Fatty Acids and Lipotoxicity in Obesity and Diabetes: Novartis Foundation Symposium 286, Volume 286. Edited by Gregory Bock and Jamie Goode Copyright Novartis Foundation 2007. ISBN: 978-0-470-05764-3
FATTY ACIDS AND LIPOTOXICITY IN OBESITY AND DIABETES
The Novartis Foundation is an international scientific and educational charity (UK Registered Charity No. 313574). Known until September 1997 as the Ciba Foundation, it was established in 1947 by the CIBA company of Basle, which merged with Sandoz in 1996, to form Novartis. The Foundation operates independently in London under English trust law. It was formally opened on 22 June 1949. The Foundation promotes the study and general knowledge of science and in particular encourages international co-operation in scientific research. To this end, it organizes internationally acclaimed meetings (typically eight symposia and allied open meetings and 15–20 discussion meetings each year) and publishes eight books per year featuring the presented papers and discussions from the symposia. Although primarily an operational rather than a grant-making foundation, it awards bursaries to young scientists to attend the symposia and afterwards work with one of the other participants. The Foundation’s headquarters at 41 Portland Place, London W1B 1BN, provide library facilities, open to graduates in science and allied disciplines. Media relations are fostered by regular press conferences and by articles prepared by the Foundation’s Science Writer in Residence. The Foundation offers accommodation and meeting facilities to visiting scientists and their societies. Information on all Foundation activities can be found at http://www.novartisfound.org.uk As one of China’s most renowned universities, Tsinghua is an important center of learning in China for fostering talent and scientific research. The centre area of the campus of Tsinghua University is situated on former imperial gardens of the Qing Dynasty. Tsinghua University was founded in 1911, originally under the name of Tsinghua Xuetang, as a preparatory school for students who would be sent by the government to study in the United States. It was renamed Tsinghua School in 1912. The university section was founded in 1925. The name National Tsinghua University was adopted in 1928 and in 1929 the Tsinghua Research Institute was set up. The Resistance War against the Japanese Invasion in 1937 forced Tsinghua to move to Kunming, where it formed, with Peking University and Nankai University, Southwest Associated University. In 1946, the University was moved back to its original location in Peking. In 1952, Tsinghua University became a polytechnic institution with an engineering focus known as China’s ‘cradle of engineers’. Since 1978, with the resumption of departments in science, economics, management and the humanities, Tsinghua University has flourished in many fields besides engineering. Tsinghua is rapidly developing into a comprehensive research university with faculties in science, engineering, humanities, law, medicine, history, philosophy, economics, management, education, and art.
Novartis Foundation Symposium 286
FATTY ACIDS AND LIPOTOXICITY IN OBESITY AND DIABETES
Copyright © Novartis Foundation 2007 Published in 2007 by John Wiley & Sons Ltd, The Atrium, Southern Gate, Chichester PO19 8SQ, UK National 01243 779777 International (+44) 1243 779777 e-mail (for orders and customer service enquiries):
[email protected] Visit our Home Page on http://www.wileyeurope.com or http://www.wiley.com All Rights Reserved. No part of this book may be reproduced, stored in a retrieval system or transmitted in any form or by any means, electronic, mechanical, photocopying, recording, scanning or otherwise, except under the terms of the Copyright, Designs and Patents Act 1988 or under the terms of a licence issued by the Copyright Licensing Agency Ltd, 90 Tottenham Court Road, London W1T 4LP, UK, without the permission in writing of the Publisher. Requests to the Publisher should be addressed to the Permissions Department, John Wiley & Sons Ltd, The Atrium, Southern Gate, Chichester, West Sussex PO19 8SQ, England, or emailed to permreq@ wiley.co.uk, or faxed to (+44) 1243 770620. This publication is designed to provide accurate and authoritative information in regard to the subject matter covered. It is sold on the understanding that the Publisher is not engaged in rendering professional services. If professional advice or other expert assistance is required, the services of a competent professional should be sought. Other Wiley Editorial Offices John Wiley & Sons Inc., 111 River Street, Hoboken, NJ 07030, USA Jossey-Bass, 989 Market Street, San Francisco, CA 94103-1741, USA Wiley-VCH Verlag GmbH, Boschstr. 12, D-69469 Weinheim, Germany John Wiley & Sons Australia Ltd, 33 Park Road, Milton, Queensland 4064, Australia John Wiley & Sons (Asia) Pte Ltd, 2 Clementi Loop #02-01, Jin Xing Distripark, Singapore 129809 John Wiley & Sons Canada Ltd, 6045 Freemont Blvd, Mississauga, Ontario, Canada L5R 4J3 Wiley also publishes its books in a variety of electronic formats. Some content that appears in print may not be available in electronic books. Novartis Foundation Symposium 286 x + 212 pages, 34 figures, 2 tables
Anniversary Logo Design: Richard J Pacifico British Library Cataloguing in Publication Data A catalogue record for this book is available from the British Library ISBN 978-0-470-05764-3 Typeset in 10.5 on 12.5 pt Garamond by SNP Best-set Typesetter Ltd., Hong Kong Printed and bound in Great Britain by T. J. International Ltd, Padstow, Cornwall. This book is printed on acid-free paper responsibly manufactured from sustainable forestry, in which at least two trees are planted for each one used for paper production.
Contents Symposium on Fatty acids and lipotoxicity in obesity and diabetes, held at Tsinghua University, Beijing, China, 17–19 October 2006 Editors: Gregory Bock (Organizer) and Jamie Goode This symposium is based on a proposal made by Peng Li Bruce M. Spiegelman
Chair’s introduction
1
Bruce M. Spiegelman Transcriptional control of energy homeostasis through the PGC1 coactivators 3 Discussion 6 Stephen O’Rahilly Human obesity and insulin resistance: lessons from experiments of nature 13 Discussion 20 Deborah M. Muoio and Timothy R. Koves dysfunction in skeletal muscle 24 Discussion 38
Lipid-induced metabolic
Alan D. Attie, Matthew T. Flowers, Jessica B. Flowers, Albert K. Groen, Folkert Kuipers and James M. Ntambi Stearoyl-CoA desaturase deficiency, hypercholesterolaemia, cholestasis and diabetes 47 Discussion 54 Karen Reue The role of lipin 1 in adipogenesis and lipid metabolism Discussion 68
58
David Carling The role of the AMP-activated protein kinase in the regulation of energy homeostasis 72 Discussion 81 Gökhan S. Hotamisligil Endoplasmic reticulum stress and inflammation in obesity and type 2 diabetes 86 Discussion 94 v
vi
CONTENTS
Ira Tabas, Tracie Seimon, Jerry Arellano, Yankun Li, Fabien Forcheron, Dongying Cui, Seongah Han, Chien-Ping Liang, Alan Tall and Domenico Accili The impact of insulin resistance on macrophage death pathways in advanced atherosclerosis 99 Discussion 109 Sandra Lobo and David A. Bernlohr Fatty acid transport in adipocytes and the development of insulin resistance 113 Discussion 121 Paul N. Black and Concetta C. DiRusso Vectorial acylation: linking fatty acid transport and activation to metabolic trafficking 127 Discussion 138 Günther Daum, Andrea Wagner, Tibor Czabany, Karlheinz Grillitsch and Karin Athenstaedt Lipid storage and mobilization pathways in yeast 142 Discussion 151 John Zhong Li and Peng Li obesity 155 Discussion 159
Cide proteins and the development of
General discussion I 162 Visualizing brown adipose tissue with FDG-PET 162 Takashi Kadowaki, Toshimasa Yamauchi, Naoto Kubota, Kazuo Hara and Kohjiro Ueki Adiponectin and adiponectin receptors in obesity-linked insulin resistance 164 Discussion 176 Gabriel Pascual, Amy L. Sullivan, Sumito Ogawa, Amir Gamliel, Valentina Perissi, Michael G. Rosenfeld and Christopher K. Glass Anti-inflammatory and antidiabetic roles of PPARγ 183 Discussion 196 Final discussion 200 Nutrition, ageing and lipotoxicity 200 Index of contributors Subject index 206
204
Participants
Alan D. Attie Department of Biochemistry, University of Wisconsin-Madison, 433 Babcock Drive, Madison, WI 53706-1544, USA David A. Bernlohr University of Minnesota, Department of Biochemistry, Molecular Biology and Biophysics, 6-155 Jackson Hall, 321 Church Street, Minneapolis, MN 55455, USA Paul N. Black Ordway Research Institute, Inc., 150 New Scotland Avenue, Albany, NY 12208, USA David Carling Cellular Stress Group, MRC Clinical Sciences Centre, Imperial College, Hammersmith Hospital Campus, Du Cane Road, London W12 0NN, UK Günther Daum Institut für Biochemie, Technische Universität Graz, Petersgasse 12/2, A-8010 Graz, Austria Diego De Mendoza Institute of Molecular and Cellular Biology of Rosario, CONICET, Universidad Nacional de Rosario, Suipacha 531, 2000-Rosario, Argentina Christopher K. Glass University of California San Diego, Department of Cellular and Molecular Medicine, 9500 Gilman Drive, GPL Rm 217, La Jolla, CA 92093-0651, USA Xiao Han The Key Laboratory of Human Functional Genomics of Jiangsu Province, Nanjing Medical University, Nanjing 210029, People’s Republic of China Gökhan S. Hotamisligil Harvard School of Public Health, Department of Genetics and Complex Diseases, Building I, Room 205, 655 Huntington Avenue, Boston, MA 02115, USA Takashi Kadowaki Department of Metabolic Diseases, Graduate School of Medicine, University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo 113-8655, Japan vii
viii
PARTICIPANTS
Jae Bum Kim Department of Biological Sciences, Seoul National University, Bldg. 20-Rm. 109, San 56-1, Sillim-Dong, Kwanak-Gu, Seoul 151-742, Korea Peng Li Department of Biological Sciences and Biotechnology, Tsinghua University, Beijing, People’s Republic of China Xiaoying Li 197 Ruijin 2nd Road, Department of Endocrinology, Ruijin Hospital, Shanghai 200025, People’s Republic of China Kan Liao Chinese Academy of Sciences, Shanghai Institute of Biochemistry and Cell Biology, 320 Yueyang Rd., Shanghai 200031, People’s Republic of China Deborah M. Muoio Sarah W. Stedman Nutrition and Metabolism Center, and Department of Pharmacology and Cancer Biology, Duke University Medical Center, Independence Park, Durham, NC 27704, USA Yi-Ming Mu Department of Endocrinology, Chinese PLA General Hospital, Fu Xing Road 28, HaiDian District, Beijing 100853, China Stephen O’Rahilly University of Cambridge, Department of Clinical Biochemistry, Level 4, Laboratory Block, Box 232, Addenbrooke’s Hospital, Hills Road, Cambridge CB2 2QR, UK Karen Reue Department of Human Genetics, Gonda Center 6506A, David Geffen School of Medicine at UCLA, 695 Charles E. Young Drive South, Los Angeles, CA 90095, USA Matthew Sabin (Novartis Foundation Bursar) Department of Endocrinology and Diabetes, Royal Children’s Hospital, Parkville 3052, Victoria, Australia Masayuki Saito Department of Nutrition, Graduate School of Nursing and Nutrition, Tenshi College, Sapporo 065-0013, Japan Yuguang Shi Department of Cellular and Molecular Physiology, Penn State University School of Medicine, Hershey, PA 17033, USA Bruce M. Spiegelman (Chair) Dana-Farber Cancer Institute, 44 Binney Street, Boston, MA 02115, USA Ira Tabas Columbia University, Department of Medicine, PH 8 East 105F, Division of Molecular Medicine, 630 West 168th Street, New York, NY 10032, USA
PARTICIPANTS
ix
Robert Yang School of Biotechnology and Molecular Biosciences, University of New South Wales, Room 206, Biological Sciences Building, Sydney 2052, Australia Chen-Yu Zhang School of Life Sciences, Nanjing University, 22 Hankou Road, Nanjing, Jiangsu 210093, People’s Republic of China
Fatty Acids and Lipotoxicity in Obesity and Diabetes: Novartis Foundation Symposium 286, Volume 286. Edited by Gregory Bock and Jamie Goode Copyright Novartis Foundation 2007. ISBN: 978-0-470-05764-3
Chair’s introduction Bruce M. Spiegelman Dana-Farber Cancer Institute and the Department of Cell Biolog y, Harvard Medical School, Boston, MA 02115, USA
I’d like to give a brief introduction to this meeting before moving to my own presentation. The context in which this meeting takes place is in the epidemic of obesity that is occurring throughout most of the world. Over the last 50 years we have seen a great increase in obesity and concomitant metabolic diseases. When I first took an interest in this area some 20 years ago, I remember discussing the epidemic of obesity in the USA. Then, shortly after, I was discussing the epidemic of obesity in the USA and other industrial and Western nations. Now, sadly, we talk about the epidemic of obesity in the whole world. China, India, Japan and Korea, for example, are also seeing an increase in obesity. There are different ways of thinking about the problem. One is the nature of the development of obesity. What is changing? Presumably human genetics haven’t changed much over the last 50 years. Furthermore, even if we put that issue aside, why does obesity lead to metabolic disease and the morbidities that follow from it? We understand that obesity is a central feature in a cluster of morbidities— hypertension, insulin resistance, atherogenic dyslipidaemia—that tend to go together. My clinical colleagues here probably have strong feelings about the extent to which these are linked together in an obligatory fashion or not. But certainly these things tend to go together, and they link to type 2 diabetes, cardiovascular disease and cancer. I work at a cancer institute, so I am particularly interested in the last point. The epidemic of obesity is leading to a great increase in human cancer, particularly of colon, prostate and breast. This has become apparent only in the last few years. Right now, the American Cancer Society considers obesity to be the second largest cause of preventable cancer in human beings, second only to cigarette smoking. Within the next 10 years obesity is expected to pass cigarette smoking as an avoidable cause of human malignancy. As therapies in cancer are improving, this benefit is being diluted by the increase in common cancers due to obesity. If type 2 diabetes and cardiovascular disease were not a frightening enough reason to try to get a grip on this problem, it is clear that it is also driving an increase in cancers. 1
2
SPIEGELMAN
There is a cluster of things that occur commonly in obesity. These include insulin resistance, lipid accumulation (so-called lipotoxicity in muscle, liver, pancreatic islets and presumably other tissues as well) and mitochondrial dysfunction linked to reduced expression of PGC1 coactivators. One of the issues we will try to address in this meeting is cause and effect in the relationship of these different components. Are these in a pathway? Are they independent variables? How do they impact on one another? Where do inflammation, ER stress and reactive oxygen species enter into this? The participants here were invited because they are among the world’s experts at helping to deconvolute these variables, or at least will be able to tell us how this deconvolution is progressing. Ultimately, we want to know what we can do about this, now and in the future. What do these pathways tell us about interventions such as lifestyle, exercise, diet and drugs? We certainly have no objection to trying to get a grip on obesity itself. I have highlighted the downstream sequelae of the metabolic syndrome, but are there approaches to the obesity problem itself? This is the focus of the meeting: trying to understand the interplay in these pathways, and then asking whether we have learned anything that will allow us some new thoughts in terms of interventions.
