility that GSK3b may function as a master regulator of MT polarity. Importantly, however, our findings reveal that although only phosphomimetic ACF7 mutants disrupt MT organization, nonphosphorylatable ACF7 mutants nonetheless impair global cell polarization and SC migration. These findings underscore the importance of regulating ACF7’s phosphorylation status in a spatially and temporally defined manner in order to establish proper directionality in cells. Consistent with this notion, GSK3b activity is specifically inhibited by CDC42 signaling at the leading edge of migrating cells (Etienne-Manneville and Hall, 2003; Sun et al., 2009) and also by Wnt signaling, which is often polarized in tissues (Nusse, 2008). Notably, Wnts have been broadly implicated in wound repair and are required for HF-SC activation at the start of the hair cycle (Blanpain and Fuchs, 2009). Although ACF7 has been implicated in recruiting the Axin-APC-GSK3b complex to the site of active Wnt signaling in the gastrulating embryo (Chen et al., 2006), an essential role in Wnt signaling would not explain why SC activation and hair cycling still occur in ACF7 cKO skin. Our findings now provide an alternative role for ACF7 in Wnt signaling, namely as a downstream sensor of GSK3b inhibition and a mobilizer of the polarized response necessary for SC migration. Collectively, our findings support a model in which upstream chemotactic cues, e.g., Wnt signaling, trigger cellular polarization, and directional movement through spatiotemporal regulation of ACF7 phosphorylation by GSK3b (Figure S7). In this model, some outcomes of Wnt signaling and GSK3b inhibition, e.g., SC activation, proliferation, and fate commitment in normal homeostasis, might still be maintained while others, e.g., SC migration during wound repair, would rely upon the ability to polarize directed migration. Future studies will determine the extent to which this pathway can explain ACF7’s broad and essential presence in tissues. In closing, our findings provide insights into the complex but important molecular machinery underlying the polarized SC migration that integrates injury-induced migratory signals, cytoskeletal remodeling, and polarized cell movements in HFSCs. Our studies also add to the repertoire of spectraplakin’s many diverse and critical functions and now pave the way for probing more deeply into the role of spectraplakins in mammalian SCs.
EXPERIMENTAL PROCEDURES Generation of Transgenic Mice and Skin Wound Healing The transgenic expression cassette was constructed so that GFP-ACF7 or GFP-ACF7(S:A) cDNA sequence was inserted 30 to the human K14 promoter and enhancer and b-globin 50 UTR, and 50 to the K14 30 UTR. Mice harboring these transgenes were engineered in Fvb/n albino mice and selected for their comparable expression relative to endogenous ACF7. Skin-wound-healing assays were performed as described (Wu et al., 2008). To monitor the migration of SCs and their progeny after wounding, we treated K15-Cre-PGR:R26-LacZ:ACF7fl/fl mice ± transgene with RU486 for 5 days starting at their first telogen (P21). At 8 weeks, mice were anesthetized and full-thickness punch wounds (6 mm) were introduced on their backs. Skin samples were then collected at 4 or 6 days after wounding and b-galactosidase activity was detected in whole-mount tissues by X-gal staining (Ito et al., 2005). For each genotype, at least ten mice were analyzed for the wound response. To monitor the activation and migration of SCs from resting HFs subjected to wound response, we took 3 mm biopsies from back skin of 2-month-old K15Cre-PGR:R26-LacZ mice that were treated with RU486. Biopsies were coated with a thin layer of matrigel for adhesion and immediately transferred to a coverslip coated with 10 mg/ml fibronectin. Skins were then incubated at room temperature for 10 min to solidify the matrigel and then exposed to rich E-media containing 15% serum and 0.3 mM calcium (Blanpain et al., 2004), which promoted HF-SC outgrowth within 2–3 days, analogous to a wound response. Live HF-SCs and progeny were identified with fluorogenic b-galactosidase substrates C12FDG or C12RG (Invitrogen, yield green or red fluorescence respectively) according to the manufacturer’s instruction. Cell Migration Assays and Time-Lapse Videomicroscopy Migration assays in 96-well chemotaxis chambers (Millipore, Billerica, MA) were carried out according to manufacturer’s instructions. Briefly, 3T3 fibroblast-conditioned medium or serum-free medium (control) was added to the lower chamber. Bulge SCs were added to the upper chamber, and, after 10 hr at 37 C, cells that migrated through the filter into the bottom chamber were collected, lysed, and stained with CyQuant GR dye (labels DNA). A fluorescence plate reader was used to quantify fluorescence intensities. To track individual HF-SC movement, skin biopsies were imaged with an Olympus phase contrast microscope for 1 day and manually tracked with National Institutes of Health’s ImageJ. Displacements along the direction facing the leading edge of the migrating front were recorded and quantified (Wu et al., 2008). To monitor MT plus end movement in live cells, we microinjected HF-SCs with plasmid encoding GFP-EB1. Six hours after injection, cells were imaged with a confocal spinning-disk microscope (Wu et al., 2008) for 5 min at 2 s/frame. Plus Tip Tracker software package (Matov et al., 2010) was used to process and track EB1 movements at the leading front. Because of MT dynamic instability, some gaps appeared between MT growth (red solid tracks in the output image of Figure 7F). Gaps could occur in either the forward or backward direction depending on underlying MT dynamics and detection performance. Forward gaps could be a MT pause (cyan dotted) or reclassified as growth (green solid). Backward gaps could be MT shrinkage/catastrophe (yellow dotted) or reclassified as pause events (blue dotted). The same color codes were used in the output movie (Movie S2), and initiation of a new growth track or gap was marked by a circle with corresponding color. Angles between MT growth track and outward direction of explants were calculated with a linear fit function of Matlab and plotted as Windrose plots. Microtubule Cosedimentation Assay, Subtilisin Treatment, and Immunoblotting MT-binding assays were performed as described (Wu et al., 2008). Equivalent amounts of pellet were analyzed by CB staining or immunoblotting. Cleavage of the unstructured tubulin C terminus was carried out by limited proteolysis of taxol-stabilized MTs with subtilisin (Knipling et al., 1999). The proteolysis reaction was stopped by adding freshly prepared 20 mM PMSF in DMSO. Subtilisin-treated MTs were pelleted by ultracentrifugation at 60,000 3 g and resuspended in MT-stabilizing buffer containing 20 mM PMSF and taxol to ensure complete removal of active protease and cleaved C-terminal tubulin
fragments. Immunoprecipitations and immunoblotting were performed as described (Wu et al., 2004).
SUPPLEMENTAL INFORMATION Supplemental Information includes Supplemental Experimental Procedures, seven figures, and two movies and can be found with this article online at doi:10.1016/j.cell.2010.12.033. ACKNOWLEDGMENTS We are grateful to M. Schober, Y.C. Hsu, E. Ezratty, and T. Chen for discussions and helpful comments and to J. Fernandez and H. Deng (Proteomics Center) for their technical assistance in mass spectrometry analysis. Valuable technical assistance was provided by N. Stokes, L. Polak, E. Wong, M Nikolova, J. Racelis, A. North, and S. Bhuvanendran. All mice used in this study were bred and maintained at the Rockefeller University Association for Assessment and Accreditation of Laboratory Animal Care Internationalaccredited Comparative Biology Center (CBC) in accordance with institutional and National Institutes of Health guidelines. This work was supported by Grant R01-AR27883 from the National Institutes of Health. E.F. is an investigator of the Howard Hughes Medical Institute. X.W. was an American Association for Cancer Research Anna D. Barker Fellow in Basic Cancer Research and the recipient of a postdoctoral fellowship from the Jane Coffin Childs Memorial Fund for Medical Research. Received: January 8, 2010 Revised: November 1, 2010 Accepted: December 17, 2010 Published: February 3, 2011 REFERENCES Akhmanova, A., and Steinmetz, M.O. (2008). Tracking the ends: a dynamic protein network controls the fate of microtubule tips. Nat. Rev. Mol. Cell Biol. 9, 309–322. Blanpain, C., and Fuchs, E. (2009). Epidermal homeostasis: a balancing act of stem cells in the skin. Nat. Rev. Mol. Cell Biol. 10, 207–217. Blanpain, C., Lowry, W.E., Geoghegan, A., Polak, L., and Fuchs, E. (2004). Self-renewal, multipotency, and the existence of two cell populations within an epithelial stem cell niche. Cell 118, 635–648. Chen, H.J., Lin, C.M., Lin, C.S., Perez-Olle, R., Leung, C.L., and Liem, R.K. (2006). The role of microtubule actin cross-linking factor 1 (MACF1) in the Wnt signaling pathway. Genes Dev. 20, 1933–1945. Etienne-Manneville, S. (2004). Cdc42—the centre of polarity. J. Cell Sci. 117, 1291–1300. Etienne-Manneville, S., and Hall, A. (2003). Cdc42 regulates GSK-3beta and adenomatous polyposis coli to control cell polarity. Nature 421, 753–756. Fathke, C., Wilson, L., Shah, K., Kim, B., Hocking, A., Moon, R., and Isik, F. (2006). Wnt signaling induces epithelial differentiation during cutaneous wound healing. BMC Cell Biol. 7, 4. Fuchs, E. (2009). The tortoise and the hair: slow-cycling cells in the stem cell race. Cell 137, 811–819. Goryunov, D., He, C.Z., Lin, C.S., Leung, C.L., and Liem, R.K. (2010). Nervous-tissue-specific elimination of microtubule-actin crosslinking factor 1a results in multiple developmental defects in the mouse brain. Mol. Cell. Neurosci. 44, 1–14. Greco, V., Chen, T., Rendl, M., Schober, M., Pasolli, H.A., Stokes, N., Dela Cruz-Racelis, J., and Fuchs, E. (2009). A two-step mechanism for stem cell activation during hair regeneration. Cell Stem Cell 4, 155–169. Ito, M., Liu, Y., Yang, Z., Nguyen, J., Liang, F., Morris, R.J., and Cotsarelis, G. (2005). Stem cells in the hair follicle bulge contribute to wound repair but not to homeostasis of the epidermis. Nat. Med. 11, 1351–1354.
Cell 144, 341–352, February 4, 2011 ª2011 Elsevier Inc. 351
Ito, M., Yang, Z., Andl, T., Cui, C., Kim, N., Millar, S.E., and Cotsarelis, G. (2007). Wnt-dependent de novo hair follicle regeneration in adult mouse skin after wounding. Nature 447, 316–320.
Rodriguez, O.C., Schaefer, A.W., Mandato, C.A., Forscher, P., Bement, W.M., and Waterman-Storer, C.M. (2003). Conserved microtubule-actin interactions in cell movement and morphogenesis. Nat. Cell Biol. 5, 599–609.
Jefferson, J.J., Leung, C.L., and Liem, R.K. (2004). Plakins: goliaths that link cell junctions and the cytoskeleton. Nat. Rev. Mol. Cell Biol. 5, 542–553.
Ro¨per, K., Gregory, S.L., and Brown, N.H. (2002). The ‘spectraplakins’: cytoskeletal giants with characteristics of both spectrin and plakin families. J. Cell Sci. 115, 4215–4225.
Karakesisoglou, I., Yang, Y., and Fuchs, E. (2000). An epidermal plakin that integrates actin and microtubule networks at cellular junctions. J. Cell Biol. 149, 195–208. Knipling, L., Hwang, J., and Wolff, J. (1999). Preparation and properties of pure tubulin S. Cell Motil. Cytoskeleton 43, 63–71. Kodama, A., Karakesisoglou, I., Wong, E., Vaezi, A., and Fuchs, E. (2003). ACF7: an essential integrator of microtubule dynamics. Cell 115, 343–354. Lauffenburger, D.A., and Horwitz, A.F. (1996). Cell migration: a physically integrated molecular process. Cell 84, 359–369. Matov, A., Applegate, K., Kumar, P., Thoma, C., Krek, W., Danuser, G., and Wittmann, T. (2010). Analysis of microtubule dynamic instability using a plusend growth marker. Nat. Methods 7, 761–768. Miller, P.M., Folkmann, A.W., Maia, A.R., Efimova, N., Efimov, A., and Kaverina, I. (2009). Golgi-derived CLASP-dependent microtubules control Golgi organization and polarized trafficking in motile cells. Nat. Cell Biol. 11, 1069–1080. Morris, R.J., Liu, Y., Marles, L., Yang, Z., Trempus, C., Li, S., Lin, J.S., Sawicki, J.A., and Cotsarelis, G. (2004). Capturing and profiling adult hair follicle stem cells. Nat. Biotechnol. 22, 411–417. Nusse, R. (2008). Wnt signaling and stem cell control. Cell Res. 18, 523–527.
Schiller, M.R. (2007). Minimotif miner: A computational tool to investigate protein function, disease, and genetic diversity. Curr Protoc Protein Sci, 2 10.1002/0471140864.ps0212s48, Unit 2.12. Siegrist, S.E., and Doe, C.Q. (2007). Microtubule-induced cortical cell polarity. Genes Dev. 21, 483–496. Stoick-Cooper, C.L., Weidinger, G., Riehle, K.J., Hubbert, C., Major, M.B., Fausto, N., and Moon, R.T. (2007). Distinct Wnt signaling pathways have opposing roles in appendage regeneration. Development 134, 479–489. Sun, D., Leung, C.L., and Liem, R.K. (2001). Characterization of the microtubule binding domain of microtubule actin crosslinking factor (MACF): identification of a novel group of microtubule associated proteins. J. Cell Sci. 114, 161–172. Sun, T., Rodriguez, M., and Kim, L. (2009). Glycogen synthase kinase 3 in the world of cell migration. Dev. Growth Differ. 51, 735–742. Tumbar, T., Guasch, G., Greco, V., Blanpain, C., Lowry, W.E., Rendl, M., and Fuchs, E. (2004). Defining the epithelial stem cell niche in skin. Science 303, 359–363. Wu, X., Suetsugu, S., Cooper, L.A., Takenawa, T., and Guan, J.L. (2004). Focal adhesion kinase regulation of N-WASP subcellular localization and function. J. Biol. Chem. 279, 9565–9576. Wu, X., Kodama, A., and Fuchs, E. (2008). ACF7 regulates cytoskeletal-focal adhesion dynamics and migration and has ATPase activity. Cell 135, 137–148.
Nusse, R., Fuerer, C., Ching, W., Harnish, K., Logan, C., Zeng, A., ten Berge, D., and Kalani, Y. (2008). Wnt signaling and stem cell control. Cold Spring Harb. Symp. Quant. Biol. 73, 59–66.
Yoshimura, T., Arimura, N., and Kaibuchi, K. (2006). Signaling networks in neuronal polarization. J. Neurosci. 26, 10626–10630.
Obenauer, J.C., Cantley, L.C., and Yaffe, M.B. (2003). Scansite 2.0: Proteomewide prediction of cell signaling interactions using short sequence motifs. Nucleic Acids Res. 31, 3635–3641.
Zaoui, K., Benseddik, K., Daou, P., Salau¨n, D., and Badache, A. (2010). ErbB2 receptor controls microtubule capture by recruiting ACF7 to the plasma membrane of migrating cells. Proc. Natl. Acad. Sci. USA 107, 18517–18522.
352 Cell 144, 341–352, February 4, 2011 ª2011 Elsevier Inc.
The RNA Exosome Targets the AID Cytidine Deaminase to Both Strands of Transcribed Duplex DNA Substrates Uttiya Basu,1,2,7,* Fei-Long Meng,1,7 Celia Keim,2,7 Veronika Grinstein,2 Evangelos Pefanis,2 Jennifer Eccleston,1 Tingting Zhang,1 Darienne Myers,1 Caitlyn R. Wasserman,1 Duane R. Wesemann,1 Kurt Januszyk,5 Richard I. Gregory,4 Haiteng Deng,3,6 Christopher D. Lima,5 and Frederick W. Alt1,* 1Howard Hughes Medical Institute, Program in Cellular and Molecular Medicine and Immune Disease Institute, Children’s Hospital Boston, Department of Genetics, Harvard Medical School, 300 Longwood Avenue, Boston, MA 02115, USA 2Department of Microbiology and Immunology, College of Physicians and Surgeons, Columbia University, New York, NY 10032, USA 3Rockefeller University, Proteomic Research Center, New York, NY 10065, USA 4Children’s Hospital Boston, Harvard Stem Cell Institute, Department of Biological Chemistry and Molecular Pharmacology, Harvard Medical School, 300 Longwood Avenue, Boston, MA 02115, USA 5Memorial Sloan Kettering Cancer Center, New York, NY 10065, USA 6School of Life Sciences, Tsinghua University, Beijing 100084, China 7These authors contributed equally to this work *Correspondence:
[email protected] (U.B.),
[email protected] (F.W.A.) DOI 10.1016/j.cell.2011.01.001
SUMMARY
Activation-induced cytidine deaminase (AID) initiates immunoglobulin (Ig) heavy-chain (IgH) class switch recombination (CSR) and Ig variable region somatic hypermutation (SHM) in B lymphocytes by deaminating cytidines on template and nontemplate strands of transcribed DNA substrates. However, the mechanism of AID access to the template DNA strand, particularly when hybridized to a nascent RNA transcript, has been an enigma. We now implicate the RNA exosome, a cellular RNA-processing/ degradation complex, in targeting AID to both DNA strands. In B lineage cells activated for CSR, the RNA exosome associates with AID, accumulates on IgH switch regions in an AID-dependent fashion, and is required for optimal CSR. Moreover, both the cellular RNA exosome complex and a recombinant RNA exosome core complex impart robust AID- and transcription-dependent DNA deamination of both strands of transcribed SHM substrates in vitro. Our findings reveal a role for noncoding RNA surveillance machinery in generating antibody diversity. INTRODUCTION Antigen-activated B lymphocytes undergo two distinct immunoglobulin (Ig) gene diversification processes, namely somatic hypermutation (SHM) and Ig heavy-chain (IgH) class switch recombination (CSR). SHM diversifies IgH and Ig light-chain (IgL) variable region exons to allow generation of B cells with
the potential to secrete higher-affinity antibodies (reviewed by Odegard and Schatz, 2006; Di Noia and Neuberger, 2007; Maul and Gearhart, 2010). IgH class switch recombination allows B cells to express different classes of antibodies with different IgH constant regions (CHs) and, as a result, different antibody effector functions (reviewed by Chaudhuri et al., 2007; Honjo et al., 2002). CSR involves joining DNA double-strand breaks (DSBs) in the large repetitive switch (S) region (Sm) that lies upstream of the Cm constant region exons to DSBs within a downstream S region (e.g., Sg1), which replaces Cm exons with a set of downstream CH exons to complete CSR (e.g., switching from IgM to IgG1). Activation-induced cytidine deaminase (AID) initiates both SHM and CSR (Muramatsu et al., 2000; Revy et al., 2000) by deaminating cytidines on, respectively, transcribed IgH or IgL variable region exons or transcribed IgH S regions (Petersen-Mahrt et al., 2002). The deaminated cytidines become targets of co-opted DNA repair pathways that lead to mutations associated with variable region exon SHM or to S region DSBs that initiate CSR (reviewed by Di Noia and Neuberger, 2007; Neuberger et al., 2003). To initiate both SHM and CSR, AID equally deaminates both template and nontemplate strands of transcribed target DNA sequences (Milstein et al., 1998; Shen et al., 2006; Xue et al., 2006). AID is a single-stranded (ss) DNA-specific cytidine deaminase that lacks activity on double-stranded (ds) DNA (Chaudhuri et al., 2003; Dickerson et al., 2003; Ramiro et al., 2003; Sohail et al., 2003). Correspondingly, SHM and CSR require transcription through duplex substrate V(D)J exons or S regions to target AID activity, consistent with transcription generating a ssDNA AID substrate (reviewed by Chaudhuri et al., 2007; Yang and Schatz, 2007; Di Noia and Neuberger, 2007; Maul and Gearhart, 2010). Transcription through mammalian S regions generates R loops in which the template strand is hybridized to the nascent transcript and the nontemplate strand is looped out as ssDNA Cell 144, 353–363, February 4, 2011 ª2011 Elsevier Inc. 353
(Daniels and Lieber, 1995; Shinkura et al., 2003; Tian and Alt, 2000; Yu et al., 2003a). In biochemical studies, transcriptiongenerated R loops within duplex substrates allow AID deamination but mainly on the looped-out nontemplate strand (Chaudhuri et al., 2003). AID that is phosphorylated on serine 38 by protein kinase A can access in vitro transcribed dsDNA SHM substrates (e.g., V(D)J exons) that lack R loop-forming ability—again, however, mainly on the nontemplate strand (Basu et al., 2005, 2008; Chaudhuri et al., 2004; Zarrin et al., 2004; Vuong et al., 2009). Ectopically expressed AID mutates transcribed substrates in bacteria and yeast but, once again, predominantly on the nontemplate strand (Go´mez-Gonza´lez and Aguilera, 2007; Ramiro et al., 2003). Thus, the mechanism by which AID accesses the template DNA strand of substrates has been a major question (e.g., Chaudhuri et al., 2007; Chelico et al., 2009; Di Noia and Neuberger, 2007; Liu and Schatz, 2009; Longerich et al., 2006; Maul and Gearhart, 2010; Pavri et al., 2010; Peled et al., 2008). Biochemical studies suggested that AID may gain access to both DNA strands of certain types of transcribed plasmids via formation of RNA polymerase-generated supercoils (Besmer et al., 2006; Shen et al., 2005; Shen and Storb, 2004). However, such a mechanism does not readily explain how AID accesses the template strand of a transcribed substrate if it is blocked by a nascent RNA transcript (Maul and Gearhart, 2010). Indeed, AID has no known activity on RNA/DNA hybrids, which form over wide regions of mammalian S regions in the form of R loops (Huang et al., 2006, 2007; Roy et al., 2008; Yu et al., 2003a) and which may also form in the context of transcription bubbles (Go´mez-Gonza´lez and Aguilera, 2007; Li and Manley, 2005). RNase H has been proposed as a candidate for exposing regions of R-looped S region template strands to AID based on biochemical ability to degrade RNA in the context of R loops (Lieber, 2010; Yu and Lieber, 2003b); however, to date, there has been no direct evidence that implicated either RNase H or other cellular RNA degradation factors in CSR. The RNA exosome is an evolutionarily conserved RNA-processing/degradation complex (reviewed by Houseley et al., 2006; Lykke-Andersen et al., 2009; Schmid and Jensen, 2008; Shen and Kiledjian, 2006). The nine-subunit core of the eukaryotic RNA exosome, which includes the CsL4, Rrp4, Rrp40, Rrp41, Rrp46, Mtr3, Rrp42, Rrp43, and Rrp45 subunits, has RNA-binding activity but lacks ribonuclease activity (Anderson et al., 2006; Greimann and Lima, 2008; Oddone et al., 2007; Figure S1 available online). RNA exosome ribonuclease activity is provided by noncore subunits, including Rrp6 and Rrp44, that have RNA 30 -50 exonuclease or endonuclease activity (Greimann and Lima, 2008; Houseley et al., 2006; Lebreton et al., 2008; Schaeffer et al., 2009; Figure S1). The mammalian RNA exosome complex interacts with cofactors, such as the TRAMP complex, that target it to particular substrates based on sequence or structural features (reviewed by Houseley et al., 2006; Houseley and Tollervey, 2008; LaCava et al., 2005). In Drosophila, the RNA exosome complex interacts with elongating RNA polymerase II (Pol II) complexes via the Spt5/6 transcription elongation cofactors (Andrulis et al., 2002), and in yeast, it binds to and removes nascent RNP complexes from template DNA (El Hage et al., 2010; Houseley et al., 2006; Houseley and Tollervey, 2008). We now describe 354 Cell 144, 353–363, February 4, 2011 ª2011 Elsevier Inc.
studies that implicate the RNA exosome as a long-speculated cofactor that targets AID deamination activity to both template and nontemplate strands of transcribed dsDNA substrates. RESULTS AID Associates with the RNA Exosome Complex To elucidate factors that might promote AID access to the template strand of transcribed substrates in the context of RNA/DNA hybrid structures, we in vitro transcribed an RGYW-rich dsDNA SHM substrate with T7 RNA polymerase in the presence of Ramos human B cell lymphoma line extracts. Transcribed DNA-protein complexes were purified via chromatographic steps that would enrich for DNA binding (CM-sepharose, DEAE cellulose) and macromolecular complex formation (gel-filtration chromatography), followed by heparin sepharose chromatography and anti-AID antibody-mediated affinity purification to enrich complexes containing AID (Figures 1A and 1B, Figure S1B, Table S1, and Extended Experimental Procedures). At each step, fractions enriched for AID deamination stimulatory activity were identified via a 3H-release assay (e.g., Figure S1B and Extended Experimental Procedures). Among proteins identified by mass spectrometric analysis of purified complexes were multiple subunits of the RNA exosome complex, including Mtr3, Csl4, Rrp43, Rrp40, and Rrp42 (Figure 1B and Table S1). To further elucidate potential functions, we assayed the ability of the AID-associated, transcribed DNA complex to enhance deamination activity of purified AID in a transcribed dsDNA SHM substrate assay (Chaudhuri et al., 2003) and found that it markedly stimulated AID activity (Figure 1C). Of note, AID association with the RNA exosome complex and purification of an AID stimulatory activity was not observed if the purification was performed from a reaction without T7 polymerase, indicating that complex formation is enhanced by transcription (Figure S1B and data not shown). To assay for AID/RNA exosome association in vivo, we immunoprecipitated AID from mouse primary splenic B cells stimulated with anti-CD40 plus IL-4 to induce AID and CSR to IgG1 and then assayed immunoprecipitates for RNA exosome subunits by western blotting. These analyses revealed that AID associated with Rrp40, Rrp46, and Mtr3 RNA exosome subunits (Figure 2A). Likewise, AID immunoprecipitated from mouse CH12F3 B lymphoma cells stimulated with anti-CD40, IL-4, and TGF-b to undergo CSR to IgA, as well as AID immunoprecipitated from the human Ramos B lymphoma cells, associated with core Rrp40 and Rrp46 subunits, as well as with the Rrp6 catalytic exosome subunit (Figure 2B). By individually expressing FLAGtagged versions of RNA exosome subunits along with AID in HEK293T cells, we observed that immunoprecipitation of any of the 11 exosome core and catalytic subunits via anti-FLAG antibodies also pulled down AID (Figure 2C). Together, our findings indicate that AID either directly or indirectly associates with the RNA exosome complex in cells. Exosome Core Subunit Rrp40 Is Required for Optimal CSR Previous work indicates that the absence of a given core exosome subunit leads to a severe defect in overall RNA exosome
A
B
A
B
C C
Figure 1. AID Forms a Transcription-Dependent Complex with RNA Exosome (A) Schematic outlining steps for enrichment of transcription-dependent AID/RNA exosome/SHM substrate complex. Details are in text and Extended Experimental Procedures. (B) Proteins enriched by purification scheme in (A) were analyzed by SDS-PAGE followed by staining with Coomassie blue. Identity of proteins from selected bands was determined by mass spectrometry; bands that contain RNA exosome subunits are indicted on the right with molecular weight markers on the left. (C) AID purified following ectopic expression in HEK293 cells was assayed in a 3H-release in vitro transcription-dependent SHM substrate assay (Chaudhuri et al., 2003; see also Figure 5) in the presence or absence of complex enriched by purification scheme in (A). Percent of total transcribed DNA substrate deaminated is presented for three separate assays. See also Figure S1 and Table S1.
function (Jensen and Moore, 2005). Therefore, to evaluate potential roles of the RNA exosome complex in CSR, we used a knockdown approach to reduce Rrp40 in the CH12F3 B lymphoma line. For this purpose, we lentivirally introduced two different shRNAs that targeted Rrp40, respectively, into CH12F3 cells to generate three different knockdown lines that had Rrp40 levels ranging from about 50% to less than 10% those of controls, including a WT line and a line harboring a nonspecific (‘‘scrambled’’) shRNA (Figure 3B and Figure S2C). Following stimulation with anti-CD40, TGFb, and IL4 for 48 hr to induce CSR to IgA, Rrp40 knockdown lines consistently displayed reduced CSR, with levels ranging from 30%–50% of those of controls (Figures 3A and 3C and Figure S2B). However, the various Rrp40 knockdown lines proliferated similarly to controls after stimulation (Figure 3E and Figures S2A and S2E). In addition, the knockdown and control lines expressed similar levels of Im and Ia transcripts and similar levels of AID; whereas there were variations in transcript levels from clone to clone, there was no correlation with Im and Ia transcripts
Figure 2. AID Complexes with RNA Exosome Subunits In Vivo (A) AID immunoprecipitates from extracts of CSR-activated AID-deficient and wild-type B cells were assayed for Rrp40, Rrp46, Mtr3, and AID (indicated on the right) via western blotting. The left two lanes show western blotting of total extract, and the right two lanes show western blotting of immunoprecipitated products. (B) AID immunoprecipitates (‘‘Anti-AIDIP’’ lanes) or control reactions without anti-AID antibody (‘‘AbIP’’ lanes) from Ramos (‘‘Ramos’’) and CSR-activated CH12F3 cells (‘‘CH12F3 Sti’’) were assayed for Rrp46, Rrp40, Mtr3, and AID (indicated on right) by western blotting. Unstimulated CH12F3 cells (‘‘CH12F3 unsti’’) were used as a negative control. (C) AID was coexpressed with individual FLAG epitope-tagged RNA exosome subunits in HEK293T cells. Lanes from left to right represent cells transfected with empty vector (‘‘vector’’) as a control or the individual FLAG epitope-tagged subunit indicated at the top. The top two panels show western blotting of total extract (‘‘input’’), and the bottom three panels show western blotting with indicated antibodies (anti-AID, anti-Rrp40, and anti-Rrp6) following immunoprecipitation with anti-Flag antibodies. Asterisks indicate bands corresponding to Flag-tagged exosome subunits. A background band corresponding to the anti-Flag Ig light chain also is indicated (‘‘Background’’).
and Rrp40 levels (Figure 3D and Figure S2F). Finally, similar knockdowns of the Mtr3 RNA exosome core subunit also led to decreased CSR without markedly affecting cell proliferation, AID levels, or Im and Ia transcription (Figure S3). Together, these findings demonstrate that physiological levels of RNA exosome core subunits are required for efficient CSR. Cell 144, 353–363, February 4, 2011 ª2011 Elsevier Inc. 355
A
B
D
C
E
Figure 3. RNA Exosome Subunit Rrp40 Is Required for Normal CSR (A) CH12F3 cells lentivirally infected with a scrambled short-hairpin plasmid (NS) or with shRNA against Rrp40 (shRrp40) were either not stimulated (‘‘Unsti’’) or stimulated (‘‘Sti’’) for 2 days with anti-CD40, IL4, and TGFb and analyzed for IgA CSR by flow cytometry. ShRrp40-1 and shRrp40-2 are independent shRrp40expressing CH12F3 isolates. Results are representative of eight experiments; additional experiments are shown in Figures S2A and S2B. (B) NS, shRrp40-1, and shRrp40-2 expressing stimulated and unstimulated CH12F3 isolates (shown in A) were assayed for Rrp40 and AID by western blotting. Results are representative of four experiments; additional experiment is shown in Figure S2. (C) Average levels and standard deviation from the mean of CSR to IgA from three independent experiments (one shown in A) performed simultaneously with unstimulated (‘‘unsti’’) NS and stimulated (‘‘Sti’’) NS, shRrp40-1 (‘‘#1’’), and ShRrp40-2 (‘‘#2’’) CH12F3 isolates. Five additional experiments gave similar results (Figures S2A and S2B). (D) Total cellular RNA from three independently stimulated samples of indicated CH12F3 isolates (the ones used for C) was assayed for Im transcripts (left) and Ia transcripts (right) via quantitative RT-PCR. Average and standard deviation from the mean is shown for the three separate experiments. An additional experiment based on northern or RT-PCR is shown in Figure S2F. (E) Growth curves of stimulated (NS) shRrp40-1 and shRrp40-2 CH12F3 cells calculated from three independent sets of three experiments (one used for C and others shown Figure S2A) with a fourth set of three experiments indicated in Figure S2E. Values represent average and standard deviation from the mean. See also Figure S2 and Figure S3.
Association of Rrp40 with S Regions in B Cells Activated for CSR If the RNA exosome functions to target AID activity, it should be found in association with transcribed S regions in B cells 356 Cell 144, 353–363, February 4, 2011 ª2011 Elsevier Inc.
activated for CSR. To evaluate this possibility, we performed chromatin immunoprecipitation (ChIP) assays to test whether Rrp40 associates with transcribed Sm sequences in activated CH12F3 cells before and after stimulation for CSR to IgA. We first
A
C
B
D
Figure 4. RNA Exosome Subunit Rrp40 Is Recruited to S Regions (A) Ch12F3 cells were either stimulated with TGFb, IL4, and CD40 or kept unstimulated for 48 hr. Subsequently, Rrp40 was immunoprecipitated from cell extracts under chromatin immunoprecipitation (ChIP) conditions and immunoprecipitates analyzed for Rrp40 by western blotting with anti-Rrp40. (B) The ‘‘ChIPed’’ Rrp40-DNA complex from Ch12F3 cells was processed to isolate bound DNA. ChIPed DNA was tested for Sm and Sa sequences via q-PCR. The average and standard deviations from the mean for three separate ChIP experiments are shown (see also Figure S4). Numbers indicate average fold changes comparing stimulated and unstimulated samples. Unstimulated samples were arbitrarily normalized as 1 (Experimental Procedures). (C) Rrp40 ChIPs were performed on extracts from primary splenic B cells stimulated for 2 days with anti-CD40 plus IL4. Sm and Sg1 were tested via semiquantitative PCR; results are shown for two independent ChIP samples for each genotype. A 5-fold serial dilution of inputs is shown with the highest input concentration corresponding to 1/20 of total input. (D) ChIPed DNA from activated splenic B cells was tested for Sm and Sg1 via q-PCR. Numbers indicate average fold changes comparing WT and AID/ samples. AID/ samples were arbitrarily normalized as 1 (Experimental Procedures). Values represent the average and standard deviation from the mean for three experiments. See Figure S4 for more details.
employed western blotting to confirm that Rrp40 is specifically precipitated with an anti-Rrp40 antibody, but not control IgG, under ChIP conditions (Figure 4A). After processing immunoprecipitates for isolation of bound DNA, we utilized quantitative PCR (q-PCR) to determine levels of Rrp40 bound to Sm and Sa. These analyses demonstrated enrichment of Sm and, to a lesser extent, Sa in the anti-Rrp40 ChIPs from stimulated versus unstimulated CH12F3 cells (Figure 4B and Figure S4). As there is some Sm transcription in nonactivated B cells (Muramatsu et al., 2000), the question arises as to whether transcription per se is sufficient to recruit the RNA exosome to S regions. To explore this question, we assayed for Rrp40 recruitment to Sm and Sg1 in WT and AID-deficient primary B cells activated with anti-CD40 and IL-4, which induce germline Sg1 transcription in both cell types (Muramatsu et al., 2000). Consistent with targeting dependent on germline transcription, Rrp40 was recruited to Sm and Sg1
in the activated WT B cells (Figures 4C and 4D and Figure S4). Of note, however, Rrp40 was not measurably recruited to Sm and Sg1 in AID-deficient B cells. Together, our results indicate that the RNA exosome complex is recruited to transcribed S regions in B cells activated for CSR in an AID-dependent fashion. RNA Exosome Stimulates AID Activity on Template and Nontemplate Strands To further evaluate the potential ability of the cellular RNA exosome complex to act as an AID cofactor, we substantially purified this complex from cell-free nuclear extracts prepared from HEK293T cells that expressed a FLAG epitope-tagged Rrp6 exosome subunit. In this purification, we maintained relatively low salt concentrations to prevent disaggregation of protein complexes (Figure S5). To test activity, we added varying amounts of the exosome-enriched extract to a 3H-uracil-release in vitro transcription-dependent AID deamination assay, which measures overall SHM substrate deamination (Figure 5A). In this assay, T7 polymerase transcription of the SHM substrate leads to little or no AID deamination activity, and addition of partially purified RNA exosome extract in the absence of AID also gives no deamination activity on the T7 transcribed substrate (Figure 5B). However, addition of both AID and partially purified RNA exosome led to substantial deamination of the transcribed substrate (Figure 5B), with activity appearing to be roughly within a range similar to that observed with phosphorylated AID and RPA (Basu et al., 2005, 2008; Chaudhuri et al., 2004; see below). The AID deamination stimulatory activity observed in these extracts is likely mediated by the exosome complex, as we found that deamination activity cofractionated with the RNA exosome during purification (Figure S5). Likewise, we found similar results when we purified the exosome complex from HEK293T cells via an approach in which affinity purification of the complex was performed with antibodies against endogenous Rrp40 (Figure S5D). Finally, we found that the RNA exosome also stimulated AID deamination of a transcribed dsDNA core Sm substrate and a synthetic R loop-forming substrate (Figure 5B). Because the RNA exosome can associate with Pol II transcription complexes and remove nascent transcripts from transcribed DNA (El Hage et al., 2010), we considered it as a candidate AID cofactor for template DNA strand deamination. To test this possibility, we performed in vitro transcription-dependent AID dsDNA SHM substrate deamination assays in which the Southern blotting readout reveals deamination of either template or nontemplate strands, respectively (Figure 5C) (Chaudhuri et al., 2004). In this assay, no deamination of either strand was observed when only AID was added in the presence or absence of T7 polymerase (Figure 5D and Figure S6A). As observed previously, addition of PKA (to phosphorylate AID on S38) and RPA along with T7 polymerase and AID led to deamination of the nontemplate strand, but not the template strand (Figure 5D and Figure S6A). Strikingly, addition of both AID and partially purified RNA exosome (in the absence of RPA or PKA) to the T7 transcription reaction led to deamination of both strands of the SHM substrate (Figure 5D and Figures S6A and S6E). In most assays, activity on both template and nontemplate strands, respectively, was robust, as evidenced by greatly diminished Cell 144, 353–363, February 4, 2011 ª2011 Elsevier Inc. 357
A
C
Figure 5. Cellular RNA Exosome Augments Transcription-Dependent AID Deamination Activity on Template and Nontemplate DNA Strands
B
D
levels of full-length substrate strands (Figure 5D and Figure S6A). The RNA exosome also enhances AID template strand deamination activity on transcribed core Sm substrate and synthetic R loop-forming substrates (Figure 5B and Figures S6B and S6C). Together, our findings indicate that the endogenous RNA exosome complex can function as a stimulatory cofactor for AID deamination activity on both strands of transcribed duplex SHM substrates. Recombinant Core RNA Exosome Stimulates AID-Dependent Deamination on Template and Nontemplate Strands The partially purified endogenous RNA exosome preparation should contain the core structural complex as well as catalytic subunits (e.g., Rrp6, which was used to pull down the complex) and potentially other cofactors or contaminating proteins. To unequivocally evaluate the ability of the RNA exosome to stimulate AID activity on transcribed SHM substrates and to also 358 Cell 144, 353–363, February 4, 2011 ª2011 Elsevier Inc.
(A) Schematic representation of 3H release assay for AID deamination of transcribed dsDNA SHM substrate. (B) (Top) Results of 3H release assay in which a SHM substrate was transcribed by T7 polymerase (T) in the presence of purified AID (AID), purified HEK293 RNA exosome (Exo293), or both. (Middle) Results of 3H release assay in which a synthetic R loop-forming substrate was transcribed by T7 polymerase (T) in the presence of purified AID (AID), recombinant RNA exosome (rExo), or both. (Bottom) Results of 3H release assay in core Sm substrate were transcribed by T7 polymerase (T) in the presence of purified AID (AID), purified HEK293 RNA exosome (Exo293), or both. In all three panels, values represent average and standard deviation from the mean from three independent experiments. (C) A schematic representation of an assay for measuring strand-specific AID deamination of a transcribed dsDNA SHM. The location of template (T) and nontemplate (NT) strand probes is indicated. (D) The strand specificity of RPA-dependent or RNA exosome-dependent DNA deamination was analyzed by the assay in (C) using either nontemplate (left) or template (right) strand-specific probes. Reactions contained AID, T7 polymerase (T), RPA, and PKA or purified HEK293 exosome (‘‘Exosome’’) as indicated. See also Figure S5 and Figure S6.
obtain insight into potential mechanisms, we assayed a nine-subunit recombinant human RNA exosome core complex reconstituted from subunits that were individually purified subsequent to expression in bacteria (Figure 6A) (Greimann and Lima, 2008; Liu et al., 2006). We found that, when titrated for optimal activity (Figure S6E), the recombinant RNA exosome core complex robustly stimulated AID-dependent deamination of transcribed SHM substrates, as measured by the H3-uracil release assay (Figure 6B and Figure S6F). In fact, maximum deamination levels stimulated by the recombinant core complex were roughly similar to those obtained with partially purified RNA exosome complex (Figure 6B and Figure S6F). Moreover, the recombinant RNA exosome core complex also stimulated AID-dependent deamination of both template and nontemplate strands of the SHM substrate (Figure 6C and Figure S6G) and synthetic R loop-forming substrates (Figure S6C). Thus, the core RNA exosome complex alone, in the absence of the RNAase catalytic subunits, promotes AID- and transcription-dependent SHM substrate deamination. To further characterize potential functions of the core RNA exosome, we assayed individual core subunits (Figure 6D and Figure S7). Individual subunits were assayed at approximately
A
C
B
D
the same molarity as when assayed in the context of the core complex. Some, but not all, recombinant core RNA exosome subunits stimulated AID- and transcription-dependent SHM substrate deamination activity; however, the level of stimulation was low compared to that of the full core complex (Figure 6D and Figure S7A). The RNA-binding subunits Rrp40, Rrp46, Rrp41, and Mtr3 showed low levels of activity (Figure 6D and Figures S7A and S7C). On the other hand, other core subunits with demonstrated RNA-binding activity and/or RNA-binding domains, along with the Rrp6 ribonuclease subunit, lacked detectable activity, indicating that the observed activity is not a property of all such proteins. DISCUSSION Enigma of AID Targeting of the Template Substrate Strand Recent studies revealed a mechanism by which AID can be brought to transcribed substrates in the context of RNA polymerase II pausing (Pavri et al., 2010). Once recruited to the substrate, transcription-based mechanisms also provide a potential means of providing AID with a ssDNA substrate in the context of a duplex substrate via looping out of the nontemplate strand (Chaudhuri et al., 2003, 2004; Go´mez-Gonza´lez and Aguilera, 2007; Ramiro et al., 2003; Yu and Lieber, 2003). However, AID
Figure 6. Recombinant Core RNA Exosome and Individual Subunits Stimulate AID Activity (A) Coomassie blue stained polyacrylamide gel electrophoresis analysis of the nine-subunit recombinant RNA exosome complex generated from individual subunits. (B) Comparative analysis of the ability of RNA exosome complex from 293T cells (Exo293) and recombinant RNA core exosome complex (rExo) to stimulate AID deamination activity as measured by 3H release/SHM substrate assay in Figure 5A. See Extended Experimental Procedures and Figure S6 for details. Optimal amounts of RNA exosome293T and recombinant RNA exosome core complex were used (Figure S6 and Extended Experimental Procedures). (C) Assay of recombinant exosome complex to stimulate strand-specific AID deamination of a transcribed SHM substrate via assay in Figure 5C. Nontemplate and template strand deamination are shown on left and right panels, respectively. Reactions contained either AID, recombinant core RNA exosome (rExo9wt), or both (AID + rExo9wt) as indicated. (D) Individual recombinant RNA exosome subunits were assayed for ability to promote AID deamination of a T7-transcribed SHM substrate as outlined in Figure 5A. Added exosome components are indicated. Control reactions with AID alone or AID plus T7 polymerase are on the left. A positive control with complete recombinant core RNA exosome (rExo9) plus T7 and AID is on the right. For (B) and (D), values represent the average and standard deviation from the mean for three independent experiments. See also Figure S7 and Figure S8.
equally deaminates both substrate DNA strands during CSR and SHM (Milstein et al., 1998; Shen et al., 2006; Xue et al., 2006). In the latter context, the mechanism by which AID generates DSB intermediates that are substrates for CSR (Petersen et al., 2001; Schrader et al., 2005; Wuerffel et al., 1997) and for chromosomal translocations (Ramiro et al., 2004; Robbiani et al., 2008) requires targeting of both DNA strands (e.g., Xue et al., 2006; reviewed by Di Noia and Neuberger, 2007). Thus, the mechanism by which AID accesses the template DNA strand has been a major question (e.g. Chaudhuri et al., 2007; Chelico et al., 2009; Di Noia and Neuberger, 2007; Liu and Schatz, 2009; Longerich et al., 2006; Maul and Gearhart, 2010; Pavri et al., 2010; Peled et al., 2008). We now show that the core RNA exosome complex promotes AID deamination of both template and nontemplate strands of in vitro transcribed SHM substrates. Moreover, in B cells activated for CSR, the RNA exosome complex associates with AID and accumulates on S regions in an AID-dependent manner. Finally, integrity of the RNA exosome complex is required for optimal CSR. Thus, the RNA exosome is a long-sought cofactor that can target AID activity to both template and nontemplate strands of transcribed SHM and CSR targets. Implications of AID/RNA Exosome Biochemistry Our in vivo knockdown, ChIP, and physical association studies provide strong biological evidence that the RNA exosome Cell 144, 353–363, February 4, 2011 ª2011 Elsevier Inc. 359
functions in CSR. However, such studies often provide limited mechanistic insight. In addition, as elimination of RNA exosome core subunits is cell lethal in yeast, it may be difficult to achieve complete knockdowns or knockouts to fully test the extent of exosome function in promoting AID activity in vivo. Classical biochemistry has been extremely useful for elucidating potential functions of essential replication, repair, and recombination factors (e.g., Chaudhuri et al., 2004; Dzantiev et al., 2004; Genschel and Modrich, 2003; Maldonado et al., 1996; Stillman and Gluzman, 1985). In this context, our biochemical studies reveal that the core RNA exosome robustly targets AID to both strands of T7 polymerase-transcribed SHM substrates in vitro. Clearly, T7 polymerase is structurally and mechanistically distinct from mammalian RNA PoI II, and the in vitro system that we have employed does not operate in the chromatin context in which SHM and CSR occur in vivo. Also, the concentration of the RNA exosome and of other factors in our assays may be greater than their cellular levels. But again, it is just these aspects of the biochemical assay that allow us to define what the RNA exosome has the potential of achieving. Additional mechanisms and factors (e.g., AID) likely will work to specifically recruit the RNA exosome to transcribed SHM and CSR targets in vivo (see below), potentially increasing local concentration to levels sufficient to achieve activities observed in vitro. The evolutionarily conserved RNA exosome complex is required for degradation and/or processing of various RNAs, including noncoding RNAs generated in cells due to error-prone transcription or from transcription initiated by cryptic promoters (Houseley et al., 2006; Jensen and Moore, 2005; Lykke-Andersen et al., 2009; Preker et al., 2008). In bacteria, the exosome core complex contains ribonuclease activity; however, in yeast and mammalian cells, the core subunits lack such activity and generally have been considered structural scaffolds (Greimann and Lima, 2008; Houseley et al., 2006; Jensen and Moore, 2005). In eukaryotes, noncore exosome subunits or cofactors provide ribonuclease activity (Callahan and Butler, 2010; Greimann and Lima, 2008; Houseley et al., 2006; Lebreton et al., 2008; Schaeffer et al., 2009; Staals et al., 2010; Tomecki et al., 2010). Our biochemical data demonstrate that the recombinant nine-subunit exosome core, in the absence of catalytic subunits, robustly targets AID to both strands of T7 transcribed substrates. The precise molecular mechanism by which the RNA exosome core achieves this activity remains to be determined. In theory, it may compete with duplex DNA for binding the nascent transcript, thereby displacing it from the template strand, while also forming a scaffold that stabilizes the transcriptionally opened template to expose both DNA strands. Our finding that certain core exosome subunits have low but significant ability to enhance AID deamination of in vitro transcribed substrates is intriguing. Currently, however, there is no clear correlation between any particular known subunit activity and which subunits promote low-level AID access. We previously have shown that RPA, a ssDNA-binding protein involved in replication and repair, binds AID phosphorylated on S38 and, thereby, targets AID to transcribed SHM substrates in vitro, albeit mostly to the nontemplate strand (Basu et al., 2005; Chaudhuri et al., 2004). A substantial amount of evidence supports the significance of the AID/RPA interaction for AID 360 Cell 144, 353–363, February 4, 2011 ª2011 Elsevier Inc.
function during CSR and SHM, both in augmenting AID deamination activity and also potentially in recruiting downstream repair pathways involved in further processing AID-initiated lesions (Basu et al., 2005, 2008; Cheng et al., 2009; McBride et al., 2006, 2008; Pasqualucci et al., 2006). Of note, RNA exosomemediated AID-targeting activity in our in vitro SHM assay does not require AID phosphorylation on S38 or RPA, demonstrating that the RNA exosome function in AID targeting is distinct. However, these two AID-targeting mechanisms still might be complementary in vivo (see below). Potential Mechanism by Which the RNA Exosome Promotes AID Activity In Vivo Both CSR and SHM in activated B cells require transcription through target S regions or variable region exons. Such transcriptional targeting was proposed to result from association of a mutator factor with stalled Pol II within the target sequence (Peters and Storb, 1996). Over the years, this model has been supported by numerous findings, including the following: AID, now known to be the hypothetical mutator, is targeted by transcription (reviewed by Chaudhuri et al., 2007; Yang and Schatz, 2007; Di Noia and Neuberger, 2007); AID associates with Pol II (Nambu et al., 2003); and Pol II indeed stalls during transcription through S regions (Rajagopal et al., 2009; Wang et al., 2009), potentially augmented by transcribed S regions forming secondary structures such as R loops (reviewed by Maul and Gearhart, 2010). Recent studies have implicated the Spt5 transcriptional elongation cofactor in recruiting AID to stalled Pol II and, thereby, bringing AID to transcribed genomic targets (Pavri et al., 2010). However, the AID/Spt5/Pol II interaction did not explain how AID actually accesses its target sequences once there and, in particular, how AID accesses the template DNA strand (Pavri et al., 2010), a process in which we now have implicated the RNA exosome. The recruitment of the RNA exosome to S regions in B cells activated for CSR, in theory, might involve a putative Spt5/RNA exosome interaction (Andrulis et al., 2002) or generation of the noncoding S region transcripts (Lennon and Perry, 1985 Lutzker and Alt, 1988), the latter of which might attract the RNA exosome complex (reviewed by Houseley et al., 2006; Jensen and Moore, 2005; Schmid and Jensen, 2008). However, the marked AID dependence of exosome recruitment to targeted S regions in activated B cells rules out Spt5 interaction or germline transcription alone for RNA exosome recruitment. Though AID might directly recruit the RNA exosome complex, we thus far have not found interaction with individual purified subunits, raising the additional possibility of indirect recruitment, for example, via an AID/transcription-related complex. We propose a working model for how the RNA exosome may enhance targeting of AID to both DNA strands of its in vivo substrates (Figure S8). This model is based on our current biochemical and cellular findings regarding relationships between AID and the RNA exosome, known aspects of AID function and recruitment to transcribed target sequences, and known RNA exosome properties. To enhance AID activity on the template strand, the RNA exosome must in some way remove the template RNA. As the RNA exosome is not known to engage RNA substrates that lack a free single-stranded 30
end, cotranscriptional RNA exosome activity in which an RNA in the RNA/DNA hybrid is still attached to the RNA polymerase would seem unlikely. Therefore, we suggest that the RNA exosome would have its relevant activity on stalled Pol II units that backtrack to reveal a free 30 end (Adelman et al., 2005) (Figure S8A). In this context, stalled RNA polymerase recruits AID via the Spt5 transcription factor (Pavri et al., 2010). Thus, Pol II stalling would indirectly recruit the RNA exosome via AID (Figure S8B). Likewise, Pol II stalling would also provide a suitable substrate to engage the RNA exosome and allow it to enhance AID substrate activity (Figure S8B), potentially via mechanisms mentioned in the preceding section of the Discussion or below. We further suggest that RPA might bind to and stabilize ssDNA targets opened up by the RNA exosome and facilitate S38-phosphorylated AID accumulation via direct interaction (Basu et al., 2005, 2008; Vuong et al., 2009) (Figures S8C and S8D). In addition, bound RPA may facilitate engagement of downstream repair pathways (Chaudhuri, Khuong, and Alt, 2004; Vuong et al., 2009). We must note that it remains possible, in vivo, that the catalytic subunit of the RNA exosome could provide ribonuclease activity that would degrade the nascent transcript from the DNA/RNA hybrids on transcribed AID substrates and, thereby, further contribute to exposing the template strand to AID activity (Figure S8E). Finally, we do not rule out the possibility that RNA exosome, in the context of an AID complex, acquires unique properties as compared to properties already described for the RNA exosome and that these unique properties are relevant to enhancing AID activity.
EXPERIMENTAL PROCEDURES Recombinant Plasmids, Antibodies, and Proteins RNA exosome subunits, AID, and RPA were expressed as described (Greimann and Lima, 2008; Chaudhuri et al., 2003, 2004; Stillman and Gluzman, 1985). AID antibodies were generated as described (Chaudhuri et al., 2003), and anti-exosome subunit antibodies were purchased from Genway or AbCam. Exosome subunit and complex purifications were as described (Greimann and Lima, 2008). Details are in Extended Experimental Procedures.
AID Activity Assays The 3H release assay was performed essentially as described (Chaudhuri et al., 2003; Basu et al., 2005). The strand-specific AID and transcriptiondependent dsDNA deamination assays were performed as described (Chaudhuri et al., 2003, 2004). Details are in Extended Experimental Procedures.
AID Complex Purification from Ramos Cells We used a combination of DNA affinity, size chromatography, and affinity purification for isolating AID complexes. We analyzed components of the complex via mass spectrometry. For details, see Extended Experimental Procedures.
RNA Exosome Purification from 293T Cells We generated protein extracts from HEK293 cells expressing FLAG epitopetagged Rrp6. We used a combination of size chromatography, glycerol gradient sedimentation, and FLAG affinity purification to isolate a protein complex enriched with RNA exosome complex. We also isolated RNA exosome from HEK293 cells using the same approach except that the affinity purification step employed anti-Rrp40 antibodies. For details, see Extended Experimental Procedures.
Immunoprecipitation of AID/RNA Exosome Complex from B Cells and HEK293 Cells We utilized previously published protocols to immunoprecipitate the AID complex from B cells (Chaudhuri et al., 2004; Basu et al., 2005) or used similarly adapted protocols for immunoprecipitating FLAG-exosome subunits from 293 cells. For details, see Extended Experimental Procedures. CSR Analysis of Rrp40-Deficient CH12F3 Cells We performed shRNA-mediated knockdowns in Ch12F3 according to TRC shRNA library protocols. Knockdown efficiencies were measured by detecting levels of Rrp40 in these cells by western blotting. We determined levels of germline S region transcripts in Rrp40-deficient cells via previously published protocols (Muramatsu et al., 2000). For details, see Extended Experimental Procedures. Chromatin Immunoprecipitation Assays We utilized adaptations of previously published protocols (Chaudhuri et al., 2004) for isolating chromatin immunoprecipitates of Rrp40. See Extended Experimental Procedures for details.
SUPPLEMENTAL INFORMATION Supplemental Information includes Extended Experimental Procedures, eight figures, and one table and can be found with this article online at doi:10.1016/j. cell.2011.01.001. ACKNOWLEDGMENTS We thank Bjoern Schwer, Jason Rodriguez, Stephen Goff, Bruce Stillman, and Tasuku Honjo for providing reagents and/or protocols. We also thank Bjoern Schwer and Cosmas Giallourakis for critically reading the manuscript. This work was supported by National Institute of Health grants AI31541 (to F.W.A.) and GM079196 (to K.J. and C.D.L.); by Trustees of Columbia University Faculty startup funds (to U.B.); and by funds from Regeneron Pharmaceuticals (in support of E.P.). F.-L.M. is a fellow of Cancer Research Institute of New York. C.K. was supported by a Cancer Biology training grant CA09503-23. K.J. is a fellow of the American Cancer Society (PF-10-23601-RMC). U.B. is a Special Fellow of the Leukemia and Lymphoma Society of America and the recipient of a New Investigator Award from the Leukemia Research Foundation. F.W.A. is an Investigator of the Howard Hughes Medical Institute. Received: September 30, 2010 Revised: December 27, 2010 Accepted: December 31, 2010 Published online: January 20, 2011 REFERENCES Adelman, K., Marr, M.T., Werner, J., Saunders, A., Ni, Z., Andrulis, E.D., and Lis, J.T. (2005). Efficient release from promoter-proximal stall sites requires transcript cleavage factor TFIIS. Mol. Cell 17, 103–112. Anderson, J.R., Mukherjee, D., Muthukumaraswamy, K., Moraes, K.C., Wilusz, C.J., and Wilusz, J. (2006). Sequence-specific RNA binding mediated by the RNase PH domain of components of the exosome. RNA 12, 1810–1816. Andrulis, E.D., Werner, J., Nazarian, A., Erdjument-Bromage, H., Tempst, P., and Lis, J.T. (2002). The RNA processing exosome is linked to elongating RNA polymerase II in Drosophila. Nature 420, 837–841. Basu, U., Chaudhuri, J., Alpert, C., Dutt, S., Ranganath, S., Li, G., Schrum, J.P., Manis, J.P., and Alt, F.W. (2005). The AID antibody diversification enzyme is regulated by protein kinase A phosphorylation. Nature 438, 508–511. Basu, U., Wang, Y., and Alt, F.W. (2008). Evolution of phosphorylationdependent regulation of activation-induced cytidine deaminase. Mol. Cell 32, 285–291.
Cell 144, 353–363, February 4, 2011 ª2011 Elsevier Inc. 361
Besmer, E., Market, E., and Papavasiliou, F.N. (2006). The transcription elongation complex directs activation-induced cytidine deaminase-mediated DNA deamination. Mol. Cell. Biol. 26, 4378–4385. Callahan, K.P., and Butler, J.S. (2010). TRAMP complex enhances RNA degradation by the nuclear exosome component Rrp6. J. Biol. Chem. 285, 3540–3547. Chaudhuri, J., Basu, U., Zarrin, A., Yan, C., Franco, S., Perlot, T., Vuong, B., Wang, J., Phan, R.T., Datta, A., et al. (2007). Evolution of the immunoglobulin heavy chain class switch recombination mechanism. Adv. Immunol. 94, 157–214. Chaudhuri, J., Khuong, C., and Alt, F.W. (2004). Replication protein A interacts with AID to promote deamination of somatic hypermutation targets. Nature 430, 992–998. Chaudhuri, J., Tian, M., Khuong, C., Chua, K., Pinaud, E., and Alt, F.W. (2003). Transcription-targeted DNA deamination by the AID antibody diversification enzyme. Nature 422, 726–730. Chelico, L., Pham, P., Petruska, J., and Goodman, M.F. (2009). Biochemical basis of immunological and retroviral responses to DNA-targeted cytosine deamination by activation-induced cytidine deaminase and APOBEC3G. J. Biol. Chem. 284, 27761–27765. Cheng, H.L., Vuong, B.Q., Basu, U., Franklin, A., Schwer, B., Astarita, J., Phan, R.T., Datta, A., Alt, F.W., and Chaudhuri, J. (2009). Integrity of the AID serine-38 phosphorylation site is critical for class switch recombination and somatic hypermutation in mice. Proc. Natl. Acad. Sci. USA 106, 2717–2722. Daniels, G.A., and Lieber, M.R. (1995). RNA:DNA complex formation upon transcription of immunoglobulin switch regions: implications for the mechanism and regulation of class switch recombination. Nucleic Acids Res. 23, 5006–5011. Di Noia, J.M., and Neuberger, M.S. (2007). Molecular mechanisms of antibody somatic hypermutation. Annu. Rev. Biochem. 76, 1–22. Dickerson, S.K., Market, E., Besmer, E., and Papavasiliou, F.N. (2003). AID mediates hypermutation by deaminating single stranded DNA. J. Exp. Med. 197, 1291–1296. Dzantiev, L., Constantin, N., Genschel, J., Iyer, R.R., Burgers, P.M., and Modrich, P. (2004). A defined human system that supports bidirectional mismatchprovoked excision. Mol. Cell 15, 31–41. El Hage, A., French, S.L., Beyer, A.L., and Tollervey, D. (2010). Loss of Topoisomerase I leads to R-loop-mediated transcriptional blocks during ribosomal RNA synthesis. Genes Dev. 24, 1546–1558. Genschel, J., and Modrich, P. (2003). Mechanism of 50 -directed excision in human mismatch repair. Mol. Cell 12, 1077–1086. Go´mez-Gonza´lez, B., and Aguilera, A. (2007). Activation-induced cytidine deaminase action is strongly stimulated by mutations of the THO complex. Proc. Natl. Acad. Sci. USA 104, 8409–8414. Greimann, J.C., and Lima, C.D. (2008). Reconstitution of RNA exosomes from human and Saccharomyces cerevisiae cloning, expression, purification, and activity assays. Methods Enzymol. 448, 185–210. Honjo, T., Kinoshita, K., and Muramatsu, M. (2002). Molecular mechanism of class switch recombination: linkage with somatic hypermutation. Annu. Rev. Immunol. 20, 165–196. Houseley, J., LaCava, J., and Tollervey, D. (2006). RNA-quality control by the exosome. Nat. Rev. Mol. Cell Biol. 7, 529–539. Houseley, J., and Tollervey, D. (2008). The nuclear RNA surveillance machinery: the link between ncRNAs and genome structure in budding yeast? Biochim. Biophys. Acta 1779, 239–246. Huang, F.T., Yu, K., Balter, B.B., Selsing, E., Oruc, Z., Khamlichi, A.A., Hsieh, C.L., and Lieber, M.R. (2007). Sequence dependence of chromosomal R-loops at the immunoglobulin heavy-chain Smu class switch region. Mol. Cell. Biol. 27, 5921–5932. Huang, F.T., Yu, K., Hsieh, C.L., and Lieber, M.R. (2006). Downstream boundary of chromosomal R-loops at murine switch regions: implications for the mechanism of class switch recombination. Proc. Natl. Acad. Sci. USA 103, 5030–5035.
362 Cell 144, 353–363, February 4, 2011 ª2011 Elsevier Inc.
Jensen, T.H., and Moore, C. (2005). Reviving the exosome. Cell 121, 660–662. LaCava, J., Houseley, J., Saveanu, C., Petfalski, E., Thompson, E., Jacquier, A., and Tollervey, D. (2005). RNA degradation by the exosome is promoted by a nuclear polyadenylation complex. Cell 121, 713–724. Lebreton, A., Tomecki, R., Dziembowski, A., and Se´raphin, B. (2008). Endonucleolytic RNA cleavage by a eukaryotic exosome. Nature 456, 993–996. Lennon, G.G., and Perry, R.P. (1985). C mu-containing transcripts initiate heterogeneously within the IgH enhancer region and contain a novel 50 -nontranslatable exon. Nature 318, 475–478. Li, X., and Manley, J.L. (2005). Inactivation of the SR protein splicing factor ASF/SF2 results in genomic instability. Cell 122, 365–378. Lieber, M.R. (2010). The mechanism of double-strand DNA break repair by the nonhomologous DNA end-joining pathway. Annu. Rev. Biochem. 79, 181–211. Liu, M., and Schatz, D.G. (2009). Balancing AID and DNA repair during somatic hypermutation. Trends Immunol. 30, 173–181. Liu, Q., Greimann, J.C., and Lima, C.D. (2006). Reconstitution, activities, and structure of the eukaryotic RNA exosome. Cell 127, 1223–1237. Longerich, S., Basu, U., Alt, F., and Storb, U. (2006). AID in somatic hypermutation and class switch recombination. Curr. Opin. Immunol. 18, 164–174. Lykke-Andersen, S., Brodersen, D.E., and Jensen, T.H. (2009). Origins and activities of the eukaryotic exosome. J. Cell Sci. 122, 1487–1494. Lutzker, S., and Alt, F.W. (1988). Structure and expression of germ line immunoglobulin gamma 2b transcripts. Mol. Cell. Biol. 8, 1849–1852. Maldonado, E., Shiekhattar, R., Sheldon, M., Cho, H., Drapkin, R., Rickert, P., Lees, E., Anderson, C.W., Linn, S., and Reinberg, D. (1996). A human RNA polymerase II complex associated with SRB and DNA-repair proteins. Nature 381, 86–89. Maul, R.W., and Gearhart, P.J. (2010). AID and somatic hypermutation. Adv. Immunol. 105, 159–191. McBride, K.M., Gazumyan, A., Woo, E.M., Barreto, V.M., Robbiani, D.F., Chait, B.T., and Nussenzweig, M.C. (2006). Regulation of hypermutation by activation-induced cytidine deaminase phosphorylation. Proc. Natl. Acad. Sci. USA 103, 8798–8803. McBride, K.M., Gazumyan, A., Woo, E.M., Schwickert, T.A., Chait, B.T., and Nussenzweig, M.C. (2008). Regulation of class switch recombination and somatic mutation by AID phosphorylation. J. Exp. Med. 205, 2585–2594. Milstein, C., Neuberger, M.S., and Staden, R. (1998). Both DNA strands of antibody genes are hypermutation targets. Proc. Natl. Acad. Sci. USA 95, 8791–8794. Muramatsu, M., Kinoshita, K., Fagarasan, S., Yamada, S., Shinkai, Y., and Honjo, T. (2000). Class switch recombination and hypermutation require activation-induced cytidine deaminase (AID), a potential RNA editing enzyme. Cell 102, 553–563. Nambu, Y., Sugai, M., Gonda, H., Lee, C.G., Katakai, T., Agata, Y., Yokota, Y., and Shimizu, A. (2003). Transcription-coupled events associating with immunoglobulin switch region chromatin. Science 302, 2137–2140. Neuberger, M.S., Harris, R.S., Di Noia, J., and Petersen-Mahrt, S.K. (2003). Immunity through DNA deamination. Trends Biochem. Sci. 28, 305–312. Oddone, A., Lorentzen, E., Basquin, J., Gasch, A., Rybin, V., Conti, E., and Sattler, M. (2007). Structural and biochemical characterization of the yeast exosome component Rrp40. EMBO Rep. 8, 63–69. Odegard, V.H., and Schatz, D.G. (2006). Targeting of somatic hypermutation. Nat. Rev. Immunol. 6, 573–583. Pasqualucci, L., Kitaura, Y., Gu, H., and Dalla-Favera, R. (2006). PKA-mediated phosphorylation regulates the function of activation-induced deaminase (AID) in B cells. Proc. Natl. Acad. Sci. USA 103, 395–400. Pavri, R., Gazumyan, A., Jankovic, M., Di Virgolio, M., Kein, I., AnsarahSobrinho, C., Resch, W., Yamane, A., Reina San-Martin, B., Barreto, V., et al. (2010). Activation-induced cytidine deaminase targets DNA at sites of RNA polymerase II stalling by interaction with Spt5. Cell 143, 122–133.
Peled, J.U., Kuang, F.L., Iglesias-Ussel, M.D., Roa, S., Kalis, S.L., Goodman, M.F., and Scharff, M.D. (2008). The biochemistry of somatic hypermutation. Annu. Rev. Immunol. 26, 481–511.
Shen, H.M., Tanaka, A., Bozek, G., Nicolae, D., and Storb, U. (2006). Somatic hypermutation and class switch recombination in Msh6(-/-)Ung(-/-) doubleknockout mice. J. Immunol. 177, 5386–5392.
Peters, A., and Storb, U. (1996). Somatic hypermutation of immunoglobulin genes is linked to transcription initiation. Immunity 4, 57–65.
Shen, V., and Kiledjian, M. (2006). A view to a kill: structure of the RNA exosome. Cell 127, 1093–1095.
Petersen, S., Casellas, R., Reina-San-Martin, B., Chen, H.T., Difilippantonio, M.J., Wilson, P.C., Hanitsch, L., Celeste, A., Muramatsu, M., Pilch, D.R., et al. (2001). AID is required to initiate Nbs1/gamma-H2AX focus formation and mutations at sites of class switching. Nature 414, 660–665.
Shinkura, R., Tian, M., Smith, M., Chua, K., Fujiwara, Y., and Alt, F.W. (2003). The influence of transcriptional orientation on endogenous switch region function. Nat. Immunol. 4, 435–441.
Petersen-Mahrt, S.K., Harris, R.S., and Neuberger, M.S. (2002). AID mutates E. coli suggesting a DNA deamination mechanism for antibody diversification. Nature 418, 99–103.
Sohail, A., Klapacz, J., Samaranayake, M., Ullah, A., and Bhagwat, A.S. (2003). Human activation-induced cytidine deaminase causes transcription-dependent, strand-biased C to U deaminations. Nucleic Acids Res. 31, 2990–2994.
Preker, P., Nielsen, J., Kammler, S., Lykke-Andersen, S., Christensen, M.S., Mapendano, C.K., Schierup, M.H., and Jensen, T.H. (2008). RNA exosome depletion reveals transcription upstream of active human promoters. Science 322, 1851–1854.
Staals, R.H., Bronkhorst, A.W., Schilders, G., Slomovic, S., Schuster, G., Heck, A.J., Raijmakers, R., and Pruijn, G.J. (2010). Dis3-like 1: a novel exoribonuclease associated with the human exosome. EMBO J. 29, 2358–2367.
Rajagopal, D., Maul, R.W., Ghosh, A., Chakraborty, T., Khamlichi, A.A., Sen, R., and Gearhart, P.J. (2009). Immunoglobulin switch mu sequence causes RNA polymerase II accumulation and reduces dA hypermutation. J. Exp. Med. 206, 1237–1244. Ramiro, A.R., Jankovic, M., Eisenreich, T., Difilippantonio, S., Chen-Kiang, S., Muramatsu, M., Honjo, T., Nussenzweig, A., and Nussenzweig, M.C. (2004). AID is required for c-myc/IgH chromosome translocations in vivo. Cell 118, 431–438. Ramiro, A.R., Stavropoulos, P., Jankovic, M., and Nussenzweig, M.C. (2003). Transcription enhances AID-mediated cytidine deamination by exposing single-stranded DNA on the nontemplate strand. Nat. Immunol. 4, 452–456. Revy, P., Muto, T., Levy, Y., Geissmann, F., Plebani, A., Sanal, O., Catalan, N., Forveille, M., Dufourcq-Labelouse, R., Gennery, A., et al. (2000). Activationinduced cytidine deaminase (AID) deficiency causes the autosomal recessive form of the Hyper-IgM syndrome (HIGM2). Cell 102, 565–575. Robbiani, D.F., Bothmer, A., Callen, E., Reina-San-Martin, B., Dorsett, Y., Difilippantonio, S., Bolland, D.J., Chen, H.T., Corcoran, A.E., Nussenzweig, A., and Nussenzweig, M.C. (2008). AID is required for the chromosomal breaks in c-myc that lead to c-myc/IgH translocations. Cell 135, 1028–1038. Roy, D., Yu, K., and Lieber, M.R. (2008). Mechanism of R-loop formation at immunoglobulin class switch sequences. Mol. Cell. Biol. 28, 50–60. Schaeffer, D., Tsanova, B., Barbas, A., Reis, F.P., Dastidar, E.G., SanchezRotunno, M., Arraiano, C.M., and van Hoof, A. (2009). The exosome contains domains with specific endoribonuclease, exoribonuclease and cytoplasmic mRNA decay activities. Nat. Struct. Mol. Biol. 16, 56–62. Schmid, M., and Jensen, T.H. (2008). The exosome: a multipurpose RNAdecay machine. Trends Biochem. Sci. 33, 501–510. Schrader, C.E., Linehan, E.K., Mochegova, S.N., Woodland, R.T., and Stavnezer, J. (2005). Inducible DNA breaks in Ig S regions are dependent on AID and UNG. J. Exp. Med. 202, 561–568.
Stillman, B.W., and Gluzman, Y. (1985). Replication and supercoiling of simian virus 40 DNA in cell extracts from human cells. Mol. Cell. Biol. 5, 2051–2060. Tian, M., and Alt, F.W. (2000). Transcription-induced cleavage of immunoglobulin switch regions by nucleotide excision repair nucleases in vitro. J. Biol. Chem. 275, 24163–24172. Tomecki, R., Kristiansen, M.S., Lykke-Andersen, S., Chlebowski, A., Larsen, K.M., Szczesny, R.J., Drazkowska, K., Pastula, A., Andersen, J.S., Stepien, P.P., et al. (2010). The human core exosome interacts with differentially localized processive RNases: hDIS3 and hDIS3L. EMBO J. 29, 2342–2357. Vuong, B.Q., Lee, M., Kabir, S., Irimia, C., Macchiarulo, S., McKnight, G.S., and Chaudhuri, J. (2009). Specific recruitment of protein kinase A to the immunoglobulin locus regulates class-switch recombination. Nat. Immunol. 10, 420–426. Wang, L., Wuerffel, R., Feldman, S., Khamlichi, A.A., and Kenter, A.L. (2009). S region sequence, RNA polymerase II, and histone modifications create chromatin accessibility during class switch recombination. J. Exp. Med. 206, 1817–1830. Wuerffel, R.A., Du, J., Thompson, R.J., and Kenter, A.L. (1997). Ig Sgamma3 DNA-specifc double strand breaks are induced in mitogen-activated B cells and are implicated in switch recombination. J. Immunol. 159, 4139–4144. Xue, K., Rada, C., and Neuberger, M.S. (2006). The in vivo pattern of AID targeting to immunoglobulin switch regions deduced from mutation spectra in msh2-/- ung-/- mice. J. Exp. Med. 203, 2085–2094. Yang, S.Y., and Schatz, D.G. (2007). Targeting of AID-mediated sequence diversification by cis-acting determinants. Adv. Immunol. 94, 109–125. Yu, K., Chedin, F., Hsieh, C.L., Wilson, T.E., and Lieber, M.R. (2003). R-loops at immunoglobulin class switch regions in the chromosomes of stimulated B cells. Nat. Immunol. 4, 442–451.
Shen, H.M., Ratnam, S., and Storb, U. (2005). Targeting of the activationinduced cytosine deaminase is strongly influenced by the sequence and structure of the targeted DNA. Mol. Cell. Biol. 25, 10815–10821.
Yu, K., and Lieber, M.R. (2003). Nucleic acid structures and enzymes in the immunoglobulin class switch recombination mechanism. DNA Repair (Amst.) 2, 1163–1174.
Shen, H.M., and Storb, U. (2004). Activation-induced cytidine deaminase (AID) can target both DNA strands when the DNA is supercoiled. Proc. Natl. Acad. Sci. USA 101, 12997–13002.
Zarrin, A.A., Alt, F.W., Chaudhuri, J., Stokes, N., Kaushal, D., Du Pasquier, L., and Tian, M. (2004). An evolutionarily conserved target motif for immunoglobulin class-switch recombination. Nat. Immunol. 5, 1275–1281.
Cell 144, 353–363, February 4, 2011 ª2011 Elsevier Inc. 363
Structural Basis of the 9-Fold Symmetry of Centrioles Daiju Kitagawa,1,5,6 Ioannis Vakonakis,2,4,5 Natacha Olieric,3,5 Manuel Hilbert,2,3,5 Debora Keller,1 Vincent Olieric,2 Miriam Bortfeld,3 Miche`le C. Erat,4 Isabelle Flu¨ckiger,1 Pierre Go¨nczy,1,5,* and Michel O. Steinmetz3,5 1Swiss Institute for Experimental Cancer Research (ISREC), School of Life Sciences, Swiss Federal Institute of Technology (EPFL), CH-1015 Lausanne, Switzerland 2Swiss Light Source 3Biomolecular Research Paul Scherrer Institut, 5232 Villigen PSI, Switzerland 4Department of Biochemistry, University of Oxford, Oxford OX1 3QU, UK 5These authors contributed equally to this work 6Present address: Center for Frontier Research, National Institute of Genetics, Mishima, Shizuoka 411-8540, Japan *Correspondence: pierre.gonczy@epfl.ch DOI 10.1016/j.cell.2011.01.008
SUMMARY
The centriole, and the related basal body, is an ancient organelle characterized by a universal 9-fold radial symmetry and is critical for generating cilia, flagella, and centrosomes. The mechanisms directing centriole formation are incompletely understood and represent a fundamental open question in biology. Here, we demonstrate that the centriolar protein SAS-6 forms rod-shaped homodimers that interact through their N-terminal domains to form oligomers. We establish that such oligomerization is essential for centriole formation in C. elegans and human cells. We further generate a structural model of the related protein Bld12p from C. reinhardtii, in which nine homodimers assemble into a ring from which nine coiled-coil rods radiate outward. Moreover, we demonstrate that recombinant Bld12p self-assembles into structures akin to the central hub of the cartwheel, which serves as a scaffold for centriole formation. Overall, our findings establish a structural basis for the universal 9-fold symmetry of centrioles. INTRODUCTION Centrioles are fundamental for the assembly of cilia and flagella across eukaryotic evolution (reviewed in Azimzadeh and Marshall, 2010). In addition, centrioles are important for assembling the centrosome, the major microtubule organizing center (MTOC) of animal cells, and as such, they are critical for genome stability. As anticipated from these important roles, aberrations in centriole structure or function are implicated in a number of disease conditions, including ciliopathies, male sterility, primary microcephaly, and cancer (reviewed in Nigg and Raff, 2009). Therefore, increased understanding of centriole biology is expected to also result in important clinical implications. 364 Cell 144, 364–375, February 4, 2011 ª2011 Elsevier Inc.
Centrioles, and the related basal bodies, are barrel-shaped microtubule-containing structures characterized by a universal 9-fold radial symmetry that they also impart to cilia and flagella (reviewed in Azimzadeh and Marshall, 2010). In most species, the centriole is organized around a cartwheel that comprises a central hub 25 nm in diameter from which nine spokes radiate outward and connect to nine microtubule blades (reviewed in Strnad and Go¨nczy, 2008). The molecular and structural principles directing the universal 9-fold symmetry of the cartwheel and the centriole remain to be discovered. The genetic material duplicates once and only once per cell cycle, and so do centrioles. In contrast to the mechanisms governing DNA replication, however, those at the root of centriole formation are poorly understood. This is despite the fact that five proteins that are essential for centriole formation have been identified initially in Caenorhabditis elegans (Dammermann et al., 2004; Delattre et al., 2004; Kemp et al., 2004; Kirkham et al., 2003; Leidel et al., 2005; Leidel and Go¨nczy, 2003; O’Connell et al., 2001; Pelletier et al., 2004). Relatives of these components are present and similarly required for centriole formation across eukaryotic evolution, indicating that they constitute an ancient core module that is essential for centriole formation (reviewed in Nigg and Raff, 2009; Strnad and Go¨nczy, 2008). Among these five components, SAS-6 is of particular interest to consider for investigating the mechanisms governing centriole formation for a number of reasons. First, proteins of the SAS-6 family are required for the earliest steps of centriole formation from Chlamydomonas reinhardtii to Homo sapiens (Culver et al., 2009; Leidel et al., 2005; Nakazawa et al., 2007; Rodrigues-Martins et al., 2007; Strnad et al., 2007; Yabe et al., 2007). Second, overexpression of SAS-6 proteins induces the formation of multiple new centrioles adjacent to the existing one in human cells (Strnad et al., 2007), as well as centriole amplification and de novo formation in Drosophila melanogaster (Rodrigues-Martins et al., 2007). Furthermore, combined overexpression in Drosophila spermatocytes of DSas-6 and the interacting protein Ana2 results in the formation of structures that resemble the cartwheel (Stevens et al., 2010). Third, SAS-6 proteins localize to the cartwheel in C. reinhardtii and
Figure 1. Domain Organization and Molecular Model of C. elegans SAS-6
A
(A) Schematic representation of C. elegans SAS-6. N, N-terminal domain; CC, coiled coil; C, C terminus. Numbers above the schematic correspond to amino acids. (B) Rotary metal shadowing electron micrographs of ceFL, ceN-CC, and ceCC specimens. Arrowheads indicate globular domains. Scale bar, 50 nm. (C and D) CD spectrum (C) and thermal unfolding profile recorded by CD at 222 nm (D) of the ceCC fragment. The data support the formation of a highly helical structure with moderate thermal stability. (E) MALS analysis of the ceCC fragment. The UV absorbance profile of size exclusion chromatography (black line) is overlaid with the molecular weight (50 kDa) estimation by MALS (gray line). (F) ceCC dilution series monitored by CD at 222 nm. The gray solid line represents the fit to the data (open circles) using a monomer-dimer model. (G) SDS-PAGE of the ceCC fragment run under reduced (+bMe) and nonreducing (bMe) conditions. Arrowheads point to protein bands corresponding to monomeric (M) and disulfide-linked dimeric (D) forms of ceCC. (H) Molecular model of SAS-6 homodimer. Each monomeric subunit is composed of a globular N-terminal domain, a coiled-coil domain that forms a parallel dimer, and a poorly structured C-terminal part. See also Figure S1.
B
C
D
F
G
E
H
self-assembly of SAS-6 homodimers is at the root of the universal 9-fold symmetry of the cartwheel and thus of centrioles. RESULTS
Tetrahymena thermophila (Kilburn et al., 2007; Nakazawa et al., 2007), to the proximal part of the new centriole in H. sapiens (Kleylein-Sohn et al., 2007; Strnad et al., 2007), and to the functionally related central tube in C. elegans (Dammermann et al., 2008; Pelletier et al., 2006). Together, these observations suggest that proteins of the SAS-6 family are somehow important for the onset of centriole formation, although whether they can initiate this process on their own or must rely on additional factors to do so is not known. Overall, although it has been hypothesized that SAS-6 proteins may be critical for forming the central hub of the cartwheel (Strnad and Go¨nczy, 2008), the actual mechanisms by which they ensure cartwheel assembly and thus centriole formation have remained elusive. In this study, using a combination of biophysical, biochemical, structural, and cell biological approaches, we establish that
Structural and Biophysical Characterization of C. elegans SAS-6 We first set out to characterize the structure of C. elegans SAS-6 to uncover the mechanisms by which it contributes to centriole formation. Proteins of the SAS-6 family comprise an N-terminal domain with the evolutionarily conserved PISA motif, followed by a segment with a predicted coiled coil and a less-conserved C-terminal region predicted to be disordered (Figure 1A). We expressed and purified soluble SAS-6 full-length (ceFL), the N terminus plus the coiled coil (ceN-CC), or the coiled-coil domain alone (ceCC) (Figure 1A and Figures S1A and S1B available online) and analyzed them by biophysical and structural methods. Inspection by electron microscopy revealed an 35 nm elongated rod in all three constructs, which fits the predicted length of the SAS-6 coiled coil (220 residues 3 0.1485 nm [axial raise per residue] = 32.7 nm) (Figure 1B). Full-length SAS-6 and ceN-CC were decorated with a globular head-like moiety at one end (Figure 1B, arrowheads), which is absent in ceCC, indicating that it corresponded to the N-terminal domain of SAS-6. No significant difference could be observed between ceFL and ceN-CC (Figure 1B), supporting the prediction that the C terminus does not adopt a globular structure. We analyzed the ceCC fragment further to uncover its stability and molecular architecture. Circular dichroism (CD) Cell 144, 364–375, February 4, 2011 ª2011 Elsevier Inc. 365
spectroscopy revealed a far-ultraviolet spectrum and a cooperative thermal unfolding profile that is characteristic of moderately stable a-helical coiled-coil structures (Figures 1C and 1D) (Steinmetz et al., 1998). To assess the oligomerization state of the coiled-coil domain, we conducted multiangle light scattering (MALS) experiments, which yielded a molecular mass that is consistent with a dimer (50 kDa versus a ceCC monomeric mass of 27.4 kDa; Figure 1E). The stability of the ceCC coiledcoil dimer was estimated by measuring the change in CD signal at 222 nm upon dilution. Fitting of the data revealed a dissociation constant, Kd, of 0.9 ± 0.1 mM (Figure 1F). To determine the relative orientation of the two ceCC monomers within the dimer, we performed SDS-PAGE analysis under nonreducing conditions. Cys204 is the only cysteine residue in the coiled coil and is predicted to occupy a heptad a core position, such that the ceCC fragment should form a disulphide bond only if the two fragments are in a parallel and in-register configuration (Figures S1C and S1D). As shown in Figure 1G, ceCC was indeed crosslinked under nonreducing conditions, indicating a parallel arrangement of monomers in SAS-6 homodimers (Figure 1H). Next, we determined the structure of the N-terminal globular domain of C. elegans SAS-6 (ceN; Figures S1A and S1B) by X-ray crystallography. We obtained crystals of a ceN variant and solved its structure to 2.1 A˚ resolution (Table S1). The asymmetric unit of the crystal contained a dimer of ceN monomers with local 2-fold symmetry (ceN-dimer) (Figure 2A). The fold of ceN is reminiscent of that of the XRCC4 family of DNA repair proteins (Junop et al., 2000). We noted that a striking interaction interface in the ceN-dimer was mediated by I154 at the tip of the b6-b7 loop of one monomer, which was inserted deeply into a hydrophobic cavity of the second monomer (Figures 2B and 2C). Both I154 and the residues shaping the hydrophobic cavity are well conserved among SAS-6 orthologs (Figure 2B and Figure S2), suggesting functional relevance. Analytical ultracentrifugation (AUC) experiments conducted at 300 mM protein concentration demonstrated that the ceN fragment could also form a dimer in solution (Figure 2D). Isothermal titration calorimetry (ITC) experiments yielded a Kd for the N-N interaction of 110 ± 30 mM (Figure 2E and Figure S3A), two orders of magnitude higher than that of the ceCC coiled coil. To address whether I154 mediates ceN-dimer formation, we substituted this residue for the charged residue glutamate (ceN[I154E]). Although the conformation of the domain was not altered by this mutation (Figures S3B and S3C), AUC experiments revealed that this change abrogated dimer formation (Figure 2D). We conclude that I154 is critical for mediating the N-N interaction. Interestingly, inspection of the ceN-dimer structure suggested that the b6-b7 loop encompassing I154 might promote an interaction between SAS-6 homodimers, as this residue is located diametrically across the ceN domain’s C terminus, which proceeds into the coiled coil (Figure 2A). To test this hypothesis, we conducted AUC experiments with the ceN-CC fragment, which could be more readily expressed and purified in an intact form than ceFL (Figure S3D). AUC of ceN-CC conducted at 200 mM protein concentration revealed the presence of higherorder oligomers besides dimers (Figure 2F). In contrast, a mutant in which I154 had been exchanged by glutamate (ceN-CC [I154E]) only formed dimers (Figure 2F). 366 Cell 144, 364–375, February 4, 2011 ª2011 Elsevier Inc.
Together, our structural and biophysical data establish that assembly of higher-order SAS-6 oligomeric structures occurs in two steps. First, elongated SAS-6 homodimers assemble, driven by the strong interaction between the helices of the two-stranded parallel coiled coil. Second, oligomers of SAS-6 homodimers assemble, a step that is mediated by the weaker interaction between pairs of N-terminal globular domains located in adjacent homodimers. Biological Significance of SAS-6 Oligomerization To investigate the biological significance of the oligomerization of SAS-6 homodimers mediated by the N-N interaction, we generated transgenic worms expressing GFP fused to SAS-6[I154E] engineered so as to be resistant to RNAi directed against endogenous SAS-6 (GFP-SAS-6RR[I154E]) (Dammermann et al., 2008). A similar approach was utilized to replace I154 by glycine, a smaller and noncharged residue, thus generating GFP-SAS-6RR[I154G]. Upon sas-6(RNAi) in an otherwise wild-type background, the two paternally contributed centrioles split from one another and assembled a bipolar spindle at the end of the one-cell stage (Movie S1). In contrast, a monopolar spindle usually assembled in each blastomere at the end of the second cell cycle (Figure 3G). In sas-6(RNAi) embryos expressing RNAi-resistant wild-type SAS-6 fused to GFP (GFP-SAS-6RR), 40% of embryos underwent bipolar spindle assembly in each blastomere at the end of the second cell cycle (Figures 3A and 3G and Movie S2) (Kitagawa et al., 2009). This reflected rescue of centriole formation, as demonstrated by the presence of the centriolar protein SAS-4 in each spindle pole (Figure 3D). Partial rescue to only 40% is likely due to GFP at the N terminus interfering with the function of SAS-6 and to levels of the fusion protein being lower than that of the endogenous protein (Figure 3H) (see also Kitagawa et al., 2009). Importantly, there was no rescue of centriole formation in sas-6 (RNAi) embryos expressing GFP-SAS-6RR[I154E] or GFP-SAS6RR[I154G] (Figures 3B, 3C, and 3E–3G and Movie S3). This was not due to differences in expression levels; in fact, GFP-SAS6RR[I154E] and GFP-SAS-6RR[I154G] were expressed at slightly higher levels than wild-type GFP-SAS-6RR (Figure 3H). We conclude that I154 is essential for centriole formation in C. elegans. We then addressed whether the importance of oligomerization mediated by the N-N interaction is evolutionarily conserved. To this end, we analyzed the human protein HsSAS-6, in which the residue corresponding to C. elegans I154 is F131 (Figure S2). We generated constructs in which wild-type or F131E mutant HsSAS-6 was fused to GFP and expressed from a doxycyclineinducible promoter (Bach et al., 2007). The fusion constructs do not contain the 30 UTR of the endogenous gene, so we targeted this region using siRNAs to deplete solely endogenous HsSAS6 without affecting the GFP fusion proteins (Figure S3E). Whereas 95% of control mitotic cells harbored the usual number of R 4 centrioles marked by the EF-hand protein centrin (Figure 3L), this was the case for only 10% of mitotic cells treated with siHsSAS6-30 UTR (Figures 3I and 3L). Expression of wild-type HsSAS-6GFP in cells treated with siHsSAS-6-30 UTR resulted in substantial rescue of centriole formation, with > 80% of mitotic cells harboring R 4 centrioles (Figures 3J and 3L). By contrast, cells expressing HsSAS-6[F131E]-GFP and subjected to siHsSAS-630 UTR did not exhibit rescue (Figures 3K and 3L).
A
B
D
C
E
F
Figure 2. Structural Analysis of C. elegans SAS-6 N-Terminal Domain (A) Two overall views of the ceN-dimer structure seen in the asymmetric unit of the crystal 90 apart. Monomers A and B (in cartoon representation) are colored in light gray and yellow, respectively. Secondary structure elements and the N and C termini are assigned. Loop a2-b5, which is unique to C. elegans, is not seen in the electron density presumably due to disorder and is indicated by a dashed line. Each monomer displays two a helices that cap the end of a two-stranded b sheet sandwich. The PISA motif spans region b3 to a2, with evolutionarily conserved residues in this region contributing to the protein core as well as to a predominantly hydrophobic cavity between a1 and a2 (see also Figure S2). The locations of loops b6-b7 are indicated by arrows. (B) Structure of the ceN-dimer, with monomer A shown as surface representation. Highly conserved residues are colored dark green, and mostly conserved residues are colored bright green. I154 of monomer B is depicted as stick representation. (C) Close-up views of the interaction network observed at the dimer interface in cartoon (main chains) and stick (contacting residues) representations. Oxygen and nitrogen atoms are colored in red and blue, respectively, and carbon atoms are colored in light gray (monomer A) or yellow (monomer B). (D) Sedimentation velocity AUC analysis of ceN (red) and ceN[I154E] (blue) fragments. The peak labeled ‘‘Monomer’’ corresponds to a molecular weight of 20 kDa, which is consistent with the molecular weight of the ceN[I154E] monomer. The peak labeled ‘‘Dimer’’ corresponds to a molecular weight of 40 kDa. Protein concentration was 300 mM. (E) Dissociation isotherm obtained by ITC for ceN. A 1.6 mM ceN solution was injected stepwise into buffer. Shown are the integrated heat changes upon dilution. The solid red line represents the fit to the data (open circles) assuming dissociation of ceN dimers into monomers. (F) Sedimentation velocity AUC analysis of ceN-CC (red) and ceN-CC[I154E] (blue) fragments. The peak labeled ‘‘Dimer’’ corresponds to a molecular weight of 90 kDa, which is consistent with the formation of ceN-CC[I154E] dimers. The broad profile observed for ceN-CC (labeled ‘‘Higher-order oligomers’’) suggests formation of higher-order oligomers beyond dimers. Protein concentration was 200 mM. See also Figure S2 and Figure S3.
Cell 144, 364–375, February 4, 2011 ª2011 Elsevier Inc. 367
A
D
B
C
E
F
G
H
I
J
K
L
Together, these experiments demonstrate that a singly conserved residue mediating the SAS-6 N-N interaction is essential for centriole formation in C. elegans and in human cells, indicating that the capacity to oligomerize is critical for the function of SAS-6 proteins across evolution. Root of the 9-Fold Symmetry of Centrioles To understand the function of SAS-6 oligomerization for centriole formation at the structural level, we investigated the molecular properties of a SAS-6 protein from an organism in which the cartwheel has a canonical structure. This is the case in human cells (Guichard et al., 2010), but recombinant HsSAS-6 and fragments 368 Cell 144, 364–375, February 4, 2011 ª2011 Elsevier Inc.
Figure 3. Functional Analysis of SAS-6 in C. elegans and Human Cells (A)–(F) Anterior is to the left and scale bar is 10 mm. (A–C) Images at the end of the second cell cycle from representative DIC recordings of embryos treated with sas-6(RNAi) and expressing GFP-SAS-6RR (A), GFP-SAS-6RR[I154E] (B), or GFP-SAS-6RR[I154G] (C). Elapsed time after pronuclear meeting is indicated in minutes and seconds; arrowheads indicate centrosomes. (D–F) Embryos during mitosis of the second cell cycle treated with sas-6(RNAi) and expressing GFP-SAS-6RR (D), GFPSAS-6RR[I154E] (E), or GFP-SAS-6RR[I154G] (F) stained with antibodies against a-tubulin (green) and SAS-4 (red); DNA in blue. Insets show an 2.5-fold magnified view of one MTOC. Note that GFP-SAS-6RR[I154E] and GFP-SAS-6RR[I154G] are not present at centrioles (data not shown), presumably because they fail to be incorporated as a result from the lack of oligomerization. (G) Quantification of experiments illustrated in (A)–(C). The percentages of embryos with four cells at the end of the second cell cycle are indicated (n = 31 for wild-type, n = 37 for GFP-SAS-6RR, n = 50 for GFP-SAS-6RR[I154E], and n = 35 for GFP-SAS-6RR[I154G]). Shown are the mean percentages ± SEM from two independent experiments. (H) Western blot analysis of GFP-SAS-6RR, GFP-SAS-6 [I154E], or GFP-SAS-6RR[I154G] embryonic extracts probed with SAS-6 antibodies to reveal both endogenous protein (filled arrowhead) and GFP fusions (open arrowhead). (I–K) Metaphase U2OS, iU2OS:HsSAS-6-GFP, and iU2OS:HsSAS-6[F131E]-GFP cells transfected with siRNAs targeting the 30 UTR of endogenous HsSAS-6 (siHsSAS-630 UTR), induced concomitantly with doxycycline, fixed after 48 hr, and stained with antibodies against centrin (red) and GFP (green); DNA in blue. Scale bar, 10 mm. Insets show magnified view of the delineated regions; scale bar in insets, 1 mm. Whereas the vast majority of mitotic cells expressing HsSAS-6[F131E]-GFP did not exhibit centriolar GFP (see C), a centriolar signal was detected earlier during the cell cycle in most cells (data not shown), suggestive of a failure in stable incorporation as a result of the lack of oligomerization. (L) Percentage of cells in mitosis (prophase to metaphase) with four or more centrioles after 48 hr treatment with Stealth RNAi Low GC negative control or siHsSAS-6-30 UTR (n = 135 for U2OS + control siRNA, n = 236 for U2OS + 30 UTR siRNA, n = 226 for U2OS + 30 UTR siRNA + HsSAS-6-GFP, and n = 160 for U2OS + 30 UTR siRNA + HsSAS-6[F131E]-GFP). Data from at least three independent experiments (R50 cells/experiment) are shown; error bar indicates SEM. See also Figure S3, Movies S1, Movie S2, and Movie S3.
thereof were not soluble (data not shown). By contrast, we were able to produce soluble recombinant proteins from C. reinhardtii Bld12p (Nakazawa et al., 2007), which has the same domain organization as other SAS-6 orthologues (Figure S4A). Like for C. elegans SAS-6, we started by producing a fragment encompassing the N-terminal domain (denoted crN; Figures S4A and S4B) and solved its structure to 2.1 A˚ resolution by X-ray crystallography (Table S1). We found that the asymmetric unit of the crystal contained three equivalent crN dimers (denoted the crN-dimer hereafter). The monomers in each dimer are related by local 2-fold symmetry (Figure 4A). The overall structure and organization of the crN-dimer, as well as the F145 residue
A
B
C
D
Figure 4. Structural Analysis of C. reinhardtii Bld12p (A) Two overall views of the crN-dimer structure 90 apart and superimposed onto the ceN-dimer structure. Monomers A and B are depicted in cartoon representations and colored in dark gray and red (crN-dimer) and light gray and yellow (ce-N dimer), respectively. The global superimposition yielded a root-mean-square deviation of 1.6 A˚ for 217 backbone atoms. (B) Two overall views of the crCC-dimer structure 90 apart. Monomers A and B are colored in magenta and light pink, respectively. (C) Close-up views of the interaction network seen at the crCC-dimer interface in cartoon (main chains) and stick (contacting residues) representations. Key secondary structure elements are assigned. Oxygen and nitrogen atoms are colored in red and blue, respectively. Carbon atoms are colored in magenta and light pink. (D) Superimposition of monomer B of the crN-dimer onto monomer A of the crCC-dimer. The resulting assembly was used as a template for building the Bld12p ring structure shown in Figure 5. See also Figure S2 and Figure S4.
engaged at the N-N interface and corresponding to I154 of C. elegans SAS-6, are similar to that of the C. elegans ceN-dimer (Figure 4A and Figure S4C). The stability of the crN-dimer in solution was assessed by ITC, and the Kd was determined to be 60 ± 20 mM (Figure S4D), which is similar to that of the ceN-dimer from C. elegans (see Figure 2E). Overall, these results indicate that there is strong structural conservation among N-terminal
domains of SAS-6 proteins across evolution. In addition, they demonstrate that the function of the critical residue within the b6-b7 loop mediating the interaction between pairs of N-terminal domains is likewise conserved. To investigate the structural organization of the Bld12p N-terminal domains in the context of the two-stranded parallel coiled coil, we produced a fragment in which the crN variant Cell 144, 364–375, February 4, 2011 ª2011 Elsevier Inc. 369
A
3
B
4 2
8 nm
5 crCC-dimer 40° 1
3.5 nm
23 nm
5 nm
90°°
crN-dimer
6
9
7 8 Figure 5. Structural Model of C. reinhardtii Bld12p Ring Oligomer (A) Two views of the crCC-dimer building block 90 apart. (B) Nine crCC-dimers were associated such that their N-terminal domains interact as observed in the crN-dimer (Figure 4D). The resulting 9-fold symmetric ring oligomer exhibits a diameter of 23 nm and a thickness of 3.5 3 5 nm. The long axis of the coiled-coil domains are in plane with and radiate out from the ring. See also Figure S5.
was extended by the first six heptad repeats of the Bld12p coiled coil (crN-6HR; Figures S4A and S4B). AUC experiments conducted at 150 mM protein concentration revealed that crN-6HR forms higher-order oligomeric species (Figure S4E). However, a mutant in which F145 was substituted for glutamate (crN6HR[F145E]) formed only dimers, as revealed by AUC and MALS experiments (Figures S4E and S4F). The stability of crN-6HR[F145E] was assessed by CD, which yielded a Kd of 0.5 ± 0.1 mM (Figure S4G). We solved the structure of Bld12p crN-6HR[F145E] to 3.0 A˚ resolution by X-ray crystallography. The asymmetric unit of the crystal revealed a dimer (denoted the crCC-dimer hereafter; Figure 4B). Dimerization is brought about by interactions between the two a3 helices, which establish a parallel, two-stranded coiled coil through knobs-into-hole packing of the residues occupying the heptad a and d core positions. The relative orientation of the two N-terminal domains is maintained by predominantly hydrophobic interactions formed between residues of their b3-b4 loops and residues from both coiled-coil a3 helices (Figure 4C). 370 Cell 144, 364–375, February 4, 2011 ª2011 Elsevier Inc.
Having the structures of both the crN-dimer and the crCCdimer of Bld12p allowed us to build a structural model of higher-order oligomers using both dimer interfaces. Strikingly, when crCC-dimers were associated such that their N-terminal domains interact as observed in the crN-dimer (Figure 4D and Figure S5), we obtained a ring with a 9-fold symmetry (Figure 5; see Experimental Procedures for full description of the modeling). In this structural model, the long axes of the coiledcoil domains are in plane with and radiate out from the ring, which is 3.5 3 5 nm in thickness and 23 nm in mean diameter (Figure 5B). A key prediction of our structural model is that Bld12p possesses properties to self-assemble into a ring with 9-fold symmetry. We tested this hypothesis by performing electron microscopy experiments with bacterially expressed Bld12p. As the full-length protein exhibited unspecific aggregation (data not shown) and as the C-terminal part of Bld12p is not evolutionarily conserved and is predicted to be largely disordered, we produced a Bld12p fragment encompassing the N-terminal
A
B
Figure 6. Electron Microscopy of C. reinhardtii Bld12p
C
N
C
E Frequency (n=52)
D
14
42
12
11
10 8 6 4 2 0 0
20
40
60
80
(A–D) Rotary metal shadowing electron micrographs of crN-CC specimens. Schematic interpretations of the specimens are indicated. Note that not all nine spokes represented in the schematic of (D) are unambiguously discerned in the electron micrographs, presumably because some of them are perturbed during sample preparation. The arrow in (A) highlights the head-like moieties of crN-CC. Scale bars, 50 nm. (E) Histogram representation of angles measured between the two legs of the V-shaped crN-CC specimens shown in (B) (n = 51). (F) Histogram representation of mean diameters measured from crN-CC ring oligomers shown in (D) (n = 73). The majority of rings possess a diameter of 22 nm, which is in good agreement with the 23 nm diameter determined from the 9-fold symmetric structural ring model of Bld12p (Figure 5). Note also that a minor fraction of crN-CC rings displayed a lower than 22 nm diameter, which probably correlates with a different number of crN-CC dimers. See also Figure S6.
Angle (deg) Frequency (n=73)
F
20
22
2 nm
16 12 8 4 0
and coiled-coil domains (crN-CC; Figures S4A and S4B). Electron microscopy revealed that crN-CC is an elongated 40 nm rod that displays a globular head-like moiety at one extremity (Figure 6A). The overall organization of crN-CC is similar to that of the C. elegans SAS-6 homodimer, with the rod corresponding to the two-stranded parallel coiled coil and the head moiety to the two N-terminal domains (compare Figure 6A with Figure 1B). Strikingly, at increased concentrations, crN-CC could associate in a head-to-head fashion to form an overall V-shaped structure (Figure 6B). The angle between the two legs of the V was determined to be 42 ± 11 (Figure 6E), which suggestively corresponds to approximately one-ninth of 360 . Remarkably, we found in addition that crN-CC further assembled into higher-order oligomers (Figure 6C) and could form ringlike structures from which emanated spokes corresponding to the coiled-coil domains (Figure 6D and Figure S6A). The mean diameter of the central ring was 22 ± 2 nm (Figure 6F), which is similar to that of the crCC-dimer ring model (Figure 5B) and of the central hub of the C. reinhardtii cartwheel (Cavalier-Smith, 1974). In contrast, no higher-order assemblies were obtained with an crN-CC mutant in which glutamate was substituted for F145 (crN-CC[F145E]; Figure S6B), demonstrating that this
residue is critical for forming V-shaped structures and ring oligomers. Consistent with the findings with crN-CC, the shorter crN-6HR fragment also formed predominantly rings with a diameter similar to 15 20 25 30 that of crN-CC, although radial spokes Diameter (nm) were not observed in this case given the small size of the crN-6HR coiled coil (Figures S6C and S6D). Collectively, these data demonstrate that Bld12p self-assembles into ringlike structures from which emanate radial spokes. DISCUSSION The 9-fold symmetry of centrioles, cilia, and flagella has fascinated biologists since it was discovered decades ago with the advent of electron microscopy. The mechanisms at the origin of this remarkable 9-fold symmetry have inspired many hypotheses (reviewed in Strnad and Go¨nczy, 2008). For instance, because duplication of the centriole occurs once per cell cycle, as is the case for replication of the genetic material, it has been proposed that centriole formation may similarly rely on nucleic acids (reviewed in Marshall and Rosenbaum, 2000). Our work demonstrates that a protein-based mechanism is sufficient to account for an initial step of centriole formation, as the self-assembly properties of SAS-6 generate a molecular architecture with a 9-fold symmetry that bears striking resemblance with the cartwheel. The cartwheel has been perhaps best described in C. reinhardtii and consists of a central hub from which emanate nine spokes capped by a pinhead-like structure (Figure 7) (Cavalier-Smith, 1974). The cartwheel is the first structure with a 9-fold symmetry apparent at the onset of Cell 144, 364–375, February 4, 2011 ª2011 Elsevier Inc. 371
Microtubules 2
3
Pinhead 1
4
Central hub 5
9
Spokes
6
8 7
50 nm Figure 7. Structural Model of the SAS-6-Based Cartwheel in the Context of the Centriole In this model, viewed from the proximal end, the SAS-6 coiled-coil domains would contribute substantially to the formation of the spokes of the cartwheel and possibly connect to the pinheads. The coiled-coil domains were extended to the size expected from structure prediction, and their lengths fit well with the 40 nm length measured from electron micrographs of crN-CC specimens (Figure 6A). However, we note that the protein Bld10p from C. reinhardtii, which localizes to the pinhead of the cartwheel, also contributes to the formation of the spokes (Hiraki et al., 2007). Scale bar, 50 nm.
centriole formation, which has led to the suggestion that it acts as a scaffold onto which centriolar microtubules then assemble (reviewed in Strnad and Go¨nczy, 2008). Support for this view has come notably from the analysis of bld12 mutants in which the cartwheel is missing (Nakazawa et al., 2007). In most cells null for bld12 function, basal bodies are fragmented into pieces, indicating that the cartwheel is required for centriole formation. Interestingly, in addition, the rare mutant cells that harbor basal bodies exhibit defects in the 9-fold symmetry, with the number of microtubule blades varying from 7 to 11. This observation strongly supports the notion that the cartwheel is critical for dictating the 9-fold symmetry of centrioles. Our findings elucidate the structural basis of the cartwheel and thus of the 9-fold symmetry of centrioles. We first establish that proteins of the SAS-6 family form coiled-coil-mediated homodimers. Our elongated molecular model of SAS-6 and Bld12p homodimers is in contrast to the proposal that Drosophila DmSas6 exhibits a globular arrangement (Gopalakrishnan et al., 2010). Although this may reflect a Drosophila-specific feature, we note that all proteins of the SAS-6 family contain a predicted coiled-coil domain that is expected to form an extended rod (Carvalho-Santos et al., 2010). Our work further reveals that interaction between homodimers mediated by adjacent N-terminal domains results in the oligomerization of SAS-6 homodimers. Strikingly, in addition, recombinant Bld12p homodimeric building blocks self-assemble into an 22 nm ring from which the coiled-coil domains emanate radially. The overall appearance of Bld12p oligomers in our electron micrographs, 372 Cell 144, 364–375, February 4, 2011 ª2011 Elsevier Inc.