Fatty Acids and Lipotoxicity in Obesity and Diabetes: Novartis Foundation Symposium 286, Volume 286. Edited by Gregory Bock and Jamie Goode Copyright Novartis Foundation 2007. ISBN: 978-0-470-05764-3
Transcriptional control of energy homeostasis through the PGC1 coactivators Bruce M. Spiegelman Dana-Farber Cancer Institute and the Department of Cell Biolog y, Harvard Medical School, Boston, MA 02115, USA
Abstract. The PGC1 transcriptional coactivators are major regulators of several crucial aspects of energy metabolism. PGC1α controls many aspects of oxidative metabolism, including mitochondrial biogenesis and respiration through the coactivation of many nuclear receptors, and factors outside the nuclear receptor family. ERR α , NRF1 and NRF2 are key targets of the PGC1s in mitochondrial biogenesis. We have recently addressed the question of the role of PGC1 coactivators in the metabolism of reactive oxygen species (ROS). We now show that PGC1α and β are induced when cells are given an oxidative stressor, H 2O2 . In fact, experiments with RNAi for the PGC1s show that the ability of ROS to induce a ROS scavenging programme depends entirely on the PGC1s. This includes genes encoding mitochondrial proteins like SOD2, but also includes cytoplasmic proteins like catalase and GPX1. Cells lacking PGC1α are hypersensitive to death from oxidative stress caused by H 2O2 or paraquat. Mice deficient in PGC1α get excessive neurodegeneration when given kainic acid-induced seizures or MPTP, which causes Parkinsonism. These data show that the PGC1s are important protective molecules against ROS generation and damage. The implications of this for diabetes and neurodegenerative diseases will be discussed. 2007 Fatty acids and lipotoxicity in obesity and diabetes. Wiley, Chichester (Novartis Foundation Symposium 286) p 3–12
Oxidative metabolism is crucial for most living systems, and the majority of oxidative metabolism in eukaryotic cells occurs in the mitochondria. Mitochondria generate ATP through the function of the electron transport system, whereby the passage of high energy electrons down this chain is coupled to the extrusion of protons across the inner membrane of the mitochondria. This proton gradient can be dissipated by passage through complex V of the electron transport chain, which couples this proton movement to the phosphorylation of ADP to ATP. Recent data have implicated mitochondrial dysfunction in a large number of important human diseases, including neurodegeneration, heart failure and 3
4
SPIEGELMAN
diabetes. The skeletal muscle of humans with type 2 diabetes, glucose intolerance or a family history of diabetes all have a reduced expression of multiple genes of the mitochondrial oxidative phosphorylation (OXPHOS) system, and their dominant regulators, the peroxisome proliferator-activated receptor γ (PPARγ ) coactivator 1 (PGC1) coactivators (Mootha et al 2003, Patti et al 2003). In this paper, I review recent data related to the role of PGC1s, especially PGC1α , in energy metabolism related to diabetes and neurodegeneration. PGC1α was discovered as a binding partner and coactivator of PPARγ in brown fat (Puigserver et al 1998). It is induced in this tissue by exposure of animals to cold, a condition which activates the thermogenic function of brown fat tissue. Functional studies in our group and others showed that PGC1α increased mitochondrial OXPHOS gene expression and the expression of UCP1 when expressed in white fat cells in culture or in vivo (Puigserver et al 1998, Tiraby et al 2003). Detailed analysis of the effects of PGC1α on mitochondria indicated that it could increase the expression of a wide variety of mitochondrial genes, whether encoded in the nuclear or mitochondrial genomes (Wu et al 1999). This ability to activate a broad mitochondrial programme results, in large measure, from the ability of PGC1α to coactivate ERRα , NRF1 and NRF2 (Handschin et al 2003, Wu et al 1999, Mootha et al 2004, Schreiber et al 2004). The ability of PGC1α , and its closest homologue PGC1β, to induce respiration in muscle cells was examined (St. Pierre et al 2003). Both coactivators increase respiration greatly, as they induce mitochondrial biogenesis. However, PGC1α increases the fraction of uncoupled respiration compared to controls, wheras PGC1β -induced respiration has the same relative proportions of coupled and uncoupled respiration as cells expressing a GFP control. Regarding the role of PGC1α in skeletal muscle, the gene sets activated are not restricted to mitochondria. Transgenic expression of PGC1α stimulates a broad programme of fibre-type switching from type IIb fibres to type IIa and type I. These oxidative fibres include more mitochondria but also include myosin heavy chain (MHC) type IIa, I and myoglobin. These data are likely to be highly relevant from a physiological perspective because PGC1α is expressed at highest levels in soleus muscle, which is very rich in type I fibres (Lin et al 2002) and is induced in rodents and humans by exercise (reviewed in Handschin & Spiegelman 2006). PGC1s and disease Our recent studies have focused on the potential role of PGC1α in the context of tissue degeneration and wasting, especially skeletal muscle and the brain. Since PGC1α mediates many of the effects of motor nerves on skeletal muscle relating to mitochondria and fibre-type switching, we have asked whether PGC1α might mediate a key function of motor nerve activity: the suppression of muscle
PGC1 COACTIVATORS
5
atrophy. Indeed, if skeletal muscle is denervated, skeletal muscle loses mass coincident with shrinkage in the diameter of muscle fibres. This is also seen in rodents and humans if limbs suffer disuse. This loss of muscle mass is a catabolic process and is associated with induction of a set of genes termed ‘atrogenes’ that include E-3 ubiquitin ligases such as atrogin and MURF (Lecker et al 2004). We were able to show that dissection of the sciatic nerve causes a loss of muscle fibre diameter of 50% within 12 days in control mice. However, transgenic expression of PGC1α causes an almost complete suppression of muscle atrophy, as well as a significant reduction of the induction of the set of atrogenes (Sandri et al 2006). This suppression of the atrogenes by PGC1α is likely to derive, at least in part, from a suppression of FOXO3 action by PGC1α . Many forms of neurodegeneration have been associated with mitochondrial dysfunction and increased oxidative damage (Lin & Beal 2006). Given that mice mutated in PGC1α have a severe neurodegeneration in the striatum (Lin et al 2004), we recently studied the role of PGC1 coactivators and the metabolism of reactive oxygen species (ROS). PGC1α and PGC1β mRNA are co-induced in cells treated with H 2O2 with the gene sets of protection from ROS, including SOD1 and SOD2, catalase, GPX1, UCP1 and UCP3. Studies with RNAi directed against PGC1α show that this coactivator is required to get full expression of the antioxidant programme (St. Pierre et al 2006). Similar results were obtained with cells mutated in PGC1α . The induction of PGC1α by an oxidative stressor was stimulated, at least in part, by the increased binding of phosphorylated CREB to the PGC1α promoter. The neuroprotective and anti-ROS effects of endogenous PGC1α could be illustrated by treating the PGC1α knockout mice with agents that induce oxidative stress and neurodegeneration. MPTP (1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine) and kainic acid induced much more degeneration in the dopaminergic centres of the substantia nigra, and hippocampus, respectively. These degenerative processes were associated with greater levels of stable markers of oxidative damage, such as nitrotyrosylation of proteins and 8-OXO-guanine in DNA. Gain of function studies in cultured cells show that elevation of PGC1α levels above those of wild-type nerve cells give an increased resistance to death by the oxidative stressors H2O2 and paraquat. Future studies will involve finding known drugs or chemical compounds that can elevate PGC1α in many tissues, especially brain and skeletal muscle. These will then be tested in models of neurodegeneration, muscle wasting and muscle dystrophies, and type 2 diabetes. References Handschin C, Spiegelman BM 2006 PGC-1 coactivators, energy homeostasis, and metabolism. Endocr Rev 27:728–735
6
SPIEGELMAN
Handschin C, Rhee J, Lin J, Tarr PT, Spiegelman BM 2003 An autoregulatory loop controls peroxisome prolifrator-activated receptor γ coactivator 1α expression in muscle. Proc Natl Acad Sci USA 100:7111–7116 Lecker SH, Jagoe RT, Gilbert A et al 2004 Multiple types of skeletal muscle atrophy involve a common program of changes in gene expression. FASEB J 18:39–51 Lin J, Tarr P, Puigserver P et al 2002 Transcriptional coactivator PGC-1alpha drives the expression of slow-twitch muscle fibres. Nature 418:797–801 Lin J, Wu PH, Tarr PT et al 2004 Defects in adaptive energy metabolism with CNS-linked hyperactivity in PGC-1alpha null mice. Cell 119:121–135 Lin MT, Beal MF 2006 Mitochondrial dysfunction and oxidative stress in neurodegenerative diseases. Nature 443:787–795 Mootha VK, Lindgren CM, Eriksson KF et al 2003 PGC-1alpha-responsive genes involved in oxidative phosphorylation are coordinately downregulated in human diabetes. Nat Genet 34:267–273 Mootha VK, Handschin C, Arlow D et al 2004 Errα and Gabpa/β specify PGC-1α -dependent OXPHOS gene expression that is altered in diabetic muscle. Proc Natl Acad Sci USA 101:6570–6575 Patti ME, Butte AJ, Crunkhorn S et al 2003 Coordinated reduction of genes of oxidative metabolism in humans with insulin resistance and diabetes: Potential role of PGC1 and NRF1. Proc Natl Acad Sci USA 100:8466–8471 Puigserver P, Wu Z, Park CW, Graves R, Wright M, Spiegelman BM 1998 A cold inducible coactivator of nuclear receptors linked to adaptive thermogenesis. Cell 92:829–839 St. Pierre J, Lin J, Krauss S et al 2003 Bioenergetic analysis of peroxisome proliferator-activated receptor γ coactivators 1α and 1β (PGC-1α and PGC-1β ) in muscle cells. J Biol Chem 278:26597–26603 St. Pierre J, Drori D, Uldry M et al 2006 Suppression of reactive oxygen species and neurodegeneration by the PGC-1 transcriptional coactivators. Cell 127:397–402 Sandri M, Lin J, Handschin C et al 2006 PGC-1alpha protects skeletal muscle from atrophy by suppressing Fox03 action and atrophy-specific gene transcription. Proc Natl Acad Sci USA 103:16260–16265 Schreiber SN, Emter R, Hock MB et al 2004 The estrogen-related receptor alpha (ERRalpha) functions in PPARgamma coactivator 1alpha (PGC-1alpha)-induced mitochondrial biogenesis. Proc Natl Acad Sci USA 101:6472–6477 Tiraby C, Tavernier G, Lefort C et al 2003 Acquirement of brown fat cell features by human white adipocytes. J Biol Chem 278:33370–33376 Wu Z, Puigserver P, Andersson U et al 1999 Mechanisms controlling mitochondrial biogenesis and respiration through the thermogenic coactivator PGC-1. Cell 98:115–124
DISCUSSION Attie: I’m interested in the islet results in PGC1α mutant mice. You said that there is defective insulin secretion. Which secretagogues is this in response to? Spiegelman: Glucose at this point. These are reproducible data, but we haven’t broadened this work out yet. Attie: When you isolate the islets do they all look the same? Spiegelman: I gather they all look abnormal. Attie: What is the insulin content of those islets?
PGC1 COACTIVATORS
7
Spiegelman: I believe this is reduced. They are big and lousy, with reduced insulin content. Bernlohr: Could you say a bit about peroxisomal formation in the muscle knockout? Spiegelman: I don’t know anything about it. Bernlohr: You could argue that it may mount some compensatory response to increased peroxisomal oxidation. Is PGC1α involved in the transcription of peroxisomal components? Spiegelman: That is a good question, because it generates ROS. I haven’t done this. Perhaps the peroxisomes are trying to make up for the mitochondrial deficiency, and this is then turning into a ROS generating machine. Bernlohr: I don’t know whether those regulatory circuits are coupled or potentially uncoupled and overlapping. Shi: The knockout mice have brain dysfunction. How about their islet function? Spiegelman: It looks quite normal. Shi: What is your explanation for this? I would expect a defect in islet function, since islet β cells share many common features with the brain and mitochondrial oxidative phosphorylation plays an important role in regulating glucose-stimulated insulin secretion, i.e. the second phase insulin secretion. Here you have a musclespecific knockout showing a different phenotype to that of the whole body knockout. It would be interesting to investigate the phenotypes of islet-specific knockout mice. Spiegelman: We are seeing several phenomena like this. The total knockout has some things going on systemically that the muscle knockout doesn’t. The total knockout has constitutive activation of AMP kinase in the muscle. The musclespecific knockout does not. There is something about the combination of hyperactivity with bad mitochondria. We get this tremendous activation of AMP kinase in muscle, but the muscle-specific knockouts are slightly hypo-active, with somewhat reduced respiration and no activation of AMP kinase. I think the musclespecific knockout is really the naked PGC1 phenotype as far as muscle goes. The total knockout is fascinating but there are so many things going on in this animal it is very hard to work it all out. When we do tissue-specific knockouts we almost always see something different from the total knockout. Shi: What happens to the body weight in the muscle-specific knockout mice? Spiegelman: It tends to be slightly low. There is almost a sense of cachexia. The data I showed you are exciting, but even the muscle-specific knockout is complicated. Personally, I am very interested in the heterozygotes, where we see glucose intolerance. In the total muscle knockout there is fibre-type switching and several things going on. This isn’t what human patients with glucose intolerance have, and there is still a lot going on in those mice. The heterozygotes have no fibre type
8
SPIEGELMAN
switching, mild deficiency in mitochondrial biology, and we still see some glucose intolerance. This is the model I like for the human disease. Kim: What about the PGC1 function and ROS in several different tissues? On the basis of your data it seems that ROS plays a role in inducing PGC1, but PGC1 has several different functions from tissue to tissue. Do you think that PGC1 has a role in ROS scavenging in every tissue? Spiegelman: In the total knockout of PGC1, there is partial deficiency of the antiROS enzymes in all tissues that we have examined. It is not really a muscle or brain specific thing. In general SOD1, SOD2, GPX and catalases all tend to be expressed at 60–80% of what they would be in wild-type, a bit like the mitochondrial genes. And at least as far as we have looked, their inducibility is lost. More than the outright deficiency of the anti-ROS, the bad effect of PGC1α loss is the loss of responsiveness of these enzymes to ROS. These are adaptive systems in mammals. Glass: I have a question relating to a theme that will come up later in the meeting: that of the relationship between inflammation and insulin resistance. What are your thoughts on the impact of inflammation on the function of PGC1α and its peers? We think of inflammatory stimuli affecting the phosphorylation of IRS1, and how this could have an impact on downstream insulin signalling, but it would also appear that inflammation could affect the function of PGC1. What are your thoughts on this? Spiegelman: That’s a very good point. In fact, we see inflammatory changes with deficiency in PGC1. In the total knockout all the inflammatory markers are up in muscle. In fact, we even wonder whether the weird islets in the knockout might be because TNF, IL1 and IL6 are going up. I don’t know whether this happens in the heterozygote yet. Glass: That is putting inflammation downstream of PGC1. I wasn’t even thinking of this: I was considering inflammation affecting PGC1 function. Spiegelman: We keep seeing these cycles with ROS both upstream and downstream. Some of the transcription factors are both upstream and downstream. Everything we see in this system looks like a cycle: almost everything downstream is also upstream. I know you have been doing some work with PGC1s and inflammation as well. Glass: Yes. In the macrophage, if we remove PGC1 and look at gene expression, many inflammatory mediators are up-regulated. This would be consistent with what you are finding. Spiegelman: A simple model that we could tie together would be to say that in the muscle-specific knockout there is something inflammatory going on. Something is getting to the β cell—inflammation, or lipotoxicity or some combination thereof. It would be exciting if the PGC1 defect in muscle can play out and affect the biology of the islets. Of course, in the diabetes field there are those who concen-
PGC1 COACTIVATORS
9
trate on insulin resistance and those who focus on islets. There hasn’t been that much tying things together. It would be great if PGC1 in the mitochondrial pathway in muscle really does have a way to talk to the islets. Diabetes has to involve defects in both. O’Rahilly: The strongest data from those papers (Patti et al 2001, Mootha et al 2003) were Mary Elizabeth Patti’s in which they specifically chose insulin-resistant individuals. I’d like to press you a bit on this link between PGC1 deficiency, mitochondrial function and insulin resistance. There seems to be an idea that it is not quite as linear a pathway as some have thought. There are also a couple of critical papers that don’t tend to be quoted. Firstly, humans with inherited mitochondrial diabetes develop severe β cell dysfunction but their insulin sensitivity in muscle is very well preserved (Maassen et al 2004). Secondly there is a body of evidence emerging from transgenic mice suggesting that primary disorders of mitochondrial oxidative phosphorylation in skeletal muscle do not result in muscle or whole body insulin resistance (see Wredenberg et al 2006). Spiegelman: But I have to say that what for you and many people in the field is a mitochondrial/insulin resistance hypothesis, we call the PGC1 hypothesis. It could just be that the PGC1s have so many different functions, the mitochondrial deficiency was really a surrogate. O’Rahilly: Out there in the broad community there is a very simple linear view. Spiegelman: I don’t assume that the defects we are seeing are due to the deficiency in mitochondria. They could be. We are not against that idea, but we know that there is a lot more going on. Inflammatory markers are going up. Lipid oxidation is going to be affected. It is a linked pathway, and deconvoluting these things is not trivial. But I don’t assume that it is the mitochondria. Hotamisligil: A little bit of hydrogen peroxide is needed to signal to most of the tyrosine kinase receptors. Could this be the reason why that if you regulate the production of ROS robustly, you might compromise what you would otherwise see in insulin action? Spiegelman: It is possible: this system has tone, but it doesn’t have to be set at zero ROS. Clearly, having too much ROS is bad, but as you say there are papers showing the beneficial effects of certain ROS. The tone in this homeostatic system isn’t zero or infinity, but it should be set at some appropriate level. Shi: It is generally believed that an increase in mitochondrial activity will generate ROS. In your case you have PGC1 doing the opposite. Spiegelman: This is a common misconception, and I am glad you asked this. It is not true that all mitochondrial activity increases ROS. I know people say it, but it isn’t true. What is true is that most ROS come from mitochondria. But there are separate variables that tell a mitochondrion to make more or less ROS. Probably the single biggest variable is membrane potential. If it gets too high mitochondria