as well as that of our structural ring model, is remarkably similar to the cartwheel comprising a central hub and radial spokes, as observed in vivo (Figure 7) (Cavalier-Smith, 1974). Although new centrioles form next to the old ones in most proliferating cells, once per cell cycle, centrioles can also assemble de novo, for instance in multiciliated epithelial cells or after ablation of the resident centrioles (Khodjakov et al., 2002; Loncarek and Khodjakov, 2009; Dirksen, 1991). Moreover, the cartwheel also possesses self-assembling properties (Gavin, 1984). Our findings provide an attractive mechanism for how de novo centriole formation may be achieved, as self-assembly of SAS-6 proteins is sufficient to mediate formation of a structure that bears resemblance to the cartwheel. Although centrioles can assemble de novo in some cases, they form strictly in the vicinity of the old centriole in most proliferating cells. We speculate that this may reflect the fact that the vicinity of the old centriole is a favorable environment for promoting self-assembly of SAS-6 proteins, perhaps because of the local enrichment of other centrosomal components. Alternatively, phosphorylation of SAS-6, for instance as is known to occur in C. elegans through the action of the kinase ZYG-1 (Kitagawa et al., 2009), could regulate the formation or stability of SAS-6 oligomers. Regardless, it will be of utmost interest to elucidate how the basic ring of SAS-6 homodimers is stabilized so that it can promote the formation of a mature centriole. In light of the importance of regulated centrosome duplication in genome stability (reviewed in Nigg and Raff, 2009), the structural information uncovered in this study, and in particular the identification of the residues mediating interaction between adjacent SAS-6 N-terminal domains, represents a promising avenue to modulate centriole formation for therapeutic purposes. Furthermore and in conclusion, because these residues are well conserved among SAS-6 orthologs, we propose that the self-assembly of SAS-6 homodimers into a 9-fold symmetric ring structure is a fundamental property at the root of the universal 9-fold symmetry of centrioles. EXPERIMENTAL PROCEDURES Protein Preparation and Biophysical Characterization Standard cloning and recombinant protein production in bacteria is described in the Supplemental Information. Protein identity was confirmed by ESI-TOF mass spectrometry and concentrations estimated by UV at 280 nm. CD spectra were collected at 10 C at a protein concentration of 25 mM in 20 mM Na2HPO4, 150 mM NaCl (pH 7.4) (PBS) using a Chirascan spectropolarimeter (AppliedPhotophysics) with a 0.1 cm path length. Thermal stability experiments were performed using a 1 C/min temperature ramp between 10 C and 90 C and monitored by CD at 222 nm. The dissociation constant of ceCC and crN-6HR[F145E] was determined by monitoring the CD signal at 222 nm and at 20 C after buffer signal subtraction in a dilution series. The samples were reduced with DTT prior to data acquisition to ensure that no covalent dimers remained. The concentration-dependent mean residue elipticity at 222 nm was fit to a two-state association model to obtain the Kd. MALS was performed in PBS supplemented with 1 mM DTT using an S-200 analytical size exclusion chromatography column connected in-line to miniDAWN TREOS light scattering and Optilab T-rEX refractive index detectors (Wyatt Technology). Samples of 2–4 mg/ml concentration were used. AUC experiments were performed at 20 C using 0.15–0.3 mM proteins in 20 mM Tris-HCl (pH 7.5), 150 mM NaCl, 2 mM TCEP, and 1% glycerol using a ProteomeLab XL-I analytical ultracentrifuge (Beckman). All sedimentation velocities were recorded by measuring absorbance at 280 or 290 nm, with
200 scans every 4 min at 35000 rpm. Data were processed using SEDFIT (Schuck, 2000). Partial specific volume was calculated from the amino acid sequence. ITC experiments were performed at 7 C using an ITC200 system (Microcal). 1.0–1.6 mM samples of ceN and crN in 20 mM sodium phosphate (pH 7) supplemented with 100 mM NaCl and 1 mM DTT (ceN) or 12 mM HEPES (pH 7) supplemented with 100 mM NaCl and 0.7 mM bMe (crN) were loaded for stepwise injection into sample buffer alone. The resulting heats were integrated using Origin (OriginLab) and fit with the two-step dissociation model provided by the software package. Cysteine crosslinking of SAS-6 ceCC was performed using protein samples of 20 mM concentration in PBS buffer without DTT. Substantial crosslinked dimer formation was observed on nonreducing SDS-PAGE after overnight incubation at 20 C. Electron Microscopy Electron micrographs were taken in a Philips Morgagni TEM operated at 80 kV equipped with a Megaview III CCD camera. Protein samples (0.1–1 mg/ml) in PBS were supplemented with glycerol to a final concentration of 30%. Samples were subsequently sprayed onto freshly cleaved mica and rotary shadowed in a BA 511 M freeze-etch apparatus (Balzers) with platinum/carbon at an elevation angle of 3 –5 (Fowler and Aebi, 1983). Mean diameters of individual crN-CC and crN-6HR specimens were determined by taking the arithmetic middle of the outer and inner diameter of the ring specimens. Structure Determination Structure solution by X-ray crystallography is described in full in the Supplemental Information. In brief, crystals of the C. elegans ceN fragment in which Ser123 was mutated to glutamate (ceN[S123E]) (Kitagawa et al., 2009) diffracted to 2.1 A˚ resolution. Phase information was obtained by SAD using NdCl3 derivatized crystals and the structure refined to final Rwork/Rfree values of 21.0%/25.7%. Crystals of the C. reinhardtii crN and crN-6HR[F145E] fragments diffracted to 2.1 and 3.0 A˚ resolution, respectively. The structures of both proteins were solved by molecular replacement and refined to final Rwork/Rfree values of 18.1%/21.8% (crN) and 19.6%/22.9% (crN-6HR[F145E]). See Table S1 for data collection and refinement statistics. X-ray data were collected at beamlines X06DA and X06SA of the Swiss Light Source (Paul Scherrer Institut, Villigen, Switzerland). Modeling Structure determination of the crCC-dimer and the crN-dimer of C. reinhardtii Bld12p revealed two distinct types of interfaces. Analysis of the three crN-dimers (denoted AB, CD, and EF) within the asymmetric unit of the crystal revealed small differences between them (rmsd values of 0.5–1.0 A˚). crCCdimers were continuously associated such that their N-terminal domains interact as observed for the AB, CD, or EF crN-dimers, resulting in flat lefthanded helices with 10–11 dimers per turn, with diameters of 23–27 nm and pitches of 80–165 A˚. In these assemblies, the coiled coils radiate out from and are nearly perpendicular to the helix axis. To assist modeling of a 9-fold symmetric ring, a planar wheel with spokes every 40 was generated, and a Ca model of the crCC-dimer structure was positioned with its 2-fold axis aligned with one spoke. Radial position (along the spoke) and orientation (rotation around the spoke axis) were optimized such that the resulting N-N interaction with a neighboring crCC-dimer generated by a 40 rotation became as close as possible to that observed for crN-dimers. The fit was assessed by comparing the generated N-N interaction with the structures of the AB, CD, and EF crN-dimers. After optimization, superposition of the generated ‘‘40 -model’’ with the AB, CD, or EF crN-dimers yielded rmsd values of 1.3 A˚, 1.7 A˚, and 1.3 A˚, respectively. Figure S5B shows the superposition of the optimized 40 -model with the CD crN-dimer. The small differences between the N-N contact generated as described above in the model and that observed in the crystal structure makes the existence of a ring very plausible. In reality, structural adjustments are expected to be distributed more globally and over many degrees of freedom and to not be locally concentrated in the interface between pairs of N-terminal domains as in the simplified modeling approach. In particular, small changes in
the coiled coil as well as in the coiled coil-N-terminal domain interfaces would also be expected. Nematode Strains and RNA Interference For the experiments with the RNAi-resistant strains, GFP-SAS-6RR (Dammermann et al., 2008) and all other strains were maintained according to standard procedures. For generating GFP-SAS-6RR[I154E] and GFP-SAS-6RR[I154G] transgenic lines, appropriate primers (sequences available upon request) were used to PCR-amplify sas-6 cDNA, replacing the ATT that normally codes I154 by GAA or GGA, respectively, and cloning the resulting fragments into pIC26, a pie-1-based vector containing a rescuing unc-119 cDNA (gift from Karen Oegema). Sequence-verified plasmids were bombarded, yielding two integrated strains for both strains. RNAi-mediated inactivation was performed by soaking (Maeda et al., 2001). In brief, L4 larvae were placed in a solution containing in vitro synthesized dsRNAs targeting a portion of sas-6 corresponding to the engineered RNAiresistant construct (Dammermann et al., 2008), incubated for 24 hr at 20 C, and allowed to recover for 12 hr at 20 C before analysis. Cell-cycle progression in C. elegans early-stage embryos was monitored by time-lapse differential interference contrast (DIC) microscopy, recording one image every 5 s at 23 C. Indirect Immunofluorescence and Western Blot Analysis for C. elegans Embryos were fixed and stained essentially as described (Leidel et al., 2005). In brief, embryos were methanol fixed for < 3 min and blocked in 3% bovine serum albumin (BSA) for > 20 min prior to incubation with primary antibodies overnight at 4 C. Primary antibodies were 1:800 SAS-4 (rabbit) (Leidel and Go¨nczy, 2003) and 1:200 a-tubulin (mouse, DM1a, Sigma). Secondary antibodies were goat anti-mouse coupled to Alexa 488 and goat anti-rabbit coupled to Alexa 568 (Molecular Probes), both used at 1:500. Slides were counterstained with 1 mg/ml Hoechst 33258 (Sigma) to reveal DNA. Indirect immunofluorescence was imaged on a Leica SP2 confocal microscope. Optical sections were acquired every 0.25–0.3 mm, and planes containing centrioles were projected together. A similar procedure was applied for microtubules and DNA. Images were processed using ImageJ and Adobe Photoshop, preserving relative image intensities within a series. For western blot analysis, transgenic worms expressing GFP-SAS-6RR, GFP-SAS-6RR[I154E], or GFP-SAS-6RR[I154G] were collected in Laemmli SDS sample buffer, boiled, and subjected to SDS-PAGE, and signal intensities were analyzed after western blotting using 1:200 SAS-6 antibody (Leidel et al., 2005). HRP-conjugated anti-rabbit antibodies (Amersham) were utilized as secondary at 1:5000. The signal was detected with chemiluminescence (Roche or Pierce). Cell Culture and Transfections U2OS cells were obtained from the EACC and maintained in McCoy’s 5A GlutaMAX medium (Invitrogen) supplemented with 10% fetal bovine serum (FBS) for U2OS cells or tetracycline-negative FBS (Brunschwig) for the inducible episomal cell lines (iU2OS). To generate such iU2OS lines, U2OS cells were transfected with pEBTet-HsSAS-6-GFP or pEBTet-HsSAS-6 [F131E]-GFP using Lipofectamine2000 (Invitrogen). Transfected cells were selected with 1 mg/ml puromycin 1 day after transfection and amplified. Early passage cells were used, inducing expression with 1ug/ml doxycycline for 48 hr. Endogenous HsSAS-6 was depleted using a Stealth RNAi siRNA (Invitrogen) targeting the 30 UTR of HsSAS-6 (5-GAGCUGUUAAAGACUGGAUACUUUA-3). Stealth RNAi siRNA negative control LO GC (Invitrogen) was used as a control. siRNA transfection was performed using Lipofectamine RNAiMax (Invitrogen) according to the manufacturer’s protocol, and cells were analyzed 48 hr after siRNA treatment. Cell-Extract Preparation and Biochemical Assays Cells were collected, washed in PBS, and lysed on ice for 30 min in lysis buffer (15 mM Tris-HCl [pH 7.5], 150 mM NaCl, 2.5 mM MgCl2, 0.5% NP-40, 50 mM NaF, and 0.2 mM orthovanadate; Complete Mini Protease Inhibitor Cocktail [Roche Diagnostics]). Lysates were cleared by centrifugation for 15 min at
Cell 144, 364–375, February 4, 2011 ª2011 Elsevier Inc. 373
13,000 3 g at 4 C and the supernatant collected. SDS-PAGE was performed using 4%–15% polyacrylamide gradient gels (BioRad), followed by transfer on nitrocellulose membrane (Amersham). The membrane was probed with mouse HsSAS-6 antibody (Santa Cruz, 1:1000) or rabbit Actin antibody (Abcam, 1:2000), followed by incubation with their respective HRP-conjugated secondary (Promega) and the signal detected with chemiluminescence. Immunofluorescence and Microscopy for Human Cells U2OS cells grown on glass coverslips were fixed for 7–10 min in –20 C methanol, washed in PBS, and blocked in 1% bovine serum albumin and 0.05% Triton X-100 in PBS. Cells were incubated 2 hr at room temperature or overnight at 4 C with primary antibodies, washed three times for 5 min in PBST (0.05% Triton X-100 in PBS), incubated 45 min at room temperature with secondary antibodies, stained with 1 mg/ml Hoechst 33258, washed three times in PBST, and mounted. Primary antibodies were 1:4000 mouse centrin (20H5; gift from Jeffrey L. Salisbury) and 1:500 rabbit GFP (gift from Viesturs Simanis). Secondary antibodies were 1:1000 goat anti-rabbit coupled to Alexa 488 and 1:1000 goat anti-mouse coupled to Alexa 568. For quantification of centrioles, mitotic cells (prophase to metaphase) with similar cytoplasmic GFP expression were used; highly expressing cells that often harbored GFP aggregates were not retained for analysis. Imaging was done on a Zeiss LSM710 confocal microscope. Optical sections were acquired every 0.12 mm, and planes containing centrioles were projected together. Images were processed using ImageJ and Adobe Photoshop, preserving relative image intensities within a series. ACCESSION NUMBERS Coordinates have been deposited in the Protein Data Bank with accession codes 3PYI (ceN), 3Q0Y (crN), and 3Q0X (crCC). SUPPLEMENTAL INFORMATION Supplemental Information includes Extended Experimental Procedures, six figures, three movies, and one table and can be found with this article online at doi:10.1016/j.cell.2011.01.008. ACKNOWLEDGMENTS We are grateful to Fritz Winkler (PSI) for help with the modeling and critical reading of the manuscript; to Vesna Oliveri, Ursula Sauder, and Gianni Morson (ZMB, University of Basel) for excellent support and access to the electron microscope; to Meitian Wang, Andrea Prota, and Daniel Frey (PSI) for help with crystallization and X-ray data collection; to Davide Demuras and Graham Knott (BioEM core facility, EPFL), as well as Petr Leiman (EPFL), for help with EM analysis; to Christine Wandrey (EPFL) for AUC analysis; to Coralie Busso (EPFL) for help in generating the C. elegans transgenic lines; to Diego Chiappe and Marc Moniatte (Proteomic core facility, EPFL) for mass spectrometry; and to Katayoun Afshar, Virginie Hachet, and Joachim Lingner for helpful comments on the manuscript. D.K. held postdoctoral fellowships from the JSPS and EMBO (ALTF-667-2007). I.V. was partly supported by the Wellcome Trust Career Development Fellowship program. M.C.E. was supported by a FP7 Marie Curie Fellowship. This work was supported also by grants to P.G. from Oncosuisse (OCS KLS 02024-02-2007) and from the ERC (AdG 233335) as well as by grants from the Swiss National Foundation to P.G. and M.S. (Sinergia CRSII3_125463). We are also indebted to Clemens Schulze-Briese (PSI) for generous support. Received: December 8, 2010 Revised: January 4, 2011 Accepted: January 5, 2011 Published online: January 27, 2011
Bach, M., Grigat, S., Pawlik, B., Fork, C., Utermo¨hlen, O., Pal, S., Banczyk, D., Lazar, A., Scho¨mig, E., and Gru¨ndemann, D. (2007). Fast set-up of doxycycline-inducible protein expression in human cell lines with a single plasmid based on Epstein-Barr virus replication and the simple tetracycline repressor. FEBS J. 274, 783–790. Carvalho-Santos, Z., Machado, P., Branco, P., Tavares-Cadete, F., Rodrigues-Martins, A., Pereira-Leal, J.B., and Bettencourt-Dias, M. (2010). Stepwise evolution of the centriole-assembly pathway. J. Cell Sci. 123, 1414–1426. Cavalier-Smith, T. (1974). Basal body and flagellar development during the vegetative cell cycle and the sexual cycle of Chlamydomonas reinhardii. J. Cell Sci. 16, 529–556. Culver, B.P., Meehl, J.B., Giddings, T.H., Jr., and Winey, M. (2009). The two SAS-6 homologs in Tetrahymena thermophila have distinct functions in basal body assembly. Mol. Biol. Cell 20, 1865–1877. Dammermann, A., Mu¨ller-Reichert, T., Pelletier, L., Habermann, B., Desai, A., and Oegema, K. (2004). Centriole assembly requires both centriolar and pericentriolar material proteins. Dev. Cell 7, 815–829. Dammermann, A., Maddox, P.S., Desai, A., and Oegema, K. (2008). SAS-4 is recruited to a dynamic structure in newly forming centrioles that is stabilized by the gamma-tubulin-mediated addition of centriolar microtubules. J. Cell Biol. 180, 771–785. Delattre, M., Leidel, S., Wani, K., Baumer, K., Bamat, J., Schnabel, H., Feichtinger, R., Schnabel, R., and Go¨nczy, P. (2004). Centriolar SAS-5 is required for centrosome duplication in C. elegans. Nat. Cell Biol. 6, 656–664. Dirksen, E.R. (1991). Centriole and basal body formation during ciliogenesis revisited. Biol. Cell 72, 31–38. Fowler, W.E., and Aebi, U. (1983). Preparation of single molecules and supramolecular complexes for high-resolution metal shadowing. J. Ultrastruct. Res. 83, 319–334. Gavin, R.H. (1984). In vitro reassembly of basal body components. J. Cell Sci. 66, 147–154. Gopalakrishnan, J., Guichard, P., Smith, A.H., Schwarz, H., Agard, D.A., Marco, S., and Avidor-Reiss, T. (2010). Self-assembling SAS-6 multimer is a core centriole building block. J. Biol. Chem. 285, 8759–8770. Guichard, P., Chre´tien, D., Marco, S., and Tassin, A.M. (2010). Procentriole assembly revealed by cryo-electron tomography. EMBO J. 29, 1565–1572. Hiraki, M., Nakazawa, Y., Kamiya, R., and Hirono, M. (2007). Bld10p constitutes the cartwheel-spoke tip and stabilizes the 9-fold symmetry of the centriole. Curr. Biol. 17, 1778–1783. Junop, M.S., Modesti, M., Guarne´, A., Ghirlando, R., Gellert, M., and Yang, W. (2000). Crystal structure of the Xrcc4 DNA repair protein and implications for end joining. EMBO J. 19, 5962–5970. Kemp, C.A., Kopish, K.R., Zipperlen, P., Ahringer, J., and O’Connell, K.F. (2004). Centrosome maturation and duplication in C. elegans require the coiled-coil protein SPD-2. Dev. Cell 6, 511–523. Khodjakov, A., Rieder, C.L., Sluder, G., Cassels, G., Sibon, O., and Wang, C.L. (2002). De novo formation of centrosomes in vertebrate cells arrested during S phase. J. Cell Biol. 158, 1171–1181. Kilburn, C.L., Pearson, C.G., Romijn, E.P., Meehl, J.B., Giddings, T.H., Jr., Culver, B.P., Yates, J.R., III, and Winey, M. (2007). New Tetrahymena basal body protein components identify basal body domain structure. J. Cell. Biol. 178, 905–912. Kirkham, M., Mu¨ller-Reichert, T., Oegema, K., Grill, S., and Hyman, A.A. (2003). SAS-4 is a C. elegans centriolar protein that controls centrosome size. Cell 112, 575–587.
REFERENCES
Kitagawa, D., Busso, C., Flu¨ckiger, I., and Go¨nczy, P. (2009). Phosphorylation of SAS-6 by ZYG-1 is critical for centriole formation in C. elegans embryos. Dev. Cell 17, 900–907.
Azimzadeh, J., and Marshall, W.F. (2010). Building the centriole. Curr. Biol. 20, R816–R825.
Kleylein-Sohn, J., Westendorf, J., Le Clech, M., Habedanck, R., Stierhof, Y.D., and Nigg, E.A. (2007). Plk4-induced centriole biogenesis in human cells. Dev. Cell 13, 190–202.
374 Cell 144, 364–375, February 4, 2011 ª2011 Elsevier Inc.
Leidel, S., and Go¨nczy, P. (2003). SAS-4 is essential for centrosome duplication in C elegans and is recruited to daughter centrioles once per cell cycle. Dev. Cell 4, 431–439. Leidel, S., Delattre, M., Cerutti, L., Baumer, K., and Go¨nczy, P. (2005). SAS-6 defines a protein family required for centrosome duplication in C. elegans and in human cells. Nat. Cell Biol. 7, 115–125. Loncarek, J., and Khodjakov, A. (2009). Ab ovo or de novo? Mechanisms of centriole duplication. Mol. Cells 27, 135–142. Maeda, I., Kohara, Y., Yamamoto, M., and Sugimoto, A. (2001). Large-scale analysis of gene function in Caenorhabditis elegans by high-throughput RNAi. Curr. Biol. 11, 171–176. Marshall, W.F., and Rosenbaum, J.L. (2000). Are there nucleic acids in the centrosome? Curr. Top. Dev. Biol. 49, 187–205. Nakazawa, Y., Hiraki, M., Kamiya, R., and Hirono, M. (2007). SAS-6 is a cartwheel protein that establishes the 9-fold symmetry of the centriole. Curr. Biol. 17, 2169–2174.
Pelletier, L., O’Toole, E., Schwager, A., Hyman, A.A., and Mu¨ller-Reichert, T. (2006). Centriole assembly in Caenorhabditis elegans. Nature 444, 619–623. Rodrigues-Martins, A., Bettencourt-Dias, M., Riparbelli, M., Ferreira, C., Ferreira, I., Callaini, G., and Glover, D.M. (2007). DSAS-6 organizes a tube-like centriole precursor, and its absence suggests modularity in centriole assembly. Curr. Biol. 17, 1465–1472. Schuck, P. (2000). Size-distribution analysis of macromolecules by sedimentation velocity ultracentrifugation and lamm equation modeling. Biophys. J. 78, 1606–1619. Steinmetz, M.O., Stock, A., Schulthess, T., Landwehr, R., Lustig, A., Faix, J., Gerisch, G., Aebi, U., and Kammerer, R.A. (1998). A distinct 14 residue site triggers coiled-coil formation in cortexillin I. EMBO J. 17, 1883–1891. Stevens, N.R., Roque, H., and Raff, J.W. (2010). DSas-6 and Ana2 coassemble into tubules to promote centriole duplication and engagement. Dev. Cell 19, 913–919.
Nigg, E.A., and Raff, J.W. (2009). Centrioles, centrosomes, and cilia in health and disease. Cell 139, 663–678.
Strnad, P., and Go¨nczy, P. (2008). Mechanisms of procentriole formation. Trends Cell Biol. 18, 389–396.
O’Connell, K.F., Caron, C., Kopish, K.R., Hurd, D.D., Kemphues, K.J., Li, Y., and White, J.G. (2001). The C. elegans zyg-1 gene encodes a regulator of centrosome duplication with distinct maternal and paternal roles in the embryo. Cell 105, 547–558.
Strnad, P., Leidel, S., Vinogradova, T., Euteneuer, U., Khodjakov, A., and Go¨nczy, P. (2007). Regulated HsSAS-6 levels ensure formation of a single procentriole per centriole during the centrosome duplication cycle. Dev. Cell 13, 203–213.
Pelletier, L., Ozlu¨, N., Hannak, E., Cowan, C., Habermann, B., Ruer, M., Mu¨llerReichert, T., and Hyman, A.A. (2004). The Caenorhabditis elegans centrosomal protein SPD-2 is required for both pericentriolar material recruitment and centriole duplication. Curr. Biol. 14, 863–873.
Yabe, T., Ge, X., and Pelegri, F. (2007). The zebrafish maternal-effect gene cellular atoll encodes the centriolar component sas-6 and defects in its paternal function promote whole genome duplication. Dev. Biol. 312, 44–60.
Cell 144, 364–375, February 4, 2011 ª2011 Elsevier Inc. 375
O-GlcNAc Transferase Catalyzes Site-Specific Proteolysis of HCF-1 Francesca Capotosti,1 Sophie Guernier,1 Fabienne Lammers,1 Patrice Waridel,2 Yong Cai,3,5 Jingji Jin,3,5 Joan W. Conaway,3,4 Ronald C. Conaway,3,4 and Winship Herr1,* 1Center
for Integrative Genomics, University of Lausanne, Ge´nopode, 1015 Lausanne, Switzerland Analysis Facility, Center for Integrative Genomics, Faculty of Biology and Medicine, University of Lausanne, 1015 Lausanne, Switzerland 3Stowers Institute for Medical Research, Kansas City, MO 64110, USA 4Department of Biochemistry and Molecular Biology, University of Kansas Medical Center, Kansas City, KS 66160, USA 5Present address: College of Life Sciences, Jilin University, 2699 Qianjin Street, Changchun, Jilin Province 130012, China *Correspondence:
[email protected] DOI 10.1016/j.cell.2010.12.030 2Protein
SUMMARY
The human epigenetic cell-cycle regulator HCF-1 undergoes an unusual proteolytic maturation process resulting in stably associated HCF-1N and HCF-1C subunits that regulate different aspects of the cell cycle. Proteolysis occurs at six centrally located HCF-1PRO-repeat sequences and is important for activation of HCF-1C-subunit functions in M phase progression. We show here that the HCF-1PRO repeat is recognized by O-linked b-Nacetylglucosamine transferase (OGT), which both O-GlcNAcylates the HCF-1N subunit and directly cleaves the HCF-1PRO repeat. Replacement of the HCF-1PRO repeats by a heterologous proteolytic cleavage signal promotes HCF-1 proteolysis but fails to activate HCF-1C-subunit M phase functions. These results reveal an unexpected role of OGT in HCF-1 proteolytic maturation and an unforeseen nexus between OGT-directed O-GlcNAcylation and proteolytic maturation in HCF-1 cell-cycle regulation. INTRODUCTION The informational complexity of organisms can be greatly enhanced by the posttranslational modification of proteins. Modifications that involve the addition of a chemical group, such as phosphorylation, acetylation, methylation, glycosylation, and ubiquitination, are generally reversible, whereas modification by proteolytic processing is considered irreversible. The variety of modification forms and their regulatory outcomes provides diverse strategies for the control of protein function. Among modifications involving the addition of a chemical group, glycosylation is the most abundant (Nalivaeva and Turner, 2001) and, of the different types of glycosylation, the monoaddition of O-linked b-N-acetylglucosamine (O-GlcNAc) moieties on serine and threonine residues is the most important to regulate the function of metazoan cytosolic and nuclear proteins (Torres 376 Cell 144, 376–388, February 4, 2011 ª2011 Elsevier Inc.
and Hart, 1984; Wells and Hart, 2003). The broad presence of O-GlcNAc-modified proteins involved in many fundamental cellular processes, including cell morphogenesis, cell signaling, apoptosis, and transcription, suggests important physiological roles for O-GlcNAcylation (Lazarus et al., 2006). O-GlcNAcylation exhibits similarities to phosphorylation, such as common target residues (i.e., serines and threonines), proteins, and dynamic nature (Hart et al., 2007; Wells and Hart, 2003; Wells et al., 2001), but differs in that O-GlcNAcylation is achieved by a single cellular enzyme: O-GlcNAc transferase (OGT) (Haltiwanger et al., 1992). OGT is a ubiquitously expressed and conserved enzyme that uses the donor substrate UDP-GlcNAc for glycosylation. It is composed of two important regions: an N-terminal 13.5 tetratricopeptide-repeat (TPR) superhelical structure (Jinek et al., 2004), involved in substrate recognition (Lubas and Hanover, 2000), and a C-terminal catalytic domain (Kreppel and Hart, 1999). The catalytic domain contains the UDP-GlcNAc binding site but the precise mechanism of O-GlcNAc transfer to substrates is not known (Clarke et al., 2008; Martinez-Fleites et al., 2008). OGT forms stable associations with numerous proteins, such as the histone acetyltransferase MOF (Cai et al., 2010), the protein phosphatase PP1 (Wells et al., 2004), the transcriptional corepressor Sin3A (Yang et al., 2002), and the cell-cycle regulator HCF-1 (Wysocka et al., 2003; Mazars et al., 2010), the subject of this study. HCF-1 is a transcriptional coregulator conserved among animal species that was first discovered as a host-cell factor for human herpes simplex virus infection (Kristie et al., 2010; Wysocka and Herr, 2003). In vertebrates, HCF-1 undergoes an unusual process of proteolytic maturation. It is synthesized as a large precursor protein that is subsequently cleaved at a series of six centrally located 26 amino acid repeats called HCF-1PRO (PRO for proteolysis) repeats. The resulting heterogeneous collection of N- (HCF-1N) and C- (HCF-1C) terminal subunits remains noncovalently associated (Kristie et al., 1995; Wilson et al., 1993, 1995). An important cellular function of human HCF-1 is the regulation of the cell cycle: the HCF-1N subunit promotes passage through the G1 phase (Goto et al., 1997; Julien and Herr, 2003; Tyagi and Herr, 2009) and the HCF-1C
A
HCF-1KEL
Basic
HCF-1PRO
Acidic Fn3
HCF-1 HCF-1PRO rep1 HCF-1PRO rep2 HCF-1PRO rep3 HCF-1PRO rep4 HCF-1PRO rep5 HCF-1PRO rep6
H. sapiens
L R H Q R R
V V G V V V
C C C C C C
S S S S S S
N N N N N N
P P P P P P
P P P P P P
C C C C C C
E E E E E E
T T T T T T
H H H H H H
E E E E E E
T T T T T T
G G G G G G
T T T T T T
T T T T T T
N N N N H H
T T T T T T
A A A A A A
T T T T T T
T T T T T T
T A A S A V
V T M N T T
V S S A S S
A N S G N N
Probability
B
T V Q T Q Q
N LS NLS 2035
C
Cleavage region
Threonine region
A20 HCF-1PRO rep2:
WT V R V C
S N P P C E T H E
T G T T N T
A23 T T
T S N
8 9 10 11 12 13 14 15 16 17 18 19 21 22
24 25 26
–
0 1
2
3
4
5
6
7
In vitro data: In vivo data: = strong cleavage inhibition
= intermediate cleavage inhibition
Figure 1. Sequence and In Vitro Mutational Analysis of HCF-1PRO Repeats (A) Schematic representation of the human HCF-1 protein and sequence alignment of the six human HCF-1PRO repeats. The black arrowhead indicates the cleavage site. (B) HCF-1PRO-repeat sequence conservation. WebLogo (Crooks et al., 2004) was obtained by the alignment of all HCF-1PRO repeats from six different species: human, mouse, Xenopus tropicalis, X. laevis, Fugu rubripes, and Danio rerio. (C) In vitro alanine-scan mutagenesis. The wild-type HCF-1PROrep2 precursor (lane 0) as well as individual HCF-1PROrep2 alanine mutants (lanes 1–19, 21, 22, and 24–26) were incubated with HeLa nuclear extract. The gray arrowheads indicate the positions of the two HCF-1PROrep2 alanines that were not mutated. The precursor (–) and the visible N-terminal cleavage product (d) are indicated. Colored dots indicate mutations with strong (blue) or intermediate (cyan) effects on either in vitro or in vivo (Wilson et al., 1995) cleavage.
subunit ensures proper mitosis and cytokinesis in M phase (Julien and Herr, 2003, 2004). Interestingly, HCF-1 proteolytic maturation is necessary to activate human HCF-1C-subunit function in M phase (Julien and Herr, 2003). Curiously, the mechanism of HCF-1 proteolytic maturation has diverged in evolution. Whereas HCF-1PRO-repeat-dependent HCF-1 maturation is apparently shared by, and restricted to, all vertebrate species, in some invertebrate species (e.g., Drosophila) HCF-1 homologs also undergo proteolytic maturation but by a different mechanism involving the protease Taspase1 (Capotosti et al., 2007). In vertebrates, Taspase1 is responsible for proteolytic maturation of the mixed-lineage leukemia (MLL) histone methyltransferase protein (Hsieh et al., 2003) but not of HCF-1 (Capotosti et al., 2007). How HCF-1PRO-repeat-dependent proteolytic maturation is achieved
remains to be clarified, as evidence supporting both autocatalytic (Vogel and Kristie, 2000) and nonautocatalytic (Capotosti et al., 2007) mechanisms has been reported. We demonstrate here a connection between HCF-1PROrepeat-induced HCF-1 proteolytic maturation and OGT, providing an unexpected link between reversible and irreversible posttranslational modification pathways to control protein function. RESULTS The HCF-1PRO Repeat Represents an Elaborate Cleavage Signal Figure 1A shows the close similarity among the six human HCF-1PRO repeats and the site of proteolytic cleavage, a glutamic acid at position 10 (E10) (arrowhead; Kristie et al., 1995; Wilson Cell 144, 376–388, February 4, 2011 ª2011 Elsevier Inc. 377
po r in e
B
al lo xa st n au ro s
A
NE: – HCF-1rep: 1 23 4 5 6
+
+
+
2
3
4
-
HCF-1 HCF-1rep123
1
C
D HCF-1rep123
siRNA: A: – Luc OGT
siRNA:
–
HCF-1repXXX
Luc OGT
–
Luc
OGT
M
OGT T
–130 – 95 – 72
αTub..
– 55
* *
-
M
– 95 – 72
*
– 55
* *
1
2
3
1
2
3
4
5
6
Figure 2. O-GlcNAcylation Is Important for HCF-1PRO-Repeat Cleavage (A) Schematic representation of full-length HCF-1 and HCF-1rep123. (B) Inhibition of HCF-1PRO-repeat cleavage. HCF-1rep123 synthesized in RRL was incubated with buffer (lane 1), HeLa nuclear extract (NE; lane 2), or NE pretreated with alloxan (lane 3) or staurosporine (lane 4). (C) OGT depletion by siRNA. 293 cells were either mock treated (lane 1) or treated with luciferase- (Luc; lane 2) or OGT- (lane 3) specific siRNAs. OGT and a-tubulin (aTub.) levels were visualized by immunoblot. (D) Effect of OGT depletion on HCF-1PRO-repeat cleavage. Cells treated as in (C) were transiently transfected with the HCF-rep123 (lanes 1–3) and HCF-1repXXX (lanes 4–6) expression constructs. Precursor cleavage was analyzed by a-HA tag immunoblot. The bracket indicates the O-GlcNAc-modified HCF-1rep123 precursor. Empty and filled arrowheads indicate the O-GlcNAc-modified HCF-1rep123 precursor and HCF-1PRO-repeat 1 cleavage product, respectively. Asterisks indicate nonspecific bands. In (B) and (D), precursor proteins (–) and cleavage products (d) are indicated.
et al., 1993, 1995). A sequence logo generated from a multiple vertebrate species alignment highlights the extensive conservation of the HCF-1PRO repeats (Figure 1B). For a protease recognition sequence, the HCF-1PRO repeat is large. Indeed, in vivo assay of a complete set of HCF-1PROrepeat 2 alanine-scan mutants revealed an 18 amino acid core required for full cleavage activity (Wilson et al., 1995). Using a previously described human HeLa cell extract assay (Capotosti et al., 2007), we tested cleavage of the same set of alanine-scan mutants in vitro, as shown in Figure 1C. The effect of the alanine substitutions was very similar to what was previously observed in vivo (Figure 1C), indicating that the in vitro assay faithfully reflects normal HCF-1 proteolytic maturation. The mutational analysis revealed a C-terminal HCF-1PRO-repeat region with a number of conserved and functionally important threonine residues. We refer to this region as the ‘‘threonine region’’ and the sequences surrounding the cleavage site as the ‘‘cleavage region’’ (see Figure 1C). 378 Cell 144, 376–388, February 4, 2011 ª2011 Elsevier Inc.
A Role for OGT in HCF-1PRO-Repeat Proteolysis In Vitro and In Vivo Three observations led us to ask whether protein modification might be important for HCF-1PRO-repeat cleavage: (1) the resistance of the HeLa cell proteolytic activity to fractionation, suggesting multiple components (F.C. and W.H., unpublished results); (2) a retarded electrophoretic mobility of the uncleaved HCF-1PRO-repeat precursor after in vitro cleavage reaction (Capotosti et al., 2007), a property often indicative of posttranslational modification; and (3) the critical threonine region given that threonines can be sites for phosphorylation and O-GlcNAcylation. We therefore tested whether a general inhibitor of phosphorylation (staurosporine) or of O-GlcNAcylation (alloxan) affects processing in vitro of the HCF-1-derived precursor called HCF-1rep123 (Figure 2A). As shown in Figure 2B, alloxan, but not staurosporine, reduced HCF-1rep123 cleavage (compare lanes 3 and 4 with lane 2), suggesting a relationship between O-GlcNAcylation and HCF-1PRO-repeat
cleavage and a possible role of OGT in HCF-1 proteolytic maturation. To test this hypothesis, we asked whether depletion of OGT in vivo affects HCF-1rep123 processing. OGT can be significantly depleted after 48 hr of human 293 cell treatment with an OGTspecific (lane 3) but not an irrelevant (lane 2) siRNA duplex (Figure 2C). We assayed the effect of OGT depletion on cleavage of the wild-type HCF-1rep123 precursor and an HCF-1repXXX mutant with alanine-substituted E10 cleavage-site residues (E10A), synthesized during the last 24 hr of 48 hr siRNA treatment. HCF-1repXXX was not affected by OGT depletion (Figure 2D, lanes 4–6). In contrast, with HCF-1rep123, OGT depletion caused a specific (compare lanes 1–3) reduction of the HCF-1PRO-repeat 1 cleavage product (filled arrowhead) and of a slower migrating— possibly modified—form of the precursor (open arrowhead). (Note that the products of HCF-1PRO-repeat 2 and 3 cleavage are obscured by other species in this experiment.) Together, these results suggest a role for OGT and possibly O-GlcNAcylation in proteolytic processing of HCF-1. The HCF-1rep123 Precursor Is O-GlcNAcylated in an HCF-1PRO-Repeat-Dependent Manner To test for a link between O-GlcNAcylation and proteolysis, we assayed glycosylation of HCF-1rep123 and HCF-1repXXX, together with the HCF-1PRO-repeat 1-containing HCF-1rep1XX protein, in vivo. After synthesis in 293 cells, a-HA tag immunopurified recombinant proteins (Figure 3Aa) were probed for O-GlcNAcylation by immunoblot (Figure 3Ab). Consistent with HCF-1PRO-repeat-dependent O-GlcNAcylation, the HCF1rep123 and HCF-1rep1XX, but not HCF-1repXXX, precursors and cleavage products were readily recognized by the a-O-GlcNAc antibody (Figure 3Ab; lanes 1–3). The slower migration of the precursor observed here and in Figure 2D (see brackets) probably results from the O-GlcNAcylation as opposed to phosphorylation, as it was insensitive to phosphatase treatment (see Figure S1 available online). The HCF-1PROrepeat dependence of HCF-1rep123 O-GlcNAcylation was unexpected because we had hypothesized that glycosylation might activate HCF-1rep123 proteolysis, not that the HCF-1PRO repeat might activate HCF-1rep123 glycosylation. HCF-1PRO Repeats Can Induce Multiple O-GlcNAc Modifications in the HCF-1N Subunit HCF-1 is O-GlcNAcylated (Wilson et al., 1993; Wysocka et al., 2003) at multiple sites (Mazars et al., 2010; Wang et al., 2010). Here we mapped O-GlcNAcylation sites in purified HCF-1rep123 and HCF-1repXXX precursors after in vivo synthesis. The HCF-1rep123 material (Figure 3B, lane 1) produced a doublet of faster- (a) and slower- (a0 ) migrating forms upon silver staining. In contrast, with HCF-1repXXX, the faster-migrating form (b) predominated (lane 2). These three polypeptides were individually analyzed by tandem mass spectrometry, and glycosylated and nonglycosylated peptides were identified and quantified (see Table S1). Figure 3C shows the results for one peptide (HCF-1 residues 724–770). This peptide from samples a and b was predominantly unmodified, whereas from sample a0 it was predominantly O-GlcNAcylated, consistent with HCF-1PROrepeat-dependent HCF-1 O-GlcNAcylation.
Figure 3D shows the results of our HCF-1rep123 O-GlcNAc modification mapping, together with sites mapped by Wang et al. (2010) within this region. Interestingly, the mapped sites exclude the HCF-1PRO-repeats but include extensive N-terminal O-GlcNAc modifications. Thus, counter to our aforementioned rationale, the HCF-1PRO-repeat threonine region is apparently not O-GlcNAc modified. To obtain a global appreciation of HCF-1 O-GlcNAcylation, we analyzed O-GlcNAcylation of endogenous HCF-1, as shown in Figure 3E. Immunopurified endogenous HCF-1 was subjected to immunoblot analysis with antibodies specific to either the HCF-1C or HCF-1N subunit, or O-GlcNAc (lanes 1–3). The resulting O-GlcNAc staining pattern matches the HCF-1N but not the HCF-1C (lane 4)-subunit profile, suggesting that HCF-1 O-GlcNAcylation is largely limited to the HCF-1N subunit (see also Mazars et al., 2010; Wang et al., 2010). OGT Recognizes the HCF-1PRO Repeat The dependence of HCF-1 O-GlcNAcylation on active HCF-1PRO repeats led us to ask whether the HCF-1PRO repeat also promotes OGT binding to HCF-1. Analysis of the HCF-1rep123 immunoprecipitates from Figure 3A (see also Figure 4Aa) showed that OGT—present in the original extracts (Figure 4Ab)—was present in the HCF-1rep123 but not mock samples (Figure 4Ac, compare lanes 1 and 4). Indeed, OGT bound even better to the HCF-1rep1XX and HCF-1repXXX precursors carrying the E10A cleavage-site mutation (lanes 2 and 3). Thus, curiously, although inhibiting HCF-1PRO-repeat cleavage and HCF-1rep123 glycosylation, the E10A mutation enhances OGT association with the HCF-1PRO-repeat precursor. These results led us to ask whether OGT association with HCF-1 can be mediated by the HCF-1PRO repeat. For this, we tested OGT association with full-length HCF-1 and the four HCF-1 mutants illustrated in Figure 4B: (1) HCF-1PROMUT carrying the E10A cleavage-site mutation in all six HCF-1PRO repeats (the HCF-1nc mutant of Vogel and Kristie, 2006); (2) HCF-1DPRO, lacking the HCF-1PRO-repeat region (Wilson et al., 1995), and (3) HCF-1DPROrep2 and (4) HCF-1DPROrep2E10A carrying a single wild-type or mutated HCF-1PRO repeat 2. Figure 4C shows the (1) levels of wild-type and mutant HCF-1 synthesis after transient expression (Figure 4Ca), (2) levels of endogenous OGT (Figure 4Cb), and (3) levels of OGT coprecipitated with the various HCF-1 proteins (Figure 4Cc). As with the HCF-1rep123 proteins, OGT was recovered with full-length HCF-1 and significantly better with HCF-1PROMUT (lanes 2 and 3). Furthermore, the E10A HCF-1DPROrep2E10A mutant bound OGT better than its wildtype HCF-1DPROrep2 counterpart (lanes 5 and 6). In contrast, there was no evident OGT association in the absence of an HCF-1PRO repeat (see HCF-1DPRO; lane 4). These results indicate that the HCF-1PRO-repeat sequence itself is the target for OGT binding and that the E10A mutation improves binding. Threonine-Region Mutations Disrupt OGT Association with HCF-1PRO Repeat 2 The E10A HCF-1PRO-repeat substitution lies within the HCF1PRO-repeat cleavage region. To test the effect of mutations in the threonine region, we assayed endogenous OGT association with wild-type or mutated HCF-1PRO repeats (repeat 2) Cell 144, 376–388, February 4, 2011 ª2011 Elsevier Inc. 379
a' a
M
a
-
B –72 –55
α-HA
HCF-1repXXX
HCF-1rep123
Mo Mock
HCF-1rep1XX HC
HC HCF-1repXXX
B HCF-1rep123 HC
A
b 1
2
C Peptide 724-770 724-770: LVTSADGKPTTIITTTQASGA LVTSADGKPTTIITTTQASGAGT..... .....KPTILGISSVSPSTTKPGTTTIIK .....KPTILGISSVSPSTTKPGT
–36
-
b α-O-GlcNAc
*
–72
O-GlcNAc
a
–55
0
35%
7%
70%
1
26%
15%
11%
2
16%
28%
10%
3
23%
50%
9%
–36
S1150
T887
T881 T861
b
T885
4
S891
3
T877 T878
T694
T726
D
2
S727 T739 T765 T T760 T771 T779 T800 S806 T 808
1
a'
Peptide coverage:
E
α-HCF-1c α-HCF-1 α-HCF-1c α -HCF-1N α -HCF-1c (red) Immunoblot: α-HCF-1c (N13) α α-O-GlcNAc -O-GlcNAc O α-O-GlcNAc α -O-GlcNAc O (gre (green) M (H12) 250– 130– 95 –
1
2
3
4
α-HCF-1c α-HCF-1 1c IP (C18) H12 HCF-1 N13
C18
Figure 3. Analysis of HCF-1rep123 O-GlcNAcylation (A) HCF-1PRO-repeat-dependent HCF-1rep123 O-GlcNAcylation. 293 cells were mock transfected (lane 4), or transfected with different HCF-1rep123 expression vectors (lanes 1–3; see Figure 2A). a-HA immunoprecipitated proteins were visualized by a-HA tag (panel a) or a-O-GlcNAc (panel b) immunoblot. Precursor proteins (–), cleavage products (d), and O-GlcNAc-modified proteins (bracket) are indicated. The asterisk indicates an immunoglobulin heavy chain. See also Figure S1. (B) Purification of HCF-1PRO-repeat precursors. HCF-1rep123 (lane 1) and HCF-1repXXX (lane 2) were synthesized in 293 cells, immunoprecipitated, and visualized by silver staining. Different HCF-1rep123 (a and a0 ) and HCF-1repXXX (b) forms are indicated. (C) Relative abundance of peptides corresponding to HCF-1 residues 724–770 with 0, 1, 2, or 3 O-GlcNAc moieties attached in samples a, a0 , and b from (B) (gray boxes indicate the predominant peptide species for each sample; see Table S1 for the complete list of O-GlcNAcylated peptides). (D) Schematic representation of mapped HCF-1rep123 O-GlcNAc sites. HCF-1rep123 regions covered by the MS analysis are indicated below the diagram. Filled and empty squares indicate certain or potential O-GlcNAcylation residues, respectively. Yellow dots indicate Wang et al. (2010)-identified O-GlcNAcylated residues. (E) Endogenous HCF-1 O-GlcNAcylation. HCF-1 was immunoprecipitated from 293 cells with the C18 a-HCF-1 antibody and visualized by H12 (lane 1), N13 (lane 2), a-O-GlcNAc (lane 3), or combined H12 and a-O-GlcNAc (lane 4) immunoblot. Regions recognized by the different a-HCF-1 antibodies are indicated below.
380 Cell 144, 376–388, February 4, 2011 ª2011 Elsevier Inc.
a
Mock
B
M
-
–72 –55
α-HA
HA IP
HCF-1repXXX
HCF-1rep1XX
HCF-1rep123
A
HCF-1WT HCF-1PROMUT
–36 b
c HA IP
–130 –95 1
2
3
HCF-1ΔPRO α-OGT
–130 –95
TI
HCF-1ΔPRO rep2 HCF-1ΔPRO rep2 E10A
4
–130 –95
HA IP 1
2
3
4
5
6
E10A/T17-22A
T17-22A
E10A
WT
–
ΔPRO rep2
ΔPRO
PROMUT
WT
c
b
c
c
α-OGT
–130 –95
TI
b
M
–130
HA IP
–130
TI
–130
HA IP 1
2
3
4
α-OGT
–250 –130
b
a
M
PROrep2:
α-HA
–
HCF-1
α-HA
HCF-1: a – HA IP
ΔPRO rep2 E10A
D
C
5
7
Figure 4. Interaction between OGT and the HCF-1PRO Repeats (A) OGT coimmunoprecipitation with wild-type and mutant HCF-1PRO-repeat constructs. Immunoprecipitated samples from Figure 3A (and reshown in panel a) were probed for OGT coimmunoprecipitation as described below. IP, immunoprecipitate; TI, total input. (B) Schematic representation of different HCF-1PRO-repeat mutants. Dots indicate E10A-mutated HCF-1PRO repeats. (C) OGT interaction with HCF-1PRO-repeat mutants. 293 cells were mock transfected (lane 1) or transfected with the different HCF-1PRO-repeat mutant constructs (lanes 2–6). Unlabeled bands represent nonspecific species. In (A) and (C), precursor proteins (–) and cleavage products (d) are indicated. (D) Effect of mutations in HCF-1PRO repeat 2 on interaction with OGT. 293 cells were mock transfected (lane 1), or transfected with a wild-type (lane 2) or mutated (lanes 3–5) Oct-1/rep2 precursor with the indicated HCF-1PROrep2. In (A), (C), and (D), immunoprecipitated samples were visualized by a-HA tag (panel a) and a-OGT (panel c) immunoblot. Total input (TI) samples visualized by a-OGT immunoblot are shown in panel b.
embedded in the heterologous protein Oct-1 (Oct-1/rep2; Wilson et al., 1995) as shown in Figure 4D. In this assay, wildtype HCF-1PRO-repeat 2–OGT interaction was not detectable (lane 2), but the E10A mutation increased OGT association sufficiently to make it clearly detectable (lane 3). In contrast, alanine substitution of four essential threonines (T17, T19, T21, and T22) in the threonine region (T17–22 mutation) did not increase OGT HCF-1PRO-repeat binding (lane 4), and indeed inhibited binding when combined as a double mutant with the E10A mutation (compare lanes 3 and 5). Thus, OGT association with the HCF-1PRO repeat is influenced by mutations in both the cleavage and threonine regions. OGT Cleaves the HCF-1PRO Repeat In Vitro The aforementioned results suggest that OGT has an intimate role in HCF-1PRO-repeat cleavage. To define this role further,
we began by testing the ability of recombinant human OGT (rhOGT) purified after synthesis in either insect Sf9 cells (Inssynt) or Escherichia coli (Bacsynt) (see Figures S2A and S2B) to induce in vitro cleavage of the HCF-1rep123 precursor in the presence of the OGT cofactor UDP-GlcNAc. As shown in Figure 5A (lanes 1–3), both these rhOGT preparations can cleave the HCF-1rep123 precursor synthesized by in vitro translation in a rabbit reticulocyte lysate (RRL). (The HCF-1rep123 fragment mobility difference in lanes 2 and 3 probably reflects the greater O-GlcNAc transferase activity of the Inssynt versus Bacsynt rhOGT) (Figure S2C). In the RRL, the HCF-1rep123 precursor is O-GlcNAcylated during synthesis (data not shown). To avoid this glycosylation, we synthesized the precursor in a plant wheat germ extract (WGE) which lacks OGT activity (Starr and Hanover, 1990). This precursor was also effectively cleaved by the rhOGT preparations (compare lanes 4–6 and 1–3), showing that Cell 144, 376–388, February 4, 2011 ª2011 Elsevier Inc. 381
C
HCF-1rep1 M
synt
72–
∗
36 –
2
3
4
5
6
D
– ∗
72– 55 –
–
–
55 –
1
72– 55 –
E10A
–
Bac
WT
rhOGT : –
Ins synt
Ins Bac – synt synt y synt y y synt y Ins Bac
+ Pefabloc
–
Bac synt
WGE
RRL rhOGT :
+ ST078925
GST
– OGT
HCF-1rep1
HCF-1rep123
Figure 5. Recombinant OGT Is Sufficient to Induce HCF-1PRO-Repeat Cleavage In Vitro
HCF-1rep1
+ alloxan
B
– UDP-GlcNAc
A
1
2
1
Cleavage region wt V R V C S N P P C
2
3
4
5
6
7
Threonine region A20 A23
E T H E T G T T
N T T T T S N
–
HCF-1PRO rep2: 0 1 2 3
4 5 6
7
8
9
10 11 12 13 14 15 16 17 18 19 21 22 24 25 26
OGT: NE:
TPR
CD
N458 R637 D925
R sO G T
TP
8A 45
7A 63
R
92
T
D
W
M
flOGT
– 130
sOGT
– 72
TPR
– 55
∗
– 130 – 72
sOGT
– 55
CKII
– 36
c
–
1
2
3
4
5
precursor O-GlcNAcylation prior to the cleavage reaction is not required for HCF-1PRO-repeat cleavage. These results led us to test for HCF-1PRO-repeat cleavage by OGT directly. To this end, we used an HCF-1PRO-repeat substrate and OGT both synthesized in and purified from bacteria. Owing to the insolubility of the HCF-1rep123 precursor synthesized in bacteria, we designed the smaller HCF-1 precursor illustrated in Figure 5B spanning a single HCF-1PRO repeat 1 (either wild-type or E10A mutant) fused to GST (GST-HCF-1rep1). As shown in Figure 5B, Bacsynt OGT cleaved (as well as O-GlcNAcylated; Figure S2D) the wild-type GST-HCF-1rep1 precursor, showing that OGT itself induces HCF-1PRO-repeat cleavage. OGT Glycosylation Activity Is Important for HCF-1PRO-Repeat Cleavage To probe the importance of the O-GlcNAcylation activity of OGT for HCF-1PRO-repeat proteolysis, we determined the 382 Cell 144, 376–388, February 4, 2011 ª2011 Elsevier Inc.
6
7
α-O-GlcNAc HCF-1rep123
b
α-OGT
flOGT
rhOGT:
–
a
5A
F
E
TPR sOGT
= intermediate cleavage inhibition
N
= strong cleavage inhibition
(A) Recombinant human OGT (rhOGT) cleavage of in vitro synthesized HCF-1PRO-repeat precursors. The HCF1rep123 precursor was synthesized in vitro using RRL (lanes 1–3) or WGE (lanes 4–6) and incubated with buffer (lanes 1 and 4), or rhOGT synthesized and purified from insect cells (Inssynt; lanes 2 and 5) or from bacterial cells (Bacsynt; lanes 3 and 6). See also Figures S2A–S2C. (B) rhOGT cleavage of a bacterially synthesized HCF1PRO-repeat precursor. The GST-HCF-1rep1 precursor (see schematic) synthesized in and purified from bacteria and incubated without (lane 1) or with (lane 2) Bacsynt rhOGT. HCF-1PRO-repeat 1 cleavage was analyzed by a-GST immunoblot. The asterisk indicates an N-terminal precursor truncation and the bracket indicates the O-GlcNAc-modified form of the precursor. See also Figure S2D. (C) rhOGT HCF-1PRO-repeat cleavage characterization. GST-HCF-1rep1 precursors containing wild-type (WT) or E10A-mutated HCF-1PRO repeat 1 were subjected to in vitro cleavage in the presence (+) or absence (–) of the indicated factors. The asterisk indicates an N-terminal precursor truncation. (D) rhOGT cleavage of HCF-1PRO-repeat 2 alanine-scan mutants. The analysis was performed as in Figure 1C using Bacsynt rhOGT and WGE synthesized precursors. Dots indicate mutations with strong (blue) and intermediate (cyan) effects on cleavage. The results with HeLa cell nuclear extract (NE) from Figure 1C are shown for reference. (E) Schematic representation of full-length (flOGT), truncated, and alanine-substitution rhOGT constructs. (F) The wild-type and indicated OGT mutants were synthesized and purified from E. coli and then visualized by a-OGT immunoblot (panel a). The O-GlcNAc transferase activity of rhOGT molecules was determined by OGT self-glycosylation and CKII O-GlcNAcylation using the a-O-GlcNAc antibody (panel b). Proteolytic activity was determined by HCF-1rep123 in vitro cleavage (panel c). The asterisk indicates a nonspecific band. In (A)–(D) and (F), the precursor proteins (–) and cleavage products (d) are indicated.
dependence of proteolysis on the UDP-GlcNAc cofactor and its sensitivity to an OGT inhibitor. As shown in Figure 5C, both Bacsynt (lane 1) and Inssynt (lane 7) OGT were active on the bacterially synthesized wild-type (top), but not E10A point mutant (bottom), HCF-1rep1 precursor, but removal of either UDP-GlcNAc or OGT prevented cleavage (top, lanes 2 and 3). The requirement for two components—OGT and its UDPGlcNAc cofactor—for HCF-1PRO-repeat proteolysis explains the resistance of the HeLa cell HCF-1PRO-repeat proteolytic activity to fractionation. Alloxan, used to inhibit O-GlcNAcylation in Figure 2B, is a relatively nonspecific inhibitor that apparently mimics the uracil moiety of the UDP-GlcNAc cofactor (Konrad et al., 2002). Here, to inactivate OGT glycosylation, we additionally used the more specific inhibitor ST078925 (Gross et al., 2005), which also inhibited HCF-1PRO-repeat proteolysis (compare lane 1 with lanes 4 and 5). In contrast, as described previously for the HeLa cell activity (Capotosti et al., 2007), OGT-induced
HCF-1PRO-repeat proteolysis is resistant to the serine protease inhibitor Pefabloc (lane 6), showing that it differs from the HCF-1 autocatalytic activity described by Vogel and Kristie (2000).
A
sHCF-1ΔPROrep2 rhOGT: –
0' 30' 1h 2h 4h 8h
–
OGT HCF-1PRO-Repeat Cleavage Is Dependent on the Complete HCF-1PRO-Repeat Signal In Figure 5D, we show that the Bacsynt rhOGT cleavage pattern of the set of 24 HCF-1PRO-repeat 2 alanine-scan mutants synthesized in WGE is essentially the same as that with the HeLa cell nuclear extract (compare with Figure 1C). Thus, OGT itself appears to recognize the large HCF-1PRO-repeat signal and to reproduce entirely the HeLa cell proteolytic activity.
1
2
3
4
5
6
7
B Both the TPR and Catalytic Domains of OGT Are Important for HCF-1PRO-Repeat Proteolysis To identify elements of OGT required for HCF-1PRO-repeat proteolysis, we assayed the activities of a set of Bacsynt rhOGT mutants previously characterized for O-GlcNAc transferase activity on either heterologous substrates or for self-glycosylation (Clarke et al., 2008; Gross et al., 2005; Hanover et al., 2003; Martinez-Fleites et al., 2008). As shown in Figure 5E, we analyzed two truncations: one containing the TPR region and one corresponding to an OGT splice variant (sOGT) lacking most of the TPR region (Hanover et al., 2003). We also analyzed three OGT alanine-substitution mutants (Clarke et al., 2008): N458A between the TPR and catalytic domain (CD); R637A in the CD; and D925A in the UDP-GlcNAc binding pocket. Bacsynt wild-type and mutant rhOGT proteins (Figure 5Fa) were tested for their O-GlcNAc transferase activity on the heterologous substrate casein kinase II (CKII) as well as by self-glycosylation (Figure 5Fb). Only the wild-type and N458A mutant displayed evident CKII O-GlcNAc transferase activity (Figure 5Fb, lanes 2 and 5). Consistent with its reported activities (Hanover et al., 2003; Clarke et al., 2008), the sOGT protein did not exhibit CKII O-GlcNAc transferase activity but displayed enhanced self-glycosylation activity (Figures 5Fa and 5Fb, compare lanes 7 and 2), whereas the TPR mutant was inactive. Among the set of rhOGT mutants, there was a perfect correlation between the ability to O-GlcNAcylate the heterologous CKII substrate and to induce HCF-1rep123 proteolysis (Figure 5Fc). In contrast, there was not such a correlation with self-glycosylation activity, as the sOGT form did not display HCF-1rep123 proteolytic activity (Figures 5Fb and 5Fc, lane 7). These results suggest (1) that both the TPR and CD regions of OGT are required for HCF-1PRO-repeat proteolysis and (2) that residues involved in O-GlcNAc transferase activity are important for HCF-1PROrepeat proteolysis. OGT-Induced O-GlcNAcylation Is Not Sufficient to Induce HCF-1PRO-Repeat Proteolysis Although OGT appears to induce HCF-1PRO-repeat proteolysis on its own, the aforementioned studies do not discriminate between whether (1) OGT O-GlcNAcylation of HCF-1 induces HCF-1PRO-repeat autoproteolysis or (2) OGT possesses a previously unrecognized site-specific proteolytic activity that can induce cleavage of the HCF-1PRO repeats either entirely on its own or in collaboration with the HCF-1PRO repeat.
OGT +/– Alx Addition: 0h
1h
8h sHCF-1ΔPROrep2
End reaction (hr):
8
1
8
8
8
Addition of Alx (hr): OGT:
– –
– +
1 – + +
0 +
–
1
2
3
4
5
Figure 6. The O-GlcNAc Modification of the HCF-1PRO-Repeat Precursor Is Not Sufficient to Induce Proteolysis (A) Time course of sHCF-1DPROrep2 precursor treatment with Bacsynt rhOGT. (B) OGT inhibition during sHCF-1DPROrep2 cleavage. The sHCF-1DPROrep2 precursor was treated as indicated in the schematic experimental flow chart (top). The sHCF-1DPROrep2 precursor synthesized in WGE was incubated with Bacsynt rhOGT without (lanes 2 and 4) or with (lanes 3 and 5) addition of alloxan (Alx) and terminated at the times indicated. Precursor protein (–), both cleavage products (d), and O-GlcNAc-modified proteins (brackets) are indicated. See also Figure S3.
To discriminate between these two possibilities, we developed an assay in which glycosylation and proteolysis could be clearly monitored simultaneously. This assay takes advantage of a ‘‘short’’ version, called sHCF-1DPROrep2, of the HCF-1DPROrep2 precursor shown in Figure 4B; it exhibits a clear electrophoretic mobility shift upon O-GlcNAcylation (see Figure S3). As shown in Figure 6A, with this precursor, within a 30 min incubation with the Bacsynt rhOGT (lane 3), a retarded O-GlcNAcylated form is apparent (indicated by the bracket). This O-GlcNAc-modified precursor increased for up to 2 hr and then slowly disappeared as two cleavage products accumulated—a slower-migrating N-terminal and faster-migrating C-terminal cleavage product (indicated by dots). We used the clear appearance of both O-GlcNAcylated precursor and Cell 144, 376–388, February 4, 2011 ª2011 Elsevier Inc. 383
cleavage products to probe the direct or indirect role of OGT in HCF-1PRO-repeat proteolysis. We used alloxan to block OGT activity at different time points and then examined the effects on HCF-1PRO-repeat proteolysis. We initiated a series of sHCF-1DPROrep2 cleavage reactions by addition of OGT (see Figure 6B, top). OGT was then inactivated by addition of alloxan at either the start of the reaction (0 hr), which prevented its function throughout the experiment (lane 5), or after 1 hr, at which point there was initial sHCF-1DPROrep2 precursor O-GlcNAcylation but little cleavage (lane 3). Two control reactions were terminated at 1 hr (lane 2) or 8 hr (lane 4) without the addition of alloxan. These control reactions showed that, after 1 hr, the O-GlcNAcylated precursor and some cleavage products appeared (compare lanes 2 and 1) and, after 8 hr, the cleavage products became more abundant whereas the O-GlcNAcylated precursor did not (lane 4). Addition of alloxan after 1 hr (lane 3) resulted in arresting the reaction such that after 8 hr the O-GlcNAcylation and cleavage pattern were essentially identical to that after 1 hr (compare lanes 2–4). Thus, there is no apparent cleavage of the O-GlcNAcylated sHCF-1DPROrep2 precursor in the absence of OGT function. These results suggest that the O-GlcNAcylated HCF-1 precursor is not potentiated for independent autoproteolysis. We conclude that OGT possesses a proteolytic activity which may be wholly contained within OGT or function in intimate collaboration with the HCF-1PRO repeat (see Discussion). A Heterologous Protease Cleavage Site Fails to Restore HCF-1C-Subunit Function Given the evident unusual nature of HCF-1 proteolytic maturation, we asked whether this specific process is important for the HCF-1C-subunit function in M phase (see Introduction). For this, we replaced the HCF-1PRO-repeat region with the two natural Taspase1 protease cleavage sites found in MLL (Hsieh et al., 2003) to create HCF-1DPROMLL (see Figure 7A). Given that Taspase1 naturally cleaves the Drosophila HCF-1 homolog, this experiment addresses whether the appearance of HCF-1PROrepeat-induced HCF-1 maturation was evolutionarily significant. We first tested for OGT interaction with the HCF-1DPROMLL mutant. As shown in Figure 7B, HCF-1DPROMLL is effectively cleaved in 293 cells (Figure 7Ba, lane 4) but, like HCF-1DPRO, is defective for OGT binding (Figure 7Bc, lanes 2–4). Thus, replacement of the HCF-1PRO repeats with the MLL Taspase1 cleavage sites separates HCF-1 proteolysis from evident OGT binding. We then asked whether the HCF-1DPROMLL protein is able to rescue the M phase binucleation defect induced by loss of HCF-1 processing (Julien and Herr, 2003). We produced a cell line expressing an siRNA-resistant HCF-1DPROMLL protein, selectively depleted the endogenous HCF-1 protein by siRNA, and counted the number of binucleated cells. As shown in Figure 7C, only the wild-type full-length HCF-1 (WT) and to a lesser extent the HCF-1C subunit (C600) but not the HCF-1DPRO protein were able to rescue the binucleation phenotype induced by the lost of endogenous HCF-1. (The inability of the HCF-1DPRO protein to rescue the binucleation phenotype is not owing to the absence of the HCF-1PRO-repeat region per se because the HCF-1PROMUT protein shown in Figure 4B also displays binucleation defects; S.G. and W.H., unpublished results.) Interestingly, the cleaved 384 Cell 144, 376–388, February 4, 2011 ª2011 Elsevier Inc.
HCF-1DPROMLL protein is not able to rescue the binucleation defect, suggesting that HCF-1PRO-repeat-mediated HCF-1 proteolytic maturation—and probably HCF-1N O-GlcNAcylation—are important to activate the M phase functions of the HCF-1C subunit. DISCUSSION We have investigated the mechanisms of HCF-1 proteolytic maturation and identified a link between OGT-induced HCF-1 proteolysis and O-GlcNAcylation. OGT binds and cleaves the HCF-1PRO repeats while inducing O-GlcNAcylation of the HCF1N subunit. This link is associated with activation of the M phase regulatory functions of HCF-1. These activities of OGT connect the class of reversible posttranslational modifications such as glycosylation with essentially irreversible proteolytic processing. How we arrived at these conclusions was circuitous. We began by hypothesizing that HCF-1PRO-repeat cleavage involves modification of a series of highly conserved and important threonines. Indeed, modification—O-GlcNAcylation—is important, but the HCF-1PRO repeat is apparently not the target of this modification. Instead, it is the binding site for OGT, which then promotes O-GlcNAcylation elsewhere as well as cleavage of its own binding site. A Bipartite Association of OGT with the HCF-1PRO Repeat The analysis of OGT binding to the HCF-1PRO repeat suggests that it recognizes both the cleavage and threonine regions: thus, OGT displays enhanced association with a cleavageregion mutant (E10A), and a threonine-region mutant (T17–22A) counteracts this enhancement (see Figure 4). The inhibition of OGT association by the threonine-region mutant suggests that OGT does not recognize and bind the cleavage region effectively on its own. We suggest that the threonine region represents a site for stable OGT interaction. The enhanced association of OGT with the E10A mutant was not expected. Given that the E10 residue is immediately adjacent to the cleavage site, we suggest that this residue associates with the catalytic center of OGT. If OGT association with the HCF-1PRO repeat is reduced upon proteolytic cleavage, then inhibition of cleavage by the E10A mutation could enhance OGT binding to the HCF-1PRO repeat by retaining OGT on the noncleavable site. Thus, OGT is essentially ‘‘trapped’’ on the E10A mutant. We find intriguing the bipartite TPR and CD structure of OGT and the bipartite threonine and cleavage structure of the HCF1PRO repeat. We suggest a one-to-one correspondence between the OGT TPR region and HCF-1PRO-repeat threonine region for binding and OGT CD region and HCF-1PRO-repeat cleavage region for HCF-1PRO-repeat proteolysis. This hypothesis can explain the surprisingly large sequence required for HCF-1PRO-repeat-induced proteolysis. Although we initiated our studies hypothesizing that the threonine region would serve as a site for modification to activate HCF-1PRO-repeat proteolysis, we now imagine that the inverse might be true: O-GlcNAcylation of the threonine region inhibits OGT binding to the HCF-1PRO repeat, which could be used to inhibit HCF-1 proteolytic maturation.
ΔPROMLL
–
HCF-1:
M
–
–250 –130
HA IP HCF-1C600 b
c
CS1 CS2
–130
HA IP
HCF-1ΔPRO MLL
1
2
3
α-OGT
–130
TI
HCF-1ΔPRO
α-HA
a
HCF-1wt
ΔPRO
B
wt
A
4
C 26 24
% binucleated cells
22
Luc siRNA HCF-1 siRNA
20 18 16 14 12 10 8 6 4 2
ΔPRO
ΔPROMLL
11%
C600
HCF-1-dependent binucleation:
wt
HeLa
0
1%
5%
12%
11%
D a OGT
HCF-1PRO repeats + + + + +
–
– – – –
OGT
+ + + + +
OGT
–
–
– – – –
1
+ + +
– – – –
+ + +
–
– –
– –
Normal M phase
3
2
b Taspase1 CS + + + + +
–
– – – –
Taspase1 Taspase1
+ + + + +
–
– – – –
+ – – + + – – + +–
Aberrant M phase
Figure 7. OGT-Mediated HCF-1PRO-Repeat Proteolysis Is Important for HCF-1C Function (A) Schematic representation of different HCF-1 recombinants. CS1 and CS2 indicate the two Taspase1 cleavage sites of the MLL protein (Hsieh et al., 2003). (B) OGT association assay with HCF-1 protein carrying the MLL Taspase1 sites. Lanes 1–3 are as shown in Figure 4C. Lane 4: 293 cells transfected with the HCF-1DPROMLL construct. Immunoprecipitated samples were visualized by a-HA tag (panel a) and a-OGT (panel c) immunoblot. Total input (TI) samples were visualized by a-OGT immunoblot (panel b). Precursor proteins (–) and cleavage products (d) are indicated. (C) Quantification of binucleated cells in the parental Flp-In-HeLa cells as well as in the indicated cell lines after control (Luc; green) or HCF-1 (red/orange) siRNA (data are represented as mean ± SD). In orange is indicated the background level of binucleation for each sample. The percentages below the histogram indicate differences between the Luc and HCF-1 siRNA for each sample. (D) Proposed model for OGT-induced HCF-1PRO-repeat proteolytic maturation showing potential consequences for HCF-1-protein conformation and function (see text). Kelch indicates an N-terminal Kelch-repeat domain.
Cell 144, 376–388, February 4, 2011 ª2011 Elsevier Inc. 385
OGT: An Unusual Protease The large sequence required for HCF-1PRO-repeat-induced proteolysis had suggested that the HCF-1PRO repeat would be cleaved by an unusual mechanism. The discovery that OGT is responsible for HCF-1PRO-repeat cleavage is consistent with this hypothesis. We imagine two principal ways in which OGT might cleave the HCF-1PRO repeat: OGT could possess a catalytic center for proteolysis or function as a coprotease to cleave HCF-1 in collaboration with the HCF-1PRO repeat. Relative to the first hypothesis, we have not been able to identify a protease-like domain. Nevertheless, OGT is distantly related to enzymes that, like proteases, are hydrolases (Wrabl and Grishin, 2001), and thus OGT might possess an unusual proteolytic catalytic center. In the second preferred hypothesis, OGT functions in intimate collaboration with the HCF-1PRO repeat for cleavage. For example, OGT might stimulate HCF-1PRO-repeat cleavage similarly to E. coli RecA-induced autocleavage of the LexA repressor (Luo et al., 2001). Here, RecA binding to LexA induces a conformational change of LexA, allowing for a repositioning of the LexA active and cleavage sites for autocleavage. By analogy, OGT binding to the HCF-1PRO repeat might lead to a change in conformation that stimulates hydrolysis, for example by activating the essential E10 residue at the cleavage site. We also imagine that GlcNAc moieties on HCF-1 could in turn play a specific role in cleavage. For example, GlcNAc could be bound by—or remain bound to—OGT via its UDP-GlcNAc pocket and, as such, either directly (e.g., participate in catalysis) or indirectly (e.g., play a conformational role) stimulate HCF-1PRO-repeat cleavage. Such a mechanism could ensure coordinate HCF-1 O-GlcNAcylation and cleavage.
A Proteolytic Cleavage Signal Regulates Posttranslational Modification: A Model for HCF-1 Maturation Proteolytic maturation of HCF-1 was known to be important for the biological functions of HCF-1. It is now evident that HCF-1 proteolytic maturation involves not only proteolysis but also O-GlcNAc modification, and that the interplay between these two modifications is what appears to regulate HCF-1 function in M phase. We propose in Figure 7Da a three-step mechanism for HCF-1 maturation: OGT (1) associates with the elaborate HCF-1PRO-repeat signal, (2) O-GlcNAcylates the HCF-1N subunit, and (3) cleaves the protein. In contrast, we suggest that Taspase1-induced HCF-1 cleavage circumvents the O-GlcNAcylation step (Figure 7Db) and that this step is critical for activation of the HCF-1C subunit for M phase progression. The HCF-1N basic and HCF-1C acidic regions are known to associate noncovalently with one another (Wilson et al., 2000). Perhaps repeated O-GlcNAc modification within and around the HCF-1N basic region induced by the presence of multiple HCF-1PRO repeats disrupts this HCF-1N and HCF-1C interaction, functionally liberating transcriptional activation (in orange; Luciano and Wilson, 2002) and chromatin association (in red; Julien and Herr, 2004) domains present in the HCF-1C acidic region. 386 Cell 144, 376–388, February 4, 2011 ª2011 Elsevier Inc.
Why Involve OGT in HCF-1 Maturation? OGT has been hypothesized to serve as a sensor for the metabolic state of the cell by responding to glucose levels (Butkinaree et al., 2010; Love et al., 2010). Because HCF-1 promotes cell proliferation and cell proliferation is associated with enhanced nutrient supplies that generate glucose, we suggest that a role for OGT in activation of HCF-1 via proteolytic maturation involving O-GlcNAcylation can link the cellular activities of HCF-1 to the metabolic state of the cell. EXPERIMENTAL PROCEDURES Plasmid Constructs and DNA Template Preparation Plasmids and DNA templates are described in Extended Experimental Procedures. Cell Culture, Extract Preparation, and Plasmid and siRNA Transfections Human HeLa and 293 cells were grown on plates at 37 C in DMEM with 10% FBS. Nuclear extracts for in vitro cleavage assay were prepared from spinner HeLa cells as previously described (Dignam et al., 1983). Plasmid and siRNA transfections were performed with Lipofectamine or Oligofectamine, respectively, as described (Invitrogen) (see also Extended Experimental Procedures). Synthesis and Purification of Recombinant Proteins from Bacterial and Insect Cells For GST–HCF-1rep1WT and GST–HCF-1rep1E10A synthesis in E. coli, BL21 (DE5) cells were grown at 22 C until an OD600 of 0.6 before induction with 0.4 mM IPTG for 4 hr. Cells were lysed with PBS supplemented with lysozyme, and lysates were centrifuged at 13,000 3 g for 30 min. SDS was added to a final concentration of 1% and lysates were boiled for 10 min. SDS was then diluted to 0.1% with PBS and the lysates were incubated with Ni beads overnight at 4 C. Proteins were eluted with 500 mM imidazole and dialyzed against PBS. Wild-type and mutant S-tagged human OGTs were synthesized in E. coli BL21(DE5) cells. Cells were grown at 25 C until an OD600 of 0.6, at which point OGT synthesis was induced with 0.4 mM IPTG and cells were incubated at 16 C for 16 hr. The OGT proteins were purified using the S-tag purification system, as recommended (Novagen). Wild-type Flag-tagged OGT was synthesized in insect Sf9 cells, lysates were prepared, and proteins were immunoprecipitated using immobilized a-Flag magnetic beads (Sigma) and eluted using 0.2 mg/ml of 13 Flag peptide as described (Cai et al., 2006, 2010). In Vitro Cleavage and O-GlcNAc Transferase Assays In vitro HCF-1PRO-repeat cleavage was performed in 100 mM HEPES (pH 7.9), 5 mM MgCl2, 20 mM KCl, 5 mM DTT, and 10% sucrose at a final volume of 30 ml at 37 C for 16 hr or as specified (Capotosti et al., 2007; Hsieh et al., 2003). For the rhOGT in vitro cleavage assay, 1.5 mg of Inssynt or Bacsynt rhOGT preparation was used and the cleavage buffer was supplemented with 5 mM UDP-GlcNAc. The presence of DTT was critical for OGT-induced HCF-1PROrepeat cleavage. Bacterially purified GST-HCF-1rep1 in vitro cleavage was performed with 500 ng precursor protein. For inhibitor treatment, reactions prior to precursor addition were treated for 30 min at RT with either 2.5 mM alloxan (Konrad et al., 2002), 15 mM staurosporine (Gani and Engh, 2010), 0.5 mg/ml Pefabloc (Vogel and Kristie, 2000), or 500 mM ST078925. The ST078925 inhibitor (4-[(3-cyano-4-(2-thienyl)-2-5,6,7,8-tetrahydroquinolylthio)methyl] benzoic acid; TimTec) is an analog substitute of the validated inhibitor 6 from Gross et al. (2005). For phosphatase treatment, in vitro cleaved HCF-1repXX3 precursor was incubated for 2 hr at 37 C with 400 U of l phosphatase (PPase) in the absence or presence of 50 mM EDTA. The O-GlcNAc transferase assay was performed as described (Clarke et al., 2008) using 2500 U of the CKII substrate. Nondenaturating Immunoprecipitation For a-HA tag immunoprecipitation, cells were lysed in 0.5% NP40, 10 mM TrisCl (pH 8.0), 150 mM NaCl, 5 mM MgCl2 (NP40 buffer; Misaghi et al.,
2009). Lysates were incubated with a-HA agarose beads (Sigma) overnight at 4 C, beads were washed extensively in NP40 buffer, and immunoprecipitated material was eluted in SDS-PAGE loading buffer. For endogenous HCF-1 immunoprecipitation, 1.5 mg of 420 mM KCl HeLa nuclear extract was diluted to 100 mM KCl and incubated overnight at 4 C with 10 ml of C18 a-HCF-1C antibody or 10 ml of nonimmune sera. Samples were analyzed by immunoblot. See Extended Experimental Procedures for the origin of antibodies used and procedure for immunoblot analysis.
O-GlcNAcylation Analysis by LC-MS/MS HA-tagged HCF-1rep123 plasmid expression vectors were transfected into ten 15 cm dishes of 293 cells for 48 hr. Cells were lysed in 8 ml NP40 buffer. The lysates were adjusted to 1% SDS and boiled for 10 min. The SDS concentration was subsequently adjusted to 0.1% by dilution with NP40 buffer, and lysates were incubated with a-HA agarose beads overnight at 4 C. For proteomic analysis, see Extended Experimental Procedures.
Generation of Cell Lines and Binucleation Analysis To create the recombinant HCF-1 Flp-In-HeLa cells, 0.4 mg of pCDNA5/HAFLAG/FRT vector containing genes of interest was cotransfected with 3.6 mg of the Flpase expression vector pOG44 into Flp-In-HeLa cells (a kind gift of S.S. Taylor) with Lipofectamine, and transfected cells were selected in hygromycin-containing medium (300 mg/ml) for 10 days as described (Invitrogen). Three independent clones for each recombinant HCF-1 construct were selected for analysis. Endogenous HCF-1 was depleted by double siRNA treatment for 50 hr and binucleated cells were quantitated by staining with a-tubulin and a-HCF-1C H12—to determine HCF-1 depletion—antibodies and DAPI as previously described (Julien and Herr, 2003). One hundred cells were counted per siRNA depletion and each independent clone was analyzed in triplicate. Data were summarized as mean ± standard deviation (SD).
SUPPLEMENTAL INFORMATION Supplemental Information includes Extended Experimental Procedures, three figures, and one table and can be found with this article online at doi:10.1016/ j.cell.2010.12.030.
ACKNOWLEDGMENTS We thank T.M. Kristie and J.A. Hanover for the HCF-1nc and the bacterial OGT expression vectors, respectively; M. Boutros and A. Wilson for the HCF1DPROrep2 construct; R.S. Haltiwanger for the a-OGT antibody; E. Julien and D.J. Aufiero for the N13 and C18 antibodies; S.S. Taylor for the Flp-In-HeLa cells; M. Quadroni for advice with the mass spectrometry analysis; L. Capponi, P. L’Hoˆte, and M. McLaird for technical support; and J.H. Reina for art design. We thank N. Hernandez, J. Michaud, and S.H. Verhelst for critical reading of the manuscript; and V. Zoete, O. Michielin, N. Hernandez, J. Mu¨ller, P. Schneider, and K. Struhl for discussion. This study was supported by the Stowers Institute for Medical Research and the Helen Nelson Medical Research Fund at the Greater Kansas City Community Foundation to J.W.C. and R.C.C. and by the Swiss National Science Foundation, Oncosuisse, and the University of Lausanne to W.H. Received: August 20, 2010 Revised: November 22, 2010 Accepted: December 16, 2010 Published: February 3, 2011 REFERENCES Butkinaree, C., Park, K., and Hart, G.W. (2010). O-linked b-N-acetylglucosamine (O-GlcNAc): extensive crosstalk with phosphorylation to regulate signaling and transcription in response to nutrients and stress. Biochim. Biophys. Acta 1800, 96–106.
Cai, Y., Jin, J., Gottschalk, A.J., Yao, T., Conaway, J.W., and Conaway, R.C. (2006). Purification and assay of the human INO80 and SRCAP chromatin remodeling complexes. Methods 40, 312–317. Cai, Y., Jin, J., Swanson, S.K., Cole, M.D., Choi, S.H., Florens, L., Washburn, M.P., Conaway, J.W., and Conaway, R.C. (2010). Subunit composition and substrate specificity of a MOF-containing histone acetyltransferase distinct from the male-specific lethal (MSL) complex. J. Biol. Chem. 285, 4268–4272. Capotosti, F., Hsieh, J.J., and Herr, W. (2007). Species selectivity of mixedlineage leukemia/trithorax and HCF proteolytic maturation pathways. Mol. Cell. Biol. 27, 7063–7072. Clarke, A.J., Hurtado-Guerrero, R., Pathak, S., Schuttelkopf, A.W., Borodkin, V., Shepherd, S.M., Ibrahim, A.F., and van Aalten, D.M. (2008). Structural insights into mechanism and specificity of O-GlcNAc transferase. EMBO J. 27, 2780–2788. Crooks, G.E., Hon, G., Chandonia, J.M., and Brenner, S.E. (2004). WebLogo: a sequence logo generator. Genome Res. 14, 1188–1190. Dignam, J.D., Lebovitz, R.M., and Roeder, R.G. (1983). Accurate transcription initiation by RNA polymerase II in a soluble extract from isolated mammalian nuclei. Nucleic Acids Res. 11, 1475–1489. Gani, O.A., and Engh, R.A. (2010). Protein kinase inhibition of clinically important staurosporine analogues. Nat. Prod. Rep. 27, 489–498. Goto, H., Motomura, S., Wilson, A.C., Freiman, R.N., Nakabeppu, Y., Fukushima, K., Fujishima, M., Herr, W., and Nishimoto, T. (1997). A singlepoint mutation in HCF causes temperature-sensitive cell-cycle arrest and disrupts VP16 function. Genes Dev. 11, 726–737. Gross, B.J., Kraybill, B.C., and Walker, S. (2005). Discovery of O-GlcNAc transferase inhibitors. J. Am. Chem. Soc. 127, 14588–14589. Haltiwanger, R.S., Blomberg, M.A., and Hart, G.W. (1992). Glycosylation of nuclear and cytoplasmic proteins. Purification and characterization of a uridine diphospho-N-acetylglucosamine:polypeptide b-N-acetylglucosaminyltransferase. J. Biol. Chem. 267, 9005–9013. Hanover, J.A., Yu, S., Lubas, W.B., Shin, S.H., Ragano-Caracciola, M., Kochran, J., and Love, D.C. (2003). Mitochondrial and nucleocytoplasmic isoforms of O-linked GlcNAc transferase encoded by a single mammalian gene. Arch. Biochem. Biophys. 409, 287–297. Hart, G.W., Housley, M.P., and Slawson, C. (2007). Cycling of O-linked b-Nacetylglucosamine on nucleocytoplasmic proteins. Nature 446, 1017–1022. Hsieh, J.J., Cheng, E.H., and Korsmeyer, S.J. (2003). Taspase1: a threonine aspartase required for cleavage of MLL and proper HOX gene expression. Cell 115, 293–303. Jinek, M., Rehwinkel, J., Lazarus, B.D., Izaurralde, E., Hanover, J.A., and Conti, E. (2004). The superhelical TPR-repeat domain of O-linked GlcNAc transferase exhibits structural similarities to importin a. Nat. Struct. Mol. Biol. 11, 1001–1007. Julien, E., and Herr, W. (2003). Proteolytic processing is necessary to separate and ensure proper cell growth and cytokinesis functions of HCF-1. EMBO J. 22, 2360–2369. Julien, E., and Herr, W. (2004). A switch in mitotic histone H4 lysine 20 methylation status is linked to M phase defects upon loss of HCF-1. Mol. Cell 14, 713–725. Konrad, R.J., Zhang, F., Hale, J.E., Knierman, M.D., Becker, G.W., and Kudlow, J.E. (2002). Alloxan is an inhibitor of the enzyme O-linked N-acetylglucosamine transferase. Biochem. Biophys. Res. Commun. 293, 207–212. Kreppel, L.K., and Hart, G.W. (1999). Regulation of a cytosolic and nuclear O-GlcNAc transferase. Role of the tetratricopeptide repeats. J. Biol. Chem. 274, 32015–32022. Kristie, T.M., Pomerantz, J.L., Twomey, T.C., Parent, S.A., and Sharp, P.A. (1995). The cellular C1 factor of the herpes simplex virus enhancer complex is a family of polypeptides. J. Biol. Chem. 270, 4387–4394. Kristie, T.M., Liang, Y., and Vogel, J.L. (2010). Control of a-herpesvirus IE gene expression by HCF-1 coupled chromatin modification activities. Biochim. Biophys. Acta 1799, 257–265.
Cell 144, 376–388, February 4, 2011 ª2011 Elsevier Inc. 387
Lazarus, B.D., Love, D.C., and Hanover, J.A. (2006). Recombinant O-GlcNAc transferase isoforms: identification of O-GlcNAcase, yes tyrosine kinase, and tau as isoform-specific substrates. Glycobiology 16, 415–421.
Vogel, J.L., and Kristie, T.M. (2006). Site-specific proteolysis of the transcriptional coactivator HCF-1 can regulate its interaction with protein cofactors. Proc. Natl. Acad. Sci. USA 103, 6817–6822.
Love, D.C., Krause, M.W., and Hanover, J.A. (2010). O-GlcNAc cycling: emerging roles in development and epigenetics. Semin. Cell Dev. Biol. 21, 646–654.
Wang, Z., Udeshi, N.D., Slawson, C., Compton, P.D., Sakabe, K., Cheung, W.D., Shabanowitz, J., Hunt, D.F., and Hart, G.W. (2010). Extensive crosstalk between O-GlcNAcylation and phosphorylation regulates cytokinesis. Sci. Signal. 3, ra2.
Lubas, W.A., and Hanover, J.A. (2000). Functional expression of O-linked GlcNAc transferase. Domain structure and substrate specificity. J. Biol. Chem. 275, 10983–10988. Luciano, R.L., and Wilson, A.C. (2002). An activation domain in the C-terminal subunit of HCF-1 is important for transactivation by VP16 and LZIP. Proc. Natl. Acad. Sci. USA 99, 13403–13408. Luo, Y., Pfuetzner, R.A., Mosimann, S., Paetzel, M., Frey, E.A., Cherney, M., Kim, B., Little, J.W., and Strynadka, N.C. (2001). Crystal structure of LexA: a conformational switch for regulation of self-cleavage. Cell 106, 585–594. Martinez-Fleites, C., Macauley, M.S., He, Y., Shen, D.L., Vocadlo, D.J., and Davies, G.J. (2008). Structure of an O-GlcNAc transferase homolog provides insight into intracellular glycosylation. Nat. Struct. Mol. Biol. 15, 764–765.
Wells, L., and Hart, G.W. (2003). O-GlcNAc turns twenty: functional implications for post-translational modification of nuclear and cytosolic proteins with a sugar. FEBS Lett. 546, 154–158. Wells, L., Vosseller, K., and Hart, G.W. (2001). Glycosylation of nucleocytoplasmic proteins: signal transduction and O-GlcNAc. Science 291, 2376– 2378. Wells, L., Kreppel, L.K., Comer, F.I., Wadzinski, B.E., and Hart, G.W. (2004). O-GlcNAc transferase is in a functional complex with protein phosphatase 1 catalytic subunits. J. Biol. Chem. 279, 38466–38470. Wilson, A.C., LaMarco, K., Peterson, M.G., and Herr, W. (1993). The VP16 accessory protein HCF is a family of polypeptides processed from a large precursor protein. Cell 74, 115–125.
Mazars, R., Gonzalez-de-Peredo, A., Cayrol, C., Lavigne, A.C., Vogel, J.L., Ortega, N., Lacroix, C., Gautier, V., Huet, G., Ray, A., et al. (2010). The THAP-zinc finger protein THAP1 associates with coactivator HCF-1 and O-GlcNAc transferase: a link between DYT6 and DYT3 dystonias. J. Biol. Chem. 285, 13364–13371.
Wilson, A.C., Peterson, M.G., and Herr, W. (1995). The HCF repeat is an unusual proteolytic cleavage signal. Genes Dev. 9, 2445–2458.
Misaghi, S., Ottosen, S., Izrael-Tomasevic, A., Arnott, D., Lamkanfi, M., Lee, J., Liu, J., O’Rourke, K., Dixit, V.M., and Wilson, A.C. (2009). Association of C-terminal ubiquitin hydrolase BRCA1-associated protein 1 with cell cycle regulator host cell factor 1. Mol. Cell. Biol. 29, 2181–2192.
Wilson, A.C., Boutros, M., Johnson, K.M., and Herr, W. (2000). HCF-1 aminoand carboxy-terminal subunit association through two separate sets of interaction modules: involvement of fibronectin type 3 repeats. Mol. Cell. Biol. 20, 6721–6730.
Nalivaeva, N.N., and Turner, A.J. (2001). Post-translational modifications of proteins: acetylcholinesterase as a model system. Proteomics 1, 735–747.
Wrabl, J.O., and Grishin, N.V. (2001). Homology between O-linked GlcNAc transferases and proteins of the glycogen phosphorylase superfamily. J. Mol. Biol. 314, 365–374.
Starr, C.M., and Hanover, J.A. (1990). Glycosylation of nuclear pore protein p62. Reticulocyte lysate catalyzes O-linked N-acetylglucosamine addition in vitro. J. Biol. Chem. 265, 6868–6873. Torres, C.R., and Hart, G.W. (1984). Topography and polypeptide distribution of terminal N-acetylglucosamine residues on the surfaces of intact lymphocytes. Evidence for O-linked GlcNAc. J. Biol. Chem. 259, 3308–3317.
Wysocka, J., and Herr, W. (2003). The herpes simplex virus VP16-induced complex: the makings of a regulatory switch. Trends Biochem. Sci. 28, 294–304.
Tyagi, S., and Herr, W. (2009). E2F1 mediates DNA damage and apoptosis through HCF-1 and the MLL family of histone methyltransferases. EMBO J. 28, 3185–3195.
Wysocka, J., Myers, M.P., Laherty, C.D., Eisenman, R.N., and Herr, W. (2003). Human Sin3 deacetylase and trithorax-related Set1/Ash2 histone H3-K4 methyltransferase are tethered together selectively by the cell-proliferation factor HCF-1. Genes Dev. 17, 896–911.
Vogel, J.L., and Kristie, T.M. (2000). Autocatalytic proteolysis of the transcription factor-coactivator C1 (HCF): a potential role for proteolytic regulation of coactivator function. Proc. Natl. Acad. Sci. USA 97, 9425–9430.
Yang, X., Zhang, F., and Kudlow, J.E. (2002). Recruitment of O-GlcNAc transferase to promoters by corepressor mSin3A: coupling protein O-GlcNAcylation to transcriptional repression. Cell 110, 69–80.
388 Cell 144, 376–388, February 4, 2011 ª2011 Elsevier Inc.
Osh Proteins Regulate Phosphoinositide Metabolism at ER-Plasma Membrane Contact Sites Christopher J. Stefan,1,2 Andrew G. Manford,1,2 Daniel Baird,1,3 Jason Yamada-Hanff,1,4 Yuxin Mao,1 and Scott D. Emr1,* 1Weill
Institute for Cell & Molecular Biology, Department of Molecular Biology & Genetics, Cornell University, Ithaca, NY 14853, USA authors contributed equally to this work 3Present address: Novartis Institutes for Biomedical Research, 250 Massachusetts Avenue, Cambridge, MA 02139, USA 4Present address: Program in Neuroscience, Harvard Medical School, 220 Longwood Avenue, Boston, MA 02215, USA *Correspondence:
[email protected] DOI 10.1016/j.cell.2010.12.034 2These
SUMMARY
Sac1 phosphoinositide (PI) phosphatases are essential regulators of PI-signaling networks. Yeast Sac1, an integral endoplasmic reticulum (ER) membrane protein, controls PI4P levels at the ER, Golgi, and plasma membrane (PM). Whether Sac1 can act in trans and turn over PI4P at the Golgi and PM from the ER remains a paradox. We find that Sac1-mediated PI4P metabolism requires the oxysterol-binding homology (Osh) proteins. The PH domain-containing family member, Osh3, localizes to PM/ER membrane contact sites dependent upon PM PI4P levels. We reconstitute Osh protein-stimulated Sac1 PI phosphatase activity in vitro. We also show that the ER membrane VAP proteins, Scs2/Scs22, control PM PI4P levels and Sac1 activity in vitro. We propose that Osh3 functions at ER/PM contact sites as both a sensor of PM PI4P and an activator of the ER Sac1 phosphatase. Our findings further suggest that the conserved Osh proteins control PI metabolism at additional membrane contact sites. INTRODUCTION Phosphatidylinositol 4-phosphate, PI4P, serves as an essential signaling molecule at the plasma membrane (PM) and Golgi in the control of membrane trafficking, cytoskeletal organization, lipid metabolism, and signal transduction pathways (D’Angelo et al., 2008). Maintaining the proper balance of PI4P levels through the regulation of phosphoinositide (PI) kinases and phosphatases is critical. Sac1-like PI phosphatases are important regulators of PI4P turnover and are implicated in disease (Liu and Bankaitis, 2010). Yeast cells lacking Sac1 display elevated PI4P levels resulting in impaired membrane trafficking, abnormal vacuolar morphology, altered lipid metabolism, and growth defects (Foti et al., 2001; Guo et al., 1999; Rivas et al., 1999). Inactivation of Sac1 PI phosphatase activity provided by Sac1 and the synaptojanin-like proteins, Sjl2 and Sjl3, results
in massive PI4P accumulation, secretory defects, and lethality (Foti et al., 2001). In mammalian cells, depletion of Sac1 leads to elevated cellular PI4P levels, Golgi fragmentation, and defects in mitotic spindle organization (Liu et al., 2008). Loss of Sac1 in the mouse and Drosophila results in embryonic lethality (Liu et al., 2008; Wei et al., 2003). Thus, Sac1 PI phosphatases are essential for multiple, conserved cellular functions. Although Sac1 PI phosphatases are key modulators of PI4P metabolism, little is known about their regulation. Both yeast and mammalian Sac1 are integral membrane proteins localized to the endoplasmic reticulum (ER) and Golgi (Faulhammer et al., 2007; Foti et al., 2001; Nemoto et al., 2000). Upon starvation in yeast, or in the absence of growth factors in mammalian cells, Sac1 traffics from the ER to the Golgi where it antagonizes PI4P synthesis and Golgi function (Blagoveshchenskaya et al., 2008; Faulhammer et al., 2007). Although roles for Sac1 in the control of PI4P pools at the ER and Golgi are known, less is understood about how Sac1 regulates PI4P levels at the PM. Previous studies indicate that sac1D mutant yeast cells accumulate PI4P at the PM generated by the PI 4-kinase Stt4 (Baird et al., 2008; Foti et al., 2001; Roy and Levine, 2004). Yet, Sac1 is not known to traffic to the PM, nor is it known if Sac1 regulates PM PI4P pools from the ER. A recent study describing the molecular structure of the Sac1 catalytic domain suggests that Sac1 may act in trans from the ER to turn over PI4P at the PM (Manford et al., 2010). However, factors that link Sac1 PI phosphatase activity to PI4P at the PM are unknown. We find that PI4P metabolism and Sac1 function require the oxysterol-binding homology (Osh) protein family. In particular, Osh3 assembles at cortical ER (cER) structures in response to PM PI4P levels. We reconstitute Osh protein-stimulated Sac1 PI phosphatase activity against PI4P-containing liposomes in vitro. In addition the ER membrane VAP proteins, Scs2 and Scs22, control PM PI4P levels in vivo and Sac1 activity in vitro. We propose that the Osh and VAP proteins regulate Sac1 PI phosphatase activity at PM/ER membrane contact sites. We also discuss potential roles for these proteins in the control of PI metabolism at additional membrane contact sites. Membrane junctions between the ER and other organelles are sites for lipid transfer and metabolism, Ca2+ transport and signaling, and the downregulation of insulin and growth factor receptors (reviewed Cell 144, 389–401, February 4, 2011 ª2011 Elsevier Inc. 389
A
Sac1-GFP
DsRed-HDEL
nER
nER
peripheral ER
B
merge nER
peripheral ER
ER
ER
Ank
FFAT
100 nm extracellular
GOLD
PM
cytoplasm
PI4P
?
PH
ORD
PH
ORD
Osh1 Osh2
PH
ORD
Osh3
ORD
PI4P
PtdIns
Sac1
P
ctor
Stt4 PI4K
effe
PtdIns
4 μm
peripheral ER
C yeast family:
PM
PM
ORD
P
ORD ORD mammalian:
ER lumen
25
cellular PI4P levels
(% of total 3H-labeled PIs)
20 15 10
Osh4/Kes1 Osh5/Hes1 Osh6 Osh7
FFAT
PH
ER
D
Figure 1. The Yeast OSBPs Control PI4P Metabolism
DIC overlay
ORD
OSBP
in vivo PI4P measurements: 18x
26oC 38oC 6x 4.5x
13x
(A) Sac1-GFP localizes to the ER (marked by DsRed-HDEL). Peripheral ER and nuclear ER (nER) compartments are labeled. Scale bar, 4 mm. See also Figure S1B. (B) Electron microscopy of a region of a yeast spheroplast showing proximity between cER and the PM (top, scale bar, 100 nm). Model for PI4P metabolism at PM/ER membrane contact sites (bottom). (C) Diagram of the Osh proteins and OSBP. ORD, OSBP-related sterol-binding domain; FFAT, di-phenylalanines within an acidic track; PH, pleckstrin homology domain; GOLD, Golgi localization domain; Ank, ankyrin repeats. (D) Cellular PI4P levels measured by 3H-inositol labeling and HPLC analysis for wild-type, oshD [CEN OSH4], oshD [CEN osh4ts], and sac1D cells at 26 C (gray) and 38 C (black). Error bars are the SD of three experiments. See also Figure S1C and Table S3. (E) PI4P localization in wild-type, oshD [CEN OSH4], oshD [CEN osh4ts], and sac1D cells. Wildtype, oshD [CEN OSH4], and oshD [CEN osh4ts] cells carrying GFP-2xPHOsh2 were shifted to 38 C for 1 hr; sac1D cells were observed at 26 C. Scale bar, 4 mm.
7.7x
GFP-2xPHOsh2
turns over these distinct PI4P pools. Sac1 traffics between the ER and Golgi 5 1.3x to regulate PI4P levels at these intracel1x 3 lular compartments (Faulhammer et al., 2007). However, to our knowledge, Sac1 0 osh1-osh7Δ osh1-osh7Δ Wild type sac1Δ has not been found at the PM. We CEN OSH4 CEN osh4ts reasoned that ER-localized Sac1 might control PM PI4P pools at PM/ER E PM PI4P PM PI4P membrane contact sites where the PM PI4P PM PI4P Golgi peripheral ER closely apposes the PM PI4P (within 10 nm) (Figure 1B). Although Sac1 localizes to the peripheral ER (FigGolgi ure 1A), no proteins that regulate Sac1 PI4P activity at PM/ER membrane contact Wild type sac1Δ osh1-osh7Δ osh1-osh7Δ sites are known. ts CEN OSH4 CEN osh4 We hypothesized that a PI4P effector might link PI4P at the PM to Sac1 at the ER (Figure 1B). For this we focused on the PI4P-binding proteins Osh1, Osh2, in Lev, 2010). Thus, modulation of PI levels at membrane contact and Osh3, members of the oxysterol-binding protein homology sites may provide a fundamental mechanism for signaling family (Figure 1C). Osh2 and Osh3 possess PH domains that bind PI4P and localize to cortical, punctate structures (Figure 1C, between cellular membrane compartments. Figure 2A, and Figure S2 available online) (Levine and Munro, 2001; Roy and Levine, 2004). Osh1 also contains a PH domain RESULTS but localizes to the Golgi and nuclear-vacuolar membrane junction sites (Levine and Munro, 2001). Osh1, Osh2, and Osh3 are The Osh Proteins Regulate PI4P Metabolism Yeast Sac1 is an integral membrane protein that localizes to similar to mammalian oxysterol-binding protein (OSBP) that nuclear and peripheral ER compartments (Figure 1A) (Foti contains a PI4P-binding PH domain (Figure 1C). These proteins et al., 2001). Yet, PI4P is generated by the PI 4-kinases Stt4 are members of the OSBP-related protein (ORP) family that and Pik1 at the PM and the Golgi, respectively (Figure 1E) (Aud- contain sterol-binding OSBP-related domains (ORDs) (Fighya et al., 2000). It is not known whether ER-embedded Sac1 ure 1C). Among the ORP family, Osh4/Kes1 does not possess 390 Cell 144, 389–401, February 4, 2011 ª2011 Elsevier Inc.
A
DsRed-HDEL
merge
DIC overlay
wild type
Osh3-GFP
C
PM PI4P
PM PI4P
3.5 3.0 2.5
2.7x
cellular PI4P levels (% of total 3H-labeled PIs)
osh2Δ osh3Δ 4.0
GFP-2xPHOsh2 localization: wild type osh2Δ osh3Δ
DIC overlay
PM PI4P
2.0 DIC overlay
DIC overlay
1.5 1.0 0.5
Osh3-GFP Wild type 38oC
wild type
m
m
stt4ts 38oC
B
*
0
D
Figure 2. Osh2 and Osh3 Localize to PM/ER Contact Sites and Control PM PI4P (A) Osh3-GFP localization in wild-type cells coexpressing the DsRed-HDEL ER marker. Arrows show Osh3-GFP puncta associated with cER. The asterisk shows an Osh3-GFP patch not associated with the cER. See also Figure S2A. (B) Cellular PI4P levels in wild-type and osh2D osh3D mutant cells measured by 3H-inositol labeling and HPLC analysis (left). Error bars are the SD of three experiments. See also Table S3. GFP-2xPHOsh2 (PI4P) localization in wild-type or osh2D osh3D mutant cells (right). Mother cells are indicated (m). (C) Osh3-GFP localization in wild-type or stt4ts cells at 38 C. Levels were enhanced using Adobe Photoshop in the DIC overlay image for stt4 mutant cells (lower left). See also Figure S2B. (D) Growth of oshD [CEN osh4ts] cells coexpressing OSH3 alleles. Serial dilutions of cells carrying empty vector or Osh3-GFP constructs were incubated at 26 C and 38 C. Osh3-GFP mutant constructs are indicated: AAAT, substitution of phenylalanines in the FFAT motif with alanine, DPH, deletion of the PH domain. See also Figure S2C. Scale bars, 4 mm.
osh1Δ-osh7Δ osh1Δ-osh7Δ CEN osh4ts 26oC CEN osh4ts 38oC +vector alone
38 C (approximately 3-fold) (Figure S1A). This did not account for the 18-fold GFP increase in PI4P in oshD:CEN osh4ts cells FFAT GFP ORD because complete loss of Sac1 resulted AAAT in only a 13-fold increase in PI4P (sac1D ORD GFP cells) (Figure 1D). Moreover, GFP-Sac1 AAAT localized to both nuclear and peripheral GFP ORD ER compartments in wild-type and oshD:CEN osh4ts cells at 38 C (Figure S1B). The 18-fold increase in PI4P levels in oshD:CEN osh4ts mutant cells resembled levels previously reported for triple mutant sac1ts sjl2D sjl3D cells (Foti et al., 2001), suggesting that the Osh proteins may regulate multiple Sac1-like PI phosphatase activities. We also examined PI4P levels in cells lacking both Sac1 and Osh protein function to test for additive effects. PI4P levels were not further elevated in sac1ts oshD:CEN osh4ts cells compared to oshD:CEN osh4ts cells at 38 C (Figure S1C and Table S3), suggesting that Sac1 and the Osh proteins act in a common pathway. Together, these results implicate the Osh proteins in PI4P metabolism and Sac1 phosphatase function. The PI4P FLARE (fluorescent lipid-associated reporter) GFP-2xPHOsh2 is found at Golgi compartments and the PM via Pik1 and Stt4 PI 4-kinase activities, respectively, in wild-type cells (Figure 1E) (Roy and Levine, 2004). In contrast the PI4P FLARE was stabilized at the PM in oshD:CEN OSH4 and oshD:CEN osh4ts cells, although weak fluorescence at intracellular puncta was detected (Figure 1E). The PI4P FLARE was also increased at the PM in sac1D cells, consistent with elevated levels of Stt4-generated PI4P at the PM (Figure 1E) (Baird et al., 2008; Foti et al., 2001; Roy and Levine, 2004). Although previous studies have implicated one Osh family member, Osh4/Kes, in PI4P regulation at the Golgi
Osh3 constructs: FFAT
+OSH3-GFP
GOLD
+osh3ΔPH-GFP
GOLD
+osh3AAAT-GFP
GOLD
+osh3ΔPH, AAAT-GFP
GOLD
PH
PH
a PH domain but also binds PI lipids (Figure 1C) (Li et al., 2002). Osh4/Kes1 is also known to regulate PI4P levels at the Golgi (Fairn et al., 2007; Li et al., 2002). Given the properties of the Osh protein family to bind PI4P, regulate PI4P (in the case of Osh4/Kes1), and to localize to membrane contact sites, we further examined the role of the Osh proteins in PI4P metabolism. First, we measured PI4P levels in cells lacking Osh protein function. For this we utilized strains that lack the OSH1-OSH7 genes and carry wild-type OSH4 or a temperature-sensitive allele, osh4-1, on a plasmid because loss of all Osh proteins is lethal (oshD:CEN OSH4 or oshD:CEN osh4ts cells) (Beh and Rine, 2004). In 3H-inositol labeling experiments, PI4P levels were 6- to 7-fold higher than wild-type cells in oshD:CEN OSH4 and oshD:CEN osh4ts cells at the permissive temperature (26 C) (Figure 1D), suggesting that Osh proteins (other than Osh4) control PI4P levels in vivo. At the restrictive temperature for oshD:CEN osh4ts cells, PI4P levels were elevated 18-fold above wild-type levels (38 C) (Figure 1D) and were greater than PI4P levels in sac1D mutant cells (13-fold above wildtype) (Figure 1D). Expression levels of a Sac1-GFP fusion were slightly reduced in oshD:CEN osh4ts mutant cells at
ORD
Cell 144, 389–401, February 4, 2011 ª2011 Elsevier Inc. 391
A mammalian:
Figure 3. The VAP Proteins Scs2 and Scs22 Control PM PI4P Metabolism
B
FFAT PH
ORD
bound fraction
OSBP
input
GST
GST-Osh31-613
TMD COIL VAP
MSP
Scs23xHA
yeast:
ORD
MSP
PH
Osh3
GSTOsh31-613
TMD COIL Scs2/Scs22
FFAT GOLD
Coomassie gel:
FFAT
GOLD PH
Osh31-613
D
GFP-2xPHOsh2 localization: wild type PM PI4P
C 4.5
wild type scs2Δ scs22Δ
scs2Δ scs22Δ
(A) Diagrams of mammalian VAP, yeast Scs2/Scs22, OSBP, and Osh3. VAP proteins possess a major sperm protein domain (MSP), a coiled-coil, and a transmembrane domain (TMD). (B) The GST-Osh31–613 fusion binds Scs2-3xHA from solubilized yeast lysates. GST alone does not bind Scs23xHA. Total input and bound fractions are shown. (C) Cellular PI4P levels in wild-type and scs2D scs22D cells measured by 3H-inositol labeling and HPLC analysis. Error bars show SD of three independent experiments. See also Table S3. (D) Localization of GFP-2xPHOsh2 (PI4P) in wild-type and scs2Dscs22D cells. Mother cells are indicated (m). Scale bar, 4 mm. See also Figure S3.
PM PI4P Golgi PI4P
cellular PI4P levels (% of total 3H-labeled PIs)
cells (Voeltz et al., 2006). Osh3-GFP patches clustered at cER compartments (marked with PM PI4P DsRed-HDEL) adjacent to the PM (stained with 3.5 filipin) in rtn1D rtn2D yop1D triple mutant cells 2.9x 3.0 and was absent from regions lacking cER strucDIC overlay DIC overlay 2.5 tures (Figure S2A). Together, our results sug2.0 gested that Osh3 localizes to PM/ER membrane m contact sites and modulates PM PI4P levels, m 1.5 possibly by regulation of cER-localized Sac1. 1.0 m Osh3 localization did not always coincide with 0.5 cER compartments (Figure 2A, asterisk), suggesting that its assembly at PM/ER membrane 0 contact sites may be dynamic and progress in regulated stages. Thus, we tested if Osh3 localization and function required PM PI4P. Osh3-GFP (Fairn et al., 2007; Li et al., 2002), our results suggested that addi- localized diffusely in the cytoplasm in cells with impaired Stt4 PI 4kinase activity (stt4ts cells; Figure 2C), showing that Osh3 localizational Osh proteins control PI4P metabolism at the PM. tion required PM PI4P synthesis. In contrast, Osh3-GFP was present at the PM in pik1ts mutant cells impaired in PI4P synthesis Osh3 Localizes to PM/ER Contact Sites in a PI4Pat the Golgi, although Osh3 was observed at intracellular strucDependent Manner Several lines implicated Osh2 and Osh3 in the control of PI4P tures in these cells (Figure S2B). Likewise, Osh3-GFP localized metabolism at the PM. First, both bind PI4P and localize to to cortical patches in cells with reduced PI(4,5)P2 levels (mss4ts cortical structures (Figure 1C, Figure 2A, and Figure S2A) (Levine cells) (Figure S2B). A form of Osh3 lacking its PH domain and Munro, 2001; Roy and Levine, 2004; Schulz et al., 2009). (Osh3DPH-GFP) did not rescue the growth defects of oshD:CEN Second, expression of OSH2 or OSH3 from a plasmid partially osh4ts mutant cells at 38 C (Figure 2D). In addition, Osh3DPHrescued the increased PI4P levels in oshD:CEN osh4ts mutant GFP displayed increased cytoplasmic localization, compared to cells at 38 C (from 18-fold above wild-type to 5- and 8-fold, wild-type Osh3-GFP (Figure S2C). Thus, the assembly of Osh3 respectively) (Table S3). Third, osh2D osh3D double mutant cells at cortical patches required PM PI4P synthesis, consistent with displayed elevated PI4P levels, as measured by 3H-inositol its ability to bind PI4P and control PM PI4P levels. Next, we asked if other regions in Osh3 play a role in its funclabeling experiments (2.7-fold above wild-type) (Figure 2B and Table S3). Lastly, the PI4P FLARE was stabilized at the PM in tion. Osh3 contains an FFAT motif (di-phenylalanines within an osh2D osh3D double mutant cells (particularly in mother cells), acidic track) (Figures 1C, 2D, and 3A). FFAT motifs bind the MSP domains of the yeast ER membrane proteins Scs2 and compared to wild-type cells (Figure 2B). We examined if Osh3 localized to PM/ER membrane contact Scs22, orthologs of the mammalian ER membrane VAP proteins sites in order to regulate PI4P at the PM. Osh3-GFP localized (Figure 3A) (Kaiser et al., 2005; Loewen et al., 2003). A form of to cortical patches often associated with cER compartments Osh3 bearing substitutions in the FFAT motif (Osh3AAAT-GFP) marked with DsRed-HDEL (Figure 2A, arrows), as well as the partially rescued the growth defects of oshD:CEN osh4ts cells PM stained with filipin (Figure S2A). To confirm that Osh3 at 38 C (Figure 2D) and assembled at cortical patches (Figpatches associated with cER, we examined Osh3-GFP localiza- ure S2C). A previous study reported that a mutant form of tion in cells lacking the reticulons Rtn1, Rtn2, and Yop1 because Osh3 with substitutions in the FFAT motif was mislocalized the cER displays expanded sheet-like structures in these mutant from the cER (Loewen et al., 2003). However, this study used 4.0
392 Cell 144, 389–401, February 4, 2011 ª2011 Elsevier Inc.
Golgi PI4P
a GFP-Osh3 fusion that required Scs2 overexpression to target to the cER. Consistent with this previous study, we found that Osh3-GFP targeting to the cER was impaired in scs2D scs22D double mutant cells (see Figure S3D). As expected, substitution of the FFAT motif in combination with deletion of the PH domain (GFP-Osh3DPH, AAAT) (Figure 2D) resulted in inactivation of Osh3 because this mutant protein did not rescue the growth defect of oshD:CEN osh4ts cells (Figure 2D). Moreover, Osh3DPH,AAATGFP localized diffusely in the cytoplasm, and not at cortical patches (Figure S2C). Thus, Osh3 undergoes multiple interactions involving at least PI4P and the Scs2/Scs22 VAP proteins to assemble and function at PM/ER membrane contact sites. The VAP Proteins Scs2 and Scs22 Regulate PI4P Metabolism We then addressed if the ER membrane VAP proteins Scs2/ Scs22 control Osh3 function and PI4P metabolism. Because FFAT motifs from OSBP and Osh proteins bind VAP proteins (Figure 3A) (Kaiser et al., 2005; Loewen et al., 2003), we specifically confirmed interactions for Osh3 and Scs2. A GSTOsh3 N-terminal fragment containing the FFAT motif (GSTOsh31–613) (Figure 3A) bound to glutathione-sepharose beads efficiently isolated an Scs2-3xHA fusion from solubilized cell extracts (Figure 3B). GST alone was unable to bind Scs2-3xHA (Figure 3B). We then tested if the VAP proteins Scs2/Scs22 impact PI4P metabolism. PI4P levels were elevated 2.9-fold above wild-type in scs2D scs22D double mutant cells, as assessed by 3H-inositol labeling (Figure 3C and Table S3). The PI4P FLARE was also stabilized at the PM in scs2D scs22D double mutant cells, particularly in mother cells (the PI4P FLARE was also weakly observed at intracellular puncta) (Figure 3D). Thus, Osh3 and Scs2 interact, and the VAP proteins Scs2/ Scs22 control PI4P levels at the PM. We performed additional tests to address how the VAP proteins Scs2/Scs22 might control PI4P metabolism. Sac1GFP localized to nuclear and cER membrane compartments in scs2D scs22D double mutant cells (Figure S3A). Steady-state expression of Sac1-GFP and Osh3-GFP was unaffected in scs2D scs22D mutant cells, compared to wild-type cells (Figures S3B and S3C). Thus, whereas Sac1, Osh3, and the VAP proteins Scs2/Scs22 act in common to control PI4P metabolism, the VAP proteins are not required for Sac1 or Osh3 stability. Similar to the cortical localization of Osh3AAAT-GFP (Figure S2C), Osh3-GFP localized to cortical structures in scs2D scs22D mutant cells (Figure S3D). However, Osh3-GFP cortical patches did not appear to be associated with the peripheral ER in scs2D scs22D mutant cells (Figure S3D), consistent with a previous study (Loewen et al., 2003). These findings suggest that PI4P metabolism at PM/ER junctions may occur in regulated stages. Osh3 may initially assemble at the PM, possibly as a PI4P sensor via its PH domain. Subsequent interactions between Osh3 and the VAP proteins may lead to stimulation of Sac1 PI phosphatase activity at PM/ER junctions. The Osh and VAP Proteins Control Sac1 Phosphatase Activity To test if the VAP and Osh proteins regulate Sac1 activity, we set up an in vitro phosphatase assay using PI4P-containing lipo-
somes (PC:PS:PI4P) and microsomes containing integral Sac1-GFP prepared from cellular membranes (Figures 4A; Figure S4A). PI4P phosphatase activity was readily detected with wild-type microsomes (145 pmol/mg total protein/min, at 10 min) (Figure 4A) and was specifically due to Sac1-GFP because microsomes prepared from sac1D cells lacked activity (Figure 4A). We also did not detect significant phosphatase activity using Sac1-GFP microsomal preps against liposomes lacking PI4P (PC:PS:PtdIns liposomes) (Figure S4B). Thus, this in vitro assay followed microsome-embedded Sac1 activity against PI4P presented on liposomes, rather than PI species that might be present in the microsomes. To confirm that liposomes did not fuse with Sac1-GFP microsomes or that PI4P did not transfer from liposomes to microsomes, we incubated Sac1-GFP microsomes with liposomes containing 3H-labeled PI4P and PI(4,5)P2 (Figure S4C). Following a 30-min incubation at 37 C (the course of the phosphatase assays) (Figures 4A and Figure S4B), Sac1-GFP microsomes were immunoisolated using anti-GFP antibodies and protein A-coupled magnetic Dynabeads (Figure S4D). Unbound liposomes (and microsomes not containing Sac1-GFP) were subsequently sedimented by high-speed centrifugation. The amount of 3H-PI species in isolated Sac1-GFP microsomes, sedimented liposomes, and the resulting supernatant fraction was determined by liquid scintillation counting. Only a minor portion, 3.1% ± 0.9% of the total 3H-PI, associated with isolated Sac1GFP microsomes (Figure S4E, bound fraction). The bulk of 3 H-PI, 93.1% ± 6.2%, was present in the unbound fraction (Figure S4E). Of 3H-PI in the unbound fraction, 88.5% ± 1.3% sedimented upon high-speed centrifugation (Figure S4F, P100 fraction), indicating that the majority remained in a membrane bilayer. A small amount of the unbound fraction, 11.5% ± 1.2%, was found in the supernatant fraction, perhaps due to micelles that did not sediment (Figure S4F, S100 fraction). Thus, liposomes did not fuse with Sac1-GFP microsomes, and 3 H-PI species were not efficiently transferred from liposomes to Sac1-GFP microsomes, under conditions that resulted in efficient dephosphorylation of PI4P in liposomes (Figures 4A and Figure S4B). We also performed experiments in the presence of recombinant Osh3 (GST-Osh3588–996 including the FFAT and ORD; see Figure 5A). However, the distribution of 3 H-PI species was not significantly affected by the addition of GST-Osh3588–996 (Figures S4E and S4F). Similarly, purified his6-Osh4 (Figure 5A) did not significantly alter the distribution of 3H-PI species in analogous experiments (unpublished data). Thus, Osh3 and Osh4 did not enhance membrane fusion or 3 H-PI transfer between liposomes and Sac1-GFP microsomes. Following these control experiments, we addressed whether the VAP proteins controlled Sac1 PI phosphatase activity. Microsomes lacking Scs2 and Scs22 were impaired in the initial rate of PI4P turnover (more than 2-fold: 66 pmol/mg total protein/min, at 10 min, as compared to wild-type) (Figure 4A). In addition, scs2D scs22D microsomes were impaired in maximal Sac1 phosphatase activity as compared to wild-type microsomes (1232 ± 159 pmol PO4 released versus 2543 ± 274 pmol, respectively) (Figure 4A). Levels of Sac1-GFP and the ER protein Dpm1 were similar in both wild-type and scs2D scs22D microsomal preparations (Figure S4A). Thus, whereas Sac1-GFP levels Cell 144, 389–401, February 4, 2011 ª2011 Elsevier Inc. 393
PC:PS:PI4P (3:1:1) liposome PH
AT FF
RD
PI
4P
Scs 2 Scs 22
GOLD
h3 O Os
P
PI
Sac1 GFP
P
P13 membrane pellet microsome
malachite green PO4 assay
PC:PS:PI4P (3:1:1), 0.8 mM total lipid pmol PO4 released/μg protein
A
3000
wild type
2500 2000
scs2Δ scs22Δ
1500 1000 500
sac1Δ
0 0
5
10
15
20
25
30
time (min) D
PH
Osh
AT FF
3 ORD
diC8 PI4P diC8 PI
P
Sac1 GFP
P13 membrane pellet microsome
P
malachite green PO4 assay
pmol PO4 released/μg protein
GOL
Scs 2 Scs 22
B
diC8 PI4P, 60 μM final 3500
wild type
3000 2500
scs2Δ scs22Δ
2000
Figure 4. The VAP Proteins Scs2/Scs22 Control Sac1 PI Phosphatase Activity In Vitro (A) PI4P-containing liposomes were incubated with wild-type (circles), scs2D scs22D (squares), or sac1D (triangles) microsomes for the indicated times. Phosphate release was measured by a malachite green assay. Error bars show the SD of two experiments measured in duplicate. See also Figure S4. (B) Short acyl chain diC8 PI4P was incubated with wild-type (circles), scs2D scs22D (squares), or sac1D (triangles) microsomes for the indicated times. Error bars show the SD of two experiments measured in duplicate. (C) Short acyl chain diC8 PI4P was incubated with untreated (circles), NaCl-washed (triangles), or NaCl-washed microsomes in the presence of GST-Osh3588–996 (diamonds) for the indicated times. Error bars show the SD of two experiments measured in duplicate. See also Figure S5A.