10
SPIEGELMAN
will generate ROS. The real-life variable that usually controls this is ATP turnover and the rate of its utilization. When there is a mismatch between mitochondrial electron transport, and ATP production and utilization, this then sets up a situation for ROS generation. Shi: In your cell line or cell based assay, have you ever measured ATP production in response to PGC1 overexpression or down-regulation? Spiegelman: Yes, many times. Cells that are deficient in PGC1α can’t defend their ATP levels well at all. Shi: Do you understand the overall picture of your knockout? I didn’t see any data on food intake or metabolic rates. Spiegelman: The knockout mice eat normally but are lean because of increased energy expenditure. The muscle-specific knockout eats a little less and is a bit smaller and leaner, but basically it has a tendency towards lower metabolic rates. The total knockout has a tendency towards higher metabolic rates in large part because they have this Huntington-like thing going on and they move more. Hotamisligil: So is that physical activity chorea? Spiegelman: In the Huntington field they do a clasping test. Pick a mouse up by the tail and the Huntington mice will do clasping. Our mice do this. They are hyper-responsive and jumpy. People in our lab don’t need to genotype the mice because they are so obvious. Attie: One metabolite that has dramatic effects on insulin sensitivity, food intake and β cell function that is often overlooked is β -hydroxybutyrate. I would predict that your mice have a defect in ketone body oxidation. I noticed that your GTP was near normal on the chow diet, but abnormal in the high fat diet. Have you measured ketones? Spiegelman: That is a good point. I don’t know. We are always looking for plausible hypotheses. Lately we’ve been more interested in the inflammatory idea, but what you say needs consideration. P Li: Do you see this neural degeneracy in the striatum in other cell types? Spiegelman: In the total knockout there are three tissues that are morphologically abnormal. The brain (striatum in particular), brown fat (it looks like bad brown fat with a deficient thermogenic programme) and there is a tendency towards hepatic steatosis. Dan Kelly sees out-and-out hepatic steatosis in his knockout; we have seen a trend towards this. But morphologically the mice look quite good. They seem to be living fairly normal lives. Muoio: When you showed data indicating an interplay between β cell function and PGC1α expression in muscle, the first candidate mediator that came to mind was IL6, which is known to be robustly up-regulated in muscles that are experiencing energy stress (Pedersen et al 2004). Spiegelman: Is it known to have effects on islets? Muoio: In vitro experiments have shown quite convincingly that IL6 and other cytokines can impair β cell function (Zhao et al 2006, Hohmeier et al 2003). In
PGC1 COACTIVATORS
11
skeletal muscle, IL6 expression and secretion is induced by exercise. There are several reports suggesting that IL6 functions as a ‘myokine’ that communicates changes in muscle energy status, and in turn regulates metabolic programmes such as lipolysis and gluconeogenesis in adipose tissue and liver, respectively (Pedersen et al 2004). You commented that ablation of PGC1α in muscle does not lower mitochondrial number, but have you considered or examined changes in mitochondrial properties? Do you have any evidence that energy metabolism is altered in the knockout animals? Spiegelman: We have taken cells in culture and done this analysis carefully. They are struggling to maintain their ATP homeostasis. In vivo we have looked in the heart by NMR, but not in skeletal muscle. We published a paper in 2005 in which we looked at the heart in a Langendorf preparation hanging in an NMR magnet (Arany et al 2005). We showed an inability to defend ATP in real time. Muoio: What is the impact on ATP-generating pathways such as glycolysis, glucose oxidation and β oxidation? Spiegelman: In culture, glycolysis is way up. We have submitted a paper on ATP homeostasis and the cells are trying to raise glycolysis. The near normal ATP homeostasis at basal levels is at the expense of highly elevated glycolysis. Hotamisligil: Endoplasmic reticulum (ER) is also a major source for ROS. Some even propose that it might be as important as mitochondria. The two might be physically and functionally connected organelles. There is physiological ROS production, for example, during disulfide bond formation and breakdown. It could become pathological when the ER is under stress. With the neural degeneration phenotype, the muscle phenotype and the total knockout phenotype I can’t help but think there might be signs of ER stress. Spiegelman: This is a subject I don’t know a great deal about. The idea strikes me as reasonable: let’s take a look at it. Zhang: Have you looked at the glucose-stimulated insulin secretion from islets in the skeletal muscle-specific knockout? Spiegelman: Yes. It is deficient. I am not sure we’ll see this in the heterozygotes, though. Kadowaki: Reduced PGC1 expression in human skeletal muscle in type 2 diabetes may be both genetically and environmentally determined. Can you comment on the impact of lifestyle factors such as high fat diet and sedentary lifestyle on the expression of PGC1α ? Spiegelman: That is a good question. The first time I presented work on PGC1α in muscle, someone asked what would happen if we elevated it. My response was that the experiment has already been done: exercise. It is known that exercise benefits this condition. In terms of environmental factors, there are now a lot of papers on exercise and the PGC1 coactivators in humans and animals. In all cases, exercise elevates the expression of PGC1α . The simple-minded idea is that physical
12
SPIEGELMAN
movement and activity is probably a dominant player. There are papers showing that high fat diet suppresses PGC1 expression in skeletal muscle. I do believe that these are major mediators of the sedentary versus active lifestyle consequences. References Arany Z, He H, Lin J et al 2005 Transcriptional coactivator PGC-1 alpha controls the energy state and contractile function of cardiac muscle. Cell Metab 1:259–271 Hohmeier HE, Tran VV, Chen G, Gasa R, Newgard CB 2003 Inflammatory mechanisms in diabetes: lessons from the beta-cell. Int J Obes Relat Metab Disord 27 Suppl 3:S12–S16 Maassen JA, ′t Hart LM, Van Essen E 2004 Mitochondrial diabetes: molecular mechanisms and clinical presentation. Diabetes 53 Suppl 1:S103–109 Mootha VK, Lindgren CM, Eriksson KF et al 2003 PGC-1alpha-responsive genes involved in oxidative phosphorylation are coordinately downregulated in human diabetes. Nat Genet 34:267–273 Patti ME, Butte AJ, Crunkhorn S et al 2001 Coordinated reduction of genes of oxidative metabolism in humans with insulin resistance and diabetes: Potential role of PGC1 and NRF1. Proc Natl Acad Sci USA 100:8466–8471 Pedersen BK, Steensberg A, Fischer C et al 2004. The metabolic role of IL-6 produced during exercise: is IL-6 an exercise factor? Proc Nutr Soc 63:263–267 Wredenberg A, Freyer C, Sandstrom ME et al 2006 Respiratory chain dysfunction in skeletal muscle does not cause insulin resistance. Biochem Biophys Res Commun 350:202–207 Zhao YF, Feng DD, Chen C 2006 Contribution of adipocyte-derived factors to beta-cell dysfunction in diabetes. Int J Biochem Cell Biol 38:804–819
Fatty Acids and Lipotoxicity in Obesity and Diabetes: Novartis Foundation Symposium 286, Volume 286. Edited by Gregory Bock and Jamie Goode Copyright Novartis Foundation 2007. ISBN: 978-0-470-05764-3
Human obesity and insulin resistance: lessons from experiments of nature Stephen O’Rahilly University of Cambridge, Department of Clinical Biochemistry, Box 232, Addenbrooke’s Hospital, Hills Road, Cambridge CB2 2QR, UK
Abstract. The past decade or so has seen the adipocyte catapulted from a position of relative obscurity onto the centre stage of biomedical science. Having long been viewed largely as a passive storage depot for energy in times of plenty and a fuel reservoir called upon in times of need, the discovery that the adipocyte is an active participant in the control mechanisms for both energy balance and intermediary metabolism represents one of the most stunning paradigm shifts in modern mammalian biology. The normal control of energy homeostasis is now known to be highly dependent on the adipocytesecreted hormone, leptin. Defects in the leptin signalling pathway, both inherited and acquired, are now known to contribute to the important clinical problem of obesity. Dysfunction of adipocytes, in both obesity and lipodystrophies, is now considered to be critically involved in the pathogenesis of insulin resistance, the metabolic syndrome and type 2 diabetes. The range of metabolites, steroids and bioactive peptides now known to be actively produced by adipocytes and influencing organs as diverse as brain, muscle, liver and pancreatic islet has increased dramatically. Our understanding of how these are co-ordinated to regulate normal metabolism and are dysregulated in metabolic disease is still in its infancy. However what is clear is that the adipocyte, until recently the ‘Cinderella Cell’ of metabolism, has rapidly become the ‘Belle of the Ball’. 2007 Fatty acids and lipotoxicity in obesity and diabetes. Wiley, Chichester (Novartis Foundation Symposium 286) p 13–23
An improved understanding of the molecular pathogenesis of obesity, diabetes and related metabolic disorders will greatly aid improved therapy and targeted prevention of these diseases. All of these disorders have strong heritable components and any comprehensive understanding of them will require the delineation of the genetic variants which predispose to them, how those genetic variants have functional consequences and how they interact with environmental factors to result in disease. While whole-genome association studies of common metabolic phenotypes are now beginning to shed genuine light on common variants predisposing to metabolic disease, these approaches can be richly complemented by studies of humans with extreme phenotypic variation. 13
14
O’RAHILLY
The study of such individuals is valuable for a number of reasons. Firstly, phenotypic variation that is severe and early in its onset is not infrequently the result of major deleterious mutations in a single or a small number of genes, and gene discovery may be readily tractable. The discovery of humans with highly specific and deleterious defects in components of metabolic control pathways can provide valuable insights into normal human physiology. Finally such individuals can provide very useful pointers towards processes that may go wrong in more common metabolic disease. Over the past 15 years we have focused much of our research effort on developing and studying two cohorts of patients with extreme metabolic phenotypes; namely severe early-onset obesity (Farooqi & O’Rahilly 2006) and severe insulin resistance (Semple & O’Rahilly 2005) disproportionate to obesity. Through focused candidate gene approaches and aiding collaborative positional cloning efforts we have identified or contributed to the identification of a number of novel human metabolic genetic disorders. Genetic syndromes of severe obesity Leptin Congenital human leptin deficiency due to homozygous missense or nonsense mutations in leptin has, to date, been found in 13 subjects from 6 families worldwide (Montague et al 1997, Licinio et al 2004, Gibson et al 2004, personal observations). It results in a syndrome of severe early-onset obesity associated with hyperphagia, hypogonadotropic hypogonadism, other endocrine dysfunction and impairment of immune function (Farooqi et al 1999). The subcutaneous administration of recombinant leptin results in a striking amelioration of all phenotypic abnormalities (Farooqi et al 2002). Using functional magnetic resonance imaging (fMRI) in patients before and after leptin therapy, we are currently exploring the effects of leptin on human neuronal circuitry concerned with the visual presentation of food to examine whether the homeostatic adipostatic pathways interact with hedonic circuitry in the brain Leptin receptor. Until recently only a single family with congenital leptin receptor deficiency had been described (Clement et al 1998). The mutation in this family caused the production of an aberrant circulating fragment of the leptin receptor’s extracellular domain. Consequently the patient had extremely high circulating plasma leptin levels. We have previously looked at obese subjects with very high leptin levels and not found any mutations. More recently we examined a cohort of 300 children with extreme obesity selected for family history, hyperphagia and severity of obesity, and found that 3% of these had homozygous loss of function mutations in the leptin receptor (Farooqi et al 2007). Plasma leptin levels were similar in those subjects and equally obese subjects without defects in the leptin receptor, indicating that plasma leptin levels cannot
OBESITY AND INSULIN RESISTANCE
15
be used as a marker of this condition. Affected individuals show a similar phenotype to patients with leptin deficiency in terms of energy balance, hyperphagia, and reproductive and immune defects. However, even in subjects with homozygous null defects, the severity of the phenotype is less severe. This could simply reflect different genetic background but other explanations, which we are currently exploring, include the possibility that, at least when present at high levels in humans, leptin could exert some of its biological actions through a putative alternative receptor Other genetic defects causing obesity Within our cohort we have discovered many subjects with obesity caused by defects in the production, processing or action of melanocortin peptides from the POMC neurons of the arcuate nucleus of the hypothalamus. Subtle mutations in POMC, especially if they affect βMSH, can significantly predispose to obesity (Challis et al 2002, Lee et al 2006). We have discovered three cases of proprotein 1 convertase deficiency (Jackson et al 1997, 2003, personal observation) a syndrome characterized by a wide-variety of clinical abnormalities resulting from a wide ranging defect in prohormone processing as well as obesity. Melanocortin 4 receptor deficiency is the commonest monogenic cause of human obesity being present in 5–6% of children with severe early onset obesity and 0.5–1% of unselected adults with obesity (Farooqi et al 2000, 2003, Larsen et al 2005, Alharbi et al 2007). Recently, the neurotrophin brain-derived neurotrophic factor (BDNF), which acts through the TrkB receptor, has been implicated in the control of energy homeostasis (Xu et al 2003). Notably, we have recently identified children with obesity accompanied by a developmental disorder, who have mutations affecting the expression or function of these genes (Gray et al 2006a, Yeo et al 2004). Lessons learned from extreme obesity Humans can become obese through simple single-gene defects in elements of the adipostatic control pathway. Most genetic defects result in obesity through an impairment of appetite control. One of these conditions, MC4R deficiency, is one of the commonest monogenic diseases in human. Finally, the dramatic therapeutic benefits of leptin therapy in congenital leptin deficiency demonstrates the principle that human obesity can be amenable to a mechanism-based intervention. Syndromes of severe insulin resistance Patients with inherited impaired insulin action disorders exhibit a complex phenotype characterised by abnormalities directly resulting from impaired actions of insulin on metabolism and growth (hyperglycaemia and growth retardation)
16
O’RAHILLY
combined with those resulting from the adverse effects of hyperinsulinaemia (e.g. acanthosis nigricans, polycystic ovary syndrome) (Semple & O’Rahilly 2005). Two basic classes of genetic disease are emerging. Firstly there is a subgroup of patients who have primary disorders of insulin signal transduction. Secondly, there are a group of patients in whom the defective development or maintenance of normal adipose tissue results in lipodystrophy, which itself results in abnormal fuel fluxes in liver and skeletal muscle with resultant severe insulin resistance (Agarwal & Garg 2006). The paradigmatic examples of the former type of defect are the insulin receptoropathies, where genetic defects in the insulin receptor itself, or acquired auto-antibodies to the insulin receptor (so called type B insulin resistance) result in clinical syndromes of insulin resistance, the severity of which parallels the degree of insulin receptor dysfunction (Musso et al 2004). While substantive scientific effort has been expended in the elucidation of the postreceptor mechanism of insulin signalling, there has been little if any direct human genetic evidence for the relevance of any of these molecules for human insulin action. We recently described a family with a missense loss of function mutation in the serine kinase domain of AKT2 in which this mutation clearly co-segregated with extreme insulin resistance (George et al 2004). This is a very rare condition and no other families with AKT2 deficiency have been described. Patients with pseudo-acromegalic severe insulin resistance show evidence for impaired actions of insulin on metabolism but preserved growth effects of insulin (Flier et al 1993, Dib et al 1998). In that subgroup we have demonstrated a functional defect in PI3 kinase signalling, but no causative mutation has yet been found. Mutations in peroxisome proliferator-activated receptor γ (PPARγ ) also result in a syndrome of severe insulin resistance, and so far all disease-causing mutations have been found in the heterozygous state (Barroso et al 1999, Savage et al 2003). While there is some controversy in the literature (Hegele et al 2002, Agarwal & Garg 2002), in our hands all mutations that cause severe insulin resistance act as a dominant negative in reporter assays (Agostini et al 2006). While murine models of these dominant negative mutations replicate the hypertension seen in human subjects, their insulin sensitivity is preserved even on a high fat diet (Tsai et al 2004, Gray et al 2006b). However, in collaboration with Toni Vidal Puig we recently showed that on the Ob/Ob background a dominant-negative PPARγ mutation greatly hastened the development of insulin resistance and diabetes. Interestingly, the doubly mutant animals had smaller fat cell size and lower fat mass, suggesting that the PPARγ mutant prevented the full expandability of the fat mass that is characteristic of the Ob/Ob mouse, and that this was associated with a worsening rather than amelioration of insulin sensitivity. These observations are in accord with the development of severe metabolic consequences when fat cell mass is limited by other genetic and acquired phenomena (Danforth 2000). The concept of ‘adipose tissue expandability’ and its limitation as a risk
OBESITY AND INSULIN RESISTANCE
17
factor for insulin resistance and the metabolic syndrome is worthy of further exploration. PPP1R3A In 2002 we described a family in which extreme insulin resistance resulted from the co-existence of a null mutation in PPARγ with a frameshift premature stop truncation mutation in the PPP1R3A gene (Savage et al 2002). The latter encodes a muscle-specific regulatory subunit of protein phosphatase 1, an enzyme critically involved in the regulation of glycogen synthesis and breakdown. Individuals in this family who were heterozygous for only one of the two mutations appeared relatively normal. Interestingly, however, one subject with the PPARγ mutation subsequently gained 3 BMI (body mass index) units in weight due to a change in social circumstances and developed marked hypertriglyceridaemia and hyperinsulinaemia (personal observations). The PPP1R3A frameshift mutation is notable as it is present in about 1% of the UK Caucasian population. We recently studied normal volunteers who were heterozygous for this mutation using NMR measurements of in vivo glycogen synthesis. These individuals show a remarkable deficit in post-meal glycogen synthesis in skeletal muscle (unpublished observations). A mouse model of this disorder was generated in collaboration with A. De Paoli Roach (Indiana University) and it confirmed the effect of this mutation on glycogen content and glycogen synthesis in skeletal muscle (unpublished observations). However, on its own, this mutation is not sufficient to lead to insulin resistance in either mice or humans. Adiponectin in syndromes of severe insulin resistance In addition to their utility in gene discovery, the availability of a cohort of patients with molecularly defined causes of severe insulin resistance can also provide a powerful resource for exploring physiological relationships in humans. Adiponectin is an adipocyte-derived protein which circulates in high concentrations (Trujillo & Scherer 2005). It is positively associated with insulin sensitivity in human populations and the consensus is that it has some, as yet poorly defined, biological actions that promote insulin sensitization (Kadowaki et al 2006). We measured adiponectin levels in patients with a variety of inherited syndromes of severe insulin resistance. Intriguingly, while adiponectin levels were predictably very low in most patients, they were abnormally high in a single group of patients, namely those whose insulin receptor was genetically defective (Semple et al 2006). Thus, it appears that normal insulin receptor action involves the suppression of adiponectin levels. Is this due to a tonic physiological effect of the insulin receptor or a developmental anomaly present in humans with genetically defective insulin receptors? To answer that we studied patients with an acquired defect in insulin receptor
18
O’RAHILLY
function, namely type B insulin resistance, where autoantibodies to the insulin receptor severely impair insulin signalling (Semple et al 2007). When insulin receptor antibody titres are high adiponectin levels are also elevated. With immunosuppressive treatment and reduction in antibody titres and clinical improvement adiponectin levels fall dramatically. Thus, it seems that the relationship between insulin action and adiponectin is more complex than it has been previously portrayed and the suppression of adiponectin levels by normal insulin receptors is an important aspect of normal physiology. Furthermore, it is notable that patients whose insulin resistance is due to a mutation in AKT2, one of the critical PI3 kinase-dependent pathways of post-receptor insulin signalling, have suppressed adiponectin levels, a finding which strongly suggests that the signalling pathway from the insulin receptor to adiponectin secretion is not Akt dependent. Conclusions The study of patients with severe insulin resistance has resulted in the discovery of previously undescribed monogenic human diseases that have provided new insights into the roles of molecules such as AKT2 and PPARγ in human biology. These disorders also represent a paradigm for gene–gene and gene–environment interactions that will provide a useful framework when trying to dissect the more complex interactions we are likely to have to understand in more common metabolic disease References Agarwal AK, Garg A 2002 A novel heterozygous mutation in peroxisome proliferator-activated receptor-gamma gene in a patient with familial partial lipodystrophy. J Clin Endocrinol Metab 87:408–411 Agarwal AK, Garg A 2006 Genetic basis of lipodystrophies and management of metabolic complications. Annu Rev Med 57:297–311 Agostini M, Schoenmakers E, Mitchell C et al 2006 Non-DNA binding, dominant-negative, human PPARgamma mutations cause lipodystrophic insulin resistance. Cell Metab 4:303–311 Alharbi KK, Spanakis E, Tan K et al 2007 Prevalence and functionality of paucimorphic and private MC4R mutations in a large, unselected European British population, scanned by meltMADGE. Hum Mutat 28:294–302 Barroso I, Gurnell M, Crowley VE et al 1999 Dominant negative mutations in human PPARgamma associated with severe insulin resistance, diabetes mellitus and hypertension. Nature 402:880–883 Challis BG, Pritchard LE, Creemers JW et al 2002 A missense mutation disrupting a dibasic prohormone processing site in pro-opiomelanocortin (POMC) increases susceptibility to early-onset obesity through a novel molecular mechanism. Hum Mol Genet 11: 1997–2004 Clement K, Vaisse C, Lahlou N et al 1998 A mutation in the human leptin receptor gene causes obesity and pituitary dysfunction. Nature 392:398–401
OBESITY AND INSULIN RESISTANCE
19
Danforth E Jr 2000 Failure of adipocyte differentiation causes type II diabetes mellitus? Nat Genet 26:13 Dib K, Whitehead JP, Humphreys PJ et al 1998 Impaired activation of phosphoinositide 3kinase by insulin in fibroblasts from patients with severe insulin resistance and pseudoacromegaly. A disorder characterized by selective postreceptor insulin resistance. J Clin Invest 101:1111–1120 Farooqi S, O’Rahilly S 2006 Genetics of obesity in humans. Endocr Rev 27:710–718 Farooqi IS, Jebb SA, Langmack G et al 1999 Effects of recombinant leptin therapy in a child with congenital leptin deficiency. N Engl J Med 341:879–84 Farooqi IS, Yeo GS, Keogh JM et al 2000 Dominant and recessive inheritance of morbid obesity associated with melanocortin 4 receptor deficiency. J Clin Invest 106:271–279 Farooqi IS, Matarese G, Lord GM et al 2002 Beneficial effects of leptin on obesity, T cell hyporesponsiveness, and neuroendocrine/metabolic dysfunction of human congenital leptin deficiency. J Clin Invest 110:1093–103 Farooqi IS, Keogh JM, Yeo GS, Lank EJ, Cheetham T, O’Rahilly S 2003 Clinical spectrum of obesity and mutations in the melanocortin 4 receptor gene. N Engl J Med 348:1085–1095 Farooqi IS, Wangensteen T, Collins S et al 2007 Clinical and molecular genetic spectrum of congenital deficiency of the leptin receptor. N Engl J Med 356:237–247 Flier JS, Moller DE, Moses AC et al 1993 Insulin-mediated pseudoacromegaly: clinical and biochemical characterization of a syndrome of selective insulin resistance. J Clin Endocrinol Metab 76:1533–1541 George S, Rochford JJ, Wolfrum C et al 2004 A family with severe insulin resistance and diabetes due to a mutation in AKT2. Science 304:1325–1328 Gibson WT, Farooqi IS, Moreau M et al 2004 Congenital leptin deficiency due to homozygosity for the Delta133G mutation: report of another case and evaluation of response to four years of leptin therapy. J Clin Endocrinol Metab 89:4821–4826 Gray J, Yeo GS, Cox JJ et al 2006a Hyperphagia, severe obesity, impaired cognitive function, and hyperactivity associated with functional loss of one copy of the brain-derived neurotrophic factor (BDNF) gene. Diabetes 55:3366–3371 Gray SL, Nora ED, Grosse J et al 2006a Leptin deficiency unmasks the deleterious effects of impaired peroxisome proliferator-activated receptor gamma function (P465L PPARgamma) in mice. Diabetes 55:2669–2677 Hegele RA, Cao H, Frankowski C, Mathews ST, Leff T 2002 PPARG F388L, a transactivationdeficient mutant, in familial partial lipodystrophy. Diabetes 51:3586–3590 Jackson RS, Creemers JW, Ohagi S et al 1997 Obesity and impaired prohormone processing associated with mutations in the human prohormone convertase 1 gene. Nat Genet 16:303–306 Jackson RS, Creemers JW, Farooqi IS et al 2003 Small-intestinal dysfunction accompanies the complex endocrinopathy of human proprotein convertase 1 deficiency. J Clin Invest 112:1550–1560 Kadowaki T, Yamauchi T, Kubota N, Hara K, Ueki K, Tobe K 2006 Adiponectin and adiponectin receptors in insulin resistance, diabetes, and the metabolic syndrome. J Clin Invest 116:1784–1792 Larsen LH, Echwald SM, Sorensen TI, Andersen T, Wulff BS, Pedersen O 2005 Prevalence of mutations and functional analyses of melanocortin 4 receptor variants identified among 750 men with juvenile-onset obesity. J Clin Endocrinol Metab 90:219–224 Lee YS, Challis BG, Thompson DA et al 2006 A POMC variant implicates beta-melanocytestimulating hormone in the control of human energy balance. Cell Metab 3:135–140 Licinio J, Caglayan S, Ozata M et al 2004 Phenotypic effects of leptin replacement on morbid obesity, diabetes mellitus, hypogonadism, and behavior in leptin-deficient adults. Proc Natl Acad Sci USA 101:4531–4536
20
O’RAHILLY
Montague CT, Farooqi IS, Whitehead JP et al 1997 Congenital leptin deficiency is associated with severe early-onset obesity in humans. Nature 387:903–908 Musso C, Cochran E, Moran SA et al 2004 Clinical course of genetic diseases of the insulin receptor (type A and Rabson-Mendenhall syndromes): a 30-year prospective. Medicine (Baltimore) 83:209–222 Savage DB, Agostini M, Barroso I et al 2002 Digenic inheritance of severe insulin resistance in a human pedigree. Nat Genet 31:379–384 (erratum: 2002 Nat Genet 32:211) Savage DB, Tan GD, Acerini CL et al 2003 Human metabolic syndrome resulting from dominant-negative mutations in the nuclear receptor peroxisome proliferator-activated receptor-gamma. Diabetes 52:910–917 Semple RK, O’Rahilly S 2005 PPAR gamma and glucose homeostasis. In: Kumar S, O’Rahilly S (eds) Insulin resistance: insulin action and its disturbances in disease. Wiley, Chichester, p 237–267 Semple RK, Soos MA, Luan J et al 2006 Elevated plasma adiponectin in humans with genetically defective insulin receptors. J Clin Endocrinol Metab 91:3219–23 Semple RK, Halberg NH, Burling K et al 2007 Paradoxical elevation of high molecular weight adiponectin in acquired extreme insulin resistance due to insulin receptor antibodies. Diabetes 56:1712–1717 Trujillo ME, Scherer PE 2005 Adiponectin—journey from an adipocyte secretory protein to biomarker of the metabolic syndrome. J Intern Med 257:167–175 Tsai YS, Kim HJ, Takahashi N et al 2004 Hypertension and abnormal fat distribution but not insulin resistance in mice with P465L PPARgamma. J Clin Invest 114:240–249 Xu B, Goulding EH, Zang K et al 2003 Brain-derived neurotrophic factor regulates energy balance downstream of melanocortin-4 receptor. Nat Neurosci 6:736–742 Yeo GS, Connie Hung CC, Rochford J et al 2004 A de novo mutation affecting human TrkB associated with severe obesity and developmental delay. Nat Neurosci 7:1187–1189
DISCUSSION Spiegelman: What about heterozygosity for either leptin or leptin receptor mutations? O’Rahilly: It is easy to answer for leptin, because we have done the studies. All our original patients came from one small area of the Punjab. To answer the question, we got their parents, studied them in detail, and then went to places in Birmingham where there were people from the same village in the Punjab who were wild-type. This got rid of any ethnic differences. When we did this we saw there was a much higher prevalence of obesity and much higher percentage adiposity in the heterozygote carriers. For the leptin receptor it is much more difficult because we have such a range of different ethnicities and the effects are usually fairly subtle in the heterozygotes. Rudy Liebel did some nice work documenting the effects in Ob and Db heterozygotes (Chung et al 1998). They had about a 20% increase in their fat mass. Spiegelman: Coleman did the original work showing resistance to starvation in the heterozygotes. Glass: In patients with the dominant negative PPARs, did all the mutations in the ligand binding domain affect the ability of the receptor to bind ligands?
OBESITY AND INSULIN RESISTANCE
21
O’Rahilly: Yes, but to a differential degree depending on the ligand. In other words, some would bind non-TZD agonists a bit better than TZDs. One mutant that we have seems to bind farglitizar relatively normally. We are considering trying to use this as a therapy for this individual. Glass: In the cases where receptors with dominant negative mutations could still bind ligand, did they fail to couple to the standard coactivators? O’Rahilly: They all have some degree of impairment of binding of ligand, but they also have impaired coactivator activation and impaired release of co-repressors. Glass: If you look in the patients for markers of inflammation such as circulating CRP, is there have any evidence of enhanced inflammation? O’Rahilly: This work is in the pipeline. These patients come from all over the world and blood is collected in different conditions. Before we answer this I would like to study them under the same conditions. Our PPARγ -dominant negative knock-in mouse doesn’t show any increase in the mRNA levels in adipose tissue for any of the classic mediators of inflammation. Glass: I was interested in the data where you took the monocytes out of these patients and showed that they were defective for activation. Have you looked at the ability of these monocytes to respond to a proinflammatory stimulus? O’Rahilly: We haven’t done this yet, but it would be interesting. Hotamisligil: If you can’t express AP2 in response to these signals, this would result in suppression of the inflammatory responses of the macrophages. O’Rahilly: So you think the absence of AP2 itself would impair the macrophage responses? Hotamisligil: In mice at least it is very clear that if AP2 is deleted there are significantly reduced inflammatory responses in macrophages. Actually, it is surprising that this is a condition of insulin sensitivity where in response to PPARγ, there is no regulation of AP2 yet PPARγ activity in macrophages is significantly increased. So it is possible that adipocyte and macrophage PPARγ activity are regulated differently. Kadowaki: The dominant negative PPARγ mutation actually resulted in extreme insulin resistance in humans, but normal insulin sensitivity in mice. I am interested in plasma adiponectin levels in human patients and in mice overexpressing dominant negative mutant PPARγ. O’Rahilly: A few years ago we reported that the adiponectin levels were unusually low in our PPARγ dominant negative patients. Now that we have the whole group they don’t look so different from the other groups of non-PPARγ severe patients. The only ones that look totally different are the patients with insulin receptor mutations which have high adiponectin levels. All our patient groups are unusual. We may have lots of other patients in that group with mutations in as yet uncharacterized downstream pathways of PPARγ. In the PPARγ dominant negative mice
22
O’RAHILLY
that we studied with Toni Vidal-Puig, adiponectin levels were low even before they were crossed with Ob/Ob mice. Leptin levels were not remarkable in the PPARγ dominant negative mice. Kadowaki: When we first demonstrated the protection from high fat diet induced obesity in heterozygous PPARγ knockout mice (Kubota et al 1999) we proposed that these mice showed increased plasma leptin levels despite their decreased body mass index, because of their release from the PPARγ -induced leptin gene suppression. We still think that is the case. Consistent with your new data, when we recently crossed our heterozygous PPARγ knockout mice with Ob/Ob mice, the protection from obesity disappeared completely. Again, we think this is consistent with the mechanism whereby the protection comes via the increased leptin levels. I’d like to raise what I feel is an important question. Apart from the rare mutations of PPARγ mutations in humans, there is a consensus that a fairly common single nucleotide polymorphism (SNP) in the PPARγ human gene, substituting Pro at codon 12 by Ala, has been reported to decrease PPARγ activity, and this protects against type 2 diabetes (Deeb et al 1998). This PPARγ SNP decreasing activity is one of the few genes that have been repeatedly reported to protect against type 2 diabetes in humans. Spiegelman: The second part of that is well known. I am not sure about the first part, that it is decreasing PPARγ activity. I’m not sure of the effect of this polymorphism on PPARγ activity. O’Rahilly: I agree with you: from the data presented it doesn’t seem to be clear that it is a loss of function allele. Kadowaki: Masato Kasuga’s laboratory also published a paper in which they demonstrated Pro-Ala at codon 12 actually decreased PPARγ activity and decreased adipogenesis. O’Rahilly: What is likely with a subtle variant like this is that it will have differential effects on processes mediated by the N-terminus that are not simply measurable in a simple ‘up or down’ way. Hotamisligil: I have a question regarding possible differences between leptin versus leptin receptor mutations in humans. Have you had a chance to take this into a clean experimental model? O’Rahilly: We are trying. There is no point in us studying mouse models because we don’t believe that the asymmetry exists in mice, so we have to look in human. We are trying to get fresh lymphocytes from patients, which is practically challenging. We have a postdoc who is majoring on this, looking at binding of radiolabelled leptin and also transcriptional profi ling of leptin receptor null cells that are exposed to leptin. Hotamisligil: You could also look at whether, for example, one of the mutations that you see in the extracellular domain of these patients also acts as a null leptin receptor mutation in the mice.