1500 1000
sac1Δ
500
membranes, possibly through the FFATcontaining Osh proteins, Osh2 and Osh3. We then tested if the Osh proteins stimtime (min) wild type NaCl-washed microsomes C ulate Sac1 activity in vitro. Because Sac1 -/+ GST-Osh3588-996: levels were reduced in oshD:CEN osh4ts diC8 PI4P, 60 μM final GST FFAT 3500 cell lysates (Figure S1B and Figure S4A), diC8 PI4P Osh3 washed + 0.2 μM GST-Osh3588-996 ORD 3000 we did not use microsomes prepared diC8 PI from these mutant cells. However, we 2500 P untreated found that Osh3 was depleted from Sac1 2000 membrane fractions by incubation in GFP 1500 P high salt (unpublished data). Sac1-GFP 1000 microsomes were then subjected to P13 membrane pellet NaCl-washed microsomes malachite green 500 microsome a high-salt wash to deplete peripherally PO4 assay associated membrane proteins, such as 0 0 5 10 15 20 25 30 Osh3. Sac1-containing microsomes time (min) were not active against PI4P-containing liposomes following the high-salt wash. Thus, we were not able to test whether recombinant Osh3 stimulated Sac1 in trans were similar, Sac1 activity was impaired in scs2D scs22D micro- activity (unpublished data). However, NaCl-washed microsomes somes. Importantly, to our knowledge, these experiments displayed a reduced rate of diC8 PI4P turnover (more than provide the first in vitro demonstration of Sac1 activity in trans 2-fold: 600 ± 115 pmol/mg total protein/min, after 1 min, as compared to untreated microsomes at 1600 ± 206 pmol/mg total and indicate that the yeast VAP proteins promote this activity. Wild-type and scs2D scs22D microsomes demonstrated protein/min, after 1 min) (Figure 4C) and maximal Sac1 phosphasimilar activity against a soluble substrate, short acyl chain tase activity compared to untreated microsomes (1746 ± 213 diC8 PI4P (in the absence of other lipids) (Figure 4B), indicating pmol PO4 released versus 2789 ± 211 pmol, respectively, after that both microsomal preps contained similar intrinsic Sac1 30 min) (Figure 4C). To test if impaired Sac1 activity in activity. As in the PI4P liposome assay, microsomes isolated NaCl-washed microsomes was due to the loss of Osh proteins, from sac1D cells did not display activity against diC8 PI4P (Fig- purified GST-Osh3588–996 (see Figure 5B) was added back to the ure 4B). Notably, both wild-type and scs2D scs22D microsomes reaction. Strikingly, 0.2 mM GST-Osh3588–996 restored activity in displayed increased activity for the soluble diC8 PI4P substrate salt-washed microsomes to levels nearly identical to untreated (approximately 1500 pmol PO4 released/mg total protein/min, Sac1-GFP microsomes (1450 ± 155 pmol/mg total protein/min after 1 min) (Figure 4B). At later time points both preps continued after 1 min, and 3000 ± 176 pmol PO4 released after 30 min) (Figto display similar turnover rates, although slower likely due to ure 4C). Thus, recombinant Osh3 stimulated full-length Sac1 in reduced substrate concentrations. Thus, the VAP proteins microsomes, providing the first evidence that ORPs control Scs2/Scs22 may act to link Sac1 to substrates on opposing Sac1 PI phosphatase activity in vitro. 0
5
pmol PO4 released/μg protein
Scs 2 Scs 22
0
394 Cell 144, 389–401, February 4, 2011 ª2011 Elsevier Inc.
10
15
20
25
30
P
PI
Sac11-522
MW 75 kDa
P
malachite green PO4 assay
B
50 kDa
PC:PS:PI4P (3:1:1) liposomes, 0.8 mM lipid total
3000
pmol PO4 released
6 -S ac 1 1-52 2 GS T-O sh 3 58899 6 his 6 -O sh 4
PI
4P
Osh3 or Osh4 ORD
Figure 5. The ORDs Stimulate Sac1 Activity In Vitro
E. coli expressed and purified proteins
PC:PS:PI4P (3:1:1)
his
A
0.10 μM Sac11-522 +1 μM GST-Osh3588-996
2500
0.33 μM Sac11-522
2000
0.10 μM Sac11-522
1500
(A) Schematic of the reconstituted Sac1 phosphatase assay. PI4P-containing liposomes were incubated with his6-Sac11–522, GST-Osh3588–996 or his6-Osh4. Phosphate release was measured by a malachite green assay. The Coomassie-stained gel shows the recombinant proteins used in the in vitro assays. The dot indicates a GSTOsh3588–996 degradation product. (B) GST-Osh3588–996 stimulates Sac1 in vitro. Results from in vitro phosphatase assays including 0.33 mM his6Sac11–522 (squares), 0.1 mM his6-Sac11–522 (diamonds), 1 mM GST-Osh3588–996 and 0.1 mM Sac11–522 (circles), and background levels in the presence of 1 mM GSTOsh3588–996 alone (triangles). Error bars show SD of two experiments measured in duplicate. See also Figure S5B. (C) Osh4 stimulates Sac1 in vitro. Results from in vitro phosphatase assays with 0.33 mM his6-Sac11–522 (squares), 0.03 mM his6-Sac11–522 (diamonds), 0.03 mM his6-Sac11–522 and 12 mM his6-Osh4 (circles), and 12 mM his6-Osh4 alone (triangles). Error bars show SD of two experiments measured in duplicate.
1000 500
1 μM GST-Osh3588-996
Reconstitution of Sac1 Phosphatase Activation by ORDs time (min) Because the Osh3 ORD stimulated Sac1 activity in microsomes, we tested whether Osh proteins C PC:PS:PI4P (3:1:1) liposomes, 0.8 mM lipid total directly stimulate Sac1. We reconstituted Sac1 3000 phosphatase activity in vitro using purified 0.33 μM Sac11-522 2500 components at known concentrations. Sac1 0.03 μM Sac11-522 lacking its transmembrane domains was 2000 +12 μM Osh4 expressed and purified from bacteria (his61500 Sac11–522) and incubated with PI4P-containing liposomes (Figure 5A). Purified his6-Sac11–522 1000 dephosphorylated PI4P in a dose and time500 dependent manner (400 pmol/min/mg protein 0.03 μM Sac11-522 for 0.33 mM his6-Sac11–522 after 5 min) (Fig0 12 μM Osh4 0 5 10 15 20 25 30 ure 5B). At 0.1 mM his6-Sac11–522, the turnover time (min) rate was reduced to approximately 250 pmol PO4 released/min/mg protein after 5 min (Figures 5B and Figure S5D). A mutant form of Sac1 bearing substitutions in both cysteine residues To test if GST-Osh3588–996 required the VAP proteins to stimu- in the CX5R active site (his6-Sac1C392S, C395S) (Figure S5C) did late Sac1, Sac1-GFP microsomes prepared from scs2D scs22D not display phosphatase activity against PI4P-containing lipomutant cells were washed with high salt to deplete peripheral somes (Figure S5D). Importantly, addition of the Osh3 ORD membrane proteins. As expected, NaCl-washed scs2D scs22D (GST-Osh3588–996) (Figure 5B) stimulated Sac1 phosphatase microsomes displayed impaired rates of diC8 PI4P turnover, activity 3-fold against PI4P-containing liposomes, as 0.1 mM compared to untreated scs2D scs22D microsomes (more than his6-Sac11–522 in the presence of 1 mM GST-Osh3600–996 2-fold: 577 ± 151 pmol/mg total protein/min, after 1 min, as displayed a turnover rate similar to 0.33 mM his6-Sac11–522 (Figcompared to untreated scs2D scs22D microsomes at 1315 ± ure 5B). As a control, 1 mM GST alone did not increase Sac1 186 pmol/mg total protein/min, after 1 min) (Figure S5A). Salt- activity (Figure S5D). GST-Osh3588–996 did not catalyze PI4P washed scs2D scs22D microsomes also displayed reduced turnover in the absence of his6-Sac11–522 (Figure 5B) or in the maximal Sac1 phosphatase activity, compared to untreated presence of inactive his6-Sac1C392S, C395S (Figure S5D). scs2D scs22D microsomes (1364 ± 223 pmol PO4 released versus Because the ORD is common to all ORPs, we addressed if 2652 ± 206 pmol, respectively, after 30 min) (Figure S5A). Interest- Sac1 activation is a conserved function for the Osh proteins. ingly, 0.2 mM GST-Osh3588–996 restored full activity in salt-washed We then tested if full-length Osh4/Kes1 stimulated Sac1 phosscs2D scs22D microsomes (1489 ± 145 pmol/mg total protein/min phatase activity in vitro. In these assays we lowered the after 1 min, and 2994 ± 176 pmol PO4 released after 30 min) (Fig- his6-Sac11–522 concentration (to 0.033 mM with activity near ure S5A). Thus, the Osh3 ORD stimulated Sac1 activity against background levels) (Figure 5C). In the presence of 12 mM his6Osh4, 0.033 mM his6-Sac11–522 activity increased 10-fold over diC8 PI4P independently of the VAP proteins Scs2 and Scs22. 0
5
10
15
20
25
30
pmol PO4 released
0
Cell 144, 389–401, February 4, 2011 ª2011 Elsevier Inc. 395
Osh4 structure bound to sterol:
B
Figure 6. PI Function
E. coli expressed & purified proteins: -O sh 4 3E 6
his
6
his
6 -O sh 4
Δ1-29
L111D
his
-S ac 1 1-52
2
A
3E R236E K242E K243E
C
in vitro phosphatase assay: 10 min, 37oC PC:PS:PI4P
2000
ORD
4P
wt or mutant Osh4 PI
pmol PO4 released
2500
1500
Binding
Regulates
Osh4
(A) The structure of Osh4/Kes1 (Im et al., 2005) is shown with mutants used in this study. Osh4D1–29 (orange) and Osh4L111D (green) are defective in sterol binding (Im et al., 2005). Osh43E (R236E, R242E, R243E; blue) is impaired in PI binding (see Figures S6A and S6B) (Li et al., 2002). Cholesterol bound within the structure is shown in yellow. (B) Coomassie-stained gel showing recombinant, purified his6-Sac11–522 and wild-type and mutant his6-Osh4 proteins. (C) In vitro phosphatase assays using 0.1 mM his6Sac11–522 and wild-type and mutant Osh43E proteins (1.2 mM). PI4P liposomes (PC:PS:PI4P, 3:1:1) were tested. Error bars show the SD of two experiments each measured in duplicate. (D) In vitro phosphatase assays using 0.1 mM his6Sac11–522 and wild-type and mutant Osh43E proteins (1.2 mM) against short acyl chain diC8 PI4P. Error bars show the SD of two experiments each measured in duplicate. See also Figure S6C.
P
PI
Sac11-522 1000
liposomes, compared to phosphatidylinositol-containing liposomes (FigureS6A). GST-Osh3588–996 also sedimented malachite green 0 PO4 assay more readily with PI4P-containing lipo+ Osh4 + 3E somes than it did with PtdIns-containing 0.10 μM Sac11-522 liposomes (Figure S6A). We tested if PI + 1.2 μM Osh4 or Osh43E binding was necessary for Osh4 function in vitro and in vivo. Substitution of three D basic residues (R236E, K242E, K243E; in vitro phosphatase assay: 10 min, 25oC termed 3E) (Figure 6A) impairs Osh4 PI binding (Li et al., 2002). The mutant his68000 wt or mutant Osh43E protein was stably expressed 7000 Osh4 diC8 PI4P and purified (Figure 6B). We confirmed 6000 that his6-Osh43E was defective in PI ORD 5000 binding using an overlay assay (FigP 4000 ure S6B). In contrast, wild-type his61-522 Sac1 Osh4 bound PI3P, PI4P, and PI(4,5)P2 3000 PI (in order of binding affinity) (Figure S6B). 2000 P Notably, his6-Osh43E did not stimulate 1000 Sac1 activity against PI4P in liposomes malachite green 0 (Figure 6C). In addition, his6-Osh43E was PO4 assay + Osh4 + 3E impaired in Sac1 activation against short 0.10 μM Sac11-522 acyl chain diC8 PI4P, as compared to + 1.2 μM Osh4 or Osh43E wild-type his6-Osh4 (2-fold versus 6-fold stimulation, respectively) (Figure 6D). Consistent with these in vitro results, the course of the experiment, reaching that of 0.33 mM his6- cellular PI4P levels were 17-fold above wild-type in oshD:CEN Sac11–522 (Figure 5C). In addition, at high concentrations osh4ts cells coexpressing Osh43E at 38 C (Table S3), indicating (300 mM) of short acyl chain diC8 PI4P, 1.2 mM his6-Osh4 stimu- that PI binding is essential for Osh4 function. Because the Osh proteins bind sterol lipids in vitro (Im et al., lated 0.1 mM his6-Sac11–522 activity greater than 6-fold (Figures 6D and Figure S6C). Thus, the ORDs from Osh3 and Osh4 2005; Raychaudhuri et al., 2006), we tested if sterol lipids regulate Osh-stimulated Sac1 activity. Addition of 250 mM cholesterol directly activated Sac1 in vitro. (>800-fold above the Kd of Osh4 for cholesterol) (Im et al., 2005) did not significantly affect Osh4-stimulated Sac1 activity against PI and Sterol Binding Control Osh4 Function Osh4/Kes1 binds PI lipids in vitro (Li et al., 2002). Consistent with diC8 PI4P (Figure S6C). We also measured in vivo PI levels in this, his6-Osh4 sedimented more efficiently with PI4P-containing oshD:CEN osh4ts cells coexpressing wild-type OSH4 or the P
0.33 μM Sac11-522
0.10 μM Sac11-522
pmol PO4 released
0.10 μM Sac11-522
500
396 Cell 144, 389–401, February 4, 2011 ª2011 Elsevier Inc.
A
Figure 7. The ORD Interacts with Sac1 and Model for Regulation of PI4P Metabolism by Osh3, Scs2/Scs22, and Sac1 at PM/ER Membrane Contact Sites
IP: α-HA
Osh7-3xHA: Sac1-13xmyc: + Sec61-13xmyc: -
+ +
-
+
-
+
+
Sac1-13xmyc
WB: α-myc
Sec61-13xmyc
WB: α-HA
Osh7-3xHA 5% inputs Sac1-13xmyc
WB: α-myc
Sec61-13xmyc
Emission (counts/second x 104)
B
NBDSac 1-5 1 22
Sac1C341NBD + lipos + GST-PHFAPP1
7 6
Sac1C341NBD + lipos
5
Sac1C341NBD + lipos + GST-Osh3588-996
4 3 2 1
Sac1C341NBD buffer
PC:PS:PI4P liposome (0.1 mM total)
Sac1 1-52
ORD
(A) Sac1-13xmyc specifically interacts with Osh73xHA. Membrane fractions from cells expressing Osh7-3xHA and Sac1-13xmyc or Sec61-13xmyc were incubated with cross-linkers, solubilized, immunoprecipitated with anti-HA beads, and analyzed by immunoblotting to detect Osh7-Sac1 complexes. See also Figure S7A. (B) GST-Osh3588–996 induces a conformational shift in the Sac1 catalytic domain. Emission scans of 0.06 mM Sac1C341NBD in buffer (red), in the presence of liposomes (PC:PS:PI4P, 3:1:1, 0.1 mM lipid total; dark gray), liposomes and 0.8 mM GST-Osh3588–996 (blue), liposomes with 0.8 mM GST-PHFAPP1 (green), and buffer alone (light gray). See also Figure S7. (C) Model for PI4P turnover at PM/ER membrane contact sites. High PM PI4P levels recruit and activate Osh3 at ER/PM contact sites. Interactions between Osh3 and the VAP proteins Scs2/Sc22 activate ER-localized Sac1.
2
levels than Osh3 (2350 Osh7 molecules/ cell versus 600 Osh3 molecules/cell) (Ghaemmaghami et al., 2003). Osh7 Emission wavelength (nm) localizes to the cER even though it lacks a PH domain and FFAT motif (Schulz C et al., 2009) (Figure 1C). A membrane Low PM PI4P High PM PI4P pellet fraction containing Osh7-3xHA extracellular extracellular and Sac1-13xmyc was incubated with PM PM cross-linkers, solubilized, and processed for co-immunoprecipitation against the inactive P PH PI P 3xHA tag on Osh7. A small fraction of Osh3: FFAT Osh3ORD PI4P PH Sac1-13xmyc specifically isolated with VAP Osh7-3xHA (Figure 7A), suggesting that proteins: ORD this interaction was transient or possibly indirect because it required crosslinking. Sac1 ER (cis) However, Sac1 is far more abundant in cells than Osh7 (48,000 Sac1 molecules/ ER lumen ER cell) (Ghaemmaghami et al., 2003); thus, only a small fraction of Sac1 is expected ER lumen to co-isolate with Osh7. Importantly, Osh7-3xHA did not crosslink to another abundant ER membrane protein, Sec61sterol-binding defective mutants, osh4D1-29 or osh4L111D (see 13xmyc (Figure 7A). We also observed crosslinking between Figure 6A) (Im et al., 2005). In oshD:CEN osh4ts cells coexpress- 3xHA-Osh3 and Sac1-13xmyc (Figure S7A). As a control the ing either Osh4D1–29 or Osh4L111D, PI4P levels were 21- and 19- ER transmembrane protein Wbp1 was not isolated with 3xHAfold above wild-type levels (Table S3). Thus, whereas cholesterol Osh3 (unpublished data). Together, these results suggested did not alter Sac1 activity in vitro, sterol binding was essential for that multiple Osh proteins interact with Sac1 and that the ORD Osh4-mediated PI4P metabolism in vivo. is responsible. Next, we tested if the ORD and Sac1 domains physically interact in vitro. We were unable to detect a stable complex The Osh Proteins Interact with the Sac1 Catalytic between the purified Sac1 domain and Osh3 or Osh4 ORDs Domain Because the Osh3 and Osh4 ORDs activated Sac1 in vitro, we in solution (unpublished data). For this reason we develexamined if Osh proteins interact with Sac1. We initially used oped assays to monitor Sac1 and ORD interactions on Osh7 for these experiments because it is expressed at higher a membrane bilayer. The Sac11–522 domain contains three Scs2
Sac
(tran1 s)
590
Scs22 Scs2
560
FF AT
530
Scs22
>10 nm
500
2-fold) (Figure S7D). Thus, the reduction in Sac1C341NBD fluorescence intensity by Osh4 was not due to decreased Sac1 domain membrane binding. Accordingly, the Osh-induced shift in Sac1C341NBD fluorescence intensity may reflect an Osh-induced conformational change in the Sac1 catalytic domain, consistent with Osh-stimulated Sac1 phosphatase activity in vitro. Moreover, an Osh-induced increase in membrane association may contribute to Sac1 activation. DISCUSSION Sac1-like phosphatases are essential regulators of PI-signaling networks. We show that the Osh proteins control Sac1 activity. Members of the ORP family are known to bind sterol and PI lipids and are implicated in disease (Ngo et al., 2010). We demonstrate a new role for this family: control of PI4P metabolism through Sac1 PI phosphatases. Our results also explain how ER-localized Sac1 can regulate PM PI4P levels in trans: via the function of Osh proteins at PM/ER membrane contact sites as sensors of PI4P at the PM and activators of the Sac1 phosphatase in the ER. 398 Cell 144, 389–401, February 4, 2011 ª2011 Elsevier Inc.
PM PI4P Metabolism Occurs at PM/ER Membrane Contact Sites Recent studies have proposed that Sac1 regulates PI4P in cis at the ER and Golgi, as well as in trans at the PM from the ER (Baird et al., 2008; Manford et al., 2010). In accord, Sac1 has been shown to regulate PI4P levels at the ER and Golgi (Faulhammer et al., 2007; Foti et al., 2001). Our findings and additional evidence indicate that ER-localized Sac1 controls PM PI4P pools. First, PI4P accumulates at the PM in cells lacking Sac1 (Figure 1E) (Roy and Levine, 2004). Second, Sac1 does not traffic to the PM and was not stabilized at the PM in an endocytosis mutant (Tahirovic et al., 2005). Likewise, cells expressing a form of Sac1 that is retained in the ER do not display elevated PI4P levels (Konrad et al., 2002). Last, we demonstrate that fulllength Sac1 in microsomes dephosphorylates PI4P in trans on distinct liposomes (Figure 4). We propose that the Osh and VAP proteins function as regulators of Sac1 activity at membrane contact sites. Notably, Osh3 localization to PM/ER membrane sites was dependent on PI4P synthesis (Figure 2C), and a mutant form lacking the PH domain was destabilized from cortical patches (Figure S2C). We suggest that the PH domain-containing protein Osh3 acts as a sensor of PI4P levels at the PM and subsequently activates the Sac1 PI phosphatase at the ER (Figure 7C). Our study also highlights a role for Scs2 and Scs22 in PI4P metabolism. Cells lacking Scs2 and Scs22 display elevated PM PI4P levels (Figure 3). Moreover, Scs2 and Scs22 stimulate Sac1 activity against PI4P in liposomes (Figure 4A). Scs2 and Scs22 may activate Osh3 at PM/ER membrane contact sites to control PM PI4P levels (Figure 7C). The yeast VAP proteins may also aid in the formation of membrane junctions. Scs2 and Scs22 have been implicated in PM/ER membrane contact site formation (Loewen et al., 2007). In addition, Scs2 binds anionic phospholipids, including PI4P, in vitro (Kagiwada and Hashimoto, 2007). Mechanisms for Regulation of Sac1 Activity by the Osh Proteins Multiple inputs control Osh protein function. We propose that PI4P binding to the Osh3 PH domain and Scs2 interactions via the FFAT motif activate Osh3 at PM/ER membrane contact sites (Figure 7C). These interactions may then facilitate interactions between the Osh3 ORD and downstream target proteins such as Sac1 (Figure 7C). Thus, Osh3 may serve as a ‘‘co-incidence detector’’ of PI4P, the VAP proteins, and Sac1. We also found that PI and sterol binding control Osh4/Kes1 function (Figure 6; Table S3). The Osh proteins control sterol localization in vivo (Beh and Rine, 2004) and transfer sterol lipids in vitro (Raychaudhuri et al., 2006). Osh4/Kes1 function required sterol binding in vivo (Table S3) but was not affected by the addition of cholesterol in vitro (Figure S6C). The concentrations of Osh4 or PI4P in our in vitro studies may have bypassed a requirement for sterol binding. In vivo, the ability of the Osh proteins to extract sterol lipids may control membrane bilayer rigidity and curvature, and thus, formation of membrane contact sites (Figure 7C). Notably, the Osh proteins can tether liposomes in vitro (Schulz et al., 2009). Alternatively, sterol binding may control Osh protein localization in vivo. Consistent with this idea, OSBP localizes to the
ER when bound to cholesterol and to the Golgi when bound to 25-hydroxycholesterol (Peretti et al., 2008). Osh4/Kes1 also binds PI lipids in vitro (Figure S6;) (Li et al., 2002). This may be a general feature of ORDs because the Osh3 ORD showed increased affinity for PI4P liposomes (Figure S6). A form of Osh4 impaired in PI binding (Figure S6) did not stimulate Sac1 activity in vitro (Figure 6). Although our results did not indicate that Osh3 transfers PI lipids (Figure S4), the Osh proteins may bind PI lipids and present substrates to Sac1 at the membrane interface. Alternatively, PI binding may be necessary for productive interactions with Sac1. This idea is supported by our findings that Osh3 and Osh4 induced changes in Sac1C341NBD spectral properties in the presence of liposomes, but PI binding-defective Osh43E was incapable of interacting with Sac1 (Figures 7 and Figure S7). The Osh-induced effects on Sac1C341NBD signal intensity may reflect a conformational change in Sac1 or a shift in the orientation of Sac1 with respect to the membrane bilayer. In addition PI binding may be necessary for Osh-stimulated Sac1 membrane association (Figure S7D).
ER/Golgi, ER/endosome, ER/lysosome). This would allow the cell to maintain a wide range of Sac1 phosphatase activity and permit use of a common phosphatase (Sac1) to regulate PI lipids at multiple sites in the cell. Membrane contact sites facilitate communication between organelles through lipid transfer and metabolism (Lev, 2010). This elegant system may allow cells to balance not only PI levels but also other lipids at ER membrane contact sites because the ER is a major site of lipid synthesis. Thus, modulation of PI levels by Sac1, the Osh proteins, and VAP proteins may allow signaling between the ER and compartments along the secretory and endocytic systems, thus linking ER lipid biosynthesis to multiple membrane compartments in the cell. Interestingly, both Sac1 and the yeast VAP protein Scs2 are implicated in phospholipid metabolism (Loewen et al., 2004; Rivas et al., 1999). In addition, Sac1 has recently been identified as a component of a protein complex at the ER involved in sphingolipid metabolism (Breslow et al., 2010). Thus, future studies on Sac1 may reveal new insights into the links between membrane homeostasis and PI-signaling pathways.
Conserved Functions for the Osh/ORP and VAP Proteins in PI Metabolism ORP and VAP proteins may share conserved roles in PI homeostasis. Each of the Osh proteins partially rescues the PI4P metabolism defects in oshD:CEN osh4ts cells (unpublished data). Both Osh3 and Osh4 stimulate Sac1 PI phosphatase activity in vitro (Figure 4, Figure 5, and Figure 6). Osh6 and Osh7 localize to the peripheral ER (Schulz et al., 2009), and Osh7 interacts with Sac1 in crosslinking experiments (Figure 7A). Osh1 localizes to nuclear-vacuole junctions to participate in lipid metabolism (Levine and Munro, 2001). Osh4 regulates PI4P at the Golgi (Fairn et al., 2007; Li et al., 2002) where it may regulate Sac1 activity. Likewise, mammalian ORP and VAP proteins have been implicated at membrane contact sites. A recent study found ORP1L and VAP at ER/endosome membrane contact sites (Rocha et al., 2009). Mammalian VAP proteins control PI4P metabolism at ER/Golgi membrane contact sites (Peretti et al., 2008). In addition, bound to 25-hydroxycholesterol, OSBP localizes to the Golgi and ER/Golgi contact sites (Peretti et al., 2008). ORP and VAP proteins may regulate additional PI signaling networks. Osh7 and Scs2 overexpression rescued the growth defects of mutant yeast cells with toxic PI3P levels due to loss of the Sac1-like PI phosphatases Ymr1 and Sjl3, myotubularin and synaptojanin orthologs (Parrish et al., 2005). Cells lacking Osh or VAP protein function also display increased cellular PI3P levels (Table S3). It is unclear which Osh proteins regulate PI3P metabolism and if Osh and VAP proteins regulate other Sac1-like PI phosphatases, such as synaptojanins or myotubularin. Yet, cells lacking Osh function accumulate PI4P levels similar to sac1 sjl2 sjl3 triple mutant cells (Figure 1) (Foti et al., 2001), suggesting that they may control Sac1-like activity encoded by the yeast synaptojanins. We propose that the Osh proteins act as sensors of PI levels and regulate the Sac1 PI phosphatase at membrane contact sites. By retaining Sac1 at the ER, cells can recruit Sac1 to distinct organelle contact sites as needed (e.g., ER/PM,
EXPERIMENTAL PROCEDURES Additional details of the Experimental Procedures are provided in the Extended Experimental Procedures. Yeast Strains, Plasmids, and Media Descriptions of strains and plasmids used in this study are in Table S1 and Table S2. Gene deletions and epitope tags were introduced by homologous recombination. Standard techniques were used for yeast and bacterial growth. Expression of recombinant proteins in bacteria is described in the Extended Experimental Procedures. Fluorescence Microscopy Microscopy was performed on mid-log cultures at the indicated temperatures. Images were obtained with a DeltaVision system equipped with an Olympus microscope, a 1003 objective, and a Cool Snap HQ digital camera. Images were deconvolved with softWoRx 3.5.0 software. All results are based on observations of >100 cells. PI Analysis Cellular PI levels were measured as described (Baird et al., 2008). Additional details are in the Extended Experimental Procedures. Protein Expression Levels Cells incubated at the indicated temperatures were harvested, lysed, and the resulting extracts analyzed by SDS-PAGE and immunoblotting with the following antibodies: a-Myc (9E10), a-HA (12CA5), a-Pep12, a-Dpm1, and a-GFP. Protein-Binding Assays For Scs2-3xHA and GST-Osh31–613 binding, GST or GST-Osh31–613 immobilized on glutathione sepharose 4B was incubated with Scs2-3xHA lysates solubilized in PBS containing 0.5% Tween 20, 0.1 mM EDTA, and protease inhibitors. Beads were washed and analyzed by SDS-PAGE and immunoblotting. For Sac1-13xmyc crosslinking to 3xHA-Osh3 and Osh7-3xHA, membrane fractions were incubated with 2 mM DSP and DTBP. Complexes were immunoisolated, reduced to cleave cross-linkers, and analyzed by SDS-PAGE and immunoblotting. Sac1 Phosphatase Assays For microsome assays, membrane fractions were sonicated in 50 mM Tris (pH 6.8), 150 mM NaCl, 2 mM DTT, 1 mM EDTA to form microsomes.
Cell 144, 389–401, February 4, 2011 ª2011 Elsevier Inc. 399
Liposomes (0.6 mM PC:0.2 mM PS:0.2 mM PI4P) were prepared by sonicating hydrated lipids in reaction buffer (50 mM Tris [pH 6.8], 150 mM NaCl, 2 mM DTT). Microsomes were incubated with liposomes or diC8 PI4P, and reactions were stopped with 50 mM N-ethylmaleimide (NEM). Phosphate release was measured by the addition of Biomol Green and measured at OD620. For assays with recombinant Sac11–522 and Osh proteins, proteins were added at indicated concentrations to liposomes or diC8 PI4P in reaction buffer at the indicated temperatures. Cholesterol-containing liposomes (0.6 mM PC:0.2 mM PS:0.2 mM PI4P:0.2 mM cholesterol) were prepared as described above. Reactions were stopped at the indicated times with NEM, and phosphate release was assayed as described above. Lipid-Binding Assays For overlay assays, lipids solubilized in chloroform:methanol were spotted onto nitrocellulose as indicated. Membranes were incubated overnight with purified proteins. Membranes were washed, and bound protein was detected by immunoblotting. For liposome sedimentation, liposomes were prepared as described, incubated with purified recombinant proteins, and centrifuged at 100,000 3 g. The supernatant and pellet fractions were resolved by SDSPAGE, and proteins were detected by Coomassie staining. NBD Protein Labeling and Fluorescence Spectroscopy Purified his6-Sac1C392S,C395S labeled with IANBD [N,N’-dimethyl-N-(iodoacetyl)-N’-(7-nitrobenz-2-oxa-1,3-diazolyl)ethylenediamine] was dialyzed to remove excess IANBD. See the Extended Experimental Procedures for further details on fluorescence spectroscopy.
SUPPLEMENTAL INFORMATION
D’Angelo, G., Vicinanza, M., Di Campli, A., and De Matteis, M.A. (2008). The multiple roles of PtdIns(4)P–not just the precursor of PtdIns(4,5)P2. J. Cell Sci. 121, 1955–1963. Fairn, G.D., Curwin, A.J., Stefan, C.J., and McMaster, C.R. (2007). The oxysterol binding protein Kes1p regulates Golgi apparatus phosphatidylinositol-4-phosphate function. Proc. Natl. Acad. Sci. USA 104, 15352–15357. Faulhammer, F., Kanjilal-Kolar, S., Knodler, A., Lo, J., Lee, Y., Konrad, G., and Mayinger, P. (2007). Growth control of Golgi phosphoinositides by reciprocal localization of sac1 lipid phosphatase and pik1 4-kinase. Traffic 8, 1554–1567. Foti, M., Audhya, A., and Emr, S.D. (2001). Sac1 lipid phosphatase and Stt4 phosphatidylinositol 4-kinase regulate a pool of phosphatidylinositol 4-phosphate that functions in the control of the actin cytoskeleton and vacuole morphology. Mol. Biol. Cell 12, 2396–2411. Ghaemmaghami, S., Huh, W.K., Bower, K., Howson, R.W., Belle, A., Dephoure, N., O’Shea, E.K., and Weissman, J.S. (2003). Global analysis of protein expression in yeast. Nature 425, 737–741. Guo, S., Stolz, L.E., Lemrow, S.M., and York, J.D. (1999). SAC1-like domains of yeast SAC1, INP52, and INP53 and of human synaptojanin encode polyphosphoinositide phosphatases. J. Biol. Chem. 274, 12990–12995. Im, Y.J., Raychaudhuri, S., Prinz, W.A., and Hurley, J.H. (2005). Structural mechanism for sterol sensing and transport by OSBP-related proteins. Nature 437, 154–158. Kagiwada, S., and Hashimoto, M. (2007). The yeast VAP homolog Scs2p has a phosphoinositide-binding ability that is correlated with its activity. Biochem. Biophys. Res. Commun. 364, 870–876. Kaiser, S.E., Brickner, J.H., Reilein, A.R., Fenn, T.D., Walter, P., and Brunger, A.T. (2005). Structural basis of FFAT motif-mediated ER targeting. Structure 13, 1035–1045.
Supplemental Information includes Extended Experimental Procedures, seven figures, and three tables and can be found with this article online at doi:10. 1016/j.cell.2010.12.034.
Konrad, G., Schlecker, T., Faulhammer, F., and Mayinger, P. (2002). Retention of the yeast Sac1p phosphatase in the endoplasmic reticulum causes distinct changes in cellular phosphoinositide levels and stimulates microsomal ATP transport. J. Biol. Chem. 277, 10547–10554.
ACKNOWLEDGMENTS
Lev, S. (2010). Non-vesicular lipid transport by lipid-transfer proteins and beyond. Nat. Rev. Mol. Cell Biol. 11, 739–750.
We thank C. Beh for strains and plasmids. We are grateful to S. Weys for technical assistance. We thank members of the S.D.E. laboratory, C. McMaster, and S. Henry for discussions, and A. Bretscher, C. Fromme, and D. Teis for comments on the manuscript. This work was supported by funds from the Weill Institute for Cell and Molecular Biology (S.D.E.). Received: June 2, 2010 Revised: September 17, 2010 Accepted: December 10, 2010 Published: February 3, 2011 REFERENCES
Levine, T.P., and Munro, S. (2001). Dual targeting of Osh1p, a yeast homologue of oxysterol-binding protein, to both the Golgi and the nucleus-vacuole junction. Mol. Biol. Cell 12, 1633–1644. Li, X., Rivas, M.P., Fang, M., Marchena, J., Mehrotra, B., Chaudhary, A., Feng, L., Prestwich, G.D., and Bankaitis, V.A. (2002). Analysis of oxysterol binding protein homologue Kes1p function in regulation of Sec14p-dependent protein transport from the yeast Golgi complex. J. Cell Biol. 157, 63–77. Liu, Y., and Bankaitis, V.A. (2010). Phosphoinositide phosphatases in cell biology and disease. Prog. Lipid Res. 49, 201–217. Liu, Y., Boukhelifa, M., Tribble, E., Morin-Kensicki, E., Uetrecht, A., Bear, J.E., and Bankaitis, V.A. (2008). The Sac1 phosphoinositide phosphatase regulates Golgi membrane morphology and mitotic spindle organization in mammals. Mol. Biol. Cell 19, 3080–3096.
Audhya, A., Foti, M., and Emr, S.D. (2000). Distinct roles for the yeast phosphatidylinositol 4-kinases, Stt4p and Pik1p, in secretion, cell growth, and organelle membrane dynamics. Mol. Biol. Cell 11, 2673–2689.
Loewen, C.J., Roy, A., and Levine, T.P. (2003). A conserved ER targeting motif in three families of lipid binding proteins and in Opi1p binds VAP. EMBO J. 22, 2025–2035.
Baird, D., Stefan, C., Audhya, A., Weys, S., and Emr, S.D. (2008). Assembly of the PtdIns 4-kinase Stt4 complex at the plasma membrane requires Ypp1 and Efr3. J. Cell Biol. 183, 1061–1074.
Loewen, C.J., Gaspar, M.L., Jesch, S.A., Delon, C., Ktistakis, N.T., Henry, S.A., and Levine, T.P. (2004). Phospholipid metabolism regulated by a transcription factor sensing phosphatidic acid. Science 304, 1644–1647.
Beh, C.T., and Rine, J. (2004). A role for yeast oxysterol-binding protein homologs in endocytosis and in the maintenance of intracellular sterol-lipid distribution. J. Cell Sci. 117, 2983–2996.
Loewen, C.J., Young, B.P., Tavassoli, S., and Levine, T.P. (2007). Inheritance of cortical ER in yeast is required for normal septin organization. J. Cell Biol. 179, 467–483.
Blagoveshchenskaya, A., Cheong, F.Y., Rohde, H.M., Glover, G., Knodler, A., Nicolson, T., Boehmelt, G., and Mayinger, P. (2008). Integration of Golgi trafficking and growth factor signaling by the lipid phosphatase SAC1. J. Cell Biol. 180, 803–812.
Manford, A., Xia, T., Saxena, A.K., Stefan, C., Hu, F., Emr, S.D., and Mao, Y. (2010). Crystal structure of the yeast Sac1: implications for its phosphoinositide phosphatase function. EMBO J. 29, 1489–1498.
Breslow, D.K., Collins, S.R., Bodenmiller, B., Aebersold, R., Simons, K., Shevchenko, A., Ejsing, C.S., and Weissman, J.S. (2010). Orm family proteins mediate sphingolipid homeostasis. Nature 463, 1048–1053.
400 Cell 144, 389–401, February 4, 2011 ª2011 Elsevier Inc.
Nemoto, Y., Kearns, B.G., Wenk, M.R., Chen, H., Mori, K., Alb, J.G., Jr., De Camilli, P., and Bankaitis, V.A. (2000). Functional characterization of a mammalian Sac1 and mutants exhibiting substrate-specific defects in phosphoinositide phosphatase activity. J. Biol. Chem. 275, 34293–34305.
Ngo, M.H., Colbourne, T.R., and Ridgway, N.D. (2010). Functional implications of sterol transport by the oxysterol-binding protein gene family. Biochem. J. 429, 13–24. Parrish, W.R., Stefan, C.J., and Emr, S.D. (2005). PtdIns(3)P accumulation in triple lipid-phosphatase-deletion mutants triggers lethal hyperactivation of the Rho1p/Pkc1p cell-integrity MAP kinase pathway. J. Cell Sci. 118, 5589– 5601. Peretti, D., Dahan, N., Shimoni, E., Hirschberg, K., and Lev, S. (2008). Coordinated lipid transfer between the endoplasmic reticulum and the Golgi complex requires the VAP proteins and is essential for Golgi-mediated transport. Mol. Biol. Cell 19, 3871–3884. Raychaudhuri, S., Im, Y.J., Hurley, J.H., and Prinz, W.A. (2006). Nonvesicular sterol movement from plasma membrane to ER requires oxysterol-binding protein-related proteins and phosphoinositides. J. Cell Biol. 173, 107–119. Rivas, M.P., Kearns, B.G., Xie, Z., Guo, S., Sekar, M.C., Hosaka, K., Kagiwada, S., York, J.D., and Bankaitis, V.A. (1999). Pleiotropic alterations in lipid metabolism in yeast sac1 mutants: relationship to ‘‘bypass Sec14p’’ and inositol auxotrophy. Mol. Biol. Cell 10, 2235–2250.
Roy, A., and Levine, T.P. (2004). Multiple pools of phosphatidylinositol 4-phosphate detected using the pleckstrin homology domain of Osh2p. J. Biol. Chem. 279, 44683–44689. Schulz, T.A., Choi, M.G., Raychaudhuri, S., Mears, J.A., Ghirlando, R., Hinshaw, J.E., and Prinz, W.A. (2009). Lipid-regulated sterol transfer between closely apposed membranes by oxysterol-binding protein homologues. J. Cell Biol. 187, 889–903. Tahirovic, S., Schorr, M., and Mayinger, P. (2005). Regulation of intracellular phosphatidylinositol-4-phosphate by the Sac1 lipid phosphatase. Traffic 6, 116–130. Voeltz, G.K., Prinz, W.A., Shibata, Y., Rist, J.M., and Rapoport, T.A. (2006). A class of membrane proteins shaping the tubular endoplasmic reticulum. Cell 124, 573–586. Wei, H.C., Sanny, J., Shu, H., Baillie, D.L., Brill, J.A., Price, J.V., and Harden, N. (2003). The Sac1 lipid phosphatase regulates cell shape change and the JNK cascade during dorsal closure in Drosophila. Curr. Biol. 13, 1882–1887.
Cell 144, 389–401, February 4, 2011 ª2011 Elsevier Inc. 401
Cells Respond to Mechanical Stress by Rapid Disassembly of Caveolae Bidisha Sinha,1,2,14 Darius Ko¨ster,1,2,14 Richard Ruez,3,4 Pauline Gonnord,3,4 Michele Bastiani,8,9 Daniel Abankwa,8,9 Radu V. Stan,10 Gillian Butler-Browne,11 Benoit Vedie,12 Ludger Johannes,3,4 Nobuhiro Morone,13 Robert G. Parton,8,9 Grac¸a Raposo,3,5,6 Pierre Sens,7 Christophe Lamaze,3,4,15,* and Pierre Nassoy1,2,15,* 1Universite ´
P. et M. Curie/CNRS UMR168 Curie, Centre de Recherche, Laboratoire Physico-Chimie 3CNRS UMR144 4Institut Curie, Centre de Recherche, Laboratoire Trafic, Signalisation et Ciblage Intracellulaires 5PICT IBiSA Institut Curie 6Centre de Recherche, Laboratoire Structure et Compartiments Membranaires, Institut Curie 26 rue d’Ulm, 75248 Paris Cedex 05, France 7ESPCI, CNRS-UMR 7083, Physico-Chimie The ´ orique, 10 rue Vauquelin, 75231 Paris Cedex 05, France 8The University of Queensland, Institute for Molecular Bioscience 9Center for Microscopy and Microanalysis Brisbane, Queensland 4072, Australia 10Dartmouth Medical School, Borwell 502W, HB7600, One Medical Center Drive, 03756 Lebanon, NH, USA 11Institut de Myologie, Ho ˆ pital Pitie´-Salpe´trie`re, UM76 UPMC, U974 Inserm, UMR7215, CNRS-AIM, 47, bld de l’hoˆpital, 75651 Paris Cedex 13, France 12Laboratoire de Biochimie, Ho ˆ pital Europe´en Georges Pompidou, 20 rue Leblanc, 75015 Paris, France 13National Center of Neurology and Psychiatry, National Institute of Neuroscience, Department of Ultrastructural Research, 4-1-1 Ogawa-Higashi, Kodaira, Tokyo 187-8502, Japan 14These authors contributed equally to this work 15These authors contributed equally to this work *Correspondence:
[email protected] (C.L.),
[email protected] (P.N.) DOI 10.1016/j.cell.2010.12.031 2Institut
SUMMARY
INTRODUCTION
The functions of caveolae, the characteristic plasma membrane invaginations, remain debated. Their abundance in cells experiencing mechanical stress led us to investigate their role in membrane-mediated mechanical response. Acute mechanical stress induced by osmotic swelling or by uniaxial stretching results in a rapid disappearance of caveolae, in a reduced caveolin/Cavin1 interaction, and in an increase of free caveolins at the plasma membrane. Tether-pulling force measurements in cells and in plasma membrane spheres demonstrate that caveola flattening and disassembly is the primary actin- and ATP-independent cell response that buffers membrane tension surges during mechanical stress. Conversely, stress release leads to complete caveola reassembly in an actin- and ATP-dependent process. The absence of a functional caveola reservoir in myotubes from muscular dystrophic patients enhanced membrane fragility under mechanical stress. Our findings support a new role for caveolae as a physiological membrane reservoir that quickly accommodates sudden and acute mechanical stresses.
Caveolae were first described in the early 1950s through the seminal electron microscopy studies of Palade and Yamada (Palade, 1953; Yamada, 1955). These characteristic 60–80 nm cup-shaped uncoated invaginations are highly enriched in cholesterol and sphingolipids (Richter et al., 2008). Present at the plasma membrane of many cells with the exception of neurons and lymphocytes, they are particularly abundant in muscle cells, adipocytes, and endothelial cells. The identification of caveolin-1 (Cav1) (Rothberg et al., 1992; Kurzchalia et al., 1992) and caveolin-2 (Scherer et al., 1996) as the main constituents of the caveolar structure was instrumental to gain insight into the cell biology, structural, and genetic features of caveolae (Stan, 2005). They have been associated with endocytosis, cell signaling, lipid metabolism, and other functions in physiological as well as in pathological conditions. Nevertheless, the role of these specialized membrane domains remains debated, and little is known about the molecular mechanisms involved in their formation and proposed functions (Parton and Simons, 2007). Recent studies have suggested that the distribution of Cav1 and caveolae-mediated signaling can be affected by external mechanical cues. In endothelial cells, chronic shear exposure activates the ERK pathway in a caveolae-dependent manner (Boyd et al., 2003; Park et al., 2000; Rizzo et al., 2003). In smooth-muscle cells, cyclic stretch can cause association of
402 Cell 144, 402–413, February 4, 2011 ª2011 Elsevier Inc.
some kinases with Cav1 (Sedding et al., 2005). To date, the role of Cav1/caveolae in mechanotransduction is mainly viewed as a downstream signaling platform, whereas their function in primary mechanosensing has not been directly addressed. A recent theoretical study has proposed that budded membrane domains like caveolae could play the role of membrane-mediated sensors and regulators of the plasma membrane tension (Sens and Turner, 2006). Endowed with a high membrane and lipid storage capacity, owing to the invaginated structure and high lipid packing, caveolae are well equipped to play such a role. We have challenged the homeostasis of the plasma membrane tension with different types of controlled mechanical stresses and analyzed the role of caveolae in the cell shortterm response. We show in endothelial cells and muscle cells that functional caveolae are required to buffer the variations of membrane tension induced by sudden and transient mechanical stress via a two-step process of rapid caveola disassembly and slower reassembly. RESULTS Mechanical Stress Leads to the Partial Disappearance of Caveolae from the Plasma Membrane We examined the response of caveolae when cells were exposed to acute mechanical stresses. Osmotic swelling causes an increase of the membrane tension of cells unless some additional membrane is delivered to the cell surface (Dai and Sheetz, 1995; Dai et al., 1998; Morris and Homann, 2001). Cav1-EGFP-transfected HeLa cells were exposed to hypoosmotic medium (30 mOsm). We observed a 35% increase of the cell volume within the first 5 min and a slow decrease thereafter (Figures 1A and 1B). Upon reversing back to iso-osmolarity (300 mOsm) after 30 min of hypotonic shock, the volume decreased below the initial cell volume. These observations support the existence of a compensatory mechanism known as regulatory volume decrease, which restores the osmotic balance by activating ion channels (D’Alessandro et al., 2002). Our data, however, suggest that this process is not dominant during the first 5 min following hypo-osmotic shock. To distinguish caveolae at the plasma membrane from the internal Golgi pool of Cav1, we used total internal reflection fluorescence (TIRF) microscopy (Figure 1C and Figures S1A and S1B available online). Upon hypo-osmotic shock, we observed that the number of caveolae significantly decreased by 30% at the cell surface (Figures 1C and 1D) and that the loss correlated with the magnitude of the shock (Figure 1E). Importantly, the cell footprint and the adhesion between the cell and the glass surface were unaltered, as shown by reflection interference contrast microscopy (RICM) (Figure S1C). Because caveolae exhibit different types of dynamics at the plasma membrane (Pelkmans and Zerial, 2005), we also checked whether any particular pool was selectively affected. Within minutes of hypo-osmotic shock, slow-moving caveolae reduced their mobility (Figure S1D), and fast dynamics were abolished (Movie S1 and Movie S2), whereas caveolae displaying all kinds of mobility were reduced in number (Figures S1E and S1F). Similar results were obtained in mouse lung endothelial cells (MLEC)
(Figure S1B). Although osmotic shocks have been extensively used to mimic the osmolarity changes that cells experience (Lang et al., 1998), we sought to rule out any indirect influence of cell swelling on caveolae. We developed a stretching device based on thin transparent silicone substrates to challenge the cell membrane with a different mechanical stress. It allowed imaging of caveolae by TIRF before and after stretch and was combined with micropatterning (Chen et al., 1997) to control the cell adhesion area and its orientation with the stretching axis. The number of caveolae present at the basal footprint of Cav1-EGFP HeLa cells decreased upon stretching (Figures 1F and 1G), and the loss correlated with the extent of stretch (Figure 1H). Therefore, acute mechanical stress induced either by hypo-osmotic shock or membrane stretching leads to a rapid and significant loss of caveolae from the cell surface. We next performed electron microscopy (EM) on MLEC. These endothelial cells experience chronic cycles of shear stress from the blood flow in lungs’ vessels in vivo. MLEC immunostaining shows multiple subcellular Cav1 positive structures, which are localized predominantly at the plasma membrane and at the Golgi apparatus (Figure S3B; Murata et al., 2007). In contrast, Cav1/ MLEC derived from cells knocked out for the CAV1 gene do not present any Cav1 staining. However, Cav1-EGFP expression can be induced by transfection in WT and Cav1/ MLEC (Figure S3C). EM analysis showed a significant decrease of the number of caveolae (50%) upon a 5 min exposure of WT MLEC to a hypo-osmotic shock (Figures 2A and 2B). These data confirm the results obtained by TIRF imaging and extend our conclusions to endogenous caveolae present on the entire surface of the cell. Flattening and Disassembly of Caveolae upon Hypo-Osmotic Shock The contribution of caveolae to the general endocytic activity of the cell is believed to be minimal (Nabi and Le, 2003), and endocytosis is disfavored at high membrane tension (Dai et al., 1997). We still tested whether the loss of caveolae upon hypo-osmotic was due to increased caveola endocytosis. We used dynasore, an inhibitor of the dynamin GTPase that is involved in caveola internalization (Henley et al., 1998; Macia et al., 2006). Indeed, dynasore significantly increased the caveolae density at the plasma membrane, reflecting the efficient inhibition of caveola endocytosis (Figure 2C). However, upon hypo-osmotic shock, a similar loss of caveolae was measured (Figures 2C and 2D). We next examined Cavin1, which is part of the caveolar complex and is required to maintain caveola invagination (Hill et al., 2008; Hansen et al., 2009). Cavin1 does not bind to free Cav1 oligomers or to Cav1 present on the Golgi apparatus. We found a high level (64%) of Cav1-EGFP colocalization with Cavin1mCherry (Figure 2E). Upon hypo-osmotic shock, there was a similar or even higher loss of Cavin1-labeled structures, confirming the partial loss of caveolae (Figure 2F). We also observed a decreased colocalization with Cavin1 (35%) for the remaining caveolae, suggesting a loss of their invaginated structure. We quantified the interaction between Cav1 and Cavin1 using fluorescence lifetime imaging microscopy (FLIM). When Cavin1 is present in caveola, the close proximity of mRFP-Cav3 and Cavin1-EGFP results in FRET and a decrease in the EGFP Cell 144, 402–413, February 4, 2011 ª2011 Elsevier Inc. 403
A
B Volume (au)
Iso
Hypo
1.2
1.0 0.7 0
10
Rec
20
30
Time (min)
D Iso
Hypo
Norm malized no. of caveolae
C
E 1.2
1.2
0.8
1.0
0.0
Iso
0.6 Iso Hypo
300
200
100
0
Osmolarity (mOsm)
Hypo
G Normalized o. of caveolae no
F
08 0.8
0.4
H
1.2
1.2 1.0
0.8
0.8
0.4
06 0.6 0.0 -
+
Stretch
0.0
0.2
0.4
(L-L0)/L0
Figure 1. Mechanical Stress Induces Partial Disappearance of Caveolae (A) YZ maximum-intensity projection of confocal stacks of Cav1-EGFP HeLa cells. Projection of four cells under iso-osmotic conditions (Iso), hypo-osmotic conditions (Hypo, 5 min), and 3 min after returning to iso-osmolarity (Rec). Scale bar, 5 mm. Dashed lines mark out the initial cell boundary. (B) Volume of Cav1-EGFP HeLa cells tracked from (and normalized to) iso-osmotic conditions through hypo-osmotic shock (onset: t = 0 min) and upon returning to iso-osmolarity (t 29 min). Arrow indicates return to iso-osmolarity. Data were derived from multiple measurements (n = 5) in three independent experiments. Error bars represent standard deviations (SD). (C) TIRF images of Cav1-EGFP HeLa cells under iso-osmotic conditions (Iso) and after 4 min hypo-osmotic shock (Hypo). Dotted line marks out the cell footprint. Scale bar, 5 mm. (D) Change in the number of caveolae for single Cav1-EGFP HeLa cells after hypo-osmotic shock (Hypo) normalized to the number counted before hypo-osmotic shock (Iso) (n = 18). Error bars represent SD (p = 4 3 1011). (E) Evolution of the loss of caveolae per cell with decreasing osmolarity. The same Cav1-EGFP HeLa cells were exposed to decreasing osmolarities during 1 min for each osmolarity. From correlation analysis, the loss of caveolae is positively correlated with the decrease in external osmolarity (r2 = 0.85). Error bars represent SD (n = 3). (F) TIRF images of a Cav1-EGFP HeLa cell on the stretching device at 0% (left) and 20% stretch (right). Dotted lines mark out cell boundaries before and after stretch. Scale bar, 5 mm. (G) Change in the number of caveolae for single Cav1-EGFP HeLa cells after stretching (15% ± 1%) normalized to the number counted before stretching. Data are derived from multiple measurements (n = 7; p = 0.00033) in seven independent experiments. Error bars represent SD. (H) Evolution of the number of caveolae for single Cav1-EGFP HeLa cells stretched to different lengths characterized by (L L0)/L0 wherein L0 and L are the initial and final lengths of the cell footprint in the stretching direction. Each point is measured on a single cell. The number of caveolae is found to be negatively correlated to the extent of stretch (n = 7; r2 = 0.85) as measured in seven independent experiments.
fluorescence lifetime (Abankwa et al., 2008; Hill et al., 2008). There was a significant increase in the fluorescence lifetime upon hypo-osmotic shock, indicating the dissociation of Cavin1 from Cav1 (Figure 2G). We also performed Cav1 immuno-EM on MLEC before and after hypo-osmotic shock (Figure 3A and Figure S2A). We found a 10-fold increase in the number of gold particles associated with Cav1 in noncaveolar membranes after 5 min of hypo-osmotic shock (Figure 3B). Upon returning to isoosmolarity, cells recovered the initial number of caveolae, and 404 Cell 144, 402–413, February 4, 2011 ª2011 Elsevier Inc.