OBESITY AND INSULIN RESISTANCE
23
O’Rahilly: The ones we are using are premature stops or deletions of multiple nucleotides near the N-terminus. Unless there is some read-through transcription, which we will check out, it is likely that they are real nulls and it’s hard to conceive how these mutations could result in anything else. In the lymphocytes we will check for any full-length wild-type leptin receptor. Shi: It has been argued for years that there is good fat versus bad fat. Have you tried to give your lipodystrophy patients ice cream versus fish oil, for example, to see whether the response is different in terms of insulin? O’Rahilly: That’s a great question and one we would be keen to pursue. Sabin: I have seen many children in our obesity clinic, some of whom are only just into the obese threshold, and yet they seem to have all complications of the metabolic syndrome and insulin resistance. Others seem to be able to handle their fat very well and are severely obese with few complications. In the severely insulin resistant children with dyslipidaemia and hypertension who are just into the obese threshold, should we be looking at things like PPARγ and PPP1R3? How common do you think this double hit is? O’Rahilly: That’s a good question. I think this whole issue of adipose tissue expandability and capacity is interesting: is there any way that we can spot these at-risk children? Are there markers within adipose tissue that would tell you that this adipose tissue is working as hard as it can? I would love to be able to tell you what the quantitative genetics of these diseases are, but all we are finding is rare individual examples. Translating this into common metabolic disease is still tough. References Chung WK, Belfi K, Chua M et al 1998 Heterozygosity for Lep(ob) or Lep(rdb) affects body composition and leptin homeostasis in adult mice. Am J Physiol 274:R985–990 Deeb SS, Fajas L, Nemoto M et al 1998 A Pro12Ala substitution in PPARgamma2 associated with decreased receptor activity, lower body mass index and improved insulin sensitivity. Nat Genet 20:284–287 Kubota N, Terauchi Y, Miki H et al 1999 PPAR gamma mediates high-fat diet-induced adipocyte hypertrophy and insulin resistance. Mol Cell 4:597–609
Fatty Acids and Lipotoxicity in Obesity and Diabetes: Novartis Foundation Symposium 286, Volume 286. Edited by Gregory Bock and Jamie Goode Copyright Novartis Foundation 2007. ISBN: 978-0-470-05764-3
Lipid-induced metabolic dysfunction in skeletal muscle Deborah M. Muoio and Timothy R. Koves Departments of Medicine and Pharmacolog y & Cancer Biolog y, and Sarah W. Stedman Nutrition and Metabolism Center, Duke University, Durham, NC 27710, USA
Abstract. Insulin resistance is a hallmark of type 2 diabetes and commonly observed in other energy-stressed settings such as obesity, starvation, inactivity and ageing. Dyslipidaemia and ‘lipotoxicity’—tissue accumulation of lipid metabolites—are increasingly recognized as important drivers of insulin resistant states. Mounting evidence suggests that lipid-induced metabolic dysfunction in skeletal muscle is mediated in large part by stress-activated serine kinases that interfere with insulin signal transduction. However, the metabolic and molecular events that connect lipid oversupply to stress kinase activation and glucose intolerance are as yet unclear. Application of transcriptomics and targeted mass spectrometry-based metabolomics tools has led to our fi nding that insulin resistance is a condition in which muscle mitochondria are persistently burdened with a heavy lipid load. As a result, high rates of β -oxidation outpace metabolic flux through the TCA cycle, leading to accumulation of incompletely oxidized acyl-carnitine intermediates. In contrast, exercise training enhances mitochondrial performance, favouring tighter coupling between β -oxidation and the TCA cycle, and concomitantly restores insulin sensitivity in animals fed a chronic high fat diet. The exercise-activated transcriptional co-activator, PGC1α , plays a key role in co-ordinating metabolic flux through these two intersecting metabolic pathways, and its suppression by overfeeding may contribute to obesity-associated mitochondrial dysfunction. Our emerging model predicts that muscle insulin resistance arises from mitochondrial lipid stress and a resultant disconnect between β -oxidation and TCA cycle activity. Understanding this ‘disconnect’ and its molecular basis may lead to new therapeutic targets for combating metabolic disease. 2007 Fatty acids and lipotoxicity in obesity and diabetes. Wiley, Chichester (Novartis Foundation Symposium 286) p 24–46
As obesity rates continue to surge worldwide, so too does the need for gaining deeper insights into the molecular underpinnings of attendant metabolic defects such as insulin resistance, a condition in which peripheral tissues are rendered unresponsive to both the glucose lowering and antilipolytic properties of the hormone. The onset of insulin resistance, which typically occurs subsequent to weight gain and/or prolonged physical inactivity, increases the risk of developing 24
SKELETAL MUSCLE DYSFUNCTION
25
diabetes, heart disease and hypertension, and thus represents a major pathophysiological link between obesity and the metabolic syndrome. In the context of overnutrition, the progression of insulin resistance occurs in close association with systemic derangements in lipid homeostasis. An increasing number of investigators suggest that insulin resistance stems from the failure of adipocytes to adequately sequester excess fuel. This in turn leads to hyperlipidaemia and ectopic storage of neutral lipids in non-adipose tissues, a phenomenon now commonly referred to as ‘lipotoxicity’ (Lee et al 1994, Unger 2002). Research over the past two decades has indeed established a convincing connection between glucose intolerance, dyslipidaemia and increased triacylglycerol storage in tissues such as skeletal muscle, liver, heart and pancreas. Skeletal muscle is a principal insulin target tissue—reduced insulin sensitivity in muscle leads to diminished postprandial glucose clearance and thereby contributes to systemic hyperglycaemia. Excess lipid supply to the muscle forces accumulation of fatty acid-derived metabolites that interfere with both insulin signalling and glucose utilization. Under normal/healthy conditions, binding of insulin to its receptor initiates a series of protein phosphorylations that culminate with GLUT4 translocation and increased glucose transport. This cascade depends on tyrosine (activating) phosphorylation of proximal signalling molecules such as the insulin receptor (IR) and insulin receptor substrate 1 (IRS1) (Saltiel & Pessin 2002). Mounting evidence from both human and animal studies suggests that lipid-induced insulin resistance is mediated in large part by serine phosphorylation of the IR and IRS1 (Hirosumi et al 2002, Perseghin et al 2003, Saltiel & Pessin 2002, Shoelson et al 2003), and that these modifications are executed by stress-activated serine kinases such as protein kinase C (PKC) and cJun terminal kinase (JNK) (Shulman 2000, Yu et al 2002, Griffin et al 1999). Despite intense investigation, a clear understanding of the molecular and/or metabolic networks that link lipid oversupply to stress kinase activation and muscle insulin resistance has remained elusive. One prominent theory suggests that these signalling events arise from impaired mitochondrial uptake and oxidation of fatty acids. As a result, the acyl-CoAs derived from circulating lipids (triacylglycerol and non-esterified fatty acids) accumulate in the cytoplasm and are preferentially used for synthesis of complex lipids such as diacylglycerol (DAG) and ceramide, which putatively activate insulin inhibitory serine kinases, including PKC (Fig. 1). The notion that reduced β -oxidation represents a primary metabolic lesion in obesity stems largely from reports showing that muscles from obese humans exhibit diminished rates of fat oxidation in the fasted state (Kelley et al 1999, Blaak 2004, Hulver et al 2003). It is noteworthy, however, that these studies relied on indirect calorimetry and/or tracer methodologies to assess mitochondrial lipid catabolism, both of which use CO2 production as the familiar endpoint of fuel oxidation. Recent studies using mass spectrometry-based metabolic profiling led to our
26
MUOIO & KOVES
FIG. 1. Molecular mechanisms connecting lipid overload to muscle insulin resistance. The prevailing view of ‘lipotoxicity’ predicts that fatty acyl-CoAs (LC-CoAs) derived from circulating lipids are diverted away from mitochondrial degradation (grey X) and towards storage, primarily in the form of intramuscular triacylglycerol (mTAG). Together, reduced carnitine palmitoyltransferase 1 (CPT1) activity, low β -oxidative capacity and increased mTAG levels force cytosolic accumulation of (LC-CoAs). This in turn provides a surfeit of substrate for complex lipids, such diacylglycerol and ceramide, which together with the LC-CoAs, are thought to activate stress-induced serine kinases that impede insulin signalling. Alternatively, emerging evidence indicates that over-nutrition results in increased β -oxidation but without co-ordinated up-regulation of the tricarboxylic acid (TCA) cycle and electron transport system (ETS) (white X). In the latter model, excessive incomplete β -oxidation and/or accompanying mitochondrial energy stress give rise to serine kinase activation and impaired insulin action. IR, insulin receptor; NEFA, non-esterified fatty acids.
discovery that assessment of CO2 production does not necessarily provide a fully informative view of fatty acid catabolism (An et al 2004, Koves et al 2005a). In contrast to the foregoing paradigm, we find that mitochondrial uptake and β oxidation of fatty acids is enhanced by obesity and overnutrition, thereby imposing a persistent lipid burden on muscle mitochondria. Accordingly, this paper will summarize evidence that skeletal muscle adapts to obesity by boosting β -oxidative machinery and increasing fat catabolism. Emphasis will centre on recent application of targeted metabolomics technologies to gain a more comprehensive view of
SKELETAL MUSCLE DYSFUNCTION
27
obesity-associated glucose intolerance, as well as its reversal by habitual exercise. We will also provide an overview of key transcriptional programmes that give rise to metabolic conditions that antagonize, or conversely enhance, insulin sensitivity. Finally, we will describe emerging evidence suggesting that the road to dietinduced insulin resistance might actually travel through the mitochondria (Fig. 1). Transcriptional adaptation to increased lipid supply In skeletal muscle, fatty acids not only provide an important metabolic substrate that permits sustained contractile activity, but they also trigger and/or participate in many of the signalling pathways that contribute to training-induced adaptations in energy metabolism. On the other hand, fatty acids and their lipid by-products are thought to play a major role in the pathogenesis of metabolic disease. Research from our laboratory and many others indicates that both adaptive and maladaptive responses to fatty acids are driven in large part by the peroxisome proliferatoractivated receptor (PPAR) family of nuclear hormone receptors. These receptors function as molecular sensors of the cellular lipid milieu (Gilde & Van Bilsen 2003). They are bound and activated by fatty acids and other lipid metabolites, and in turn initiate transcriptional programmes that regulate fatty acid metabolism. There are three known PPAR subtypes; PPARs α and δ, which are the major subtypes expressed in muscle, primarily target pathways involved in lipid catabolism (Gilde & Van Bilsen 2003, Muoio et al 2002), and PPARγ, which is expressed mainly in adipose tissue and regulates programmes of adipogenesis, lipogenesis and inflammation (Rosen et al 1999). The advent of cDNA microarray chips and real-time quantitative PCR technology has permitted comprehensive analysis of how the skeletal muscle transcriptome responds to a heavy influx of fatty acid. To this end, we analysed gene expression profi les in muscle from rodents that were subjected to a diverse set of physiological challenges known to raise circulating levels of free fatty acids, including overnight starvation, chronic (12 week) high-fat feeding and acute exercise. Results from the analyses showed that like starvation and exercise, chronic high fat feeding results in robust induction of PPAR-targeted genes associated with lipid transport and trafficking, β -oxidation, glucose sparing and mitochondrial uncoupling. Likewise, induction of a similar set of genes was observed in cultured rat L6 myocytes that were rendered insulin resistant by exposure to 200–500 µ M fatty acids for 24–48 h. Thus, whereas enhanced β -oxidative capacity is typically thought to confer protection against metabolic disease, these results suggested that high rates of β -oxidation might be detrimental under some circumstances. Considering that obesity and exercise induce contrasting changes in insulin sensitivity, we proceeded to survey mRNA expression patterns to identify genes
28
MUOIO & KOVES
that were differentially regulated by high fat feeding compared to exercise. Prominent among the targets identified was PPARγ coactivator 1α (PGC1α ), a transcriptional co-activator that functions as a master regulator of mitochondrial biogenesis and energy homeostasis (Puigserver et al 1998). In skeletal muscle, PGC1α has been shown to stimulate mitochondrial proliferation via co-activation of the nuclear respiratory factor (Lin et al 2002) and to regulate genes involved in oxidative phosphorylation through interactions with oestrogen-related receptor α (Mootha et al 2004). PGC1α also co-activates the PPARs (Puigserver & Spiegelman 2003), thereby regulating pathways of lipid metabolism. Consistent with other reports, we found that exercise increases mRNA levels of PGC1α and that the gene is more abundant in red/oxidative compared to white/ glycolytic muscles. Remarkably, PGC1α expression decreased 70% in response to high fat feeding. Denervation (a model of muscle disuse) caused a similar suppression of PGC1α (Koves et al 2005a, Sparks et al 2005). Thus, in the context of overnutrition, obesity and/or inactivity, lipid-induced PPAR activation increases expression of the β -oxidative machinery, but paradoxically, this occurs in the face of reduced PGC1α activity and a corresponding decrease in several nuclear encoded mitochondrial genes (Koves et al 2005a, Sparks et al 2005). These results prompted speculation that the pathophysiological consequences of heightened PPAR activity might depend on the metabolic backdrop, particularly as it relates to adjustments in mitochondrial energy homeostasis. Obesity is characterized by mitochondrial lipid overload The finding that obesity-associated hyperlipidaemia leads to up-regulation of β oxidative enzymes in muscle is seemingly at odds with the theory that lipid-induced insulin resistance stems from diminished fat oxidation. To reconcile this discrepancy we have proposed that diabetic states are characterized by high rates of ‘incomplete fat oxidation’, referring to a condition in which fatty acids are only partially degraded (Fig. 2A). This emergent model holds that lipid-induced upregulation of the enzymatic machinery for β -oxidation of fatty acids is not always supported by a co-ordinated increase in downstream metabolic pathways such as the tricarboxylic acid (TCA) cycle and electron transport chain (Koves et al 2005a, Muoio & Newgard 2006). The concept of incomplete fuel oxidation was initially guided by the foregoing microarray studies, and later substantiated in experiments that contrasted the metabolic consequences of enhancing fatty acid supply to muscle via changes in fibre type, exercise, overnight starvation or high fat feeding (Koves et al 2005a, 2005b). These analyses were performed in isolated mitochondria. Most strikingly, we found that oxidation rates of [14C]palmitate (expressed per g mitochondrial protein) in mitochondria from deep red gastrocnemius were eightfold higher than those from the superficial white gastrocnemius. Similarly,
SKELETAL MUSCLE DYSFUNCTION
29
FIG. 2. Muscle acylcarnitine profi ling in diet-induced insulin resistance and exercise training. (A) Schematic of mitochondrial fatty acid oxidation. Complete oxidation of fatty acid to CO2 depends on carnitine palmitoyltransferase (CPT1), β -oxidation ( β -Ox), tricarboxylic acid (TCA) cycle activity and energy flux through the electron transport system (ETS). Acyl-carnitines, derived primarily from mitochondrial acyl-CoA intermediates, are by-products of incomplete fuel oxidation. (B) Gastrocnemius muscles were harvested from rats fed ad libitum (fed) or starved 24 h after 12 weeks on either a standard chow (SC) or high fat (HF) diet. (C) Gastrocnemius muscles were harvested from mice fed on standard chow (SC) or high fat (HF) diets for 14 weeks. During the fi nal 2 weeks of the diet half of the mice in each group were kept sedentary (Sed) or exercise trained (Ex) on a running wheel. Muscle acylcarnitine profi les were evaluated by tandem mass spectometry and are expressed as a percent of SC-fed controls. Data are from Koves et al 2005a.
30
MUOIO & KOVES
exercise training, which causes a shift in muscle fibre composition, increased mitochondrial fatty acid oxidative capacity approximately twofold over the sedentary condition. Interestingly however, unlike the exercise and fibre type models, neither of the nutritional manoeuvres enhanced the capacity of the muscle mitochondria to fully oxidize palmitate to CO2 , despite a robust induction of β -oxidative genes. In these studies we also measured radiolabel incorporation into chain-shortened acid soluble metabolites (ASM), which provides an index of incomplete fat catabolism. Compared to the control condition, both starvation and high fat feeding increased 14C labelling of the ASM fraction. Thus, mitochondria from rats fed on a high fat compared to a standard chow diet displayed similar rates of [14C]palmitate oxidation to CO2 , but accumulated more radiolabelled intermediates in an acid-soluble pool (Koves et al 2005a). Additionally, pyruvatemediated inhibition of palmitate oxidation, which was robust in normal mitochondria, was severely blunted after high fat feeding. Lastly, we found that 12 weeks of high fat feeding diminished mitochondrial respiratory capacity as measured by classic polarography (T. R. Koves and D. M. Muoio, unpublished), thereby implying critical impairments in mitochondrial function. Subsequent studies are now showing that extended high fat feeding (1 year) not only amplifies the high rates of incomplete β -oxidation, but concomitantly lowers mitochondrial potential to completely oxidize lipid substrate. Thus, similar to observations in obese humans (Kelley et al 1999, Hulver et al 2003), we too fi nd that complete oxidation of fatty acid (to CO2 ) is inefficient with prolonged obesity. However, as described below, this phenomenon appears to be linked to excessive mitochondrial fatty acid uptake rather than the converse. To gain a more global view of mitochondrial metabolism in response to lipid stress we turned to mass spectrometry-based metabolic profi ling. Consistent with other outcome measures, these analyses revealed marked accumulation of fatty acylcarnitine species (mitochondrial-derived by-products of incomplete lipid oxidation) in muscles from rats fed on a high fat diet (Koves et al 2005a). Rats fed a standard chow diet lowered muscle production of fatty acylcarnitines during transition from the fasted to the fed state, whereas those on the high-fat diet exhibited little or no change (Fig. 2B). This apparent failure to switch substrates in the fed state suggests that high fat feeding imposes a persistent lipid burden on muscle mitochondria, and moreover, reflects a pattern reminiscent of the metabolic inflexibility theory originally proposed by Kelley et al (1999). Moreover, profi ling of organic acid metabolites indicated that diet-induced obesity leads to a reduction in muscle concentrations of several TCA cycle intermediates, possibly reflecting diminished rates of glycolysis, redox inhibition of the TCA cycle and/or electrochemical pressure on the ETS (Fig. 3). We have observed similar metabolic profi les (elevated acylcarnitines and reduced organic acids) in muscle of Zucker Diabetic Fatty rats. Thus, consistent with radiotracer studies in isolated mitochondria, the
SKELETAL MUSCLE DYSFUNCTION
31
FIG. 3. Organic acid profi ling of gastrocnemius muscle. (A) Gastrocnemius muscles were harvested from rats fed on either a standard chow (SC) or high fat (HF) diet for 12 weeks, and organic acid intermediates of glucose metabolism and the tricarboxylic acid (TCA) cycle were measured by tandem mass spectometry. Metabolic flux through the TCA cycle and electron transport system (ETS) is inhibited by high NADH/NAD and ATP/ADP, respectively. Dietinduced decreases in these intermediates might reflect reduced glycolysis and/or redox (NADH) inhibition of TCA cycle enzymes such as isocitrate dehydrogenase (ICD) and α -ketoglutarate dehydrogenase ( α KGD).
emerging ‘metabolomic’ fingerprint of obesity suggests excessive β -oxidation, diminished TCA cycle activity and impaired substrate switching. Adding merit to the idea that glucose intolerance might be linked to mitochondrial lipid overload are the results from an earlier study in which genetic rescue of insulin sensitivity corresponded to a unique decrease in the concentration of one lipid-derived metabolite, β -OH-butyrylcarnitine (C4-OH), in skeletal muscle (An et al 2004). Muscle concentrations of this metabolite, which is considered to be a marker of β -oxidation and ketogenesis, correlated positively with serum levels of non-esterified fatty acids but not circulating ketones. These findings suggested that production of β -OH-butyrylcarnitine occurred locally within the muscle as a consequence of increased lipid delivery. Further studies revealed that exposure of L6 myotubes to elevated concentrations of fatty acids not only induces enzymes of fatty acid oxidation, but also increases the expression of the ketogenic enzyme
32
MUOIO & KOVES
mitochondrial HMG CoA synthase (An et al 2004) as well as the production of free ketones (unpublished). This work therefore raised the likely possibility that de novo ketogenesis (typically thought of as a hepatic programme) might be induced in skeletal muscle to provide an outlet for excess acetyl CoA that accumulates when β -oxidative flux exceeds TCA cycle activity. Exercise training promotes lipid tolerance Similar to high fat feeding, exercise training increases skeletal muscle supply and storage of lipid substrates (Goodpaster et al 2001). However, in contrast to the adverse events provoked by a high fat diet, exercise is known to promote muscle as well as whole-body metabolic fitness (Astrup 2001). These observations suggest that physical activity enhances the muscle’s capacity to cope with lipid surplus. Indeed, contractile activity leads to dramatic adjustments in oxidative fuel metabolism that are mediated in large part by increases in PGC1α expression and activity, and accompanying stimulation of mitochondrial biogenesis (Adhihetty et al 2003). These reports led us to question whether the mitochondrial abnormalities caused by lipid overload might be overcome by exercise training. Consistent with this notion, we found that a three-week exercise intervention in mice fed on a chronic high fat diet lowered muscle acylcarnitine levels (Fig. 2C), in association with increased TCA cycle activity and rescue of glucose tolerance, despite continued feeding on the fat-rich regimen (Koves et al 2005a). The metabolic benefits of exercise were also associated with restored expression of PGC1α . The results from animal studies gained further support from cell-based experiments that likewise highlighted important roles for PGC1α and the PPARs in mediating lipid-induced metabolic reprogramming (Koves et al 2005a). Similar to muscle mitochondria from high fat fed rats, L6 myocytes exposed to increasing fatty acid concentrations exhibited disproportionate increases in the rates of incomplete (assessed by measuring incorporation of the label from [14C]oleate into acid-soluble β -oxidative intermediates) relative to complete (label incorporation into CO2 ) β -oxidation of fatty acids (Fig. 4). PGC1α overexpression increased oxidative capacity, thereby permitting 14CO2 production to maintain pace with the [14C]-labelled acid-soluble β -oxidative intermediates (Figs 4A, B). In other words, the ratio of complete to incomplete β -oxidation was dramatically increased by high PCG1α activity (Fig. 4C). Accordingly, myocyte accumulation of fatty acylcarnitines decreased (Fig. 4D). Consistent with these functional assessments, cDNA microarray analyses showed that fatty acid exposure in the context of low PGC1α activity (which is observed in obesity) resulted in the induction of classic PPARtargeted genes involved in lipid trafficking, glucose sparing and β -oxidation, but with little or no change in other downstream pathways that regulate respiratory capacity. In contrast, high PGC1α expression enabled the co-ordinated induction
SKELETAL MUSCLE DYSFUNCTION
33
FIG. 4. PGC1α enhances complete oxidation of fatty acids. Fatty acid oxidation was evaluated in rat L6 myocytes treated with recombinant adenoviruses encoding β -galactosidase ( β -gal) or PGC1α , compared against a no virus control (NVC) group. 48 h after addition of virus, cells were incubated for 3 h with 100–500 µ M [14C]oleate. Complete (A) and incomplete (B) fatty acid oxidation was determined by measuring 14C label incorporation into CO2 and acid soluble metabolites (ASM), respectively. (C) The relationship between incomplete and complete fatty acid oxidation was expressed as a ratio of label incorporated into ASM divided by labelling of CO2 . (D) Incomplete fuel oxidation assessed by measurement of total acylcarnitines. Differences between groups were analysed by ANOVA and Student’s t-test, * indicates P < 0.05 comparing PGC1α to NVC and β -gal treatments, ‡ indicates P < 0.05 comparing low and high FA conditions. Data are from Koves et al 2005a.
34
MUOIO & KOVES
of β -oxidative enzymes with equally important downstream targets (e.g. TCA cycle, ETS, and NADH shuttle systems). These findings imply that PGC1α enables tighter coupling between β -oxidation and the TCA cycle. Does lipid-induced insulin resistance require mitochondrial b -oxidation? Our evolving model holds that accelerated incomplete fatty acid oxidation, resulting from a lipid-induced mismatch between β -oxidation and TCA cycle activity, gives rise to a mitochondrial-derived stress signal that impedes insulin action (Fig. 1). Evidence from other laboratories has similarly linked increased β -oxidation to insulin resistance. For example, transgenic mice with muscle-specific overexpression of PPARα display both local and systemic glucose intolerance in association with marked induction of several lipid-oxidative genes (Finck et al 2005). Notably, the diabetic phenotype of these animals was reversed by administration of oxfenicine, a compound that prevents mitochondrial uptake and oxidation of fatty acids. Equally supportive are data from studies using PPARα null mice, which exhibit decreased fat oxidation, increased circulating NEFA and elevated triacylglycerol content in several peripheral tissues (Leone et al 1999, Muoio et al 2002, Finck et al 2005). Despite this severe lipid-dysregulated phenotype, the PPARα -null mice are remarkably protected against diet-induced glucose intolerance (Finck et al 2005, Guerre-Millo et al 2001). Interestingly, we have found that the antidiabetic phenotype of PPARα null mice corresponds with reduced rates of incomplete β oxidation in response to a lipid challenge (overnight fasting). These mice represent one of the few models in which elevated serum non-esterified fatty acid levels actually accompany improved insulin sensitivity, suggesting that lipids are less destructive when β -oxidation is limited. We are currently testing this hypothesis using both genetic and pharmacological approaches in vitro and in animal models. Similar to the animal studies, exposure of rat L6 cultured myotubes to a 1 : 1 oleate: palmitate mix produced a lipid-induced phenotype marked by elevated intracellular levels of short, medium and long chain aclycarnitines and increased rates of incomplete β -oxidation, assessed by mitochondrial production of ASM and ketones. Accumulation of these intermediates depended on the presence of carnitine and was prevented by co-administration of etomoxir, which inhibits carnitine palmitoyltransferase 1 and hence disallows fatty acid entry into the mitochondria. Twentyfour hour exposure of L6 myotubes to high fatty acid levels impaired insulin signalling (assessed by phosphorylation of AKT). Notably however, the insulin desensitizing effect of lipid exposure required the presence of carnitine and was obviated by etomoxir, implying that the signalling defect depends on mitochondrial β -oxidation. Whereas a number of confl icting reports associate obesity and insulin resistance with either increased (Randle et al 1963) or decreased (Saha & Ruderman 2003,
SKELETAL MUSCLE DYSFUNCTION
35
Yu et al 2002) fat oxidation, the foregoing findings now provide a potential reconciliation for these discrepancies. The new model described in Fig. 5 holds that fuel oversupply to muscle results in enhanced fatty acid β -oxidation due to both transcriptional remodelling and increased substrate supply. However, in the absence of work (i.e. exercise), the TCA cycle not only remains inactivated at a
FIG. 5. Proposed model of lipid-induced metabolic dysfunction in muscle. Obesity and overnutrition increase fatty acid supply coming from both circulating and endogenously stored lipids. Fatty acid surplus and peroxisome proliferator-activated receptor (PPAR)-mediated activation of target genes promote β -oxidation without an accompanying increase in tricarboxylic acid (TCA) cycle enzymes. TCA cycle flux and complete fat oxidation is further hampered by a high energy redox state (rising NADH/NAD and acetyl-CoA/free CoA ratios). As a result, metabolic by-products of incomplete fatty acid oxidation (acylcarnitines, ketones and reactive oxygen species [ROS]) accumulate, and in turn activate stress kinases or other signals that impinge upon insulin action. Exercise combats lipid stress by activating PPARγ coactivator 1 α (PGC1α ), which co-ordinates increased β -oxidation with the activation of downstream metabolic pathways, thereby enhancing both mitochondrial function and complete fuel oxidation. Tighter coupling of β -oxidation and TCA cycle activity alleviates mitochondrial stress, lowers intramuscular lipids and restores insulin sensitivity. Abbreviations: ACS, acyl-CoA synthase; β -Oxd, β -oxidative enzymes; CD36/FAT, fatty acid transporter; CPT1, carnitine palmitoyltransferase 1; DAG, diacylglycerol; ETS, electron transport chain; Glut4, glucose transporter 4; TAG, triacylglycerol; IR, insulin receptor; LC-CoAs, long-chain fatty acyl-CoAs; TF, transcription factor.
36
MUOIO & KOVES
transcriptional level, but moreover, flux through the pathway is inhibited by the high energy redox state that prevails under circumstances of over-nutrition. As a result, acetyl-CoA and other acyl-CoA species accumulate while concentrations of several TCA cycle intermediates diminish (as reflected by acylcarnitine and organic acid profi ling). This environment favours generation of other lipid-derived molecules, including ketones and reactive oxygen species, that might feed back on serine kinases that oppose insulin action in an attempt to alleviate mitochondrial energy stress (Bloch-Damti & Bashan 2005). Eventually, the overburdened mitochondria begin to fail, imposing more severe impairments in oxidative phosphorylation and fuel homeostasis. In summary, rapid advances in ‘omics’ technologies are providing novel insights into the mechanisms whereby over-nutrition and exercise confer contrasting effects on metabolic health. It has become increasingly apparent that alterations in lipid homeostasis and accompanying mal/adaptations at the level of the mitochondria are central to these phenomena (Lowell & Shulman 2005). Whereas considerable attention has been focused on the potential role of glycerolipids, acyl-CoAs and ceramides as the molecular perpetrators of insulin resistance, recent evidence provides cause to re-evaluate the possibility that glucose intolerance develops as a result of excessive β -oxidation (Randle et al 1963). Further investigation is now necessary to determine whether fatty acids must penetrate the mitochondria to exert their insulin desensitizing actions on muscle; or alternatively, whether the high rates of fatty acid catabolism in the obese state are insufficient to compensate for increased lipid delivery, thereby allowing cytosolic lipid metabolites to accumulate and block insulin signalling. Perhaps these two paradigms are not mutually exclusive, but rather, coexist. Assuming that insulin resistance originally evolved as an adaptive mechanism to facilitate survival during periods of famine, it is likely that nature has devised several distinct metabolic and molecular roadways leading to the same (dys)functional endpoint. Future studies are certain to reveal additional clues regarding the interplay between skeletal muscle mitochondrial function, insulin action and the progression of metabolic disease. Acknowledgements We thank Chris Newgard and Jie An for their important contributions to the ideas presented in this paper. We are also thank the dedicated staff of the Metabolomics and Biomarker Core of the Sarah W. Stedman Nutrtion and Metabolism Center. Studies cited from the authors’ laboratory were supported by NIH grants K01 DK56112 (D.M.M.) and the American Diabetes Association (D.M.M.).