Cav1 was mainly associated with caveola (Figures 3B and 3C). Under iso-osmotic conditions, deep-etched EM showed a majority of budded caveolae with characteristic tight striated coats (Morone et al., 2006). After hypo-osmotic shock, several flat structures with loose striated coats reminiscent of formerly budded caveolae were observed. Upon iso-osmolarity recovery, all caveolae were budded (Figure 3D and Figure S2B for three-dimensional view). These findings clearly indicate that cells respond to acute mechanical membrane stress by the rapid
Figure 2. Caveolae Morphology and Cav1Cavin1 Interaction Are Lost upon HypoOsmotic Shock
B
A
No. of cavveolae per μm
0.8
Iso
04 0.4
0.0
Iso
Hypo
Hypo
Hypo
1.0 0.8 0.6
1.2 0.8 04 0.4 0.0
0
10
20
30
40
50
Time (min)
E
Cavin1-mCherry
Cav1-EGFP C
Iso
Iso Ctrl Dyn Hypo
F
Hypo No. of caveolae
Norm malized numb ber of cav
1.2
D Normalized no. o of caveolae
Ctrl Dyn
C
Cav1-EGFP Cavin1-mCherry
150 120 90 60 30 0
Iso
Hypo
G C i 1 Cavin1 Cavin1 + Cav3 *
Life-tim me (ns)
Merge
2.2 1.9
(A) Ultrathin cryosections of WT MLEC before (Iso) and 5 min after (Hypo) switch to hypo-osmotic medium examined by EM. Arrows mark out caveolae. Scale bar, 150 nm. (B) Quantification of caveolae detected per mm of plasma membrane on ultrathin cryosections of WT MLEC before (Iso) and 5 min after switch to hypo-osmotic medium (Hypo) reveals a significant decrease (p = 0.047) in the number of caveolae after hypo-osmotic shock. Total membrane used for quantification was 76 and 67 mm for iso- and hypo-osmotic conditions, respectively, imaged from different sections of multiple randomly selected cells (>10). Data represent mean ± SD. (C) Time evolution of caveolae number (± SD) detected by TIRF in control (Ctrl) and dynasoretreated cells (Dyn) normalized to the caveolae number before addition of dynasore (t = 5 min). Dynasore was added at t = 0, and hypo-osmotic shock was applied at t 45 min (30 mOsm, shaded region). Error bars represent SD; n = 4. (D) Change in number of caveolae for single cells in control (Ctrl) and in dynasore-treated cells (Dyn) after hypo-osmotic shock normalized to the number before shock (Iso) (n = 9). Error bars represent SD (p = 1 3 104). (E) TIRF images of HeLa cells expressing Cavin1mCherry and Cav1-EGFP before (Iso) and 5 min after switch to hypo-osmotic conditions (Hypo). Scale bar, 10 mm. (F) Change in number of Cav1 and Cavin-1 structures per cell before and after hypo-osmotic shock of 5 min. Error bars represent SD (n = 3). (G) HeLa cells transiently expressing Cavin1-EGFP with or without Cav3-mRFP were exposed to iso-osmotic (Iso) or hypo-osmotic (Hypo) media for 15 min and analyzed by FLIM. Data represent mean EGFP fluorescence lifetime ± standard errors (SE). n = 40–70 cells; *p < 4 3 1012.
1.6 Iso
flattening of a fraction of the caveola. We also measured the level of Cav1-EGFP diffusing freely at the membrane using fluorescence recovery after photobleaching (FRAP). At steady state, the fraction of freely diffusing Cav1 in the plasma membrane was low (10%; Figure 3E), as reported (Pelkmans et al., 2004; Hill et al., 2008). However, we measured a higher mobile fraction (30%) in cells exposed to hypo-osmotic shock for at least 10 min. This increase is likely to reflect the release of Cav1 from flattened caveolae. Caveolae Are Selectively Required for Buffering Membrane Tension Caveola flattening is likely to release the amount of membrane stored within the caveolar invagination and thereby to provide the additional membrane required to maintain membrane
Hypo
tension homeostasis during mechanical stress (Sens and Turner, 2006). We tested this hypothesis with the tether pulling technique (Dai and Sheetz, 1995), which measures the cell membrane tension. Optically trapped beads adhering to the plasma membrane served as handles to extract membrane tethers (Figure 4A, Figure S3A and Movie S3). The restoring tether force f, which was derived from the bead displacement, is an indicator of the effective membrane tension (Sheetz, 2001). Its value is proportional to the square root of the effective ~ which corresponds to the sum of the lipid bilayer tension s, tension s and the cytoskeleton-to-membrane adhesion energy ~ (Dai and Sheetz, W0. The latter term represents 75% of s 1999) and arises from all molecular interactions between membrane and cytoskeleton. The presence of exogenous proteins is thus likely to have an intricate influence on W0, as Cell 144, 402–413, February 4, 2011 ª2011 Elsevier Inc. 405
D
Iso
A
Hypo
Iso
B 100
** *
***
50
0.8
0.0
0
Iso Hypo Rec
E
IsoRec
1.00 Iso Hypo Rec
0.75
Recc
Fluorescence Reco F overy
C
No. o of caveolae per μm p
Perrcentage of e Cav1 labeling surface
Membrane Caveolae
Hypo
Rec
0.50 0.25 0 00 0.00 0
100
200
300
Time (sec) Figure 3. Caveolae Flatten Out and Disassemble upon Hypo-Osmotic Shock and Reassemble upon Recovering Iso-Osmolarity (A) Immuno-EM images of ultrathin cryosections with gold-tagged Cav1 antibody of WT MLEC under iso-osmotic (Iso), hypo-osmotic (Hypo, 5 min), and recovered iso-osmotic (Rec, 5 min) conditions. Scale bar, 150 nm. See also Figure S2A. (B) Percentage of gold particles found in caveolae and endosomal structures close to the plasma membrane versus those found in noncaveolar membranes (*p = 4 3 103; **p = 4 3 102; ***p = 1 3 103). Total membrane used for quantification was 23, 24, and 35 mm, respectively, for iso-osmotic, hypo-osmotic, and recovered iso-osmotic conditions, imaged from different sections of multiple cells. Error bars represent SE. (C) Comparison between the number of caveolae per mm of membrane before hypo-osmotic shock (Iso) and after return to iso-osmotic medium (Rec) analyzed from ultrathin cryosections of WT MLEC using 60 mm of membrane imaged from different sections (n = 8) of multiple cells. Data represent mean ± SE (p = 1 3 102). (D) Deep-etched EM images of MLECs under iso-osmotic (Iso), hypo-osmotic (Hypo, 5 min), and recovered iso-osmotic (Rec, 5 min) conditions. Scale bar, 200 nm. Left insets depict representative images of clathrin-coated pits. Right images depict representative images of caveolae. Scale bars (insets), 100 nm. See also Figure S2B. (E) Fluorescence recovery after photobleaching (FRAP) of Cav1-EGFP in HeLa cells in iso-osmotic (Iso; n = 8), hypo-osmotic (Hypo; n = 8), and recovered isoosmotic conditions (Rec; n = 8). Lines show fit for the curves to standard recovery equation. Hypo-osmotic shock results in a statistically higher fluorescence recovery (p = 2 3 104) than in iso-osmotic conditions. Data represent mean ± SE.
shown later. Tether force measurements do not enable a priori to separate s and W0. Therefore, we used the tether pulling technique as a differential assay to probe the relative changes in 406 Cell 144, 402–413, February 4, 2011 ª2011 Elsevier Inc.
tether forces under conditions that keep W0 unaltered. In particular, osmotic shocks have been assumed to mostly affect s with minor changes on adhesion (Dai et al., 1998). By quantifying the
A
C
B
D
actin bundles architecture, we found that the cortical actin cytoskeleton was unaltered within the first 5 min of osmotic shock (Figures S3D–S3F). At these timescales, variations in f thus directly mirror changes in s. We measured the tether force f0 in isotonic conditions and recorded the variations of the tether force f while MLEC were exposed to hypo-osmotic shock. Figure 4A shows representative temporal traces of the relative tether force changes, (f-f0)/f0, obtained in WT and Cav1/ MLEC. We found that, upon hypo-osmotic shock (150 mOsm), f remained almost identical to f0 in WT MLEC. In contrast, the tether force increased by 200% in Cav1/ MLEC. Our data imply that the membrane tension increase is buffered in WT MLEC by the presence of caveolae. Importantly, the expression of functional caveolae by transfection of Cav1-EGFP in Cav1/ MLEC restored membrane tension buffering. Therefore, the lack of membrane tension buffering is due to the absence of caveolae and not to other cellular structures that may have been altered in Cav1/ MLEC (Figure 4A). M-b-cyclodextrin, which flattens caveolae through cholesterol depletion (Rothberg et al., 1992), led also to membrane tension increase upon hypo-osmotic shock in WT MLEC (Figure 4A), confirming that caveola flattening is required for buffering the membrane tension surge. Finally, we have calculated that the number of lost caveolae per cell observed by TIRF and EM is in excellent agreement with the amount of released area (0.3%) required to buffer s (see Supplemental Results). We also tested whether clathrin-coated pits (CCP), another type of plasma membrane invagination, could buffer membrane
Figure 4. Caveolae Buffer the Membrane Tension Rise during Hypo-Osmotic Shock (A) Representative force curves for tethers extracted from WT MLEC, Cav1/ MLEC, Cav1/ MLEC transfected with Cav1-EGFP, and WT MLEC treated with mbCD exposed to hypoosmotic shock (150 mOsm). Hypo-osmotic shock is indicated by an arrow (break from 1.34 to 2.7 min). (B) Relative change of the mean tether force after hypo-osmotic shock (5 min) for WT (n = 9), and Cav1/ MLEC (n = 9; p = 0.01502), WT (n = 3) and Cav1/ MEFs (n = 4; p = 8 3 104), and HeLa cells (n = 4). Data represent mean ± SE. See text for details. (C) Cav3 immunostaining in differentiated WT and Cav3-P28L human myotubes. Scale bar, 5 mm. (D) Relative change of the mean tether force after hypo-osmotic shock (5 min) for WT (n = 11) and P28L myotubes (n = 12; p = 5 3 108). Data represent mean ± SE.
tension. Thus, we quantitatively analyzed the fate of CCPs upon hypo-osmotic shock in MLEC and HeLa cells. We observed that the number of CCPs lost at the membrane was, at most, one-tenth of the number of lost caveolae, both in WT and Cav1/ MLEC (Figures S4A–S4C). Accordingly, deep-etch EM showed that the structure of CCPs was not affected (Figure 3D). Additionally, under hypo-osmotic shock, membrane tension was buffered to the same extent whether clathrin was expressed or knocked down in WT MLEC (Figures S4D and S4E). In contrast, Cav1/ MLEC having CCPs could not buffer the membrane tension surge. These results rule out a contribution of CCPs in membrane tension regulation and establish caveolae as the primary stress-responsive membrane structure. Finally, we obtained similar results in Cav1-EGFP HeLa and in embryonic fibroblasts (MEF) lacking or expressing Cav1 (Figure 4B). We could further extend the physiological significance of these findings by measuring the stress reactivity of human muscle cells. Several human muscular dystrophies have been associated with mutations in Cav3, the muscular isoform of caveolin (Woodman et al., 2004), and more recently with mutations in Cavin1 (Hayashi et al., 2009). Most Cav3 mutations prevent caveolae assembly at the plasma membrane by sequestration of Cav3 in the Golgi apparatus. We studied differentiated human myotubes bearing the P28L Cav3 mutation described in familial hyperCKaemia (FHCK), a hereditary form of muscular dystrophy (Woodman et al., 2004). Muscle fibers that were isolated from these patients show a strong decrease of Cav3 expression and reduced Cav3 staining at the cell surface (Merlini et al., 2002). Accordingly, we found that Cav3 immunostaining was restricted to the Golgi apparatus in P28L Cav3 myotubes, whereas WT Cav3 was mainly found at the plasma membrane (Figure 4C). Under hypo-osmotic shock, membrane tension was increased in P28L Cav3 myotubes and buffered in WT Cell 144, 402–413, February 4, 2011 ª2011 Elsevier Inc. 407
B 1.0
1.0
0.8
0.8
Normalized no. of caveolae
Normalized no. of caveolae e
A
0.6 0.4
0.6 0.4 0.2
0.2 0.0
Iso
Ctrl
0.0
CD LatA Jas no ATP
Stretch
Hypo
C
Figure 5. Membrane Tension Surge Buffering by Caveolae Flattening Occurs in an ATP- and Actin-Independent Process
D
4
Ctrl
Crtl
-
+
CD no ATP +
+
Cav1-EGFP Cavin1-mCherry
(f-f0)/f0
3
2
1
0
Ctrl
CD Wt
no ATP
Ctrl
Cav1-EGFP Cavin1-mCherry
E
22
23
Wt - transfected -/Cav1
24
0.6
(f-f0)/f0
0.2 0.0 0.4
0.3
30 Pa 00 0.0
02 0.2 0.0
F
5 Pa
0.4
Norma alized intensity
CD no ATP -/Cav1
0
10
20
Distance (μm)
30
40
5
10
15
20
25
30
Aspiration pressure (Pa)
Cav3 myotubes (Figure 4D). P28L myotubes and Cav1/ MLEC also showed an increased tendency to membrane rupture under hypo-osmotic shock (Figure S5). This may explain the high blood level of creatine kinase found in these patients in the absence of a functional caveolar reservoir during the repeated extensionrelaxation cycles. Membrane Tension Surge Buffering by Caveola Flattening Occurs in an ATP- and Actin-Independent Process Although the actin cytoskeleton was unaffected by hypoosmotic shocks at early times, we still investigated its potential role in caveola flattening. Under hypo-osmotic shock (30 mOsm), none of the actin-perturbing drugs prevented the loss of caveolae from the membrane (Figure 5A). This was further confirmed by cell ATP depletion before hypo-osmotic shock. 408 Cell 144, 402–413, February 4, 2011 ª2011 Elsevier Inc.
35
(A) Normalized number of caveolae per cell after hypo-osmotic shock (Hypo). Reference is the number before hypo-osmotic shock (Iso) for control cells (Ctrl; n = 18). Reference is the number after the drug treatment and before hypo-osmotic shock for cytochalasin D (CD; n = 10, p = 2 3 105), latrunculin A (Lat; n = 21, p = 7 3 1011), jasplakinolide (Jas, n = 11, p = 6 3 108) treated cells, and ATP-depleted cells (no ATP; n = 10, p = 2 3 105). Data represent mean ± SD. (B) Change in number of caveolae for single HeLa Cav1-EGFP cells after stretching (15% ± 1%). Same normalization as in (A) for control cells (n = 7; p = 3 3 104), and for cells treated with cytochalasin D (CD; n = 5; p = 0.01) and ATP-depleted cells (no ATP; n = 5; p = 0.04). Data represent mean ± SD. (C) Relative change of the tether force after hypoosmotic shock (5 min) for WT and Cav1/ MLEC for control (Ctrl; n = 9 for WT and n = 5 for Cav1/; p = 0.015), cytochalasin D treated (CD; n = 9 for WT and n = 10 for Cav1/; p = 3 3 105), and ATP depleted (no ATP; n = 6 for WT n = 5 for Cav1/; p = 2 3 104) cells. Data represent mean ± SE. (D) Confocal image of a WT MLEC transfected with Cav1-EGFP (green) and Cavin1-mCherry (red) after incubation for 6 hr in PMS buffer. Scale bar, 10 mm. (E) (Top) Confocal image of a PMS positive for Cavin1-mCherry (red) and Cav1-EGFP (green) after micropipette aspiration (white lines) and formation of a membrane tether with an optically trapped bead (white disk). Scale bar, 10 mm. (Bottom) Line scans of cavin1-mCherry (red) and Cav1-EGFP (green) normalized intensity along the circumference of the PMS shown above for two aspiration pressures. Arrows indicate regions of colocalization. (F) Relative change of the tether force as a function of the micropipette aspiration pressure in PMS obtained from Cav1-GFP + Cavin1-mCherry MLEC (black squares) and from Cav1/ MLEC (Cav1/, red circles). Data were obtained in eight independent experiments and represent the mean value of 60 s measurements ± SD.
ATP depletion abolished the mobility of the internal pool of caveolae, confirming the inhibition of active cellular processes (Movie S4 and Movie S5). However, a similar loss of caveolae occurred in ATP-depleted cells (Figure 5A). Likewise, upon cell stretching, caveolae still disappeared in ATP-depleted and cytochalasin D (CD)-treated cells (Figures 5B and Figure S6A). Furthermore, membrane tension measurements under hypoosmotic shock (150 mOsm) showed that the tether force of Cav1/ MLEC still increased by 200% in cells treated with CD or depleted in ATP (Figures 5C and Figure S6B). In contrast, no significant tether force variation was measured for WT MLEC, indicating that the buffering is not dependent on ATP and actin dynamics. Importantly, the invaginated shape of caveola was unaffected by CD or ATP depletion (Figure S7). Finally, we could unambiguously establish that membrane tension buffering is an intrinsic mechanical property of caveolae by using plasma
membrane spheres (PMS). PMS are composed of plasma membrane and cytosol, whereas the subcellular compartments (Lingwood et al., 2008) and filamentous actin (Figure S6C) are excluded. PMS were prepared from MLEC transfected with Cav1-EGFP and Cavin1-mCherry. As expected, a major colocalization of Cavin1 and Cav1 was observed at the plasma membrane of the donor cell (Figure 5D). Confocal imaging revealed that Cavin1 was present inside PMS and along the membrane, where it colocalized with Cav1 as a punctuated pattern likely to reflect the incorporation of caveolar invaginations in these vesicles (Figure 5E). PMS were then mechanically stressed by micropipette aspiration (Dimova et al., 2006). Membrane tethers were pulled with optical tweezers from PMS aspirated under minimal pressure (Figure 5E). Upon increasing aspiration pressure, PMS from Cav1/ MLEC exhibited a steady tether force increase, whereas f remained constant over a significant range of aspiration pressures in PMS from WT MLEC (Figure 5F). Concomitantly, over the aspiration range corresponding to the force plateau, the number of Cavin1/Cav1 colocalizations also decreased to noise level (Figure 5E). When the aspiration was increased further, the force plateau was followed by an increase in f, which is in agreement with the complete flattening of caveolae and thus depletion of all residual membrane reservoir (Figure 5F). We also measured identical amounts of free cholesterol in WT and Cav1/ PMS (75 ± 20 nmol/mg of cell protein), indicating that the lack of buffering effect in Cav1/ PMS is not related to changes in lipid composition. These results on PMS reinforce our findings obtained with actin drugs and ATP depletion in cells and definitely establish that membrane tension buffering by caveolae flattening is a purely passive mechanism solely driven by membrane mechanics. Caveolae Reassemble in an Actin- and ATP-Dependent Process upon Stress Relaxation We next examined the behavior of caveolae when cells were left in hypo-osmotic medium (30 mOsm) for 5 min and returned to iso-osmotic medium (300 mOsm). Within 2–5 min, caveolae reappeared at the plasma membrane (Figures 3A–3D and Figure 6A). Accordingly, there was a decrease of both the fraction of mobile Cav1 (Figure 3E) and Cav1 immunogold labeling in noncaveolar membranes (Figures 3A and 3B and Figure S2A). We also measured a decrease in EGFP fluorescence lifetime, indicating the reassociation of Cav1 and Cavin1 and therefore the reassembly of functional caveolae at the cell surface (Figure 6B). On average, the initial rate of reassembly was about 10 caveolae per min per cell and required ATP (Figure 6C). In contrast to caveolae disassembly, reassembly was enhanced when actin dynamics were blocked by CD. Finally, we tested whether caveolae reassembled directly from the plasma membrane pool of free Cav1 or from other compartments. The disruption of the Golgi apparatus by Brefeldin A (BFA) did not prevent reassembly excluding a major contribution of the Cav1 Golgi pool (Figure 6D). Furthermore, we found no significant transfer of the photoconverted Golgi pool of Cav1 to the plasma membrane on returning to iso-osmolarity. However, the nonphotoconverted Cav1 present throughout the cell showed an increased punctuated localization at the plasma membrane (Figure 6E and 6F).
DISCUSSION Since their discovery in 1953, the precise role of caveolae has remained a matter of considerable debate. Freeze-fracture analysis of the surface membrane of smooth and striated muscle cells in the early 1970s led to the first hypothesis that caveolae could flatten out under stretching conditions (Dulhunty and Franzini-Armstrong, 1975; Prescott and Brightman, 1976). Later studies have also associated caveolae with the mechanosensing response of the cell (Boyd et al., 2003; Park et al., 2000; Rizzo et al., 2003; Sedding et al., 2005; Kawamura et al., 2003; Kozera et al., 2009). Whether caveolae are directly involved in the cell response to mechanical stress, and by which mechanisms, still remain unknown. In this study, we establish that the primary cell response to an acute mechanical stress occurs through the rapid flattening of caveolae into the plasma membrane. Upon cell stretching or osmotic swelling, caveolae flattening provides the additional amount of membrane that inhibits any surge of membrane tension. This response occurs also in ATP-depleted cells and in membrane-derived vesicles devoid of actin, indicating that the ability to buffer membrane tension is intrinsic to the caveola structure. These results are in line with a theoretical model proposing that invaginated domains can serve as a membrane reservoir to regulate the membrane tension (Sens and Turner, 2006). However, in contrast to this model, which predicted that flattening of existing caveolae or budding of new caveolae was driven by the deviation of the membrane tension with respect to the resting tension, our data indicate that the regulatory mechanism is asymmetric. Whereas stress-induced flattening of caveolae is truly passive, the reassembly of new caveolae is assisted by ATP and actin dynamics. The energy landscape for flat and invaginated caveola is thus characterized by an activation energy. The transition between the two forms thus requires an energy barrier to be overcome or lowered. In this model, the flat configuration would be stabilized upon application of a mechanical force, whereas the formation of invaginated caveola is energetically favored through interactions with cortical actin in the presence of ATP. This is consistent with the known role of actin dynamics and various kinases in caveola function (Pelkmans et al., 2005; Pelkmans and Zerial, 2005). Our results further indicate that clathrin-coated pits, the other abundant invaginations present at the plasma membrane, are not involved in the fast response to membrane strain. Whereas CCPs exchange coat proteins within seconds, most caveolae are rather stable at the plasma membrane. The stability of caveolae combined with their low endocytic activity allows this membrane reservoir to be readily available to respond to sudden mechanical strains. The stability of the caveolae reservoir is a striking feature of cells that experience pulsating mechanical stresses in their lifetime (muscle cells, cardio-myocytes, and endothelial cells) because they all express very high numbers of caveolae. Indeed, we could associate the lack of Cav3 expression at the surface of myotubes with impaired membrane tension buffering in patients bearing the Cav3 P28L mutation found in a form of human muscular dystrophy. Interestingly, P28L myotubes appear to be more fragile than WT myotubes when exposed to acute osmotic shock. Cell 144, 402–413, February 4, 2011 ª2011 Elsevier Inc. 409
A
B
C
D
E
F
We have shown that the timescales of caveolae and actin cortex responses to osmotic shock are well separated (5 min, respectively). Though the immediate response to mechanical stress relies primarily on caveola flattening, it is likely that other processes such as endocytosis, exocytosis, and actin dynamics may prolong or complete the initial response at longer times. Whether and how caveolae also contribute to long timescale regulation remain to be investigated. In endothelial cells, the application of chronic and repetitive shear stress tensions results in a several-fold increase in the number of caveolae at the plasma membrane through the mobilization of the Cav1 pool associated with the Golgi complex (Boyd et al., 2003; Park et al., 1998). In agreement with the slow delivery rate of Cav1 from the Golgi complex (Tagawa et al., 2005), we show that caveolae are reassembled independently from the Golgi complex when the mechanical stress is relaxed. It is thus important to distinguish between the short-term 410 Cell 144, 402–413, February 4, 2011 ª2011 Elsevier Inc.
Figure 6. Caveolae Reassembly at the Plasma Membrane upon Recovering IsoOsmolarity (A) TIRF imaging of caveolae reassembly after shifting from hypo-osmotic (Hypo) to iso-osmotic (Rec) conditions. Dotted line marks out the cell boundary. Scale bar, 5 mm. (B) HeLa cells transiently expressing Cavin1-EGFP with or without Cav3-mRFP were exposed to hypo-osmotic medium (Hypo) for 15 min followed by iso-osmotic medium (Rec) for 10 min and were analyzed by FLIM. Data represent mean EGFP fluorescence lifetime ± SE (n = 40–60 cells; *p = 8 3 107). (C) After hypo-osmotic shock (15 min), cells were returned to iso-osmolarity (t = 0 min). Data represent the number of caveolae per cell for cytochalasin D-treated (CD), ATP-depleted (no ATP), and control cells (Ctrl). Error bars represent SD. (D) (Left) TIRF images show a typical BFA-treated cell after 15 min hypo-osmotic shock (Hypo) and after 3 min of recovering iso-osmolarity. Scale bar, 10 mm. (Right) Quantification of the number of caveolae before hypo-osmotic shock, after 15 min hypo-osmotic shock, and after 3 min of recovery to iso-osmotic conditions in BFA-treated (n = 3) and control cells (n > 10). *p = 8 3 1029; **p = 1 3 1022. Error bars represent SD. (E) Cav1-Dendra2 photoconversion from green to red fluorescence at the Golgi apparatus (rectangle) of HeLa cells followed through hypoosmotic shock and recovery to iso-osmotic conditions. Arrow marks out the section of the plasma membrane used for plotting intensity profiles. (F) Intensity profiles of green (Cav1-Dendra2) and red (photoconverted Cav1-Dendra2) fluorescence at the plasma membrane before (top) and after (bottom) switch from hypo-osmotic to iso-osmotic conditions.
mechanical function of caveolae and the long-term adaptation of the cell to chronic stress. In this context, we also analyzed the contribution of caveolae to the setting of the membrane tension under resting conditions. At steady state, we measured a lower membrane tension in Cav1/ than in WT MLEC; however, it was identical in WT and Cav1/ MEF, albeit to a lower level (Figure S8A), raising the possibility of a peculiarity of MLEC. Furthermore, the resting tension was drastically affected by m-b-cyclodextrin, cytochalasin D, or ATP depletion treatments in the different cell types, independently from the expression of caveolae (Figure S8B). Although caveolae play a key role in cell tension homeostasis, their direct contribution to the resting tension remains intricate. As previously mentioned, the dynamic membrane-to-cytoskeleton adhesion, which is the main contribution to the apparent membrane tension, is likely to be regulated by cell line-dependent compensatory mechanisms.
Our study establishes a new physiological mechanism by which cells can respond immediately to sudden variations in membrane tension induced by acute mechanical stress (Figure 7). The different proposed roles of caveolae should therefore be reconsidered through this unique ability to respond to mechanical stress, especially in situations in which cells experience physiological or pathological membrane strains such as osmotic swelling, shear stress, or mechanical stretching. EXPERIMENTAL PROCEDURES A list of chemicals and materials can be found in the Extended Experimental Procedures.
Figure 7. Cells Respond to Acute Mechanical Stresses by Rapid Disassembly and Reassembly of Caveolae In resting conditions, caveolae present at the plasma membrane are mostly budded. Magnification shows oligomerized Cav1 and Cavin1 in the caveolar structure. Upon acute mechanical stress (hypo-osmotic shock or stretching), caveolae flatten out in the plasma membrane to provide additional membrane and buffer membrane tension. Magnification shows disassembly and diffusion of Cav1 in the plasma membrane and loss of interaction between Cav1 and Cavin1. Return to resting conditions allows the reassembly of the caveolar structure together with Cavin1 interaction. This cycle represents the primary cell response to an acute mechanical stress.
The well-conserved scaffolding domain of caveolin (CSD) has been involved in both the assembly of caveolae and the interaction with several signaling effectors in vitro (Parton et al., 2006; Parton and Simons, 2007). Because several of these effectors have been associated with mechanotransduction (Vogel and Sheetz, 2006), stress-induced disassembly of caveolae may generate mechanosensitive signals mediating the short- and long-term cell response to mechanical challenges. It is tempting to speculate that the mechanical release of free Cav1 oligomers could favor the interaction between signaling effectors and CSD, which is otherwise hidden in the caveolar structure (Kirkham et al., 2008; Parton et al., 2006). This mechanosensitive signaling would be terminated through the reassembly of free Cav1 oligomers into caveolae when the mechanical stress is relaxed. Endocytosis, which is favored for the retrieval of the excess of membrane during tension relaxation, may also contribute to signaling termination through the internalization of free Cav1 and its degradation in the endolysosomal pathway. The recently characterized Cavin1 protein, which was first described as a transcription factor (Jansa et al., 2001), may also contribute to mechanosignaling regulation through the release from flattened caveola. Indeed, the redistribution of Cav1 to noncaveolar portions of the plasma membrane, the increased mobility of Cav1, and the decreased association of Cav1 and Cavin1 upon hypo-osmotic treatment are all consistent with the effect of Cavin1 knockdown on these parameters (Hill et al., 2008), suggesting that dissociation of the Cav1-cavin module may be crucial in the caveolar response.
Cell Culture and Treatments MLEC (Murata et al., 2007) were maintained in EGM-2/20% FBS medium. They were transfected with AMAXA HUVEC nucleofector kit and used for experiments after 24–72 hr. HeLa cells stably transfected with Cav1-EGFP (Pinaud et al., 2009) and MEFs were maintained in DMEM/10% FBS. HeLa cells were transfected with FuGENE or Lipofectamine 2000 and used after 16 hr. Human muscle cells were maintained in X medium (64% DMEM, 16% 199 Medium, 20% FBS, 2.5 ng/ml HGF, 50 mg/ml gentamycin + 107 M dexamethasone). For differentiation, cells were grown on collagen type I coated surface for 7–10 days in DMEM supplemented with 50 mg/ml gentamycin, 10 mg/ml of bovine insulin, and 100 mg/ml of human apotransferrin. For dynasore treatment, cells were washed thrice with PBS2+ (Phosphate Buffer Saline + 1.5 mM Ca2+ + 1.5 mM Mg2+) and overlaid with 80 mM dynasore in PBS2+. Cytochalasin D (CD), Latrunculin A (Lat), and Jasplakinolide (Jas) were used for 15 min at 37 C at 5 mg/ml, 1 mM, or 1 mM, respectively in normal medium. ATP depletion was performed by incubating the cells for 30 min at 37 C in PBS2+ with 10 mM deoxy-D-glucose and 10 mM NaN3. For disrupting the Golgi apparatus, cells were pretreated for 50 min with 10 mg/ml BFA. Cells were incubated for 50 min in PBS2+ supplemented with 5 mM mbCD to extract cholesterol. Hypo-osmotic shock was performed on pretreated cells by using growth medium diluted appropriately in deionized water (1:9 dilution for 30 mOsm hypo-osmotic shock and 1:1 for 150 mOsm hypo-osmotic shock). The concentration of all drugs was maintained during the hypo-osmotic shock as well as during recovery. Membrane Tether Extraction and Force Measurements Plasma membrane tethers were pulled out from cells by a concanavalin A-coated bead trapped in optical tweezers (Cuvelier et al., 2005). After extraction, the tether was held at a constant length between 5 and 150 mm, and tether forces were measured from the detected position of the bead after calibration of the optical trap. Samples were maintained at 37 C throughout. PMS Formation and Micropipette Aspiration Plasma membrane spheres (PMS) were generated by a protocol adapted from Lingwood et al. (2008). Cells were grown on glass coverslips and were incubated for 6–8 hr in PBS2+ supplemented with 10 mM MG132. Individual PMS were selected for micropipette aspiration experiments as described previously for lipid vesicles (Sorre et al., 2009). In brief, PMS were held with a micropipette (diameter 3 mm) under slight aspiration. By partially entering the pipette, the membrane is strained. Tether forces were measured as explained above, and the aspiration of the PMS was gradually increased. Fluorescence Imaging and Analysis TIRF microscopy was performed using a 1003, 1.45 NA objective and an EMCCD camera (Hamamatsu Photonics, Japan) in a Zeiss Axiovert 200 microscope. Confocal imaging was performed on a Nikon A1R microscope with a 1003, 1.4 NA objective at 1 airy unit pinhole aperture. TIRF-FRAP experiments (5 3 5 mm bleaching region) were performed on a Nikon Eclipse 2000 microscope equipped with an EMCCD camera (Roper Scientific, Tucson, AZ). Cells were maintained at 37 C during imaging. Confocal images of PMS were taken in a Z section of width 0.4 mm around the equatorial plane. Image
Cell 144, 402–413, February 4, 2011 ª2011 Elsevier Inc. 411
analysis (caveolae detection, volume measurement, FRAP, and colocalization) was done using Labview 8.5 and Vision 8.5 (National Instruments, Austin, TX) as detailed in the Extended Experimental Procedures. FLIM microscopy was performed as described previously (Hill et al., 2008).
REFERENCES
Cell Stretching Cells were grown on a rectangular PDMS sheet (thickness 100 mm, dimensions 12 3 7 mm) decorated with adhesive micropatterns and stretched uniaxially using a custom-built device equipped with a motorized linear actuator (PI, Karlsruhe, Germany) and a temperature controller. Imaging was done on the TIRF microscope. Strains were applied in 5 s, and the number of caveolae was measured from images taken within 1–3 min of stretching.
Boyd, N.L., Park, H., Yi, H., Boo, Y.C., Sorescu, G.P., Sykes, M., and Jo, H. (2003). Chronic shear induces caveolae formation and alters ERK and Akt responses in endothelial cells. Am. J. Physiol. Heart Circ. Physiol. 285, H1113–H1122.
Immunofluorescence Studies Cells were fixed, immunolabeled, and imaged on an inverted microscope (Leica, Wetzlar, Germany). Deep-Etched EM of the Cytoplasmic Surface of MLEC Cells grown on coverslips were prepared by first unroofing their apical surface, and then fixing and freezing the basal surface. The cytoplasmic surface was deeply etched and rotary shadowed with platinum/carbon at an angle of 22 C from the surface and with carbon from the top. The replica was mounted on the sample grid and observed using a transmission electron microscope. Electron Microscopy For conventional EM, cells were grown on coverslips, fixed, dehydrated, and embedded in epon. For ultrathin cryosectioning and immunogold labeling, cells were fixed and processed for ultracryomicrotomy and were immunogold labeled. Quantification of caveolae was done by counting only the number of neck-open caveolae at the plasma membrane in EM images that had at least one caveola. SUPPLEMENTAL INFORMATION Supplemental Information includes Extended Results, Extended Experimental Procedures, eight figures, and five movies and can be found online at doi: 10.1016/j.cell.2010.12.031. ACKNOWLEDGMENTS This work was supported by the Institut Curie (PIC Division Cellulaire, Polarite´ et Cancer), Agence Nationale pour la Recherche, Fondation de France (programme tumeurs), Association Franc¸aise contre les Myopathies, the National Health and Medical Research Council of Australia and the Australian Research Council (R.G.P. and D.A.), and NIH grants HL83249 and HL92085 (R.V.S.). Microscopy (M.B.) was performed at the ACRF/IMB Dynamic Imaging Centre for Cancer Biology. D.K. was funded by a doctorate fellowship from the Institut Curie, and B.S. was funded by postdoctoral fellowships from the Institut Curie and the Association pour la Recherche sur le Cancer. D.A. is a fellow of the Swiss National Science Foundation (PA00A-111446) and is indebted to K. Alexandrov for support. The authors gratefully acknowledge the Nikon Imaging Centre at Institut Curie and the CNRS consortium CellTis. We thank F. Perez for the pDendra2-C plasmid, A. Helenius for the Cav1-EGFP plasmid and MEF cells, and F. Pinaud for Cav1-EGFP HeLa. We are also grateful to M. Piel and N. Carpi for help in micropatterning and P. Bassereau for access to the confocal microscope coupled to optical tweezers. We thank C. Viaris and C. Blouin for carefully reading the manuscript, and we are indebted to A. Bigot for providing human muscle cells. We would like to thank Eurobiobank and Pr. C. Minette for giving access to primary cells to isolate the immortalized Cav3 cell line. C.L. wishes to dedicate this work to the memory of his father, Robert Lamaze, and to all patients suffering from pancreatic cancer. Received: January 7, 2010 Revised: October 27, 2010 Accepted: December 23, 2010 Published: February 3, 2011
412 Cell 144, 402–413, February 4, 2011 ª2011 Elsevier Inc.
Abankwa, D., Hanzal-Bayer, M., Ariotti, N., Plowman, S.J., Gorfe, A.A., Parton, R.G., McCammon, J.A., and Hancock, J.F. (2008). A novel switch region regulates H-ras membrane orientation and signal output. EMBO J. 27, 727–735.
Chen, C.S., Mrksich, M., Huang, S., Whitesides, G.M., and Ingber, D.E. (1997). Geometric control of cell life and death. Science 276, 1425–1428. Cuvelier, D., Dere´nyi, I., Bassereau, P., and Nassoy, P. (2005). Coalescence of membrane tethers: experiments, theory, and applications. Biophys. J. 88, 2714–2726. Dai, J., and Sheetz, M.P. (1995). Regulation of endocytosis, exocytosis, and shape by membrane tension. Cold Spring Harb. Symp. Quant. Biol. 60, 567–571. Dai, J., and Sheetz, M.P. (1999). Membrane tether formation from blebbing cells. Biophys. J. 77, 3363–3370. Dai, J., Ting-Beall, H.P., and Sheetz, M.P. (1997). The secretion-coupled endocytosis correlates with membrane tension changes in RBL 2H3 cells. J. Gen. Physiol. 110, 1–10. Dai, J., Sheetz, M.P., Wan, X., and Morris, C.E. (1998). Membrane tension in swelling and shrinking molluscan neurons. J. Neurosci. 18, 6681–6692. D’Alessandro, M., Russell, D., Morley, S.M., Davies, A.M., and Lane, E.B. (2002). Keratin mutations of epidermolysis bullosa simplex alter the kinetics of stress response to osmotic shock. J. Cell Sci. 115, 4341–4351. Dimova, R., Aranda, S., Bezlyepkina, N., Nikolov, V., Riske, K.A., and Lipowsky, R. (2006). A practical guide to giant vesicles. Probing the membrane nanoregime via optical microcopy. J. Phys. Condens. Matter 18, S1151– S1176. Dulhunty, A.F., and Franzini-Armstrong, C. (1975). The relative contributions of the folds and caveolae to the surface membrane of frog skeletal muscle fibres at different sarcomere lengths. J. Physiol. 250, 513–539. Hansen, C.G., Bright, N.A., Howard, G., and Nichols, B.J. (2009). SDPR induces membrane curvature and functions in the formation of caveolae. Nat. Cell Biol. 11, 807–814. Hayashi, Y.K., Matsuda, C., Ogawa, M., Goto, K., Tominaga, K., Mitsuhashi, S., Park, Y.E., Nonaka, I., Hino-Fukuyo, N., Haginoya, K., et al. (2009). Human PTRF mutations cause secondary deficiency of caveolins resulting in muscular dystrophy with generalized lipodystrophy. J. Clin. Invest. 119, 2623–2633. Henley, J.R., Krueger, E.W., Oswald, B.J., and McNiven, M.A. (1998). Dynamin-mediated internalization of caveolae. J. Cell Biol. 141, 85–99. Hill, M.M., Bastiani, M., Luetterforst, R., Kirkham, M., Kirkham, A., Nixon, S.J., Walser, P., Abankwa, D., Oorschot, V.M., Martin, S., et al. (2008). PTRF-Cavin, a conserved cytoplasmic protein required for caveola formation and function. Cell 132, 113–124. Jansa, P., Burek, C., Sander, E.E., and Grummt, I. (2001). The transcript release factor PTRF augments ribosomal gene transcription by facilitating reinitiation of RNA polymerase I. Nucleic Acids Res. 29, 423–429. Kawamura, S., Miyamoto, S., and Brown, J.H. (2003). Initiation and transduction of stretch-induced RhoA and Rac1 activation through caveolae: cytoskeletal regulation of ERK translocation. J. Biol. Chem. 278, 31111–31117. Kirkham, M., Nixon, S.J., Howes, M.T., Abi-Rached, L., Wakeham, D.E., Hanzal-Bayer, M., Ferguson, C., Hill, M.M., Fernandez-Rojo, M., Brown, D.A., et al. (2008). Evolutionary analysis and molecular dissection of caveola biogenesis. J. Cell Sci. 121, 2075–2086. Kozera, L., White, E., and Calaghan, S. (2009). Caveolae act as membrane reserves which limit mechanosensitive I(Cl,swell) channel activation during swelling in the rat ventricular myocyte. PLoS ONE 4, e8312. Kurzchalia, T.V., Dupree, P., Parton, R.G., Kellner, R., Virta, H., Lehnert, M., and Simons, K. (1992). VIP21, a 21-kD membrane protein is an integral
component of trans-Golgi-network-derived transport vesicles. J. Cell Biol. 118, 1003–1014. Lang, F., Busch, G.L., Ritter, M., Vo¨lkl, H., Waldegger, S., Gulbins, E., and Ha¨ussinger, D. (1998). Functional significance of cell volume regulatory mechanisms. Physiol. Rev. 78, 247–306. Lingwood, D., Ries, J., Schwille, P., and Simons, K. (2008). Plasma membranes are poised for activation of raft phase coalescence at physiological temperature. Proc. Natl. Acad. Sci. USA 105, 10005–10010. Macia, E., Ehrlich, M., Massol, R., Boucrot, E., Brunner, C., and Kirchhausen, T. (2006). Dynasore, a cell-permeable inhibitor of dynamin. Dev. Cell 10, 839–850. Merlini, L., Carbone, I., Capanni, C., Sabatelli, P., Tortorelli, S., Sotgia, F., Lisanti, M.P., Bruno, C., and Minetti, C. (2002). Familial isolated hyperCKaemia associated with a new mutation in the caveolin-3 (CAV-3) gene. J. Neurol. Neurosurg. Psychiatry 73, 65–67. Morone, N., Fujiwara, T., Murase, K., Kasai, R.S., Ike, H., Yuasa, S., Usukura, J., and Kusumi, A. (2006). Three-dimensional reconstruction of the membrane skeleton at the plasma membrane interface by electron tomography. J. Cell Biol. 174, 851–862. Morris, C.E., and Homann, U. (2001). Cell surface area regulation and membrane tension. J. Membr. Biol. 179, 79–102. Murata, T., Lin, M.I., Stan, R.V., Bauer, P.M., Yu, J., and Sessa, W.C. (2007). Genetic evidence supporting caveolae microdomain regulation of calcium entry in endothelial cells. J. Biol. Chem. 282, 16631–16643. Nabi, I.R., and Le, P.U. (2003). Caveolae/raft-dependent endocytosis. J. Cell Biol. 161, 673–677. Palade, G.E. (1953). An electron microscope study of the mitochondrial structure. J. Histochem. Cytochem. 1, 188–211. Park, H., Go, Y.M., St John, P.L., Maland, M.C., Lisanti, M.P., Abrahamson, D.R., and Jo, H. (1998). Plasma membrane cholesterol is a key molecule in shear stress-dependent activation of extracellular signal-regulated kinase. J. Biol. Chem. 273, 32304–32311. Park, H., Go, Y.M., Darji, R., Choi, J.W., Lisanti, M.P., Maland, M.C., and Jo, H. (2000). Caveolin-1 regulates shear stress-dependent activation of extracellular signal-regulated kinase. Am. J. Physiol. Heart Circ. Physiol. 278, H1285– H1293. Parton, R.G., and Simons, K. (2007). The multiple faces of caveolae. Nat. Rev. Mol. Cell Biol. 8, 185–194. Parton, R.G., Hanzal-Bayer, M., and Hancock, J.F. (2006). Biogenesis of caveolae: a structural model for caveolin-induced domain formation. J. Cell Sci. 119, 787–796. Pelkmans, L., and Zerial, M. (2005). Kinase-regulated quantal assemblies and kiss-and-run recycling of caveolae. Nature 436, 128–133. Pelkmans, L., Bu¨rli, T., Zerial, M., and Helenius, A. (2004). Caveolin-stabilized membrane domains as multifunctional transport and sorting devices in endocytic membrane traffic. Cell 118, 767–780.
Pelkmans, L., Fava, E., Grabner, H., Hannus, M., Habermann, B., Krausz, E., and Zerial, M. (2005). Genome-wide analysis of human kinases in clathrinand caveolae/raft-mediated endocytosis. Nature 436, 78–86. Pinaud, F., Michalet, X., Iyer, G., Margeat, E., Moore, H.P., and Weiss, S. (2009). Dynamic partitioning of a glycosyl-phosphatidylinositol-anchored protein in glycosphingolipid-rich microdomains imaged by single-quantum dot tracking. Traffic 10, 691–712. Prescott, L., and Brightman, M.W. (1976). The sarcolemma of Aplysia smooth muscle in freeze-fracture preparations. Tissue Cell 8, 241–258. Richter, T., Floetenmeyer, M., Ferguson, C., Galea, J., Goh, J., Lindsay, M.R., Morgan, G.P., Marsh, B.J., and Parton, R.G. (2008). High-resolution 3D quantitative analysis of caveolar ultrastructure and caveola-cytoskeleton interactions. Traffic 9, 893–909. Rizzo, V., Morton, C., DePaola, N., Schnitzer, J.E., and Davies, P.F. (2003). Recruitment of endothelial caveolae into mechanotransduction pathways by flow conditioning in vitro. Am. J. Physiol. Heart Circ. Physiol. 285, H1720– H1729. Rothberg, K.G., Heuser, J.E., Donzell, W.C., Ying, Y.S., Glenney, J.R., and Anderson, R.G. (1992). Caveolin, a protein component of caveolae membrane coats. Cell 68, 673–682. Scherer, P.E., Okamoto, T., Chun, M., Nishimoto, I., Lodish, H.F., and Lisanti, M.P. (1996). Identification, sequence, and expression of caveolin-2 defines a caveolin gene family. Proc. Natl. Acad. Sci. USA 93, 131–135. Sedding, D.G., Hermsen, J., Seay, U., Eickelberg, O., Kummer, W., Schwencke, C., Strasser, R.H., Tillmanns, H., and Braun-Dullaeus, R.C. (2005). Caveolin-1 facilitates mechanosensitive protein kinase B (Akt) signaling in vitro and in vivo. Circ. Res. 96, 635–642. Sens, P., and Turner, M.S. (2006). Budded membrane microdomains as tension regulators. Phys. Rev. E Stat. Nonlin. Soft Matter Physiol. 73, 031918. Sheetz, M.P. (2001). Cell control by membrane-cytoskeleton adhesion. Nat. Rev. Mol. Cell Biol. 2, 392–396. Sorre, B., Callan-Jones, A., Manneville, J.-B., Nassoy, P., Joanny, J.-F., Prost, J., Goud, B., and Bassereau, P. (2009). Curvature-driven lipid sorting needs proximity to a demixing point and is aided by proteins. Proc. Natl. Acad. Sci. USA 106, 5622–5626. Stan, R.V. (2005). Structure of caveolae. Biochim. Biophys. Acta 1746, 334–348. Tagawa, A., Mezzacasa, A., Hayer, A., Longatti, A., Pelkmans, L., and Helenius, A. (2005). Assembly and trafficking of caveolar domains in the cell: caveolae as stable, cargo-triggered, vesicular transporters. J. Cell Biol. 170, 769–779. Vogel, V., and Sheetz, M. (2006). Local force and geometry sensing regulate cell functions. Nat. Rev. Mol. Cell Biol. 7, 265–275. Woodman, S.E., Sotgia, F., Galbiati, F., Minetti, C., and Lisanti, M.P. (2004). Caveolinopathies: mutations in caveolin-3 cause four distinct autosomal dominant muscle diseases. Neurology 62, 538–543. Yamada, E. (1955). The fine structure of the gall bladder epithelium of the mouse. J. Biophys. Biochem. Cytol. 1, 445–458.
Cell 144, 402–413, February 4, 2011 ª2011 Elsevier Inc. 413
Theory
Influence of Cell Geometry on Division-Plane Positioning Nicolas Minc,1,3,4,* David Burgess,2,3 and Fred Chang1,3 1Department of Microbiology and Immunology, Columbia University College of Physicians and Surgeons, 701 W168th Street, New York, NY 10032, USA 2Department of Biology, Boston College, 528 Higgins Hall, 140 Commonwealth Avenue, Chestnut Hill, MA 02167-3811, USA 3Marine Biological Laboratory, 7 MBL Street, Woods Hole, MA 02543, USA 4Present address: Institut Curie, UMR 144 CNRS/IC, 26 rue d’Ulm, 75248 Paris Cedex 05, France *Correspondence:
[email protected] DOI 10.1016/j.cell.2011.01.016
SUMMARY
The spatial organization of cells depends on their ability to sense their own shape and size. Here, we investigate how cell shape affects the positioning of the nucleus, spindle and subsequent cell division plane. To manipulate geometrical parameters in a systematic manner, we place individual sea urchin eggs into microfabricated chambers of defined geometry (e.g., triangles, rectangles, and ellipses). In each shape, the nucleus is positioned at the center of mass and is stretched by microtubules along an axis maintained through mitosis and predictive of the future division plane. We develop a simple computational model that posits that microtubules sense cell geometry by probing cellular space and orient the nucleus by exerting pulling forces that scale to microtubule length. This model quantitatively predicts division-axis orientation probability for a wide variety of cell shapes, even in multicellular contexts, and estimates scaling exponents for length-dependent microtubule forces. INTRODUCTION The orientation of the division plane is a key element in the generation of a multicellular organism. During development, cells adopt a wide variety of geometrical configurations including spherical, ellipsoidal, and polyhedral shapes. Cell shape is thought to dictate the orientation of the division plane in many systems (Concha and Adams, 1998; Gray et al., 2004; O’Connell and Wang, 2000; Strauss et al., 2006; Thery and Bornens, 2006). This effect may guide the polarity of the initial cleavages in many developing embryos (Jenkinson, 1909). The correlation of the division plane with cell shape is described in Hertwig’s empirical rule, also referred to as the ‘‘long axis rule’’ (Hertwig, 1884): ‘‘The two poles of the division figure come to lie in the direction of the greatest protoplasmic mass.’’ The mechanism of nuclear and spindle positioning is now known to be a dynamic process that involves motor proteins, pulling and/or pushing forces 414 Cell 144, 414–426, February 4, 2011 ª2011 Elsevier Inc.
from the microtubule (MT) and/or actin cytoskeletons (Grill and Hyman, 2005; Kunda and Baum, 2009; Reinsch and Gonczy, 1998; Wuhr et al., 2009). Depending on cell type, the division plane can be set by the orientation of the nucleus during interphase or early prophase or may be modified by rotation or movement of the spindle during anaphase. How these forcegenerating systems globally sense the shape and dimensions of the cell remains an outstanding question. The single-cell sea urchin zygote is an attractive cell type for studying the effects of cell geometry. To date, many well-characterized systems for studying spindle positioning are in cells that exhibit asymmetric division, such as in C. elegans or S. cerevisiae, or in adherent mammalian cells. In these cell types, polarity cues or cell adhesion patterns appear to override geometric cues (Carminati and Stearns, 1997; Grill et al., 2001; Thery et al., 2005). In contrast, sea urchin zygotes are nonadherent, divide symmetrically, and appear to lack extrinsic polarity cues. These are spherical cells that have a highly reproducible cell size and cell-cycle timing. They have been used extensively in seminal studies in cytokinesis using physical manipulation approaches (Rappaport, 1996). Here, we introduce the use of microfabricated wells to manipulate cell geometry parameters in a systematic and quantitative manner. By placing the sea urchin eggs into these chambers, we can mold them into highly reproducible series of cell shapes. In contrast to traditional physical manipulation methods on single cells, this approach allows for rapid acquisition of large datasets suitable for quantitative analysis. We find that the ‘‘long axis rule’’ does not apply for certain cell shapes. We develop a computational model that fully predicts the preferred division plane and the probability that this axis will be chosen for any given cell shape. This work demonstrates that cell shape sensing can be explained by a simple mechanism based upon microtubule length-dependent forces. RESULTS Manipulation of Cell Shape using Microfabricated Chambers To control the geometry of sea urchin zygotes, we devised polydimethylsiloxane (PDMS) microfabricated chambers in which single eggs could be pushed into a variety of defined shapes
Figure 1. Controlling Cell Shape of Sea Urchin Embryos using Microfabricated Chambers (A) Use of microfabricated PDMS wells for manipulating the shape of sea urchin embryos. Fertilized eggs are placed into wells of different shapes that are designed all to be the volume of an egg. The depth of the chambers (h) is smaller than the egg diameter (d) so that the egg is slightly flattened into its new geometry. (B) Differential interference contrast (DIC) pictures of eggs in chambers adopting different geometries. (C) Division-plane positioning in cells with different geometries. Cells of different shapes were introduced into wells and then assayed for divisionplane positioning. The ‘‘long axis rule’’ predicts that cells will divide at the center of cell mass at an axis perpendicular to the long axis at this center. The cells in the top row follow this rule, but the cells in the bottom row do not. (D) Time-lapse sequence of an embryo shaped in a rectangular chamber, from 15 min after fertilization to completion of cytokinesis. DNA was stained with Hoechst. Note the early centering of the zygote nucleus after pronuclei migration and fusion, and the elongation of the interphase nucleus along the future division axis. (E) The orientation of the interphase nucleus predicts the future spindle axis and division plane in these cells. The relative centering and orientation at metaphase (M), anaphase (A), and cytokinesis (C) relative to interphase (I) are computed as indicated in the figure from time-lapse sequences. Error bars represent standard deviations. Scale bars, 20 mm. See also Figure S1, Figure S2, and Movie S1.
(Minc et al., 2009b) (Figure 1A and Figure S1A available online). The total volume of each chamber was kept similar to the egg volume whereas the height was smaller than the egg diameter, so that the egg was slightly flattened into its new shape, allowing a bidimensional description of the process (see below). Cells were malleable and could form relatively sharp angles (down to 5–10 mm local radii of curvature), although further deformation was limited, probably because of cortical tension (Figure 1B). By removing sea water between the PDMS array and the top coverslip, we could cause the eggs to enter the chamber and change their shape in 1–5 min (Figure 2A). The embryos were surrounded by sea water and were not deprived of oxygen, as PDMS is gas permeable. In all shapes assessed, cells went on to divide within the chambers with normal timing for at least 24 hr (Figure S1B), indicating that their general physiology was not grossly perturbed. Cell Shape Dictates Division-Plane Positioning during Interphase We monitored the positioning and orientation of the division plane in many different shapes. In some shapes, cells divided in a plane at the cytoplasmic center perpendicular to the longest axis of the cell at this center. However, in other cases, we found exceptions to the long axis rule. Cells often divided at an angle
different from that perpendicular to the cell’s longest axis, for instance in an ellipse with a small aspect ratio in which the long axis was not well defined. The rule also did not apply well to shapes such as squares or rectangles. Rectangular cells usually did not divide along the long axis, which is the diagonal in this shape, but rather along the longest axis of symmetry (Figure 1C and Figure S1C). To gain more insight into this process, we performed timelapse imaging of embryos inside chambers going through the first cell cycle (from typically 20 min after fertilization to the end of cytokinesis). We imaged chromosomes stained with DNA dye Hoechst 33342 to track cell-cycle transitions and determine the positions and orientations of nuclei and mitotic spindles. Upon fertilization, the sperm nucleus, which brings the centrosomal material (Holy and Schatten, 1997), migrates toward the cytoplasmic center while pulling the female nuclei at the same time and is centered upon nuclear fusion (Hamaguchi and Hiramoto, 1986). We found that the zygote nucleus was centered within 5% of the cytoplasmic center of mass of the cell in all shapes assessed. In 92.5% of cases (n = 80), the interphase nucleus was clearly elongated along a stable and specific axis. After nuclear envelope breakdown, the mitotic spindle was aligned along this same axis throughout metaphase, anaphase, and telophase. The cleavage furrow then Cell 144, 414–426, February 4, 2011 ª2011 Elsevier Inc. 415
Figure 2. Nuclear Centering Is Microtubule Dependent (A) Method for dynamically altering cell shape. Images of an interphase cell before and after entry into a well. The nucleus, stained with Hoechst, is positioned at the new cell center within minutes. (B) Images of interphase cells a few minutes after being pushed into chambers in the presence of 1% DMSO (control), 20 mM nocodazole, or 20 mM Latrunculin B (from top to bottom). Black arrows point to asters adjacent to the nucleus. The cell center of mass is marked by a red dot, and the nucleus is outlined in purple. (C) Quantification of nuclear position in the indicated conditions. The error in centering is defined as the ratio between the distances from the nucleus center to the cell’s center of mass to the long axis of the cell. Error bars represent standard deviations. (D) Confocal images of cells in chambers fixed and stained in situ for tubulin (green) and DNA (blue), in the presence of 1% DMSO or 20 mM nocodazole. Images are projections of Z stacks of 20 midsection slices that cover a total depth of 10 mm. **p < 0.01, Student’s t test compared with the control. Scale bars, 20 mm. See also Figure S3.
formed in a plane perpendicular to this axis (Figures 1D and 1E, Figure S2A, and Movie S1). Importantly, the furrow did not appear to reorient or reposition during contraction (Figure S2B). After cytokinesis, however, adhesion forces between the two daughter cells sometimes led to a minor (usually < 5–10 ) reorientation of the initial division axis. Computer-aided measurements showed that the position of the interphase nucleus predicted the position of the mitotic spindle and the subsequent division plane (to within 5% of the cell’s radius on average), and that orientation of the interphase nucleus also predicted the orientation of the spindle and division plane (within 10% on average) (Figure 1E). Together these data suggested that the division position and axis relative to a given geometry are set by the orientation of the nucleus during interphase or early prophase. Nuclear Centering Is Dependent on Microtubules and Not Actin The microtubule and actin cytoskeletons have been implicated in positioning and orienting the nucleus and spindle in various cell types and contexts (Carminati and Stearns, 1997; Grill et al., 2001; Schuh and Ellenberg, 2008; Tran et al., 2001). In normal urchin zygotes, interphase microtubules (MTs) are organized in two asters nucleated from two diametrically opposed zones around the nucleus (Foe and von Dassow, 2008; Holy and Schatten, 1997). F-actin appears relatively diffuse throughout the cytoplasm and is enhanced at the cell surface (Wong et al., 1997). Inhibition of MTs with nocodazole or 416 Cell 144, 414–426, February 4, 2011 ª2011 Elsevier Inc.