References Adhihetty PJ, Irrcher I, Joseph AM, Ljubicic V, Hood DA 2003 Plasticity of skeletal muscle mitochondria in response to contractile activity. Exp Physiol 88:99–107
SKELETAL MUSCLE DYSFUNCTION
37
An J, Muoio DM, Shiota M et al 2004. Hepatic expression of malonyl-CoA decarboxylase reverses muscle, liver and whole-animal insulin resistance. Nat Med 10:268–274 Astrup A 2001 Healthy lifestyles in Europe: prevention of obesity and type II diabetes by diet and physical activity. Public Health Nutr 4:499–515 Blaak EE 2004 Basic disturbances in skeletal muscle fatty acid metabolism in obesity and type 2 diabetes mellitus. Proc Nutr Soc 63:323–330 Bloch-Damti A, Bashan N 2005 Proposed mechanisms for the induction of insulin resistance by oxidative stress. Antioxid Redox Signal 7:1553–1567 Finck BN, Bernal-Mizrachi C, Han DH et al 2005 A potential link between muscle peroxisome proliferator-activated receptor- α signaling and obesity-related diabetes. Cell Metab 1:133–144 Gilde AJ, Van Bilsen M 2003 Peroxisome proliferator-activated receptors (PPARs): regulators of gene expression in heart and skeletal muscle. Acta Physiol Scand 178:425–434 Goodpaster BH, He J, Watkins S, Kelley DE 2001 Skeletal muscle lipid content and insulin resistance: evidence for a paradox in endurance-trained athletes. J Clin Endocrinol Metab 86:5755–61 Griffi n ME, Marcucci MJ, Cline GW et al 1999 Free fatty acid induced insulin resistance is associated with activation of protein kinase C theta and alterations in the insulin signaling cascade. Diabetes 48:1270–1274 Guerre-Millo M, Rouault C, Poulain P et al 2001 PPAR-alpha-null mice are protected from high-fat diet-induced insulin resistance. Diabetes 50:2809–2814 Hirosumi J, Tuncman G, Chang L et al 2002 A central role for JNK in obesity and insulin resistance. Nature 420:333–336 Hulver MW, Berggren JR, Cortright RN et al 2003 Skeletal muscle lipid metabolism with obesity. Am J Physiol Endocrinol Metab 284:E741–E747 Kelley DE, Goodpaster B, Wing RR, Simoneau JA 1999 Skeletal muscle fatty acid metabolism in association with insulin resistance, obesity, and weight loss. Am J Physiol 277: E1130–E1141 Koves TR, Li P, An J et al 2005a Peroxisome proliferator-activated receptor-gamma co activator 1alpha-mediated metabolic remodeling of skeletal myocytes mimics exercise training and reverses lipid-induced mitochondrial inefficiency. J Biol Chem 280:33588–33598 Koves TR, Noland RC, Bates AL, Henes ST, Muoio DM, Cortright RN 2005b Subsarcolemmal and intermyofibrillar mitochondria play distinct roles in regulating skeletal muscle fatty acid metabolism. Am J Physiol Cell Physiol 288:C1074–C1082 Lee Y, Hirose H, Ohneda M, Johnson JH, McGarry JD, Unger RH 1994 Beta-cell lipotoxicity in the pathogenesis of non-insulin-dependent diabetes mellitus of obese rats: impairment in adipocyte-beta-cell relationships. Proc Natl Acad Sci USA 91:10878–10882 Leone TC, Weinheimer CJ, Kelly DP 1999 A critical role for the peroxisome proliferatoractivated receptor alpha (PPARalpha) in the cellular fasting response: the PPARalpha-null mouse as a model of fatty acid oxidation disorders. Proc Natl Acad Sci USA 96:7473–7478 Lin J, Wu H, Tarr PT et al 2002 Transcriptional co-activator PGC-1 alpha drives the formation of slow-twitch muscle fibres. Nature 418:797–801 Lowell BB, Shulman GI 2005 Mitochondrial dysfunction and type 2 diabetes. Science 307:384–387 Mootha VK, Handschin C, Arlow D et al 2004 Erralpha and Gabpa/b specify PGC-1alphadependent oxidative phosphorylation gene expression that is altered in diabetic muscle. Proc Natl Acad Sci USA 101:6570–6575 Muoio DM, Newgard CB 2006 Obesity-related derangements in metabolic regulation. Annu Rev Biochem 75:367–401 Muoio DM, MacLean PS, Lang DB et al 2002 Fatty acid homeostasis and induction of lipid regulatory genes in skeletal muscles of peroxisome proliferator-activated receptor (PPAR)
38
MUOIO & KOVES
alpha knock-out mice-evidence for compensatory regulation by PPAR delta. J Biol Chem 277:26089–26097 Perseghin G, Petersen K, Shulman GI 2003 Cellular mechanism of insulin resistance: potential links with inflammation. Int J Obes Relat Metab Disord 27 Suppl 3:S6–11 Puigserver P, Spiegelman BM 2003 Peroxisome proliferator-activated receptor-gamma coactivator 1alpha (PGC-1alpha): transcriptional coactivator and metabolic regulator. Endocr Rev 24:78–90 Puigserver P, Wu ZD, Park CW, Graves R, Wright M, Spiegelman BM 1998 A cold-inducible coactivator of nuclear receptors linked to adaptive thermogenesis. Cell 92:829–839 Randle PJ, Garland PB, Hales CN, Newsholme EA 1963 The glucose fatty-acid cycle. Its role in insulin sensitivity and the metabolic disturbances of diabetes mellitus. Lancet 1:785–789 Rosen ED, Sarraf P, Troy AE et al 1999 PPAR gamma is required for the differentiation of adipose tissue in vivo and in vitro. Mol Cell Biol 4:611–617 Saha AK, Ruderman NB 2003 Malonyl-CoA and AMP-activated protein kinase: an expanding partnership. Mol Cell Biochem 253:65–70 Saltiel AR, Pessin JE 2002 Insulin signaling pathways in time and space. Trends Cell Biol 12:65–71 Shoelson SE, Lee J, Yuan M 2003 Inflammation and the IKK beta/I kappa B/NF-kappa B axis in obesity- and diet-induced insulin resistance. Int J Obes Relat Metab Disord 27 Suppl 3:S49–52 Shulman GI 2000 Cellular mechanisms of insulin resistance. J Clin Invest 106:171–176 Sparks LM, Xie H, Koza RA et al 2005 A high-fat diet coordinately downregulates genes required for mitochondrial oxidative phosphorylation in skeletal muscle. Diabetes 54:1926–1933 Unger RH 2002 Lipotoxic diseases. Annu Rev Med 53:319–36 Yu C, Chen Y, Cline GW et al 2002 Mechanism by which fatty acids inhibit insulin activation of insulin receptor substrate-1 (IRS-1)-associated phosphatidylinositol 3-kinase activity in muscle. J Biol Chem 277:50230–50236
DISCUSSION Spiegelman: This work feeds in with what we discussed earlier, emphasizing the idea of ROS. One thing you alluded to is that energy has to go somewhere, so we can never accept more or less oxidation of fatty acids as an endpoint. The question is always, where do the electrons go? They can build up the mitochondrial membrane potential and dissipate to make ATP. But if that isn’t happening then something else must be. They have to either go into ROS or something else. When you are getting the metabolomic data showing partial oxidation of fatty acids, where are the electrons going in these systems? People who are unsophisticated in energy metabolism think this is the end of the story, but it is not. Where do you think the electrons are going in this system? Muoio: Some of the excess electrons might leak from the system as ROS. We suggest that redox pressure associated with or caused by partial fat oxidation results in inefficient electron transfer through ETS complexes I and III, which in turn increases the half-life of the superoxide anion and thereby promotes oxidative
SKELETAL MUSCLE DYSFUNCTION
39
stress (Schrauwen & Hesselink 2004, Berthiaume et al 2005). In the cell culture system, we have found that exposure to fatty acids increases mitochondrial membrane potential; supporting the notion that incomplete β -oxidation might reflect a mitochondrial environment that is conducive to oxidative stress. Perhaps ROS serve as signals that link excessive fat oxidation to insulin resistance; at present, this is our leading hypothesis. In addition, it is important to recognize that the acylcarnitine by-products of β -oxidation can exit the mitochondria; therefore many of these lipid-derived carbons are redirected, bypassing both the TCA cycle and electron transport system. Spiegelman: It is a dilemma. The animals are fed on a high fat diet and they are not exercising more. You think you are getting more fat oxidation: where is the energy going? Muoio: Measurement of serum and urine acyl carnitine intermediates is commonly used as a diagnostic screen for mitochondrial type diseases. Under conditions of severe mitochondrial insufficiencies, these intermediates are exported from the tissue and can be excreted in urine. We suspect that a similar path is taken in models of obesity; in other words, some energy can be redirected to other tissues and/or lost through this route. Black: What is the composition of your high fat diet? The follow-up on this is that presumably it is not a fat diet that consists of all ‘good’ fat, so have you tried modulating the composition of the diet to be predominantly lard based or predominantly fish oil based to see the effects? Muoio: The diet we use is 45% fat, consisting mostly of saturated fatty acids such as palmitate. We have used GC/MS to measure individual circulating free fatty acids in animals fed this diet. The two predominant fatty acids that are elevated in response to the diet are palmitate and oleate, which are the same two fatty acids that we use in our cell culture system. We are also collaborating with other laboratories to compare the metabolic signature of diets that are high in trans-fatty acids or medium chain fatty acids. For example, the Surwit diet is one that is commonly used to induce obesity and insulin resistance. This diet is quite high in medium chain fatty acids and is known to induce ketogenesis. Predictably, acylcarnitine profi les from animals fed this diet are marked by an abundance of medium chain β -oxidative intermediates and accumulation of ketone-related metabolites. We have not yet examined diets high in fish oil and/or polyunsaturated fatty acids. Hotamisligil: In both cells and animals palmitate could be as strong as tunicamycin in disrupting endoplasmic reticulum (ER) function. Furthermore, if you expose a macrophage to palmitate it has almost the same effect as exposing it to lipopolysaccharide (LPS). In a dietary condition like this and the aged model, is it possible that some of the biology ascribed to mitochondria might actually be secondary defects?
40
MUOIO & KOVES
Muoio: There are a number of reports showing severe, adverse responses when cells in culture are exposed to palmitate. However, the physiological relevance of providing only palmitate to cultured cells is debatable. There are two main reasons for questioning this approach. First, in vivo, cells are always exposed to a mixture of fatty acids. Second, most studies in cell culture models have been performed in the absence or near absence of carnitine, which is synthesized in liver but stored in most tissues. When provided to cells without other unsaturated fatty acids, palmitate serves as a relatively poor substrate for TAG biosynthesis; and when carnitine is limited its use as an oxidative substrate is severely impaired. These conditions represent a severe departure from normal physiology, and in our hands, cause necrotic and/or apoptotic cell death, most likely triggered by catastrophic elevations in palmitoyl-CoA, DAG and/or ceramide. We find that addition of either carnitine or oleate can rescue cell viability. For these reasons, we have turned to using a fatty acid mixture for our in vitro studies. Zhang: The diet you used decreased the PGC1α expression levels. In our hands, oleic acid decreases PGC1α expression in a couple of cell types, including skeletal muscle cells. Palmitic acid stimulates this expression. Muoio: Using either oleate or an oleate : palmitate mix, we have not observed fatty acid-mediated suppression of PGC1α in our cell culture system. In contrast, Mary Elizabeth Patti has presented data showing that exposure of myocytes to palmitate does reduce PGC1α levels. Again, I suspect that this response might be secondary to palmitate-induced cytotoxicity. Zhang: High glucose also decreased PGC1α expression in a couple of cell types, including β cells, skeletal muscle and vessel smooth muscle. Muoio: We have performed similar studies in primary human muscle cells and rat L6 myotubes; we find no evidence of nutrient-induced changes in PGC1α expression. Spiegelman: One thing I would add for anyone doing these kinds of experiments is that cultured muscle cells don’t induce the PGC1 co-activators very well. We have the sense that in vivo a lot of the signalling is given tone from the nerves. When we use primary muscle cells fresh from the animals we can see some interesting regulation that is completely absent in C2C12 and L6 cells. Whatever else is going on, nerve tone is what gets the PGC1 gene ready to respond to things. C2C12 cells are almost null for PGC1α . If you want to study this system, primary muscle is the way to go. Zhang: I have a question about exercise. Exercise seems to increase PGC1α expression in skeletal muscle. If you increase PGC1α in the liver, this increases insulin resistance, so did exercise decrease PGC1α in the liver, or does it stay at the same level? Muoio: We haven’t evaluated this. Spiegelman: This has been done by others. The induction in exercise is specific to the muscle. The stimuli and Ca2+ fluxes are tissue autonomous.
SKELETAL MUSCLE DYSFUNCTION
41
Muoio: The program initiated by PGC1α in different cell types is quite distinct. We can envision why there might be different phenotypes in each tissue. Spiegelman: Ultimately the coactivator is a slave to the transcription factors, not the other way round. The robust gluconeogenic response seen in the liver isn’t seen in other tissues. Muoio: An important question with regard to PGC1α activity is whether or not its localization is acutely regulated. There is surfacing evidence that exercise or other stimuli might drive translocation to the nucleus. Spiegelman: I have heard that through the grapevine, but I haven’t seen the data. The rumour I heard was that it was a splice isoform, not the parent PGC1, which is almost always in the nucleus. We always fi nd PGC1 in the nucleus under normal conditions: it is a nuclear protein. Kadowaki: You mentioned that the PPARα null mice are protected from a high fat diet-induced insulin resistance. Consistent with this you showed decreased incomplete fatty acid oxidation under fasting conditions in PPARα knockouts. But on the other hand, PPARα agonists have been reported to ameliorate high fat diet-induced insulin resistant diabetes. What are the effects of PPARα agonists and complete and incomplete fatty acid oxidation, under both fed and fasted conditions? Muoio: It is important to consider that in an animal system the PPAR agonists act on several different tissues. PPARα agonists primarily target liver. There is compelling evidence that enhancement of liver β oxidation in the context of obesity and/or overnutrition protects against systemic insulin resistance. This may not be true for skeletal muscle. An important distinction between the two tissues is that liver harbours a high potential to generate ketones, providing an efficient bleed-off system for excess fatty acids. This programme is much less active in skeletal muscle mitochondria. Hotamisligil: If you set up a lipid-induced insulin resistance model, whether this is in vivo or in vitro, once these lipids are in contact with the cells you would certainly activate some of the serine kinases that are also shown in your slide. You are ascribing the molecular mechanism of insulin resistance to an abnormality of the TCA cycle of lipid flux. Can you rule out a mitochondrion-independent direct activation of these kinases? In other words, in the model that you are using, if you examine PKC or JNK activity and then fix the mitochondrial defect, does that always correlate with the regulation of stress kinases? How did you conclude that some of these are primary or secondary events? Muoio: At present, we do not have a definitive answer. Experiments in muscle cells suggest that the insulin desensitizing effects of fatty acids require mitochondrial metabolism. In these experiments, lipid-induced impairment of insulin signalling was prevented when mitochondrial uptake and oxidation was inhibited by omission of carnitine or addition of the CPT1 inhibitor, etomoxir. We are evaluating the stress kinase pathways. Fatty acid treatment activates JNK; studies to
42
MUOIO & KOVES
determine whether lipid-induced activation of JNK or other kinases requires mitochondrial oxidation are underway. Spiegelman: What we have is a package of things all going on at once, and it’s hard to work out what is the chicken as opposed to the egg. Hotamisligil: Mitochondrial dysfunction will produce ROS, ROS will activate JNK, JNK will activate cytokines and cytokines will activate ROS. It’s a cycle. Muoio: The stress kinases are most certainly considered in our model. We propose that mitochondrial events feed into those pathways. Hotamisligil: If you could find a way to dissociate these events, this would be very important and a significant advance. Spiegelman: That’s tough. I’m not sure how they could be dissociated. Glass: It may be different in different tissues. For example, there is some evidence that fatty acids might directly activate Toll receptors. There, you wouldn’t need the mitochondria. You could get these responses directly through the MyD88 pathways, but in muscle I am not sure that this pathway would be relevant. Muoio: Emerging evidence does support a role for the Toll receptors in muscle. I support the view that insulin resistance originally evolved as a fundamental survival mechanism that permits glucose sparing during conditions of energy stress. Given this view, it is likely that nature has devised redundant but mechanistically distinct signalling events that can produce the insulin resistant phenotype. Some of these signals might depend heavily on mitochondrial energy metabolism as a barometer of the local environment, whereas others stem from receptor-mediated events that are largely driven by external factors. Bernlohr: Earlier someone commented that all the metabolic processes were interconnected. Many times this comes back to changes in NADH/NAD ratio. If there is a mechanism to maintain the balance so that NAD isn’t depleted, can you ascribe the differences in insulin resistance to mitochondrial effects on these? That is, is it simply that you are altering the coenzyme level sufficiently that metabolism is slowing down or resetting appropriately? So if you use an ATP or NAD regenerating system, can you drive complete lipid oxidation under those conditions? Muoio: That’s an important question. We have been looking for assays that will allow us to assess the redox state in real time. Measuring NADH levels in isolated mitochondria or tissues is probably not so meaningful. To my knowledge, these assays aren’t available yet. An interesting observation that relates to fluctuations in redox state is that PGC1α regulates a cluster of genes encoding enzymes that participate in the shuttling of reducing equivalents between the mitochondrial and the cytosolic compartments. These include both the cytosolic and mitochondrial isoforms of malate dehydrogenase and glutamate-oxaloacetate transaminase. These genes are upregulated by exercise training and are expressed more abundantly in red compared
SKELETAL MUSCLE DYSFUNCTION
43
to white muscle (Barron et al 1998, 1999, Hittel et al 2005). Glutamate-oxaloacetate transaminase catalyses the rate-limiting step of the malate–aspartate shuttle, which plays a principal role in permitting oxidation of cytosolic reducing equivalents that are produced by glycolysis and/or via the oxidation of lactate. Importantly, this system operates as a reversible shuttle that also allows mitochondrial export of reducing equivalents. This process is likely to occur during transitions from exercise to rest, when fatty acid supply and β -oxidative production of NADH exceeds ATP demand. A high NADH/NAD ratio slows oxidative phosphorylation at several steps. By dissipating the rapid intramitochondrial accumulation of reducing equivalents that could occur upon cessation of contractile activity, the malate shuttle might play a key role in preventing a complete shutdown of oxidative metabolism. Spiegelman: In a way, this all comes back to energy balance. No matter how powerful the PGC1s are, if the energy input exceeds the energy outflow you are removing electrons from lipids, and inevitably the NADH:NAD ratio must be changed. In the cell NAD will be reduced and the electron transport system will be relatively highly reduced whenever the oxidation exceeds ATP production or ATP turnover. There will be many consequences of this. Bernlohr: The other component here is heat production. Not all the electrons are captured. Have you had a chance to measure this? Muoio: No. Spiegelman: This is like proton leak. People were hot on the UCP2/3 story a couple of years ago: there has been a lot of interest in their role as an anti-ROS mechanism. This is not necessarily thermogenesis per se, but lowering the membrane potential does lower the production of ROS. Muoio: We have measured the potential of the mitochondria to respond to uncoupling stimuli in isolated mitochondrial preparations. In some models of obesity we find that both coupled and uncoupled respiratory capacity is impaired. We have not measured heat production. Shi: I have a question about the role of CPT1 in the cycle. There is an ongoing debate in drug companies about whether it would be best to develop a CPT1 agonist or antagonist. The first idea was to use an antagonist to prevent fatty acid entering the mitochondria, thus stimulating glucose oxidation as a way to lower blood glucose for the treatment of diabetes. Now the reverse is being tried, developing an agonist to stimulate fatty acid oxidation as a way to treat obesity. On the basis of your data it seems that mitochondria are already very tired metabolising fatty acid. There is an accumulation of acyl carnitine. What do you think of this? And have you ever tested those metabolites to see whether they have an effect on glucose stimulated insulin secretion? Muoio: It is unclear whether or not the acylcarnitines function in a signalling role. There are a few reports in the literature suggesting that acyl carnitine might
44
MUOIO & KOVES
modulate Ca2+ signalling (Yamada et al 2000). With regard to using CPT1 inhibitors as an antidiabetic therapy, one potentially adverse consequence in the whole animal model would be targeting of liver CPT1. There are several genetic mouse models showing that inhibition of liver β -oxidation can result in hepatic steatosis, obesity and whole-body insulin resistance. I understand that some companies are now developing partial CPT1 antagonists that would slow β oxidation. These compounds might hold promise. Spiegelman: You always have to come back to the first law of thermodynamics. Oxidising fat in the absence of other things is not going to be beneficial. Perhaps other things happen and there are feedback systems, but the idea of just burning fat doesn’t make sense. The electrons will not disappear, and if you can’t explain where they are going to go, be nervous. Perhaps there is a compensatory increase in proton leak or uncoupling and everything will work out fine. I am agnostic on this point. But stimulating fuel oxidation in the absence of figuring out where the energy in the high energy electrons is going to go is worrisome. Muoio: In the liver, excess reducing equivalents are used to support ketogenesis. The ketogenic enzyme, β hydroxyacyl-CoA dehydrogenase uses NADH to produce β hydroxybutyrate, the main ketone in circulation. Spiegelman: I’m not sure that is enough. Kadowaki: Why do you think that PPARα agonists have an insulin sensitising effect on the liver, even though it stimulates β oxidation of lipids? Spiegelman: That might be an anti-steatotic effect. If we step away from the energetics, it could be that liver steatosis is a bad thing. We can make a unifying hypothesis: on the one hand, activation of PPARα in the liver has a beneficial effect through anti-steatosis. It also causes cancer in the rats, which could be from ROS production. You are getting β oxidation that exceeds ATP utilization, with an electron leak into the ROS, but you are also getting rid of the steatosis. Kadowaki: Intramuscular triglyceride correlates well with insulin resistance. So a PPARα agonist may also have dual effects, with beneficial effects to ameliorate triglyceride accumulation in the skeletal muscle but on the other hand it may be harmful by increasing incomplete fatty acid oxidation products. Spiegelman: Perhaps all these things occur. There could be oxidative stress, increased ROS, compensatory anti-ROS systems, scavenging enzymes and uncoupling proteins. So maybe it is OK. The body can deal with the oxidative stress. I think the liver is a nice system in that it does show both things. If we give a PPAR agonist to an obese diabetic mouse, we improve the diabetes. But it also increases oxidative stress, and causes peroxisome proliferation and hepatocarcinoma. Muoio: In the animal model the PPAR agonists are largely targeting liver β oxidation; thus the liver functions as a sink for excess fat. This appears to relieve the lipid stress on muscle.