F-actin with Latrunculin B blocks pronuclei migration or entry into mitosis, depending on the time of addition (data not shown) (Hamaguchi and Hiramoto, 1986; Schatten et al., 1986). To investigate the mechanism of nuclear centering, after pronuclear fusion, we pushed cells into chambers to dynamically alter their cell shape and then assayed for the ability of the nucleus to center relative to this new shape (Figure 2A). In control cells, the nucleus recentered at the new center of mass in less than 1–2 min (Figures 2A–2C). Addition of 20 mM nocodazole prior to the cell shape change inhibited recentering of the nucleus (Figures 2B and 2C); this nocadozole treatment led to the depolymerization of detectable MTs at this stage (Figure 2D). Depolymerization of F-actin with 20 mM Latrunculin B did not affect the process (Figure 2C and Figure S3C). Thus, nuclear centering depends on microtubules and not actin in this cell type. The Nucleus Acts as a Force Sensor We observed that nuclear shape was elongated along an axis that predicts the future spindle axis. The nucleus elongated along the new long axis in less than 3 min after the change in cell shape (Figure 2A). Nocodazole treatment prior to the shape change abolished nuclear elongation and gave rise to a spherical nucleus (Figures 2B and 2D and Figure 3A). Treatment with 20 mM Latrunculin B had a minor but significant effect on the nuclear aspect ratio (Figure 2B and Figure 3A). Depolymerization of both MTs and actin was similar to treatment with nocodazole alone, suggesting that these effects were not additive. Latrunculin B did not grossly affect MT distribution (Figure S3D). Based upon work in other cell types, F-actin could contribute to multiple aspects of nuclear elongation, including
Figure 3. Nuclear Shape Is an Indicator of Microtubule Pulling Forces (A) Quantification of nuclear shape in cells treated with the indicated drugs (as in Figure 2B). The nuclear aspect ratio is defined as the ratio between the long and short axis of the ellipsoid shape of the nucleus. The cells used for this quantification have a geometrical aspect ratio smaller than 1.5 (see 3C). Error bars represent standard deviations. (B) Nuclear shape in cells with increasing aspect ratios. Close-up images of nuclei in different cells are presented on the right. Dotted colored lines outline each nucleus. Their superimposition highlights the increase in nuclear aspect ratio as cells are more elongated. Black arrows in the brightfield picture point at aster centers on the side of the nucleus. (C) Plot of the nuclear aspect ratio as a function of the cell aspect ratio, a/b, for a series of ellipsoid and rectangular cell shapes. Each point is binned from data on 10 or more cells having the same shape, for a total number of 104 cells for the ellipses and 82 for the rectangles. Error bars represent standard deviations. The dotted lines are depicted to guide the eyes. (D) Time-lapse images of the interphase nucleus in a cell just prior to and after treatment with 20 mM nocodazole. Note that nuclear shape becomes spherical in minutes. (E) The nuclear aspect ratio as a function of time in the representative cell in Figure 2D and two other cells treated in the same manner. (F) Effect of 20 mM nocadozole on the nuclear shape in elongated cells inside chambers. Black arrows point at aster centers on the side of the nucleus and the dotted purple lines outline the nucleus. (G) Nuclear aspect ratio in cells inside chambers before and after treatment with 20 mM nocodazole (1 cell per color, n = 8 cells). The black horizontal bar represents the mean value. **p < 0.01, Student’s t test compared with the control. Scale bars, 20 mm.
the linking of the centrosomes to the nuclear envelope, mechanical properties of the nucleus, and the ability of MT motors to pull (Bettinger et al., 2004; Kunda and Baum, 2009). In these urchin cells, MTs provide the primary force that elongates the nucleus, with actin providing a minor contribution. We found that nuclear elongation was more pronounced in cells with elongated cell shapes (Figure 3B). We examined this correlation in a series of ellipsoidal and rectangular cells with increasing aspects ratios. Quantitation of the nuclear and cell aspect ratios showed a significant positive correlation between these parameters in both of these classes of shapes (R2 = 0.88 and 0.98 for ellipses and rectangles, respectively). We considered whether the degree of nuclear elongation could correspond to a stretching force exerted on the nucleus.
Previous biophysical measurements have shown that the nucleus in Xenopus behaves as an elastic material with a defined elastic surface modulus of 25 pN/mm (Dahl et al., 2004). Nocodazole treatment caused the elongated nuclear shape to become spherical within 5 min (Figures 3D–3G). This effect was independent of initial nuclear aspect ratio. This behavior illustrates the elastic nature of the nucleus and the absence of plasticity (Dahl et al., 2004, 2008; Vaziri and Mofrad, 2007). These results indicate that MTs exert forces to stretch the elastic nucleus along the long axis of the cell, and that these forces scale with the aspect ratio of the cell shape. Estimates of MT forces on the nuclear envelope can be obtained by representing the nucleus as a spherical thin elastic shell of elastic surface modulus K and radius rN. The force necessary to deform such sphere into a prolate spheroid of aspect ratio Cell 144, 414–426, February 4, 2011 ª2011 Elsevier Inc. 417
Figure 4. A Computational Model that Predicts Nuclear Orientation and Division-Plane Orientation in Response to Cell Geometry (A) Schematic 2D representation of the cellular organization used for modeling. The nucleus (gray) is located at the center of mass and oriented along an axis a. Microtubules (green) emanate from two centrosomes (orange) attached to each side of the nucleus and extend out to the cortex. The total force generated by the two MT asters along the a axis is F(a), and the total torque at the cell’s center is T(a). Inset: A microtubule in the aster has a length L and is nucleated at an angle q from the axis a. It produces a pulling force f at its nuclear attachment and a torque t at the nucleus center. The projection of the force along the axis a is denoted fp. (B) Examples of cells in specific geometries (triangles and fan) at interphase and after cytokinesis, stained with Hoechst. The nuclear and subsequent spindle orientation, a, is reported and highlighted by the yellow dotted line. (C) The plots represent the different outputs of the model for these three cells in (B) (corresponding colors). The dots on the plots position the experimental spindle orientation aexp. Note that in the three cases, aexp is close to maxima of the total normalized force and the probability density, to zeros of the total normalized torque, and to minima of the normalized potential. (D) (Left) Frequency histogram of the absolute difference between aexp and ath (calculated as the maxima of the probability density) for 77 dividing cells with different shapes. (Right) Frequency histogram of the probability density ratio for the same 77 sequences. The ratio is 1 when the experimental axis has the same probability density as the theoretical axis and 0 when the experimental axis falls in the zone where the probability density is 0. (E) Model prediction of cleavage-plane orientation probability density in the depicted shapes and experimental frequency histograms of division axis in the depicted shape. Note that the division-plane angle used here, adiv, is rotated 90 from the nuclear orientation angle a presented in panels A–D.
418 Cell 144, 414–426, February 4, 2011 ª2011 Elsevier Inc.
rN is given by (see supplemental model, Extended Experimental Procedures): F=
K rN 2=3 r 1 ; 1 + n lnðrN =rC Þ N
(1)
where y is the Poisson ratio of the material and rC is the radius of the force application zone. If we assume that the elastic surface modulus of the sea urchin nucleus is similar to that of the Xenopus nucleus, our calculations estimate that the MT-dependent forces on the nucleus range from 10 to 30 pN, depending on the shape of the cell. Microtubule Organization in Different Cell Shapes We next examined the distribution of the interphase microtubules responsible for nuclear positioning. Bright-field images show that during interphase, the nucleus is associated with two asters, one on each side of the nucleus (Figure 2B and Figures 3B and 3F). To examine MTs directly, we performed immunostaining of tubulin in cells in the chambers (Figures S3A and S3B; see Experimental Procedures). In all shapes observed, MT staining confirmed the bipolar aster organization around the nucleus (Figure 2D and Figure S3E). MTs filled the whole volume of the cell and extended out to the cortex, even in elongated cells. They appeared to emanate from the centrosome at a relatively constant angular density, and the complete aster extended out over a little more than 180 (Figure S3F). In general, MTs did not exhibit buckling or curling around the cortex. In some cases there appeared to be increased tubulin staining, often near the cortex and regions of cell elongation. This pattern may be an artifact of fixation, in which some regions of the cells may be fixed unevenly due to staining in the chambers (Figure 2D and Figure S3E); alternatively, it may represent true increased local density of MTs in these regions. Although MT probes for imaging MT organization in live cells have been recently described, that study used injection of esconsin mRNA to assay 8-cell-stage embryos (von Dassow et al., 2009); our preliminary attempts to inject live 1-cell embryos in chambers with labeled esconsin protein, without modifying their shape, have not been successful (data not shown). The distribution of MTs throughout the cytoplasm and cortex suggest that MTs may directly probe the whole cell surface or volume for shape sensing. The stretching of the nucleus and the lack of MT buckling strongly suggest that MTs are providing pulling forces on the nucleus (Figure 2D and Figure S3E). Although MT pushing, which results from MT polymerization, is a major factor in smaller cells such as fission yeast (Dogterom and Yurke, 1997; Tran et al., 2001), it may be only a minor component in these larger cells (Wuhr et al., 2009). Computational Models for Microtubule-Based Shape Sensing To understand how microtubules may globally sense cell shape for nuclear positioning, we developed quantitative models. The model aims to compute the forces and torques generated by
MTs on the nucleus during interphase and output the probability of a given orientation of the division axis in a given geometry (Thery et al., 2007). We use a two-dimensional representation in which MTs emanate from two points diametrically opposed around the nucleus, grow straight, and reach out the cortex. Each microtubule produces a pulling force, f, on its nuclear attachment site and a torque, t, at the nucleus center (Figure 4A). A necessary input of the model is to assume that the force generated by each microtubule, f, depends on its length L (the exact nature of this dependence is discussed and tested below). Through this assumption, it can be shown that the nucleus will be centered near the cytoplasmic center of mass (Bjerknes, 1986; Howard, 2006). Thus, in what follows, the nucleus is assumed to be centered at the center of mass of the geometry, and we focus our attention on describing how cell shape affects axis orientation. For each shape, we aim to compute the global force F and torque T generated as a function of the orientation angle of the stretched nuclear axis, a (Figure 4A). For each possible orientation a (a varying from 0 to p), we generate two asters of N MTs nucleated at a constant angular density r from centrosomes placed at a distance ± rN from the cell’s center of mass along the axis a. An MT orientated along the direction a + q has a length L(a,q) and generates a pulling force f(L(a,q)) on its nuclear attachment that we project on the axis a, to compute the noncompensated force fp: fp ða; qÞ = fðLða; qÞÞcosðqÞ:
(2)
The resultant total force F(a) generated by each aster on its nuclear attachment along the axis a is then obtained by summing the projected force over all MTs: F
Z2 FðaÞ =
fðLða; qÞÞcosðqÞrdq;
(3)
F 2
where F is the total angular width of the aster. The torque created by each MT at the center of mass O and projected along the z axis is in turn computed as: tða; qÞ = rN fðLða; qÞÞsinðqÞ;
(4)
which yields a total torque, T(a), F
Z2 TðaÞ = rN
fðLða; qÞÞsinðqÞrdq:
(5)
F 2
Initial tests of the model showed that, above a certain threshold, the number of MTs N (or equivalently the angular density: r = N/ F) does not impact axis definition but only affects the time required to align along this axis (see below). Thus, in what follows, we keep N as a silent parameter by normalizing
(F) (Top) DIC images of divided eggs in different geometry. (Bottom) Plot of the experimental and theoretical division-plane orientation to the y axis, sin(adiv)cos(adiv), as a function of the geometrical aspect ratio. The orientation to the y axis is 1 if all cells cleave perpendicular to the y axis, 1 if they all cleave parallel to the y axis, and 0 for a random distribution of cleavage planes. Each experimental point is averaged on at least 15 different cells in a given shape. Error bars represent standard deviations. Scale bars, 20 mm. See also Figure S4, Figure S5, and Table S1.
Cell 144, 414–426, February 4, 2011 ª2011 Elsevier Inc. 419
the total force and torque with the total force computed in a round normal cell, F0: FeðaÞ = FðaÞ=F0 and TeðaÞ = TðaÞ=ðrN F0 Þ. These normalized parameters have no units and are independent of MT number. Stable axis orientation can be identified from local minima of the normalized potential VeðaÞ computed as a primitive of TeðaÞ. The probability density associated with each orientation p(a) is then calculated by introducing a white noise in the distribution of torques as proposed in Thery et al. (2007), e V ðaÞ : pðaÞ = p0 exp C
(7)
R In this formula, p0 is adjusted so that pðaÞ = 1 and C is a dimensionless fitting parameter that includes noise strength and nuclear friction (Thery et al., 2007). The parameter C is fitted by dichotomy to match the experimental average orientation in an ellipse with a small aspect ratio (see Figure 4F). This C value, as well as the geometrical parameters, rN and F, and the total force F0 are then fixed in all simulations presented hereafter, so that the only input that varies through different simulations is the geometry of the cell (see supplemental model, Extended Experimental Procedures and Table S1). As examples, we present the results of two triangles and an elongated fan-like shape, for which we plot FeðaÞ; TeðaÞ; VeðaÞ and p(a) (Figures 4B and 4C). In these cases, the theoretical orientation angle ath, which is calculated at the highest predicted orientation probability, was within 5%–10% of the experimental orientation angle aexp measured at anaphase. Testing the Model for Different Shapes We compared theoretical predictions and experimental results of time-lapse sequences of cells in many different shapes. In general, the theoretical predictions closely approximated the experimental results. In a large dataset of 77 cells of assorted shapes, the average difference between the experimental and theoretical angle aexp and ath was 15.6 ± 10.0 (Figure 4D). In addition, we also tested whether the model predicts the probability that a specific orientation will be chosen. In shapes where length differences between different axes are small (a circle), a large difference between theoretical and experimental angles could be obtained whereas the difference in probability density is close to 0, as all orientations have an equal probability. The average ratio between the theoretical and experimental probability densities obtained from the same set of 77 cells was 0.72 ± 0.16 (Figure 4D). We also compared sets of division-plane data for each individual cell shape (Figure 4E). In each shape, the computed probability density quantitatively correlated with the experimental division-plane orientation frequency. Thus, the model was highly successful in predicting the division-plane orientations. Testing the Sensitivity of Shape Sensing We sought to determine how sensitive a cell is in sensing its shape. We generated series of datasets in which the aspect ratios of the shape were systematically altered (Figure 4F). The simplest case is a series covering the transition from a circle to ellipses with increasing aspect ratio. As expected, the division axis was purely random in a circular cell. Shaping a cell into an 420 Cell 144, 414–426, February 4, 2011 ª2011 Elsevier Inc.
ellipse with an aspect ratio as small as 1.15, however, was sufficient to significantly bias the division orientation perpendicular to the long axis. At higher aspect ratios, the average orientation saturated, and the standard deviation decreased. Similar results were seen in the transition from a square to rectangles. Another informative case consisted of varying the ratio between the two major axes of a fan-like shape, in which the transition between two orthogonal division axes can be examined. This transition showed a sharp slope at the point where the two axes have similar length, supporting the view that the shape-sensing mechanism is finely tuned. These results demonstrate that the cell can robustly sense fine differences in aspect ratio of less than 15%. In all cases, the results of the theoretical model using a single set of fixed parameters matched the experimental data with remarkable accuracy in predicting the average orientation, the standard deviation, and the precision in sensing shape. Thus this simple model, which improves upon Hertwig’s long axis rule, demonstrates that length-dependent MT forces can account for spindle orientation and cell shape sensing. Testing Mechanisms of Microtubule-Dependent Forces MT pulling forces can be generated through minus-end-directed motors such as cytoplasmic dynein that may interact with the cortex and/or with cytoplasmic elements that serve as a putative ‘‘cytoplasmic scaffold’’ (Carminati and Stearns, 1997; Gonczy et al., 1999; Hamaguchi and Hiramoto, 1986; Kimura and Onami, 2005; O’Connell and Wang, 2000; Wuhr et al., 2010). We considered various models that differ in the location and density of MT pulling motors. One distinguishing hallmark between these models is how the force scales with MT length L (Hays et al., 1982) . For instance, in a model in which motors are attached to the cortex and are in limiting concentration, the MT force is predicted to scale approximately with L2 (Grill and Hyman, 2005; Hara and Kimura, 2009; Howard, 2006). In a model in which motors are uniformly attached to the cytoplasmic matrix, the pulling force may scale with L. Thus, to test mechanisms of length-dependent forces, we used a general scaling law with the form of f(L) Lb and sought to determine an experimental value of the exponent b. This b value could be estimated by comparing our experimental data on nuclear stretching with theoretical pulling forces exerted by MTs. We focused on the data series of ellipsoid and rectangular shapes with increasing aspect ratio (Figure 3B). In both of these types of shapes, the preferred axis orientation corresponds to a = 0 and to the maximum total force as well, Femax = Feða = 0Þ (Figure 5). We thus computed for different values of b the evolution of Femax as a function of the aspect ratio a/b, in ellipses and rectangles (see different color plots in Figure 5), and calculated what value of b fits best the experimental behavior (see supplemental model, Extended Experimental Procedures). This approach leads to exponent values for each dataset: bell = 3.1 ± 0.9 and brect = 3.4 ± 1.5. Integrating MT forces on a 3D volume led to estimates of exponents of 4 to 4.5 (see supplemental model, Extended Experimental Procedures). Thus, these two independent datasets show that MT forces roughly scale with L to the power of 3 to 5. This measurement indicates nonlinearity in these MT length-scaling forces. This analysis rules out
Figure 5. Nuclear Force Scaling in Different Shapes Reveals Length Dependency of Microtubules Forces (Top) Schematic 2D representation of the cellular organization used for force scaling modeling in the series of ellipses and rectangles. The axis orientation is fixed at the stable orientation, a = 0, which also corresponds to the maximum force, Fmax. Inset: The MT pulling force, f, is set to be proportional to Lb. (Bottom) Logarithmic plots of the normalized total force as a function of the cell aspect ratio. The experimental maximum forces are computed using Equation 1 from the nuclear aspect ratio data reported in Figure 3C and normalized to the force in a round cell. Theoretical plots are generated for different values of b (different colors). The slope s is computed from fitting each plot linearly. Error bars represent standard deviations.
certain mechanisms, such as a simple model where cytoplasmic-anchored motors saturate the MT lattice, and suggests mechanisms that incorporate more complex length dependency (see Discussion). We also tested whether these different power laws could affect orientation axis, by comparing experimental division-axis distribution with theoretical density probability plots generated with different b values (Figure S4). This comparison showed that axis orientation was independent of b (for b > 0) and suggested that neither the total force intensity nor the detailed scaling of the force-generation system are critical for steadystate orientation (see also Figure 7). We additionally tested another proposed shape-sensing mechanism based on contact-angle-dependent MT forces (Tsou et al., 2003) and found that it was not consistent with our experimental results in some shapes (Figure S4). Predicting Division Axis in Multicellular Contexts To test the generality of this model, we examined whether it could predict division planes in multicellular, developmental contexts. First, we used it to predict division positions in subsequent embryonic divisions in the sea urchin. We followed the behavior of urchin embryos through multiple divisions as they were confined in PDMS chambers of different geometries (Figure 6A). Cells were shaped by contact with the wall of the chamber and by other cells in the chamber. We used our model to predict the division planes based solely upon cell shapes and found that the theoretical predictions were in excellent agreement with the experimental findings: the average difference between the experimental and theoretical division-plane orientation angle was 11.5 ± 9.1 and the mean ratio of theoretical to experimental probability densities was 0.82 ± 0.12. We note that these experiments mimic geometrical effects during embryonic development in different organisms, in which the chamber can be regarded as an artificial egg shell that constrains and shapes the dividing cells. A circular chamber
corresponds, for instance, to the situation for most deuterosomes including humans and urchins, as well as the animal pole view of meroblastic cleavages of Xenopus or Zebrafish eggs. An ellipsoidal chamber is similar to a mouse embryo or an embryo of the nematode Prionchulus, which exhibits initial symmetric cleavages (Schierenberg, 2006). The division planes with sea urchin blastomeres in the chambers reproduced with excellent fidelity the observed patterns of initial blastula divisions seen in these diverse organisms. Although cell-cell contact has been considered to provide spindle orientation cues (Goldstein, 1995; Wang et al., 1997), we noted that in certain shaped chambers (like triangles and elongated rectangles), spindle orientation did not correlate with the orientation of the first cleavage plane and thus appeared to be governed in these situations more by cell shape cues than by putative cell-cell contact signals. Next, we asked whether our model could predict spindle axis orientation within an adult tissue. Figure 6B depicts an illustration of a tissue section from the pigeon testis (Guyer, 1900), which shows the shape and spindle orientation of spermatocytes undergoing meiosis. From this image, we traced the geometry of each cell and simulated the probability density distribution of spindle orientations. Our model was able to predict the division planes accurately: the average difference between the experimental and theoretical spindle orientation angle was 8.5 ± 3.1 , and the mean ratio of theoretical to experimental probability densities was 0.79 ± 0.22 (n = 13). Thus, in these cells, spindle orientation may be determined by geometrical cues. These findings demonstrate how this model may be generally applicable in predicting division patterns and illustrate how cell shape can be a major parameter in defining cleavage patterns during development. Shape Sensing in Spindle Reorientation In many cell types, the spindle undergoes rotation during metaphase or anaphase. Although the spindle does not normally rotate, in these early divisions in sea urchins (Figure 1), we tested Cell 144, 414–426, February 4, 2011 ª2011 Elsevier Inc. 421
Figure 6. Predicting Embryonic Cleavage Patterns and Spindle Orientation in Tissues (A) Time-lapse images of urchin embryos going through the first two cleavages encased in chambers of different geometries. Each colored graph on the right corresponds to the predicted probability density computed from the shape of the cell on the left, contoured with the same color. The color blue is used for the zygote, and green and red are used for the 2-cell blastomeres. The spots on the plots correspond to the experimental division-plane axis. Note that the division-plane angle used here, adiv, is rotated 90 from the nuclear orientation angle a presented in Figures 4A–4D. (B) Drawing of meiotic spermatocytes with different shapes from a tissue section of the pigeon testis, reprinted from Guyer (1900). The corresponding computed theoretical spindle orientation probability density is plotted on the right in the corresponding color and number. Only cells depicting a spindle in the image plane are analyzed. The spots on the plots correspond to the experimental axis. Scale bars, 20 mm.
whether it could reorient upon a change in cell shape. We arrested urchin embryos in metaphase by treatment with the proteasome inhibitor MG132 and introduced them into chambers of different shapes. We observed that the metaphase spindle reoriented along an axis defined by the new cell shape (Figures 7A–7E and Figure S6). The relatively slow rotation allowed us to analyze this movement using time-lapse imaging. Rotations occurred in a relatively unidirectional and steady movement and occurred in 3 to 30 min depending on the initial axis orientation and the geometry. Spindles that were properly aligned initially did not exhibit rotations or oscillations. This process depended on microtubules but not on actin (Figures 7C, 7D, and 7E). The slow timescale for this rotation as compared to the interphase nucleus, which orients in less than 3 min (Figure 2A), suggested that the torque intensity generated by MTs was much smaller in metaphase (see supplemental model, Extended Experimental Procedures). We studied this movement using our computational model (Figure 7F), using an identical set of parameters (except for MT number, see below) and an estimate of friction of the nucleus and spindle in the cytoplasm (see supplemental model, Extended Experimental Procedures). The results of the simulations correlated well with experimental findings (Figure 7B and Figure S6). The best overall fit in the model occurred if we assumed that metaphase spindles have only around 0.6% of the number of effective MTs of the interphase arrays (at which the ratio of interphase to metaphase number of MTs is around 150; Figure 7G). The calculation is consistent with immunofluorescence images showing very few astral MTs contacting the cortex in these cells (Figures 7C and 7D) (Strickland et al., 2005). Thus, this model is applicable to spindle orientation as well as nuclear orientation. 422 Cell 144, 414–426, February 4, 2011 ª2011 Elsevier Inc.
Further, this example demonstrates how this model can be used to analyze parameters such as the rate of orientation and the number of effective MTs.
DISCUSSION A Model for Orienting the Division Axis Relative to Cell Shape By systematically studying the effects of changing cell shape, we develop and test a simple quantitative model for how cell geometry dictates the positioning and orientation of the nucleus and spindle, which subsequently positions the cleavage furrow. MTs emanating from centrosomes on the nucleus probe the dimensions of the cell and exert pulling forces that depend on MT length. The observed elongation of the nuclear envelope provides estimated forces on the order of 10–30 pN on the nuclear envelope, which scales with the aspect ratio of the cell. This force, which may represent a time average of a more dynamic molecular organization, corresponds to a relatively small number of force generators (around 10–50), as seen in other systems (Grill et al., 2003). Through modeling, we demonstrate that such length-dependent MT forces are sufficient to explain nuclear centering and orientation in these cells; more elaborate mechanisms involving intracellular gradients (Moseley and Nurse, 2010) or actin-based mechanisms (Kunda and Baum, 2009) are not needed. The cell may thus sense its shape primarily by ‘‘measuring’’ the length of its MTs that extend from the centrosome to the cell surface. Integration of these forces over the whole cell provides a mechanical ensemble that seeks to reach equilibrium.
Figure 7. Rotation of the Metaphase Spindle in Response to Cell-Shape Changes (A) Cells were blocked in metaphase by treatment with the inhibitor MG132 and then changed into a new shape by introducing them into a well. Timelapse images of spindle rotation (as shown by Hoechst staining) in a metaphase-arrested cell in a rectangular chamber are shown. (B) Examples of experimental and theoretical plots of the reorientation of the spindle axis to the theoretical force axis. The theoretical plots are computed by assuming a ratio of metaphase to interphase MT number of 1/150 (see Figure 5G). (C) Images of embryos blocked in metaphase and introduced into a chamber for 20 min, in the presence and in the absence of 20 mM nocodazole. The red point corresponds to the cell’s center of mass and the yellow dotted line corresponds to the spindle axis. (D) Confocal images of embryos blocked in metaphase, treated with 1% DMSO or 20 mM nocodazole and introduced into a chamber. After 20 min, cells were fixed and stained in situ for tubulin (green) and DNA (blue). Images are projections of Z stacks of 20 mid-section slices that cover a total depth of 10 mm. (E) Quantification of spindle centering and orientation along the theoretical force axis in response to shape changes, in the indicated conditions. The error in centering is defined as the ratio between the distances from the metaphase spindle center to the cell’s center of mass to the long axis of the cell. The orientation ratio, p(aexp) /p(ath), is 1 when the spindle axis has the same probability density as the theoretical axis and 0 when the spindle axis falls in the zone where the probability density is 0. Error bars represent standard deviations. (F) Schematic 2D representation of the cellular organization at metaphase. (G) Average error between experimental and theoretical times corresponding to 12 different cells, plotted as a function of the ratio of metaphase to interphase MT number. A positive number corresponds to an underestimation in the reorientation timing of the model whereas a negative number corresponds to an overestimation of the model. **p < 0.01, Student’s t test compared with the control. Scale bars, 20 mm. See also Figure S6.
Our theoretical model is kept as simple as possible and deliberately infers only fixed adjustable parameters, modeling all cells in different geometries in the exact same manner. The exceptional fit between our experimental data and theoretical model for a wide variety of different shapes indicates that this process is close to the theoretical limit. Our model can readily predict spindle orientation with good accuracy in other cell types and thus promises to be adaptable for predicting division orientations in many cell types in a broad variety of contexts. Celltype-specific parameters such as the relative sizes of the cell and nuclei, as well as MT organization, could even be optimized to provide the best predictability (Figure S5). These rules may
apply best to cells in which cell geometry itself plays a primary role in determining the axis of division. In addition, this model could be valuable to predict whether a given cell uses cell geometry as a primary cue and to analyze parameters in the system, such as effective MT number. Microtubule Length-Dependent Forces A key aspect of our model is the assumption that MTs exert forces that scale with MT length. Recent progress has begun to reveal molecular details of how such length-dependent forces could occur through interactions with MT motors such as dynein or depolymerizing kinesin (Gardner et al., 2008; Tischer et al., Cell 144, 414–426, February 4, 2011 ª2011 Elsevier Inc. 423
2009; Varga et al., 2009; Vogel et al., 2009). For instance, a long MT may accumulate or contact more MT pulling motors than a short one. In considering different mechanisms, an open question is whether MT forces are being generated directly by pulling motors attached on cytoplasmic elements or at the cortex (Grill and Hyman, 2005; Reinsch and Gonczy, 1998; Wuhr et al., 2009). Studies in Xenopus and sand dollar embryos show that centering mechanisms can function even if the MTs do not contact the cortex, suggesting that cytoplasmic pulling is a possible mechanism in these large cells (Hamaguchi and Hiramoto, 1986; Wuhr et al., 2010). Motors may be situated in the cytoplasm, for instance on some internal membrane component or some putative cytoplasmic ‘‘matrix.’’ Alternatively, motors may primarily associate on the MT lattice and travel to accumulate at the plus ends. Although technical limitations in the sea urchin system have limited the ability to directly image motors in this process, our experimental data coupled with computational modeling allow us to discriminate between some proposed mechanisms. Our results estimate that the forces scale in a nonlinear manner to MT length (L) to roughly L3. This finding is not consistent with basic linear models in which the number of force-generating elements is strictly proportional to MT length. One previously proposed nonlinear scaling can result from having a limited number of motors at the cortex, which pull on a fraction of astral MTs (Grill and Hyman, 2005; Howard, 2006). This view makes the aster ‘‘surface sensitive’’ and is equivalent to a force per MT scaling with L2 (Hara and Kimura, 2009). We propose a ‘‘volume sensing’’ model that is more consistent with our scaling close to the L3: motors, which are in limited numbers in the cytoplasm, encounter a given MT of length L with a probability that is proportional to the cone-shaped unit volume surrounding the MT. Additional dynamic sources of nonlinear length scaling may be positive feedback or cooperativity among multiple mechanisms. For instance, longer MTs might accumulate more motors not only because of their increased length but also because of their long lifetime (Seetapun and Odde, 2010), which can lead to increased tubulin posttranslational modifications that can help recruit more motors and increase accumulation of MT stability factors (Cai et al., 2009). Another potential factor may be the increased probability of new nucleation of noncentrosomal MTs off of longer pre-existing MTs (Janson et al., 2007). Further quantitative analysis of MTs and associated motors will help to test these proposed models. EXPERIMENTAL PROCEDURES Microchamber Fabrication and Operation Chambers containing the arrays of microwells were fabricated by rapid prototyping and PDMS technology as described in Minc et al. (2009a), Minc et al. (2009b), and Velve-Casquillas et al. (2010). A SU-8-positive master containing hundreds of posts that are 70 mm in height and of different geometries was first made by microlithography (Figure S1A). A 10:1 mixture of PDMS Sylgard 184 silicone elastomer and curing agent was poured onto the master and baked at 65 C for 4 hr. The replica was cut, peeled off the master, and activated with a plasma cleaner (Harrick Plasma). A 100 ml drop of fertilized eggs in sea water was placed onto the PDMS replica and the eggs were left to sediment by gravity for 2 min. A 22 mm2 glass coverslip was then placed on top of the suspension, and water was gently sucked from the sides of the coverslip with a kimwipe, which slowly pushed the eggs into the chambers. To perfuse
424 Cell 144, 414–426, February 4, 2011 ª2011 Elsevier Inc.
cells inside chambers, for in situ immunofluorescence or drug treatment, we used an inverse set-up: The PDMS replica was first pierced with two large holes. The drop of eggs was placed onto a 45 3 50 mm2 large coverslip and subsequently covered with the PDMS replica. Water was removed, as before, to reduce the space between the PDMS and the coverslip, which caused the eggs to adopt the shape of the microchamber (Figure S3A). Buffers, drugs, or fixatives were added slowly through the large holes and removed by sucking from the side of the PDMS replica. Cells were monitored on the microscope to ensure that no major shape change was occurring. For incubation periods longer than 1 hr the whole chamber was placed in a sealed box filled with some wet paper to limit drying. Rapid nocodazole treatment on normal spherical eggs was performed in a different set of microfluidic channels designed as in Minc and Chang (2010). Collection and Fertilization of Sea Urchin Eggs Lytechinus pictus sea urchins were purchased from Marinus Scientific. Gametes were collected by intracoelemic injection of 0.5M KCl. The eggs were resuspended and gently agitated twice in fresh sea water. Sperm were diluted 1000-fold in sea water, activated by vigorous aeration, and then added dropwise to the eggs. Fertilization was monitored after 2 min. Fertilization envelopes were subsequently removed by pouring the eggs through Nitex mesh in sea water with 5 mM PABA (4-Aminobenzoic acid). The cells were maintained at 17 C–19 C throughout the experiment. Pharmacological Inhibitors and Dyes The DNA stain Hoechst 33342 (Molecular Probes) was added at a final concentration of 1 mg/ml after the fertilization envelopes were removed. Inhibitors were added at appropriate periods of the cell cycle and incubated 5 min prior to observation. Nocodazole was used at a final concentration of 20 mM from a 1003 stock solution made fresh in DMSO. Latrunculin B (Sigma) was used at a final concentration of 20 mM from a 1003 stock in DMSO. MG132 (Sigma) was added to the eggs 30 min before fertilization at a final concentration of 50 mM from a 1003 stock in DMSO. Fixation and Staining Procedure Fixation and staining procedures were adapted from Foe and von Dassow (2008) and Strickland et al. (2004). Briefly, cells were fixed in 100 mM HEPES (pH 6.9), 50 mM EGTA, 10 mM MgSO4, 2% formaldehyde, 0.2% glutaraldehyde, 0.2% acrolein, 0.2% Triton X-100, and 400 mM dextrose for 45 min. Cells were then rinsed three times in PBT, treated with 0.1% NaBH4 in PBS to limit autofluorescence for 30 min, and finally blocked in 5% goat serum for 1 hr. Microtubule staining was performed using a primary anti-a-tubulin antibody, clone DM 1A (Sigma) at 1/8000 incubated overnight and a CY3conjugated anti-mouse secondary antibody at 1/750 (Sigma) incubated during 5 hr. Actin staining was performed using Alexa fluor phalloidin (Molecular Probes) incubated during 1 hr. Microscopy and Image Analysis Microscopy was performed with an inverted wide-field fluorescence microscope with a motorized stage (Ludl Instrument). The objectives used were either a 103 0.25 NA or a 403 0.75 NA. Confocal imaging of microtubules was performed with a laser-scanning confocal microscope (LSM 710, Zeiss) with a 403 1.3 NA oil objective. Images were acquired, processed, and analyzed with OpenLab (Improvision), Micro-manager, Image J, Zen (Zeiss), and Matlab (Mathworks). Computational Modeling All computational simulations were performed using Matlab (Mathworks). Scripts can be made available upon request. Details and tests of the different used models are provided in the Extended Experimental Procedures. SUPPLEMENTAL INFORMATION Supplemental Information includes Extended Experimental Procedures, six figures, one table, and one movie and can be found with this article online at doi:10.1016/j.cell.2011.01.016.
ACKNOWLEDGMENTS The authors acknowledge all members of the Chang and Burgess laboratories for discussions and technical assistance. We thank A. Boudaoud for careful reading of the manuscript and J. Brill for discussions. Microfabrication was made in the Columbia CEPSR clean room. Part of the microscopy was performed in the Microbiology Department confocal microscopy facility. This work, which was initiated at the Marine Biological Laboratory, was supported by National Institutes of Health (NIH) RO1 grants GM069670 and GM056836 and an Ellison Senior Scholar award to F.C. N.M. acknowledges support from the CNRS and an ANR ‘‘retour post-doctorants’’ grant ANR-10PDOC003-01. The authors dedicate this manuscript to the memory of Ray Rappaport. Received: May 25, 2010 Revised: November 9, 2010 Accepted: January 10, 2011 Published: February 3, 2011
REFERENCES Bettinger, B.T., Gilbert, D.M., and Amberg, D.C. (2004). Actin up in the nucleus. Nat. Rev. Mol. Cell Biol. 5, 410–415. Bjerknes, M. (1986). Physical theory of the orientation of astral mitotic spindles. Science 234, 1413–1416. Cai, D., McEwen, D.P., Martens, J.R., Meyhofer, E., and Verhey, K.J. (2009). Single molecule imaging reveals differences in microtubule track selection between Kinesin motors. PLoS Biol. 7, e1000216. Carminati, J.L., and Stearns, T. (1997). Microtubules orient the mitotic spindle in yeast through dynein-dependent interactions with the cell cortex. J. Cell Biol. 138, 629–641. Concha, M.L., and Adams, R.J. (1998). Oriented cell divisions and cellular morphogenesis in the zebrafish gastrula and neurula: a time-lapse analysis. Development 125, 983–994. Dahl, K.N., Kahn, S.M., Wilson, K.L., and Discher, D.E. (2004). The nuclear envelope lamina network has elasticity and a compressibility limit suggestive of a molecular shock absorber. J. Cell Sci. 117, 4779–4786. Dahl, K.N., Ribeiro, A.J., and Lammerding, J. (2008). Nuclear shape, mechanics, and mechanotransduction. Circ. Res. 102, 1307–1318. Dogterom, M., and Yurke, B. (1997). Measurement of the force-velocity relation for growing microtubules. Science 278, 856–860. Foe, V.E., and von Dassow, G. (2008). Stable and dynamic microtubules coordinately shape the myosin activation zone during cytokinetic furrow formation. J. Cell Biol. 183, 457–470. Gardner, M.K., Bouck, D.C., Paliulis, L.V., Meehl, J.B., O’Toole, E.T., Haase, J., Soubry, A., Joglekar, A.P., Winey, M., Salmon, E.D., et al. (2008). Chromosome congression by Kinesin-5 motor-mediated disassembly of longer kinetochore microtubules. Cell 135, 894–906. Goldstein, B. (1995). Cell contacts orient some cell division axes in the Caenorhabditis elegans embryo. J. Cell Biol. 129, 1071–1080.
Grill, S.W., Howard, J., Schaffer, E., Stelzer, E.H., and Hyman, A.A. (2003). The distribution of active force generators controls mitotic spindle position. Science 301, 518–521. Guyer, M.F. (1900). Spermatogenesis of normal and of hybrid pigeons. A dissertation, (Chicago, IL: The University of Chicago). http://post.queensu. ca/forsdyke/guyer01.htm. Hamaguchi, M.S., and Hiramoto, Y. (1986). Analysis of the role of astral rays in pronuclear migration in sand dollar eggs by the colcemid-UV method. Dev. Growth Differ. 28, 143–156. Hara, Y., and Kimura, A. (2009). Cell-size-dependent spindle elongation in the Caenorhabditis elegans early embryo. Curr. Biol. 19, 1549–1554. Hays, T.S., Wise, D., and Salmon, E.D. (1982). Traction force on a kinetochore at metaphase acts as a linear function of kinetochore fiber length. J. Cell Biol. 93, 374–389. Hertwig, O. (1884). Das Problem der Befruchtung une der Isotropie des Eies, eine Theory der Vererbung (Jenaische Zeitschrist). Holy, J., and Schatten, G. (1997). Recruitment of maternal material during assembly of the zygote centrosome in fertilized sea urchin eggs. Cell Tissue Res. 289, 285–297. Howard, J. (2006). Elastic and damping forces generated by confined arrays of dynamic microtubules. Phys. Biol. 3, 54–66. Janson, M.E., Loughlin, R., Loiodice, I., Fu, C., Brunner, D., Nedelec, F.J., and Tran, P.T. (2007). Crosslinkers and motors organize dynamic microtubules to form stable bipolar arrays in fission yeast. Cell 128, 357–368. Jenkinson, J.W. (1909). Experimental Embryology (Oxford, UK: Clarendon Press). Kimura, A., and Onami, S. (2005). Computer simulations and image processing reveal length-dependent pulling force as the primary mechanism for C. elegans male pronuclear migration. Dev. Cell 8, 765–775. Kunda, P., and Baum, B. (2009). The actin cytoskeleton in spindle assembly and positioning. Trends Cell Biol. 19, 174–179. Minc, N., Boudaoud, A., and Chang, F. (2009a). Mechanical forces of fission yeast growth. Curr. Biol. 19, 1096–1101. Minc, N., Bratman, S.V., Basu, R., and Chang, F. (2009b). Establishing new sites of polarization by microtubules. Curr. Biol. 19, 83–94. Minc, N., and Chang, F. (2010). Electrical control of cell polarization in the fission yeast Schizosaccharomyces pombe. Curr. Biol. 20, 710–716. Moseley, J.B., and Nurse, P. (2010). Cell division intersects with cell geometry. Cell 142, 184–188. O’Connell, C.B., and Wang, Y.L. (2000). Mammalian spindle orientation and position respond to changes in cell shape in a dynein-dependent fashion. Mol. Biol. Cell 11, 1765–1774. Rappaport, R. (1996). Cytokinesis in Animal Cells (Cambridge, UK: Cambridge University Press). Reinsch, S., and Gonczy, P. (1998). Mechanisms of nuclear positioning. J. Cell Sci. 111, 2283–2295. Schatten, G., Schatten, H., Spector, I., Cline, C., Paweletz, N., Simerly, C., and Petzelt, C. (1986). Latrunculin inhibits the microfilament-mediated processes during fertilization, cleavage and early development in sea urchins and mice. Exp. Cell Res. 166, 191–208.
Gonczy, P., Pichler, S., Kirkham, M., and Hyman, A.A. (1999). Cytoplasmic dynein is required for distinct aspects of MTOC positioning, including centrosome separation, in the one cell stage Caenorhabditis elegans embryo. J. Cell Biol. 147, 135–150.
Schierenberg, E. (2006). Embryological variation during nematode development. WormBook 2006, 1–13.
Gray, D., Plusa, B., Piotrowska, K., Na, J., Tom, B., Glover, D.M., and ZernickaGoetz, M. (2004). First cleavage of the mouse embryo responds to change in egg shape at fertilization. Curr. Biol. 14, 397–405.
Seetapun, D., and Odde, D.J. (2010). Cell-length-dependent microtubule accumulation during polarization. Curr. Biol. 20, 979–988.
Schuh, M., and Ellenberg, J. (2008). A new model for asymmetric spindle positioning in mouse oocytes. Curr. Biol. 18, 1986–1992.
Grill, S.W., and Hyman, A.A. (2005). Spindle positioning by cortical pulling forces. Dev. Cell 8, 461–465.
Strauss, B., Adams, R.J., and Papalopulu, N. (2006). A default mechanism of spindle orientation based on cell shape is sufficient to generate cell fate diversity in polarised Xenopus blastomeres. Development 133, 3883–3893.
Grill, S.W., Gonczy, P., Stelzer, E.H., and Hyman, A.A. (2001). Polarity controls forces governing asymmetric spindle positioning in the Caenorhabditis elegans embryo. Nature 409, 630–633.
Strickland, L., von Dassow, G., Ellenberg, J., Foe, V., Lenart, P., and Burgess, D. (2004). Light microscopy of echinoderm embryos. Methods Cell Biol. 74, 371–409.
Cell 144, 414–426, February 4, 2011 ª2011 Elsevier Inc. 425
Strickland, L.I., Wen, Y., Gundersen, G.G., and Burgess, D.R. (2005). Interaction between EB1 and p150glued is required for anaphase astral microtubule elongation and stimulation of cytokinesis. Curr. Biol. 15, 2249–2255. Thery, M., and Bornens, M. (2006). Cell shape and cell division. Curr. Opin. Cell Biol. 18, 648–657. Thery, M., Jimenez-Dalmaroni, A., Racine, V., Bornens, M., and Julicher, F. (2007). Experimental and theoretical study of mitotic spindle orientation. Nature 447, 493–496. Thery, M., Racine, V., Pepin, A., Piel, M., Chen, Y., Sibarita, J.B., and Bornens, M. (2005). The extracellular matrix guides the orientation of the cell division axis. Nat. Cell Biol. 7, 947–953. Tischer, C., Brunner, D., and Dogterom, M. (2009). Force- and kinesin8-dependent effects in the spatial regulation of fission yeast microtubule dynamics. Mol. Syst. Biol. 5, 250. Tran, P.T., Marsh, L., Doye, V., Inoue, S., and Chang, F. (2001). A mechanism for nuclear positioning in fission yeast based on microtubule pushing. J. Cell Biol. 153, 397–411.
Vaziri, A., and Mofrad, M.R. (2007). Mechanics and deformation of the nucleus in micropipette aspiration experiment. J. Biomech. 40, 2053–2062. Velve-Casquillas, G., Le Berre, M., Piel, M., and Tran, P.T. (2010). Microfluidic tools for cell biological research. Nano Today 5, 28–47. Vogel, S.K., Pavin, N., Maghelli, N., Julicher, F., and Tolic-Norrelykke, I.M. (2009). Self-organization of dynein motors generates meiotic nuclear oscillations. PLoS Biol. 7, e1000087. von Dassow, G., Verbrugghe, K.J., Miller, A.L., Sider, J.R., and Bement, W.M. (2009). Action at a distance during cytokinesis. J. Cell Biol. 187, 831–845. Wang, S.W., Griffin, F.J., and Clark, W.H., Jr. (1997). Cell-cell association directed mitotic spindle orientation in the early development of the marine shrimp Sicyonia ingentis. Development 124, 773–780. Wong, G.K., Allen, P.G., and Begg, D.A. (1997). Dynamics of filamentous actin organization in the sea urchin egg cortex during early cleavage divisions: implications for the mechanism of cytokinesis. Cell Motil. Cytoskeleton 36, 30–42.
Tsou, M.F., Ku, W., Hayashi, A., and Rose, L.S. (2003). PAR-dependent and geometry-dependent mechanisms of spindle positioning. J. Cell Biol. 160, 845–855.
Wuhr, M., Dumont, S., Groen, A.C., Needleman, D.J., and Mitchison, T.J. (2009). How does a millimeter-sized cell find its center? Cell Cycle 8, 1115– 1121.
Varga, V., Leduc, C., Bormuth, V., Diez, S., and Howard, J. (2009). Kinesin-8 motors act cooperatively to mediate length-dependent microtubule depolymerization. Cell 138, 1174–1183.
Wuhr, M., Tan, E.S., Parker, S.K., Detrich, H.W., 3rd, and Mitchison, T.J. (2010). A model for cleavage plane determination in early amphibian and fish embryos. Curr. Biol. 20, 2040–2045.
426 Cell 144, 414–426, February 4, 2011 ª2011 Elsevier Inc.
Theory
Control of the Mitotic Cleavage Plane by Local Epithelial Topology William T. Gibson,1,2,3,5 James H. Veldhuis,4 Boris Rubinstein,3 Heather N. Cartwright,3 Norbert Perrimon,5 G. Wayne Brodland,4 Radhika Nagpal,2 and Matthew C. Gibson3,6,* 1Program
in Biophysics of Engineering & Applied Sciences Harvard University, Cambridge, MA 02138, USA 3Stowers Institute for Medical Research, Kansas City, MO 64110, USA 4Department of Civil and Environmental Engineering, University of Waterloo, Waterloo, ON N2L 3G1, Canada 5Department of Genetics, Howard Hughes Medical Institute, Harvard Medical School, Boston, MA 02115, USA 6Department of Anatomy and Cell Biology, Kansas University Medical Center, Kansas City, KS 64110, USA *Correspondence:
[email protected] DOI 10.1016/j.cell.2010.12.035 2School
SUMMARY
For nearly 150 years, it has been recognized that cell shape strongly influences the orientation of the mitotic cleavage plane (e.g., Hofmeister, 1863). However, we still understand little about the complex interplay between cell shape and cleavage-plane orientation in epithelia, where polygonal cell geometries emerge from multiple factors, including cell packing, cell growth, and cell division itself. Here, using mechanical simulations, we show that the polygonal shapes of individual cells can systematically bias the long-axis orientations of their adjacent mitotic neighbors. Strikingly, analyses of both animal epithelia and plant epidermis confirm a robust and nearly identical correlation between local cell topology and cleavage-plane orientation in vivo. Using simple mathematics, we show that this effect derives from fundamental packing constraints. Our results suggest that local epithelial topology is a key determinant of cleavage-plane orientation, and that cleavage-plane bias may be a widespread property of polygonal cell sheets in plants and animals. INTRODUCTION The active control of the mitotic cleavage plane is crucial to numerous processes, and the consequences of cleavage-plane misorientation can be catastrophic, ranging from polycystic kidney disease and organ malformation to tumorigenesis (Baena-Lo´pez et al., 2005; Fischer et al., 2006; Gong et al., 2004; Quyn et al., 2010; Saburi et al., 2008). Although the control of cleavage-plane orientation is usually understood from a molecular viewpoint (Buschmann et al., 2006; Ferna´ndezMin˜a´n et al., 2007; Johnston et al., 2009; Siller and Doe, 2009; Speicher et al., 2008; The´ry et al., 2005; Traas et al., 1995; Vanstraelen et al., 2006; Walker et al., 2007; Wright et al., 2009), more than a century of observations show that mitotic
cells in both plants and animals tend to divide orthogonal to their geometric long axis as a default mechanism (Gray et al., 2004; Hofmeister, 1863; O’Connell and Wang, 2000; Strauss et al., 2006). In plants, the geometric location of the division plane can be predicted by cytoskeletal structures (Kost and Chua, 2002; Palevitz, 1987; Pickett-Heaps and Northcote, 1966; Sinnott and Bloch, 1940), and biophysical experiments suggest that the cytoskeleton may be involved in the process of orienting the division plane as dictated by cell geometry (Flanders et al., 1990; Goodbody et al., 1991; Katsuta et al., 1990; Lloyd, 1991). Further, in animal cells, recent studies implicate the geometry of cell-matrix adhesions as a key determinant of cleavage-plane orientation (The´ry et al., 2005, 2007). Cell geometry is thus a critical determinant of cleavage-plane orientation, at both the molecular and biophysical levels. Whereas the regulation of mitotic cleavage-plane orientation by geometric cues has been extensively probed in unicellular systems, far less attention has been given to adherent epithelial and epidermal layers. In this context, cell geometry does not exist in isolation because cell shapes emerge from the combined effects of growth, mitosis, and cellular packing. A priori, this complex interplay of biological processes, and the diverse genetic programs that have evolved to control them in plants and animals, would appear to suggest a staggering range of possible cell geometries within an epithelium. However, spatial considerations impose powerful constraints on the shapes of cells in monolayer sheets, from the distribution of polygonal cell types (Rivier et al., 1995) to their neighbor correlations (Peshkin et al., 1991) and relative sizes (Rivier and Lissowski, 1982). Indeed, empirical investigation confirms that many monolayer cell sheets across the plant and animal kingdoms converge on a default equilibrium distribution of cellular shapes, with approximately 45% hexagons, 25% pentagons, and 20% heptagons (Gibson et al., 2006; Korn and Spalding, 1973; Lewis, 1928). Such clear conservation of cellular network architecture raises the question as to whether conserved cellular division mechanisms are responsible for generating such similar packing arrangements of cells, as numerous studies have proposed (Dubertret et al., 1998; Gibson et al., 2006; Korn and Spalding, 1973; Miri and Rivier, 2006; Patel et al., 2009). The strongest Cell 144, 427–438, February 4, 2011 ª2011 Elsevier Inc. 427
A
apical
B
initial states:
relaxed states:
M
M
D
4 B’
4
E
M
M
6
6
M
M
8
8
F
C Internal pressure
basal Hookean spring Figure 1. Local Epithelial Topology Is Predicted to Influence the Geometry of an Epithelial Cell (A) A stereotypical simple columnar epithelium. Black spots represent nuclei. (B) The Drosophila wing disc epithelium, with NRG-GFP (green) marking the septate junctions. (B0 ) A planar network representation of (B). (C) A model for finding the minimum energy configuration of cell packing, based on internal pressure and ideal springs. (D–F) (Initial states, left) Initial conditions for the relaxation algorithm. Each case varies the topology of the marked cell. (Relaxed states, right) At equilibrium, cell shape is specified by a balance between pressure and tension. The central cell’s shape is strongly influenced by the labeled cell’s topology. See also Figure S1.
evidence to date that common mechanisms are used among plants and animals to generate conserved packing relationships can be found in the mitotic shift, wherein the distribution of mitotic cell shapes is shifted by a single polygon class to have a heptagonal mean, as opposed to a hexagonal mean as seen in interphase cells. Here, we use computational modeling, experimental observation, and mathematical analysis to report that, as a default property, neighbor cell shape can strongly bias cleavage-plane orientation in the monolayer cell sheets of both plants and animals. Intriguingly, we show that this bias increases the structural regularity of an epithelium by increasing the frequency of hexagons. Our analysis indicates that simultaneously, cleavage-plane bias is also involved in specifying the mitotic shift. The mechanism through which cleavage-plane bias accomplishes these effects is differential side-gaining, whereby dividing cells preferentially cleave their common interfaces with subhexagonal cells such as quadrilaterals and avoid cleaving their common interfaces with superhexagonal cells such as octagons. Together, our results suggest a common emergent mechanism in plants and animals for the control of tissue-level 428 Cell 144, 427–438, February 4, 2011 ª2011 Elsevier Inc.
architecture by packing-mediated control of the mitotic cleavage plane. RESULTS The Shape of a Cell Is Predicted to Be Influenced by Local Topology In epithelia, the tissue-level architecture at the apical junctions is a contiguous tiling of polygonal cell shapes (Figures 1A and 1B). This pattern can be described as a simple planar network wherein a cell’s number of neighbors (topology) is equivalent to its polygon class (Figure 1B0 ). To investigate the effect of polygonal cell packing on mitotic cell shape, and by extension, cleavage-plane orientation, we tested whether a cell’s long axis is systematically influenced by the polygon class of neighboring cells. To address this, we numerically solved for the minimal energy configuration of a local cellular neighborhood (Prusinkiewicz and Lindenmayer, 1990), defined to be a central mitotic cell (M) and its first-order polygonal neighbors. Geometrically, cells were idealized as polygonal prisms with constant height (Figure 1A).
long axis (solid)
A short axis (dashed)
4
5
6
7
8
B
4
4
0.8
D
Long axis division rule
Cleavage plane index
Uniform random division rule Null hypothesis
0.6
0.4
0.2
0
4
5
6
7
4
Average neighbor topology as a function of angle theta
Predicted cleavage plane bias
C
4
4
8
Neighbor cell topology
E
Average neighbor cell topology in direction theta
4
7 6.8 6.6 6.4
Observed Expected
6.2 6 5.8 5.6 5.4 5.2 5 0
45
90
(angle with respect to short axis, in degrees)
Figure 2. The Orientation of a Cell’s Short Axis Is Predicted to Correlate with Its Quadrilateral and Pentagonal Neighbors and to Anticorrelate with Heptagonal and Octagonal Neighbors (A) Neighbor cell topology, N, influences the direction of the cellular long axis (solid line) and short axis (dashed line), based on an ellipse of best-fit (red). Secondorder and higher neighbors, which are uniformly hexagonal, are not shown. For N < 6, the short axis is oriented toward the N-sided cell N; for N > 6, it is oriented perpendicular to N. (B) The attraction of the short axis to quadrilateral cells (N = 4) is robust to heterogeneity in the local cell neighborhood. (C) We computed the cleavage-plane index, or fraction of neighbors in each polygon class (black line) located adjacent to the central cell’s short axis (presumed cleavage plane). Neighbor cells having N < 6 are significantly enriched in this position. Conversely, neighbors having N > 6 are underrepresented. For comparison, for a randomly oriented division plane, all N values occur with similar frequency (green), which is close to the null hypothesis of 2/7 (red). Error bars represent ±1 standard deviation (SD). (D) We defined an acute angle, q, with respect to a cell’s short axis (dashed red line), as well as the neighbor topology in direction q (green cells). (E) On average, neighbor topology (black) is an increasing function of acute angle q. Error bars represent the standard deviation in the sample mean topology in direction q per cell (an average of the four positions on the cell cortex corresponding to the q, over 420 such cells).
For relaxation, cell mechanics were modeled in terms of a balance between edge-length tensions, described using ideal springs, and internal pressure, modeled as an ideal gas (Figure 1C). The central mitotic cell, M, was a heptagon, consistent with the fact that the average mitotic cell is seven-sided in vivo (Aegerter-Wilmsen et al., 2010; Gibson et al., 2006). Parameters were chosen to be uniform for every cell, and initial conditions were arbitrary (Figures 1D–1F). Given these choices, the effect of local topology on the shape of the central cell was an emer-
gent property of the relaxed mechanical network at equilibrium (Figures 1D–1F; Figure S1 available online; Extended Experimental Procedures). To analyze the impact of local topology on the long axis of M, we replaced one neighbor hexagon with a single N-sided cell, N. Strikingly, inserting any nonhexagonal neighbor induced a clear long axis in M, with opposite orientation of the long axis for N < 6 versus N > 6 (Figures 1D–1F; Figure 2A). Specifically, the presence of quadrilateral or pentagonal neighbors induced a long Cell 144, 427–438, February 4, 2011 ª2011 Elsevier Inc. 429
D
C
B
A
DLG ACT PH3
C’
B’
DLG PH3
I
D’
M
C
F
E 6
6
5
5 6
7 6 7
6
4
5
5 5
5
6
5
cleavage plane index =
# n-sided neighbors in cleavage plane total # n-sided neighbors
nrg-GFP 1
H
G 1200
0.8
Counts
neighbors in c.p. 800
Cleavage plane index
neighbors
0.6
0.4
Predicted
Figure 3. In Both Plants and Animals, a Dividing Cell’s Cleavage Plane Correlates with Its Quadrilateral and Pentagonal Neighbors and Anticorrelates with Heptagonal and Octagonal Neighbors (A) The Drosophila wing imaginal disc, stained with anti-DLG to mark the junctions (green) and antiPH3 to mark chromatin (blue). (B–D and B0 –D0 ) Cell division proceeds in the plane of the epithelium via a stereotyped division process including interphase (I), mitosis (M), and cytokinesis (C). Actin staining is shown in red. (E) We can infer the topological complement of neighbors, as well as the division orientation of dividing cells, from cytokinetic figures. Junctions are marked by an nrg-gfp protein trap (red). (F and G) We examined >400 such figures and sorted the neighbors by polygon class. The neighbors on the division plane (red) are a subset of the full complement of neighbors (green and red). (H) An overlay of the predicted mitotic cleavageplane bias based on our mechanical model (black), with the biases computed from both Drosophila wing disc epithelium (blue) and cucumber epidermis (red). Each is compared with the topological null hypothesis (green). Note that here the mechanical model (black) uses the empirically derived local neighborhood topologies for direct comparison with the Drosophila data (blue). Error bars represent ±1 SD. See also Figure S2 for further information.
Cucumis Drosophila Null hypothesis
chance (Figure 2C). To quantify this relationship, we defined an acute angle, q, 0.2 with respect to the presumed cleavage plane along the central cell’s short axis (see Figure 2D). On average, as a function 0 0 8 7 6 7 9 4 5 6 8 5 4 of increasing q, the neighbor polygon Neighbor cell topology Neighbor cell topology class in direction q increased monotonically (Figure 2E). Therefore, even in a heterogeneous context, the topology of a cellular neighborhood robustly and axis parallel to the NM interface, whereas heptagons and octa- systematically influenced the orientation of the long axis in gons induced a long axis orthogonal to interface NM. These a central cell. results suggest that in cell sheets, the orientation of a mitotic cell’s longest axis can be strongly influenced by the polygon Cleavage-Plane Bias in the Drosophila Wing Disc class of a single neighboring cell. As a consequence of this In both plants and animals, cells are thought to divide their long effect, neighbor cells with fewer sides (such as quadrilaterals axis by forming a cleavage plane along the short axis of the cell and pentagons) tend to lie along the shortest axis of M, which (Hofmeister, 1863; Strauss et al., 2006). If a cell’s short axis consistently bisects its cellular neighbors having the fewest is the location of the presumed cleavage plane. To test whether this effect was robust under more realistic sides (Figure 2), then mitotic division planes should be disproconditions, we numerically relaxed heterogeneous local neigh- portionately biased toward quadrilaterals and pentagons borhoods that were stochastically generated from the known in vivo. To test this, we measured the correlation between polygonal cell shape distribution of the Drosophila mela- neighbor cell polygon class and cleavage-plane orientation in nogaster wing epithelium (Figure 2B) (Aegerter-Wilmsen et al., the Drosophila wing imaginal disc (Figure 3A). Here, cell division 2010; Gibson et al., 2006). Even under these conditions, more proceeds through a stereotyped process of cell rounding at the than 70% of quadrilateral neighbors were positioned on the apical epithelial surface (Figures 3B–3D) (Gibson et al., 2006). To central cell’s short axis, double the percentage expected by define the frequency with which different classes of polygonal 400
430 Cell 144, 427–438, February 4, 2011 ª2011 Elsevier Inc.
neighbors were bisected by the cleavage plane, we examined 420 cells at the cytokinetic stage, which is the most stable and easily scored phase of mitosis (Figure 3E). For each case, we recorded the position of all primary neighboring polygons and computed the frequency with which each polygon class occupied the cleavage-plane position (Figures 3F and 3G). If the orientation of cell division were random with respect to local topology, approximately 28.6% of any given polygon class would be expected to correlate with the cleavage plane (two randomly chosen cells out of seven neighbors). However, in the wing disc, we found that more than 50% of quadrilaterals in the primary neighborhood occupied the division plane position (Figure 3H; n = 46/83). Further, octagons were anticorrelated with the division plane and occupied that position with less than 10% probability (n = 6/77). As predicted by the mechanical model, this cleavage-plane bias was monotone decreasing across all polygon types. We conclude that in the Drosophila wing disc, the polygonal topology of local neighborhoods strongly influences cleavage-plane orientation in mitotic cells. In order to test the assumption that Drosophila wing disc cells actually divide their longest axis, we next performed time-lapse analysis of proliferating Drosophila wing discs in ex vivo culture (see Movie S1; Experimental Procedures). For each of 198 mitotic cells (Figure 4A), we measured the geometric long-axis orientation during both interphase (Figure 4A0 , far left) and cytokinesis (Figure 4A0 , far right). We found a strong tendency for cells to follow a long-axis division mechanism, although with moderate noise in the orientation (Figure 4B). This tendency to divide the longest axis correlated with the interphase geometry (Figure 4B) and increased with the cell’s interphase elongation ratio (the ratio of the long axis to the short axis; Extended Experimental Procedures). For example, for the 99 cells having an elongation ratio below the median value of 1.68, the average deviation from a long-axis division mechanism was about 33 ; by contrast, for the 99 cells having an elongation ratio above the median value, the average deviation was about 21 (data not shown). This dependence on the relative axis lengths suggests that these cells might be able to adjust their spindle orientations to their newly acquired shapes following mechanical strain, as has been previously reported in cell culture and in vertebrate embryonic cells (Black and Vincent, 1988; Gray et al., 2004; O’Connell and Wang, 2000; Strauss et al., 2006). To test whether deviation from the long-axis division mechanism could explain the discrepancy between our cleavage-plane bias predictions and the empirical measurements, we incorporated the measured deviation into our original model (Figure 4C; Extended Experimental Procedures). Interestingly, when the measured deviation was incorporated, the mechanical predictions were significantly improved (compare the red and black curves in Figure 4C), closely matching the empirically measured bias (Figure 4C, blue curve). Therefore, cleavage-plane bias is likely to be robust to noise in the cleavage-plane mechanism and may be present even when cell divisions do not perfectly obey a long-axis division scheme. Cleavage-Plane Bias in Plant Epidermis Because our original predictions were mechanically motivated (Figure 1 and Figure 2), and are expected to persist even when
there is moderate noise in the cleavage plane (Figure 4), we reasoned that cleavage-plane bias should be equally likely to appear in plant tissues. To test this, we used data from F.T. Lewis’s classical study of cucumber epidermal cell topology (Cucumis sativus) to compute the probability with which an N-sided polygonal cell occupies the division plane of a mitotic neighbor (Extended Experimental Procedures) (Lewis, 1928). Remarkably, in Cucumis, the cleavage-plane bias was almost indistinguishable from that measured in the Drosophila wing disc (Figure 3H). We once again observed strong enrichment for 4-sided cells along the cleavage planes of mitotic cells, whereas 8-sided cells were underrepresented. In order to verify our inferences from Lewis’s data (1928), we also directly examined the relationship between local topology and cellular long-axis orientation in the epidermis of Cucumis (Figure S3A). From fixed samples of cucumber epidermis, we studied a population of 501 epidermal cells having the same polygonal distribution as the original population of 500 mitotic cells studied by Lewis (1928). Cells were selected in a spatially constrained, impartial manner based solely on polygon class (Extended Experimental Procedures). We next tested whether a naive long-axis division rule was sufficient to generate cleavage-plane bias in Cucumis. Based on an ellipse of best fit to each cell’s geometry (Figure S3A; Extended Experimental Procedures), we were able to reproduce not only the cleavage-plane enrichment observed in Lewis’s original data (Figure S3C) but also the inferred cleavage-plane bias (Figure S3D). Taken together, our results suggest that cleavage-plane bias occurs in polygonal cell sheets as an emergent effect of cell packing, independent of species-specific considerations. Cleavage-Plane Bias and the Topological Constraints on Cell Geometry The quantitative similarity of cleavage-plane bias in plants, animals, and in silico suggests that the underlying mechanism is geometric, rather than molecular. In fact, fundamental geometric constraints imposed by the internal angles of neighboring polygons are sufficient to explain this phenomenon. For illustration, consider the comparison between a tiling of three hexagons versus two hexagons and a square (Figures 5A and 5B). From elementary geometry, a square (N = 4) has internal angles of 90 , whereas the internal angles of a hexagon (N = 6) average 120 (for an N-sided polygon, average internal angles are 180 (N2)/N). In the context of a contiguous layer, the presence of 90 internal angles within the square forces the internal angles of the adjacent hexagon to increase to 135 (Figure 5B). Intuitively, this deformation results in elongation of these hexagons parallel to the interface with the square, thus generating a cellular long axis. The constraints imposed by the internal angles of one cell upon the long axis of its neighbor can be formalized for the arbitrary case of an N-sided cell, surrounded by N symmetric hexagonal neighbors (Figure 5C). Assume that a mitotic cell, M, is situated vertically above cell N, resulting in a horizontal interface NM of length L. In the simplest case, all side lengths, including L, are equal and without loss of generality can be set to one. Further, the internal angles aN and bM can be Cell 144, 427–438, February 4, 2011 ª2011 Elsevier Inc. 431
A
:00m
:04m
:08m
:22m
:24m
:31m
H2RFP nrg-GFP A’
Interphase long axis correlates with cleavage plane
B
Bias prediction with measured deviation
C
0
330
30 0.8
300
60
20
40
60
80
270
100 90
240
120
210
Prediction with measured deviation
Drosophila 0.6
Null hypothesis
0.4
0.2
0
150 180
Cleavage plane index
Original prediction
counts
4
n = 198 divisions
5
6
7
8
Neighbor cell topology
Figure 4. Drosophila Wing Disc Cells Approximately Obey a Long-Axis Division Rule (A) Time series analysis illustrates the process in which an interphase cell entering mitosis gradually dilates before reaching full rounding, and then subsequently undergoes cytokinesis, in an orientation approximately predicted by its interphase long axis. (A0 ) Drawings of the process described in the corresponding panels in (A), including the mitotic cell and its immediate neighbors. The long axis of the ellipse of best fit (red) is labeled with a solid line, whereas the dashed line (predicted cleavage plane) represents the short axis. (B) The eventual orientation of the cleavage plane can be predicted based on the interphase long-axis orientation. The red line (zero deviation from long-axis division) represents a perfect correlation between the interphase long axis and the long axis of the resulting cytokinetic figure. Blue bars show the number of cells (represented by radial distance from the center) that divided with a particular angular deviation from the interphase long axis. On average, the deviation was approximately 27.1 degrees. The data are represented by the first quadrant (0 to 90 ), which is also displayed symmetrically in the other three quadrants (90 to 360 ). (C) The bias curve prediction that incorporates the measured deviation of 27 from the long axis (red) is significantly closer to the empirically measured cleavageplane bias (blue) than the naive long-axis prediction is (black). A Gaussian noise model with 27 standard deviation gives a similar result (data not shown). We controlled for the influence of topological relationships by using the same local neighborhoods as were measured from the empirical data (blue). Error bars represent ±1 SD. See also the Extended Experimental Procedures and Figure S3, which suggests that a long-axis mechanism may also operate in Cucumis.
computed as a function of N. Using simple trigonometry and exploiting the symmetric configuration of neighbors, we can solve for the ratio of the horizontal axis, dm, to the vertical axis, hm, for the ellipse of best fit to cell M (Figure 5C; Extended Experimental Procedures): 432 Cell 144, 427–438, February 4, 2011 ª2011 Elsevier Inc.
p p ffi p rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi dm 1 + sin sin zsec N N N hm
(1)
In this framework, the direction of M’s short axis (presumed cleavage plane) is described by the ratio dm:hm, which the above
A
B
C
M
120°
1
120°
135° 135° 90°
120°
6
βm
E
βm
L
αn
dm 1
N
4
D
hm
F
2 1.8 1.6
dm/hm
1.4 1.2 1.0
L=1.4 L=1.0 L=0.6
0.8 0.6 0.4
4
5
6
7
8
9
Figure 5. Fundamental Packing Constraints Are Sufficient to Explain Cleavage-Plane Bias (A) Hexagons pack at 120 angles. (B) A four-sided cell distorts the internal angles of the surrounding hexagons, inducing a long axis. (C) A geometrical argument for division-plane bias. The N-sided neighbor cell influences the ratio of the horizontal axis, dm, to the vertical axis, hm, in the M-cell. When dm:hm > 1, the N-cell is in the predicted cleavage-plane position for the M-cell. (D) A plot of the ratio dm:hm for different values of N and L. Above the gray line, the N-cell is in the M-cell’s predicted division plane; the opposite is true below the gray line. (E and F) Both N and L influence the direction of the long axis in the M-cell. (E) The value of N influences the direction of the long axis in the M-cell (top cell) for constant L. (F) The long axis of the M-cell is influenced by the side length, L, for a constant N value. See also Figure S4.
equation shows is determined by the N value (Figure 5D). Geometrically, the ratio dm:hm varies with N because the length dm decreases for N > 6 and increases for N < 6 (Figure 5E). Consequently, when N > 6 (dm:hm < 1), dm forms the short axis parallel to interface NM. Conversely, if N < 6 (dm:hm > 1), then hm forms the short axis, or presumed cleavage plane, in the direction of N, perpendicular to the interface NM. Cleavage-Plane Bias Is Predicted to Be Robust to Side Length and Cell Size Differences Intuitively, differential side lengths of N-sided neighbors would also affect the short-axis orientation of M (Figure 5F). To analyze the relative contributions of angular constraints versus side lengths, consider the more realistic case when the edge lengths are nonuniform (Ls1). Here, dm:hm depends on both N and L (Figure 5D and Extended Experimental Procedures): p p ffi p rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi dm sin L + sin (2) zsec N N N hm
For the simplified case when L = 1, the direction of the short axis undergoes a 90 rotation (between horizontal and vertical) when dm:hm passes through the value 1, which corresponds to N = 6 (red line, Figure 5D). Changing the value of L changes the length dm (Figure 5F) and thus alters the N value at which this transition occurs (black lines, Figure 5D). The long-axis orientation of M is thus determined by the interplay between the polygon class and apposed side length of each neighbor, N. In the Drosophila wing disc, the value of L fluctuates by about 40% on average (Table 1). Equation (2) predicts that a 40% deviation in L value would change the point of rotation by only a single N value, suggesting that cleavage-plane bias should be noisy yet reproducible. Supporting this analysis, cell size has a surprisingly weak influence compared to polygon class in our mechanical simulations (Figure S4). Consistent with our simulations, based on liveimaging analysis of local neighborhoods surrounding dividing cells in the Drosophila wing disc epithelium, there was no discernable difference in average area for cells occupying the cleavage-plane position (Figure S4D). We conclude that Cell 144, 427–438, February 4, 2011 ª2011 Elsevier Inc. 433
Table 1. The Effective L Value Changes by Approximately 40% in Wild-Type Drosophila Tissues N Cell Polygon Class
Average Effective L Value
Standard Deviation in Effective L Value
Sample Size(Hexagonal Interfaces with N Cells)
4
1.2504
.4165
22 interfaces
5
1.1158
.4141
231 interfaces
6
1.0580
.4053
487 interfaces
7
.9237
.4081
341 interfaces
8
.9620
.5405
46 interfaces
For each value of N (column 1), the average effective L value has been computed (column 2), as well as the sample standard deviation (column 3), using empirically extracted cellular networks from the Drosophila wing imaginal disc (Extended Experimental Procedures). The sample size for each calculation is given in column 4. The effective L value, computed for hexagonal cells, is the average value of an edge shared with an Nsided neighbor, divided by the average length of the remaining 5 edges.
internal angle constraints linked to the polygon class of neighboring cells are likely the dominant cause of cleavage-plane bias, with a lesser contribution from the effects of differential side lengths. Cleavage-Plane Bias Is Predicted to Alter Global Tissue Topology Numerous recent studies have used mathematical or computational approaches to understand the equilibrium topology of proliferating epithelia (Aegerter-Wilmsen et al., 2010; Cowan and Morris, 1988; Dubertret et al., 1998; Dubertret and Rivier, 1997; Gibson et al., 2006; Korn and Spalding, 1973; Miri and Rivier, 2006; Patel et al., 2009). Intuitively, cleavage-plane bias must alter the topology of a cell sheet because it modulates the rates at which specific polygon classes gain sides due to neighbor cell mitoses. We therefore investigated the implications of cleavage-plane bias for the distribution of polygonal cell shapes. We used two distinct computational simulations informed by the empirical division parameters (Figures S2A–S2C) to model global topology with and without cleavage-plane bias (Figure 6, Figure S5, and Figure S6). For both simulation types, the cleavageplane bias values approximated those measured empirically (Figure S5F and Figure S6H). Both an abstract, topological simulator using a Monte-Carlo framework based on topological weights (Figure 6A) (Patel et al., 2009) and a mechanical model of tissue growth based on long-axis divisions (Figure 6D) (Brodland and Veldhuis, 2002) confirmed that cleavage-plane bias affects the frequency of hexagonal cells (Figures 6B and 6E). Moreover, the distribution of mitotic polygonal cells was severely disrupted in the absence of bias, resulting in decreased frequencies of heptagons and increased frequencies of octagons and nonagons (Figures 6C and 6F). Taken together, these results suggest that cleavage-plane bias is required to achieve the normal equilibrium distribution of cell shapes. DISCUSSION The results presented here raise several important questions. First, although our analysis provides a geometrical rationale 434 Cell 144, 427–438, February 4, 2011 ª2011 Elsevier Inc.
for cleavage-plane bias based on interphase polygon topology (Figure 5), we still cannot rule out the simultaneous action of molecular cues at the cell cortex. In metazoans, epithelial cells often undergo mitosis-induced deformation prior to cleavage (Figures 3C and 3C0 ; Figures 4A and 4A0 ) (Gibson et al., 2006; The´ry and Bornens, 2008), and our live-imaging results from Drosophila strongly suggest that a cellular long axis present in interphase can inform spindle orientation during mitosis (Figure 4). One intriguing possibility is that the interphase distribution of cell-cell contacts determines the localization of cortical cues important for spindle alignment, as has been previously reported (The´ry et al., 2005, 2007). For plant cells, by contrast, our results indicate that local cell packing influences, either directly or indirectly, the placement of the phragmosome and/or pre-prophase band (Pickett-Heaps and Northcote, 1966; Sinnott and Bloch, 1940). There are multiple ways in which this might be accomplished, potentially including stress- or strain-sensing mechanisms (Hamant et al., 2008; Lintilhac and Vesecky, 1984; Lynch and Lintilhac, 1997) or, more simply, based on cytoskeletal mechanisms that are able to sense cell shape (Flanders et al., 1990; Goodbody et al., 1991; Katsuta et al., 1990). To conclude, in addition to our purely geometrical interpretation, our results are also consistent with a hypothesis that both in animals and in plants, local epithelial topology may coordinately specify both the cellular long axis and the distribution of cortical determinants of the eventual cleavage plane. A second important question concerns the broader implications of cleavage-plane bias for the emergence of cell shape. Previous studies of proliferating cell sheets in Drosophila and in Cucumis have shown that the distribution of mitotic cell shapes is shifted to have a heptagonal mean, as opposed to the hexagonal mean observed in the population of cells overall (Aegerter-Wilmsen et al., 2010; Gibson et al., 2006; Lewis, 1928). Our simulations (Figures 6A and 6D) suggest that the mitotic cell distribution is disrupted in the absence of cleavage-plane bias (Figures 6C and 6F), which is consistent with the view that in both Drosophila and Cucumis, interphase cells passively gain additional sides as a consequence of neighbor cell divisions. This interpretation contrasts with the idea that the mitotic shift solely reflects modulation of the cell cycle by topology-dependent mechanical stress (AegerterWilmsen et al., 2010). Moreover, cleavage-plane bias is actually expected to synergize with the mitotic shift. By enriching for superhexagonal cells in the mitotic cell population, which are entropically favored to neighbor subhexagonal cells (Peshkin et al., 1991), the mitotic shift intuitively must amplify the effects of cleavage-plane bias. In summary, by varying the orientation of cell division based on neighbor cell geometry, cells and tissues are able to achieve increased geometric regularity via a dynamic, topology-mediated feedback and control system. Precisely how the default geometric forces that bias cleavage-plane orientation interact with other mechanisms of division-plane control (Baena-Lo´pez et al., 2005; Gong et al., 2004; Li et al., 2009; Se´galen et al., 2010; Siller et al., 2006; Willemsen et al., 2008) should be an important avenue for future research.
A
Topological Simulator
B
C
Global cell shape distribution Drosophila wing disc epithelium Divisions with cleavage plane bias Topologically random divisions
0.55 0.5
0.45 0.4 Frequency
0.4 0.35 0.3
0.35 0.3 0.25
0.25 0.2
0.2
0.15
0.15
0.1
0.1
0.05
0.05 0
0
4
(initial condition)
Drosophila wing disc epithelium Divisions with cleavage plane bias Topologically random divisions
0.5
0.45 Frequency
Mitotic cell shape distribution 0.55
5
6
7
8
9
4
10
5
D
Finite element simulator
Global cell shape distribution
E 0.5
0.5
9
10
0.45
0.4
0.4
0.35
0.35
Frequency
Frequency
8
Drosophila wing disc epithelium Long axis divisions, -T1’s Random axis, -T1’s
0.55
0.45
0.3
0.3 0.25
0.25 0.2
0.2
0.15
0.15
0.1
0.1
0.05
0.05 0
0
(initial condition)
7
Mitotic cell shape distribution
F
Drosophila wing disc epithelium Long axis divisions, -T1’s Random axis, -T1’s
0.55
6
Cell topology
Cell topology
4
5
6
7
8
9
10
4
5
6
7
8
9
10
Cell topology
Cell topology
Figure 6. Cleavage-Plane Bias Participates in Cell Shape Emergence and Is Required for Normal Cell Packing (A) The topological simulator does not model cellular mechanics but does explicitly keep track of topological neighbor relationships. Based on topological weights, the division likelihood, division symmetry, and cleavage-plane bias are approximately matched to empirically measured statistics in a Monte-Carlo framework (see Figures S2A–S2C). (B) Hexagonal frequency declines by approximately 4% in the absence of bias. Arrows highlight this difference. (C) The distribution of mitotic cells shows pronounced alterations in the absence of bias. Arrows highlight the differences. (D) The finite element simulator models cellular mechanics, division, and rearrangement. The simulator captures mechanics in terms of a net, interfacial tension, which is modeled using rod-like finite elements. Division likelihoods are informed by the empirically measured values (Figure S2A). Cleavage-plane bias approximates the empirical values and is achieved using long-axis divisions. For finite element simulations incorporating cellular rearrangements (T1 transitions), see Figure S6. (E) In the absence of bias, hexagonal frequency declines by about 4% (compare with panel B). (F) The distribution of mitotic cells again shows pronounced alterations (compare with panel C). Error bars in (B), (C), (E), and (F) represent ±1 SD. See also Figure S5 and Figure S6.
EXPERIMENTAL PROCEDURES Fly Strains To visualize the septate junctions, we used a neuroglian-gfp exon trap line, which was described in a previous study (nrg-gfp; Morin et al., 2001). To visualize the chromosomes in parallel, we generated a stock also carrying a Histone-H2 RFP marker (Schuh et al., 2007; Bloomington stock 23650). Wing Disc Sample Preparation and Imaging Wing discs were dissected from wandering 3rd instar larvae in Ringers’ solution, fixed in 4% paraformaldehyde in PBS, and mounted in 70% glycerol/PBS. For live imaging, discs were carefully dissected and placed in a 50:50 mixture of Ringer’s solution (130 mM NaCl, 5 mM KCl, 1.5 mM MgCl2) and a second
solution (adapted from Aldaz et al., 2010), consisting of 2% FBS (GIBCO) and 0.5% Pen/Strep (GIBCO; 5000 units/ml penicillin; 5000 mg/ml streptomycin) in Shields and Sang M3 Insect media (Sigma). Live discs were mounted between two pieces of Scotch double-sided tape (3M). Air bubbles were added to the medium using an insulin syringe (BD Ultra-fine with a 30-gauge needle) to potentiate gas exchange. Wing discs were maintained in culture for up to 4 hr and imaged at intervals of 15–30 s. All samples, live and fixed, were imaged on a Leica SP5 or Leica SP2 confocal microscope with a 633 glycerol or oil objective. Cucumis Sample Preparation and Imaging Epidermis was collected from freshly gathered cucumbers approximately 10 cm in length and 2 cm in diameter (Red Ridge Farm, Odessa, MO, USA).
Cell 144, 427–438, February 4, 2011 ª2011 Elsevier Inc. 435
Epidermis was peeled in thin layers and fixed in 4% paraformaldehyde in 50 mM KPO4, 5.0 mM EDTA, and 0.2% Tween20 (pH 7) for at least 2 hr at room temperature (adapted from Gallagher and Smith, 1999). Tissue pieces were then washed 2–5 times in dH2O and incubated in 5 mg/ml Calcofluor White (Sigma) in PBS for at least 10 min before imaging. Images were collected using a Zeiss LSM 510 Meta with a 203 Plan-Apochromat objective, NA 0.8. Error Bars Unless otherwise specified, error bars refer to a single standard deviation. For the case of ratio distributions, we have reported an average value of the standard deviation. This was computed as follows: the data were randomized and broken into three subsamples of equal size in order to compute an average value for the standard deviation, based on 1000 random shuffles of the data. Annotation of Drosophila Wing Disc Cytokinetic Figures in Fixed Preparations A total of 420 cytokinetic figures and their 2946 cellular neighbors were scored by hand, in multiple focal planes to ensure accuracy of topological counts. Out of the 2946 neighbors, 840, or exactly two per cytokinetic figure, were designated as being in the division-plane position. Cells were interpreted to be in the division-plane position when they occupied the majority of the cytokinetic furrow. Due to the ambiguity of division ordering, cytokinetic figures adjacent to other cytokinetic figures were not considered for analysis.
(Mathworks). See the Extended Experimental Procedures for additional information. Topological Simulations of Proliferation Proliferation was simulated in terms of a network containing exclusively tricellular nodes, with wrapping boundary conditions. All division parameters, including division likelihoods of polygonal cells, the statistical partitioning of mother cell nodes, and the likelihoods of orienting the division plane in the direction of specific polygonal neighbor cell types, are matched to the empirically measured statistics for the Drosophila wing disc (see Figures S2A–S2C). The algorithmic details are described in the Extended Experimental Procedures. Finite Element Models of Proliferating Cell Sheets The FEM simulations (Brodland and Veldhuis, 2002; Chen and Brodland, 2000) model apical contractility, cell-cell adhesion, and all other forces along the cellular edge lengths in terms of a net, interfacial tension, g, which is generated by rod-like finite elements. Proliferation is modeled in terms of long-axis divisions. Cell-cell rearrangements (T1 transitions) are permitted when cellular edge lengths shrink below a threshold value. See Figure S6 for a comparison between simulations in which T1 transitions are active, versus those for which they are inactive. Additional details are described in the Extended Experimental Procedures. SUPPLEMENTAL INFORMATION
Annotation of Fixed Drosophila Wing Disc Epithelial Cell Sheets Images of contiguous epithelial regions from Drosophila wing disc epithelia were annotated by hand using Microsoft Powerpoint. We used custom-built software to digitize the annotations for analysis in MATLAB. A total of three such cell sheets, containing 254, 195, and 233 cells, respectively, were analyzed to compute the effective L value (Figure 5C; Table 1), which is described in the text. See the Extended Experimental Procedures for additional details. Live-Imaging Analysis of Mitosis in the Drosophila Wing Disc From live movies, a total of 198 mitotic cells in the Drosophila wing disc epithelium were analyzed by hand using ImageJ. With the exception of cells located on compartment boundaries, every scoreable cell on the epithelium was used. To control for possible mechanical influences due to neighboring divisions, we did not consider dividing cells neighboring each other to be scoreable if they rounded up at the same time. Interphase geometry measurements were based on the earliest available time point (the first movie frame), except in rare cases when epithelial morphology obscured the cell in question, in which case a slightly later time point was used. The long-axis orientation of each cell was computed using ImageJ, including curvature, based on manual input from the Polygon Selections tool. The identical procedure was used for each cell at later stages, including the eventual cytokinetic figure (see Figure 4A0 for an illustration). See the Extended Experimental Procedures for additional details. Analysis of Cucumis Epidermal Cell Sheets Images of contiguous regions of Cucumis epidermis were annotated by hand using ImageJ. Cell geometry was outlined using the Polygon Selections tool, with one node placed per tri-cellular junction, except in cases of very curved cellular edges, in which additional nodes were used to increase annotation accuracy. To visualize the ellipse of best fit to cell geometry, we used a custom-made ImageJ macro. See the Extended Experimental Procedures for additional information. Algorithm for Computing the Minimal Energy Configuration for Local Cellular Neighborhoods We used a mechanical relaxation algorithm for cellular networks that has been previously described (Prusinkiewicz and Lindenmayer, 1990). For relaxation (Figure 1), cellular networks were modeled in terms of a balance between edge length tensions (described using ideal springs) and internal pressure (Figure S1). Relaxation was implemented in terms of a system of ordinary differential equations that were solved numerically using the ODE45 solver in MATLAB
436 Cell 144, 427–438, February 4, 2011 ª2011 Elsevier Inc.
Supplemental Information includes Extended Experimental Procedures, six figures, and one movie and can be found with this article online at doi: 10.1016/j.cell.2010.12.035. ACKNOWLEDGMENTS The authors gratefully acknowledge support for this research from the Stowers Institute for Medical Research and the Burroughs Wellcome Fund (to M.C.G.), from the National Science Foundation (to R.N.), from the Natural Sciences and Engineering Research Council of Canada (to G.W.B.), and from the Howard Hughes Medical Institute (to N.P.). W.T.G. was supported in part by NIH/ NIGMS Molecular Biophysics Training Grant #T32 GM008313 and CTC grant 1029 to M.C.G. Received: June 22, 2010 Revised: November 4, 2010 Accepted: December 15, 2010 Published: February 3, 2011 REFERENCES Aegerter-Wilmsen, T., Smith, A.C., Christen, A.J., Aegerter, C.M., Hafen, E., and Basler, K. (2010). Exploring the effects of mechanical feedback on epithelial topology. Development 137, 499–506. Aldaz, S., Escudero, L.M., and Freeman, M. (2010). Live imaging of Drosophila imaginal disc development. Proc. Natl. Acad. Sci. USA 107, 14217–14222. Baena-Lo´pez, L.A., Baonza, A., and Garcı´a-Bellido, A. (2005). The orientation of cell divisions determines the shape of Drosophila organs. Curr. Biol. 15, 1640–1644. Black, S.D., and Vincent, J.P. (1988). The first cleavage plane and the embryonic axis are determined by separate mechanisms in Xenopus laevis. II. Experimental dissociation by lateral compression of the egg. Dev. Biol. 128, 65–71. Brodland, G.W., and Veldhuis, J.H. (2002). Computer simulations of mitosis and interdependencies between mitosis orientation, cell shape and epithelia reshaping. J. Biomech. 35, 673–681. Buschmann, H., Chan, J., Sanchez-Pulido, L., Andrade-Navarro, M.A., Doonan, J.H., and Lloyd, C.W. (2006). Microtubule-associated AIR9 recognizes the cortical division site at preprophase and cell-plate insertion. Curr. Biol. 16, 1938–1943.
Chen, H.H., and Brodland, G.W. (2000). Cell-level finite element studies of viscous cells in planar aggregates. J. Biomech. Eng. 122, 394–401. Cowan, R., and Morris, V.B. (1988). Division rules for polygonal cells. J. Theor. Biol. 131, 33–42. Dubertret, B., and Rivier, N. (1997). The renewal of the epidermis: A topological mechanism. Biophys. J. 73, 38–44. Dubertret, B., Aste, T., Ohlenbusch, H.M., and Rivier, N. (1998). Two-dimensional froths and the dynamics of biological tissues. Phys. Rev. E Stat. Phys. Plasmas Fluids Relat. Interdiscip. Topics 58, 6368–6378. Ferna´ndez-Min˜a´n, A., Martı´n-Bermudo, M.D., and Gonza´lez-Reyes, A. (2007). Integrin signaling regulates spindle orientation in Drosophila to preserve the follicular-epithelium monolayer. Curr. Biol. 17, 683–688. Fischer, E., Legue, E., Doyen, A., Nato, F., Nicolas, J.F., Torres, V., Yaniv, M., and Pontoglio, M. (2006). Defective planar cell polarity in polycystic kidney disease. Nat. Genet. 38, 21–23. Flanders, D.J., Rawlins, D.J., Shaw, P.J., and Lloyd, C.W. (1990). Nucleusassociated microtubules help determine the division plane of plant epidermal cells: avoidance of four-way junctions and the role of cell geometry. J. Cell Biol. 110, 1111–1122. Gallagher, K., and Smith, L.G. (1999). discordia mutations specifically misorient asymmetric cell divisions during development of the maize leaf epidermis. Development 126, 4623–4633. Gibson, M.C., Patel, A.B., Nagpal, R., and Perrimon, N. (2006). The emergence of geometric order in proliferating metazoan epithelia. Nature 442, 1038–1041. Gong, Y., Mo, C., and Fraser, S.E. (2004). Planar cell polarity signalling controls cell division orientation during zebrafish gastrulation. Nature 430, 689–693. Goodbody, K.C., Venverloo, C.J., and Lloyd, C.W. (1991). Laser microsurgery demonstrates that cytoplasmic strands anchoring the nucleus across the vacuole of premitotic plant cells are under tension. Implications for division plane alignment. Development 113, 931–939. Gray, D., Plusa, B., Piotrowska, K., Na, J., Tom, B., Glover, D.M., and ZernickaGoetz, M. (2004). First cleavage of the mouse embryo responds to change in egg shape at fertilization. Curr. Biol. 14, 397–405. Hamant, O., Heisler, M.G., Jo¨nsson, H., Krupinski, P., Uyttewaal, M., Bokov, P., Corson, F., Sahlin, P., Boudaoud, A., Meyerowitz, E.M., et al. (2008). Developmental patterning by mechanical signals in Arabidopsis. Science 322, 1650–1655. Hofmeister, W. (1863). Zusatze und Berichtigungen zu den 1851 vero¨ffentlichen Untersuchungengen der Entwicklung ho¨herer Kryptogamen. Jahrbucher fu¨r Wissenschaft und Botanik 3, 259–293. Johnston, C.A., Hirono, K., Prehoda, K.E., and Doe, C.Q. (2009). Identification of an Aurora-A/PinsLINKER/Dlg spindle orientation pathway using induced cell polarity in S2 cells. Cell 138, 1150–1163. Katsuta, J., Hashiguchi, Y., and Shibaoka, H. (1990). The role of the cytoskeleton in positioning of the nucleus in premitotic tobacco by-2 cells. J. Cell Sci. 95, 413–422. Korn, R.W., and Spalding, R.M. (1973). The geometry of plant epidermal cells. New Phytol. 72, 1357–1365. Kost, B., and Chua, N.H. (2002). The plant cytoskeleton: Vacuoles and cell walls make the difference. Cell 108, 9–12. Lewis, F.T. (1928). The correlation between cell division and the shapes and sizes of prismatic cells in the epidermis of cucumis. Anat. Rec. 38, 341–376. Li, W., Kale, A., and Baker, N.E. (2009). Oriented cell division as a response to cell death and cell competition. Curr. Biol. 19, 1821–1826. Lintilhac, P.M., and Vesecky, T.B. (1984). Stress-induced alignment of division plane in plant tissues grown in vitro. Nature 307, 363–364. Lloyd, C.W. (1991). How does the cytoskeleton read the laws of geometry in aligning the division plane of plant-cells. Development Suppl. 1, 55–65. Lynch, T.M., and Lintilhac, P.M. (1997). Mechanical signals in plant development: a new method for single cell studies. Dev. Biol. 181, 246–256.
Miri, M., and Rivier, N. (2006). Universality in two-dimensional cellular structures evolving by cell division and disappearance. Phys. Rev. E Stat. Nonlin. Soft Matter Phys. 73, 031101. Morin, X., Daneman, R., Zavortink, M., and Chia, W. (2001). A protein trap strategy to detect GFP-tagged proteins expressed from their endogenous loci in Drosophila. Proc. Natl. Acad. Sci. USA 98, 15050–15055. O’Connell, C.B., and Wang, Y.L. (2000). Mammalian spindle orientation and position respond to changes in cell shape in a dynein-dependent fashion. Mol. Biol. Cell 11, 1765–1774. Palevitz, B.A. (1987). Actin in the preprophase band of Allium cepa. J. Cell Biol. 104, 1515–1519. Patel, A.B., Gibson, W.T., Gibson, M.C., and Nagpal, R. (2009). Modeling and inferring cleavage patterns in proliferating epithelia. PLoS Comput. Biol. 5, e1000412. Peshkin, M.A., Strandburg, K.J., and Rivier, N. (1991). Entropic predictions for cellular networks. Phys. Rev. Lett. 67, 1803–1806. Pickett-Heaps, J.D., and Northcote, D.H. (1966). Organization of microtubules and endoplasmic reticulum during mitosis and cytokinesis in wheat meristems. J. Cell Sci. 1, 109–120. Prusinkiewicz, P., and Lindenmayer, A. (1990). The Algorithmic Beauty of Plants (New York: Springer-Verlag). Quyn, A.J., Appleton, P.L., Carey, F.A., Steele, R.J., Barker, N., Clevers, H., Ridgway, R.A., Sansom, O.J., and Na¨thke, I.S. (2010). Spindle orientation bias in gut epithelial stem cell compartments is lost in precancerous tissue. Cell Stem Cell 6, 175–181. Rivier, N., and Lissowski, A. (1982). On the correlation between sizes and shapes of cells in epithelial mosaics. J. Phys. Math. Gen. 15, L143–L148. Rivier, N., Schliecker, G., and Dubertret, B. (1995). The stationary state of epithelia. Acta Biotheor. 43, 403–423. Saburi, S., Hester, I., Fischer, E., Pontoglio, M., Eremina, V., Gessler, M., Quaggin, S.E., Harrison, R., Mount, R., and McNeill, H. (2008). Loss of Fat4 disrupts PCP signaling and oriented cell division and leads to cystic kidney disease. Nat. Genet. 40, 1010–1015. Schuh, M., Lehner, C.F., and Heidmann, S. (2007). Incorporation of Drosophila CID/CENP-A and CENP-C into centromeres during early embryonic anaphase. Curr. Biol. 17, 237–243. Se´galen, M., Johnston, C.A., Martin, C.A., Dumortier, J.G., Prehoda, K.E., David, N.B., Doe, C.Q., and Bellaı¨che, Y. (2010). The Fz-Dsh planar cell polarity pathway induces oriented cell division via Mud/NuMA in Drosophila and zebrafish. Dev. Cell 19, 740–752. Siller, K.H., and Doe, C.Q. (2009). Spindle orientation during asymmetric cell division. Nat. Cell Biol. 11, 365–374. Siller, K.H., Cabernard, C., and Doe, C.Q. (2006). The NuMA-related Mud protein binds Pins and regulates spindle orientation in Drosophila neuroblasts. Nat. Cell Biol. 8, 594–600. Sinnott, E.W., and Bloch, R. (1940). Cytoplasmic behavior during division of vacuolate plant cells. Proc. Natl. Acad. Sci. USA 26, 223–227. Speicher, S., Fischer, A., Knoblich, J., and Carmena, A. (2008). The PDZ protein Canoe regulates the asymmetric division of Drosophila neuroblasts and muscle progenitors. Curr. Biol. 18, 831–837. Strauss, B., Adams, R.J., and Papalopulu, N. (2006). A default mechanism of spindle orientation based on cell shape is sufficient to generate cell fate diversity in polarised Xenopus blastomeres. Development 133, 3883–3893. The´ry, M., and Bornens, M. (2008). Get round and stiff for mitosis. HFSP J 2, 65–71. The´ry, M., Jime´nez-Dalmaroni, A., Racine, V., Bornens, M., and Ju¨licher, F. (2007). Experimental and theoretical study of mitotic spindle orientation. Nature 447, 493–496. The´ry, M., Racine, V., Pe´pin, A., Piel, M., Chen, Y., Sibarita, J.B., and Bornens, M. (2005). The extracellular matrix guides the orientation of the cell division axis. Nat. Cell Biol. 7, 947–953.
Cell 144, 427–438, February 4, 2011 ª2011 Elsevier Inc. 437
Traas, J., Bellini, C., Nacry, P., Kronenberger, J., Bouchez, D., and Caboche, M. (1995). Normal differentiation patterns in plants lacking microtubular preprophase bands. Nature 375, 676–677. Vanstraelen, M., Van Damme, D., De Rycke, R., Mylle, E., Inze´, D., and Geelen, D. (2006). Cell cycle-dependent targeting of a kinesin at the plasma membrane demarcates the division site in plant cells. Curr. Biol. 16, 308–314. Walker, K.L., Mu¨ller, S., Moss, D., Ehrhardt, D.W., and Smith, L.G. (2007). Arabidopsis TANGLED identifies the division plane throughout mitosis and cytokinesis. Curr. Biol. 17, 1827–1836.
438 Cell 144, 427–438, February 4, 2011 ª2011 Elsevier Inc.
Willemsen, V., Bauch, M., Bennett, T., Campilho, A., Wolkenfelt, H., Xu, J., Haseloff, J., and Scheres, B. (2008). The NAC domain transcription factors FEZ and SOMBRERO control the orientation of cell division plane in Arabidopsis root stem cells. Dev. Cell 15, 913–922. Wright, A.J., Gallagher, K., and Smith, L.G. (2009). discordia1 and alternative discordia1 function redundantly at the cortical division site to promote preprophase band formation and orient division planes in maize. Plant Cell 21, 234–247.
Resource
Reference Maps of Human ES and iPS Cell Variation Enable High-Throughput Characterization of Pluripotent Cell Lines Christoph Bock,1,2,3,4,8 Evangelos Kiskinis,2,3,5,8 Griet Verstappen,1,2,3,8 Hongcang Gu,1 Gabriella Boulting,2,3,5,6 Zachary D. Smith,1,2,3 Michael Ziller,1,2,3 Gist F. Croft,7 Mackenzie W. Amoroso,7 Derek H. Oakley,7 Andreas Gnirke,1 Kevin Eggan,2,3,5,* and Alexander Meissner1,2,3,* 1Broad
Institute, Cambridge, MA 02142, USA of Stem Cell and Regenerative Biology, Harvard University, Cambridge, MA 02138, USA 3Harvard Stem Cell Institute, Cambridge, MA 02138, USA 4Max Planck Institute for Informatics, 66123 Saarbru ¨ cken, Germany 5The Howard Hughes Medical Institute, USA 6Department of Molecular and Cellular Biology, Harvard University, Cambridge, MA 02138, USA 7Project A.L.S./Jenifer Estess Laboratory for Stem Cell Research, Departments of Pathology, Neurology, and Neuroscience, Center for Motor Neuron Biology and Disease (MNC), and Columbia Stem Cell Initiative (CSCI), Columbia University, New York, NY 10032, USA 8These authors contributed equally to this work *Correspondence:
[email protected] (K.E.),
[email protected] (A.M.) DOI 10.1016/j.cell.2010.12.032 2Department
SUMMARY
The developmental potential of human pluripotent stem cells suggests that they can produce diseaserelevant cell types for biomedical research. However, substantial variation has been reported among pluripotent cell lines, which could affect their utility and clinical safety. Such cell-line-specific differences must be better understood before one can confidently use embryonic stem (ES) or induced pluripotent stem (iPS) cells in translational research. Toward this goal we have established genome-wide reference maps of DNA methylation and gene expression for 20 previously derived human ES lines and 12 human iPS cell lines, and we have measured the in vitro differentiation propensity of these cell lines. This resource enabled us to assess the epigenetic and transcriptional similarity of ES and iPS cells and to predict the differentiation efficiency of individual cell lines. The combination of assays yields a scorecard for quick and comprehensive characterization of pluripotent cell lines. INTRODUCTION Human embryonic stem (ES) cell lines can be cultured and expanded for many passages in vitro, without losing their ability to differentiate into all three embryonic germ layers (Thomson et al., 1998). The same is true for induced pluripotent stem (iPS) cell lines, which are obtained by reprogramming somatic cells using ectopic expression of the transcription factors OCT4, SOX2, KLF4, and C-MYC (Takahashi et al., 2007) or alternative reprogramming cocktails (reviewed in Stadtfeld and
Hochedlinger, 2010). Both ES and iPS cell lines are powerful research tools and could provide substantial quantities of disease-relevant cells for biomedical research. Several groups have already used human pluripotent cell lines as a model system for dissecting the cellular basis of monogenic diseases, and the range of diseases under investigation is rapidly expanding (reviewed in Colman and Dreesen, 2009). Future applications of human pluripotent stem cell lines could include the study of complex diseases that emerge from a mixture of genetic and environmental effects; cell-based drug screening in diseaserelevant cell types; and the use of pluripotent cells as a renewable source for transplantation medicine (Colman and Dreesen, 2009; Daley, 2010; Rubin, 2008). All of these applications require the selection and characterization of cell lines that reliably, efficiently, and stably differentiate into disease-relevant cell types. However, significant variation has been observed for the differentiation efficiency of various human ES cell lines (Di Giorgio et al., 2008; Osafune et al., 2008), and further concerns have been raised about the equivalence of human ES and iPS cell lines. For example, it has been reported that human iPS cells collectively deviate from ES cells in the expression of hundreds of genes (Chin et al., 2009), in their genome-wide DNA methylation patterns (Doi et al., 2009), and in their neural differentiation properties (Hu et al., 2010). Such differences must be better understood before human ES and iPS cell lines can be confidently used for translational research. In particular, it is necessary to establish genome-wide reference maps for patterns of gene expression and DNA methylation in a large collection of pluripotent cell lines, providing a baseline against which comparisons of epigenetic and transcriptional properties of new ES and iPS cell lines can be made. Previous research has shown that human pluripotent cells exhibit highly characteristic patterns of DNA methylation and gene expression (Guenther et al., 2010; Hawkins et al., 2010; Lister et al., 2009; Mu¨ller et al., 2008). However, these studies focused on few cell lines and Cell 144, 439–452, February 4, 2011 ª2011 Elsevier Inc. 439
therefore could not systematically investigate the role of epigenetic and transcriptional variation. In order to firmly establish the nature and magnitude of epigenetic variation that exists among human pluripotent stem cell lines, three genomic assays were applied to 20 established ES cell lines (Chen et al., 2009; Cowan et al., 2004; Thomson et al., 1998) and 12 iPS cell lines that were recently derived and functionally characterized (Boulting et al., 2011). The assays performed on each cell line included DNA methylation mapping by genome-scale bisulfite sequencing, gene expression profiling using microarrays, and a novel quantitative differentiation assay that utilizes highthroughput transcript counting of 500 lineage marker genes in embryoid bodies (EBs). Collectively, our data provide a reference of variation among human pluripotent cell lines. This reference enabled us to perform a systematic comparison between ES and iPS cell lines, to identify cell-line-specific outlier genes, and to predict each cell line’s differentiation propensity into the three germ layers. Finally, we show that the differentiation propensities that we report here are highly predictive of the efficiencies by which Boulting and colleagues could direct the differentiation of the 12 iPS cell lines into motor neurons (Boulting et al., 2011). In summary, we found that epigenetic and transcriptional variation is common among human pluripotent cell lines and that this variation can have significant impact on a cell line’s utility. Our observation applies to both ES and iPS cell lines, underlining the need to carefully characterize each cell line, regardless of how it was derived. As a step toward lowering the experimental burden of comprehensive cell line characterization and to improve the accuracy over existing assays, we have combined our three genomic assays into a bioinformatic scorecard. This scorecard enables high-throughput prediction of the quality and utility of any pluripotent cell line. RESULTS A Reference of DNA Methylation and Gene Expression in Human ES Cell Lines Human ES cell lines are subject to many factors of influence that could contribute to epigenetic and transcriptional variation, such as their genetic background, differences between derivation protocols, and varying cell culture conditions. To establish a baseline of variation among high-quality pluripotent cell lines, we obtained low-passage freezes of 20 well-characterized and widely used human ES cell lines (Table S1). These cell lines were cultured for several passages under standardized conditions, and we confirmed the expression of pluripotency markers by immunostainings (Figure S1A) before collecting material for genomic analysis of DNA methylation and gene expression. DNA methylation profiling was performed by reduced-representation bisulfite sequencing (RRBS) as described previously (Gu et al., 2010; Meissner et al., 2008) and resulted in DNA methylation measurements for approximately four million individual CpG dinucleotides per cell line. The genomic coverage was sufficient to determine DNA methylation levels for three quarters of all gene promoters, the majority of CpG islands and many other genomic elements (Figures S1B and S1C). Gene expression profiling was performed using commercially available Affymetrix microarrays and gave rise to expression levels 440 Cell 144, 439–452, February 4, 2011 ª2011 Elsevier Inc.
for a total of 15,210 Ensembl genes. All data are publicly available for visual browsing and download (http://scorecard. computational-epigenetics.org/). To determine whether the global patterns of DNA methylation and gene expression would segregate ES cell lines into subclasses, we performed hierarchical clustering (Figure 1A, Table S2), which also included data from the 6 primary fibroblast cell lines as nonpluripotent controls. Two well-separated clusters emerged, one comprising all ES cell lines and the other comprising all fibroblast cell lines. Within the ES cell cluster, there was some indication that cell lines derived at the same institution cluster together (HUES cell lines versus H1, H7, and H9), which is consistent with a prior study of marker gene expression in human ES cell lines (Adewumi et al., 2007). However, this trend was mild compared to the difference between pluripotent and nonpluripotent cells and did not significantly influence the results reported below. Consistent with the overall similarity among all 20 ES cell lines, the majority of genetic loci exhibit similar DNA methylation and gene expression levels between different ES cell lines, as exemplified by the DNA methyltransferase gene DNMT3B (Figure 1B). However, a moderate number of genes show variable DNA methylation and/or gene expression levels. For example, the antioxidant gene CAT exhibits substantial and correlated variation of DNA methylation and gene expression; the developmental regulator PAX6 exhibits gene expression variation and a consistently unmethylated gene promoter; and the macrophage/granulocyte surface marker CD14 exhibits DNA methylation variation while not being expressed in any of the 20 ES cell lines (Figure 1B). Importantly, cell-line-specific differences were maintained when we collected biological replicates from different passages of the same cell line (Figure S1D). To investigate the variation observed among human ES cell lines in a more quantitative manner, we calculated, for each gene, the distribution of DNA methylation and gene expression among the 20 ES cell lines (Table S3). The resulting ‘‘reference corridor’’ quantifies the range of DNA methylation and gene expression values for a given gene (or genomic region) in a reference set of pluripotent cell lines. Any measurement that falls outside of this corridor is regarded as an outlier and could potentially affect that cell line’s functional properties. We illustrate the concept of the reference corridor using boxplots (Figure 1C), which display the median and range of observed DNA methylation/expression levels for representative genes with different degrees of variability. For each gene (or genomic region), these plots impose upper and lower thresholds between which the DNA methylation/expression levels must fall to be considered ‘‘within the range of the current ES cell reference.’’ With this reference in hand, it becomes possible to determine the number and identity of deviations in any pluripotent cell line by using a statistical outlier filter (Tukey, 1977) and to investigate the causes and potential consequences of this variation. Causes and Consequences of Epigenetic and Transcriptional Variation among Human ES Cell Lines Plotting the deviation from the ES cell reference maps for all genes confirmed our initial observation that epigenetic and transcriptional variation focuses on a subset of genes, whereas most
DNA methylation:
A
high
medium
low
Gene expression:
high
medium
low
B
C
1.0
DNA methylation
10
0.8
8
0.6
6
0.4
4
0.2
2
0.0
Gene expression
0 PAX6 DNMT3B BMP4 DAZL SNAI1 LEFTY2 CD14 TF GATA6 GAPDH SOX2 CXCL5 MEG3 S100A6 CAT
CD14 TF DAZL SOX2 BMP4 PAX6 GATA6 LEFTY2 GAPDH SNAI1 DNMT3B S100A6 CAT CXCL5 MEG3
Figure 1. DNA Methylation and Gene Expression Profiles Quantify Variation among Human ES Cell Lines (A) Joint hierarchical clustering of DNA methylation and gene expression in 20 human ES cell lines (‘‘HUESx,’’ ‘‘Hx’’) and 6 primary fibroblast cell lines (‘‘hFibx’’). Light colors indicate high levels of DNA methylation (red) or gene expression (green), and dark colors indicate low levels. Joint DNA methylation and gene expression data are available from Table S2. (B) High-resolution view of DNA methylation and gene expression at four selected genes. DNA methylation patterns are shown for the promoter regions (5 kb to +1 kb) of representative Ensembl-annotated transcripts. Each box on the left represents a single CpG dinucleotide (dark red: high methylation, light red: little or no methylation). The single boxes on the right visualize the normalized expression levels of each gene (dark green: little or no expression, light green: high expression). The DNA methylation patterns are not drawn to scale. (C) Boxplots of gene-specific DNA methylation (left) and gene expression levels (right) among 20 low-passage human ES cell lines, illustrating the concept of an epigenetic/transcriptional reference corridor. Boxplot boxes correspond to center quartiles, the median is marked by a black bar, and whiskers indicate the width of the reference corridor as defined in the Extended Experimental Procedures (i.e., value of the most extreme data point that is no more than 1.5 times the interquartile range from the box if the distance from the median exceeds a minimum threshold of 0.2 for DNA methylation and 1 for gene expression; otherwise these thresholds—which correspond to 20 percentage points for DNA methylation and a 2-fold change for gene expression—define the reference corridor). Data points that fall outside the whiskers are flagged as outliers and are suppressed in this figure; their position relative to the reference corridor is shown in Figure 4A.
genes exhibit little deviation from the reference in any of the ES cell lines (Figure 2A). Specifically, 13% of genes account for half of the total DNA methylation variation, and 20% of genes account for half of the total gene expression variation (Table S3). As one might have expected, housekeeping genes such as GAPDH were among the least variable genes between the cell lines. Similarly, we observed relatively low variation
among several genes that are highly expressed in pluripotent cell lines, including SOX2 and DNMT3B. In contrast, moderate to high levels of variation were found for several genes that regulate embryonic development and are induced upon ES cell differentiation, including GATA6, LEFTY2, and PAX6. Finally, a small number of loci exhibited highly variable DNA methylation levels between cell lines, ranging from close to 0% methylation in Cell 144, 439–452, February 4, 2011 ª2011 Elsevier Inc. 441
3500
MEG3
800 0
PAX6
0
500
1000
1500
Normal scaling of x-axis◄ ► 5-fold compressed scaling of x-axis
CAT CXCL5 GATA6 LEFTY2
600
Deviation > 0.05: 9 9% of genes 9.9%
GAPDH DNMT3B DAZL CD14 SNAI1 TF SOX2 S100A6 BMP4
400
2000
S100A6
Frequency of occurrence e
BMP4
CD14 CAT TF
200
CXCL5 LEFTY2 MEG3
Deviation < 0.05: 90 1% of genes 90.1%
2500
3000
PAX6 GAPDH SOX2 DNMT3B GATA6 DAZL SNAI1
0
Frequency of occurrence
Gene expression
DNA methylation
A
0.00
0.01
0.02
0.03
0.04
0.05
0.10
0.15
0.20
0.25
0.0
0.30
Mean absolute deviation from the ES-cell reference across all ES cell lines
DNA methylation
B
Top-1000 most variable genes
1.0
1.5
2.0
C
All genes with sufficient data
DNA methylation
4.2% 51.3%
0.5
Mean absolute deviation from the ES-cell reference across all ES cell lines
Gene expression
889
Top-1000 most variable genes (autosomes only)
111
889
Top-1000 most variable genes (autosomes only)
48.7% 95.8% 2.8-fold enrichment of co-localization over random chance (p