SKELETAL MUSCLE DYSFUNCTION
45
Spiegelman: I think that steatosis is a separate issue from the energy balance question. There is good evidence that anti-steatotic factors are good, on balance. Shi: The animal model fits with the notion of developing PPARα agonists as a way to lower free fatty acids. Spiegelman: Yes, that is where the drugs came from. Hotamisligil: You made a comment on why the short-term (as in not combined with the ageing model) lipid-induced insulin resistance was different. You mentioned that the age-related defect was different from the short-term high fat feeding. Why was this the case? Could you please expand? Muoio: In the chronic feeding model we can’t separate the ageing component from the fact that the animals are on a diet for a year. The goal of that study was to understand the consequences of having animals on a diet until midlife. One of the distinctions between the short and long term diet treatments was the impact on mitochondrial potential to completely oxidize fatty acids. With the short term diet, we found a preferential increase in incomplete oxidation but no change in complete fat oxidation to CO2 . When animals were on the diet for a year, this disconnect was exacerbated; incomplete oxidation was even more severe and there was also a marked reduction in mitochondrial capacity to oxidise fuel, particularly the fatty acids. This may have relevance to the progression of human obesity: in moderately overweight humans, we often observe an increase in fat oxidation measured by the respiratory exchange ratio. However, as obesity becomes more severe, several groups including ours have shown that muscle capacity to oxidise fat is compromised in manner reminiscent of our animal experiments. So perhaps long term exposure to energy overload eventually causes the mitochondria to fail. Hotamisligil: Is this an adaptive mechanism, at least up to a certain point? Muoio: It could be. Early adaptations to obesity may drive the β -oxidative programme to compensate for an increase in the supply of lipid fuel. With chronic exposure, the heavy load imposed on the β -oxidative pathway might promote oxidative stress. Perhaps eventual down-regulation of mitochondrial oxidation reflects an adaptation to combat some of that stress. Shi: If you repeat your data with adipocytes, what would happen? There is some evidence that if you give patients TZDs, it stimulates adipocyte metabolism. Muoio: I believe you are asking whether or not mitochondrial metabolism of fatty acids play any role in adipocyte function and cytokine secretion? That’s an interesting question that we are just beginning to address. Attie: I’d like to describe an experiment we did with Chris Newgard to ask a question about whether we can also find evidence for coordinate regulation, not in terms of a treatment effect, such as diet or exercise, but in terms of a genetic dimension. We took 288 F2 mice, segregating for obesity and diabetes. In effect, we were maximizing genetic variance across 288 animals. We performed unsupervised clustering of 67 liver metabolites, and asked whether they correlated in any
46
MUOIO & KOVES
way that made biological sense when maximal genetic variance was introduced in the sample. The answer was yes, and to a great extent. We obtained distinct clusters of amino acids, TCA cycle intermediates, fatty acids of intermediate chain length, and hydroxylated fatty acid β -oxidation intermediates. This tells us that there is coordinate regulation across the genetic dimension. We can now ask whether there are major gene loci that control these. There are some really big hot spots on various chromosomal locations. One we are focusing on is on chromosome 2, because this is where we also map insulin resistance. We have congenic mice to see whether or not this is really the same locus where we are seeing some fatty acid and amino acid phenotypes. The take-home lesson is that this metabolic profiling could be a useful tool in genetics as well. It is one thing to ask about the abundance of a transcription factor and another to ask about the activity of a transcription factor. Ultimately this could be a readout for assessing the activity of transcription factors across a genetic dimension. Hotamisligil: How does this compare with a transcriptional profi le? Instead of a metabolite profi le, if you put transcriptional profi les in those congenics, can you still identify hotspots? Attie: 60 of these 288 mice were microarray profi led, and the same hotspot came up in the transcription profi les. Overlaying the metabolic profi les on top of the mRNA profi les has turned out to be a difficult problem statistically, but it is something we are doing. We’re going to do a much larger experiment that looks at five tissues. References Barron JT, Gu L, Parrillo JE 1998 Malate-aspartate shuttle, cytoplasmic NADH redox potential, and energetics in vascular smooth muscle. J Mol Cell Cardiol 30:1571–1579 Barron JT, Gu L, Parrillo JE 1999 Relation of NADH/NAD to contraction in vascular smooth muscle. Mol Cell Biochem 194:283–290 Berthiaume JM, Oliveira PJ, Fariss MW, Wallace KB 2005 Dietary vitamin E decreases doxorubicin-induced oxidative stress without preventing mitochondrial dysfunction. Cardiovasc Toxicol 5:257–267 Hittel DS, Kraus WE, Tanner CJ, Houmard JA, Hoffman EP 2005 Exercise training increases electron and substrate shuttling proteins in muscle of overweight men and women with the metabolic syndrome. J Appl Physiol 98:168–179 Schrauwen P, Hesselink MK 2004 Oxidative capacity, lipotoxicity, and mitochondrial damage in type 2 diabetes. Diabetes 53:1412–1417 Yamada KA, Kanter EM, Newatia A 2000 Long-chain acylcarnitine induces Ca 2+ efflux from the sarcoplasmic reticulum. J Cardiovasc Pharmacol 36:14–21
Fatty Acids and Lipotoxicity in Obesity and Diabetes: Novartis Foundation Symposium 286, Volume 286. Edited by Gregory Bock and Jamie Goode Copyright Novartis Foundation 2007. ISBN: 978-0-470-05764-3
Stearoyl-CoA desaturase deficiency, hypercholesterolaemia, cholestasis and diabetes1,2 Alan D. Attie*, Matthew T. Flowers*, Jessica B. Flowers*†, Albert K. Groen‡, Folkert Kuipers§ and James M. Ntambi*† * Departments of Biochemistry, † Nutritional Sciences, University of Wisconsin-Madison, Madison, WI, USA, ‡ Academic Medical Center, Amsterdam, and § Department of Pediatrics, Center for Liver, Digestive and Metabolic Diseases, Universtiy Medical Center Groningen, Groningen, The Netherlands
Abstract. Previous studies have shown that mice deficient in Scd1 have a dramatically reduced level of liver triglyceride and an improvement in insulin sensitivity. The mice are lean and partially protected from obesity induced by leptin deficiency or high fat diets. These results predicted that Scd1−/− mice would be protected from the increased serum triglyceride levels that result from a lipogenic diet and might also might protect a diabetes-susceptible obese mouse strain from diabetes. We studied Scd1−/− mice on a very low-fat high-carbohydrate lipogenic diet. The animals were almost entirely devoid of high density lipoprotein (HDL). Nonetheless, they were hypercholesterolaemic due to a large increase in low density lipoprotein (LDL). They had a high level of serum bilirubin and bile acids, and the appearance of lipoprotein-X, indicative of cholestasis. These changes were reversible with oil containing both mono- and polyunsaturated fat, but not entirely reversible with just triolein, suggesting that Scd1 deficiency increased the requirement for polyunsaturated fat. We performed hyperinsulinemic-euglycemic clamp experiments and found that the Scd1−/− mice on a normal chow diet had dramatically improved insulin sensitivity. But, surprisingly, leptinob/ob Scd1−/− mice had worse diabetes than leptinob/ob Scd1wt/wt mice. Thus, despite its effects on insulin sensitivity, Scd1 deficiency worsens diabetes in leptin-deficient obese mice. 2007 Fatty acids and lipotoxicity in obesity and diabetes. Wiley, Chichester (Novartis Foundation Symposium 286) p 47–57 1
This article has been published separately with permission from John Wiley & Sons, Ltd., as Attie AD, Flowers MT, Flowers JB, Groen AK, Kuipers F, Ntambi JM 2007 Stearoyl-CoA desaturase activity, hypercholesterolemia, cholestasis, and diabetes. Nutr Rev 65:S35–38. 2 Some of the data discussed in this chapter have been published in Flowers JB, Rabaglia ME, Schueler KL et al 2007 Loss of stearoyl-CoA desaturase-1 improves insulin sensitivity in lean mice but worsens diabetes in leptin-deficient obese mice. Diabetes 56:1228–1239, Flowers JB, Oler AT, Nadler ST et al 2007 Abdominal obesity in BTBR male mice is associated with peripheral but not hepatic insulin resistance. Am J Physiol Endocrinol Metab 292:E936–945, and Flowers MT, Groen AK, Oler AT et al 2006 Cholestasis and hypercholesterolemia in SCD1-deficient mice fed a low-fat, high-carbohydrate diet. J Lipid Res 47:2668–2680. 47
48
ATTIE ET AL
Stearoyl-CoA desaturase (Scd1) catalyses the introduction of a double bond at the D9 position of palmitoyl-CoA (16 : 0) and stearoyl-CoA (18 : 0) to produce the monounsaturated fatty acids (MUFAs) palmitoleoyl-CoA (16 : 1) and oleoyl-CoA (18 : 1), respectively (Ntambi & Miyazaki 2003). The dietary requirement for MUFA is difficult to ascertain because of the capacity for MUFA synthesis conferred by Scd1. Scd1 is widely expressed, particularly in lipogenic tissues such as liver and adipose tissue. Like other lipogenic genes, it is positively regulated by insulin and leptin and by several known lipogenic transcription factors, sterol responsive element binding protein 1c (SREBP-1c) and carbohydrate responsive element binding protein (ChREBP). In addition, it is positively regulated by the liver-X-receptor (LXR). High-carbohydrate diets increase the expression of lipogenic enzymes, including Scd1 and the rate of triglyceride synthesis in the liver. This induction is attenuated in mice deficient in Scd1 (Miyazaki et al 2001, 2004). Scd1 deficiency was reported to reduce serum triglycerides (Attie et al 2002). However, we have found that in some mouse strains Scd1 deficiency, although it still reduces hepatic triglyceride levels, does not reduce serum triglycerides. These studies have all been conducted in animals receiving an ample supply of polyunsaturated fat. Long-term feeding of Scd1-deficient animals with lipogenic diets containing a minimum amount of polyunsaturated fat had not been tested. In addition to its effects on lipid metabolism, Scd1 deficiency has dramatic effects on adipogenesis (Ntambi et al 2002, Cohen et al 2002) and insulin sensitivity (Rahman et al 2003). We have previously speculated that hepatic lipogenesis, through competition for some gluconeogenic substrates, might decrease the rate of gluconeogenesis (Nadler & Attie 2001). Thus, the effect of Scd1 deficiency on obesity-induced diabetes could have both an anti-diabetic effect (increased insulin sensitivity) and a pro-diabetic effect (increased gluconeogenesis). We have studied obesity-induced diabetes using two mouse strains that, when made obese with the leptinob mutation, differ in diabetes susceptibility. As originally shown in the classic studies of Coleman (1978), we have found that the C57BL/6 strain is transiently and mildly hyperglycaemic when it carries the leptinob mutation. In contrast, the BTBR leptinob/ob mice are severely diabetic (Stoehr et al 2000). Gene expression studies carried out in these two strains revealed that the level of lipogenic gene expression, including Scd1, was higher in the C57BL/6 leptinob/ob mice than the BTBR leptinob/ob mice (Lan et al 2003). Similar studies in lipodystrophic mice showed a positive correlation between lipogenic gene expression, hepatic triglyceride content, and resistance to diabetes (Colombo et al 2003). Another angle on strain differences emphasized triglyceride clearance rather than triglyceride production and secretion. Studies comparing
SCD1 DEFICIENCY
49
leptin-deficient obese or lipodystrophic C57BL/6 and FVB strains found that increased triglyceride clearance in the C57BL/6 strain, effectively diverted triglyceride away from muscle (Colombo et al 2003). These studies concluded that this change in triglyceride partitioning would tend to improve whole-body insulin sensitivity even though it would worsen hepatic insulin sensitivity. Results and discussion Very low fat/high carbohydrate diet and Scd1 deficiency Scd1−/− C57BL/6 mice were obtained by backcrossing 129/sv mice carrying the null mutation into the C57BL/6 background for at least five generations. The mice were fed a standard chow diet (PMI 5008; Formulab) or very low fat (VLF; TD03045), VLF-coconut (TD03138), or VLF-canola (Harlan Teklad). When maintained on the VLF diet, the Scd1−/− mice developed hypercholesterolaemia. Their cholesterol levels were elevated 250% whereas cholesterol only increased by 20% in the wild-type animals on the same diet. Despite the increase in total cholesterol, the Scd1−/− mice had a nearly complete loss of HDL. The hypercholesterolaemia was due to a large increase in VLDL and LDL. In addition, by agarose gel electrophoresis, we were able to detect lipoprotein-X. To better understand the basis for the increase in LDL, we measured the rate of clearance of a [125I]LDL tracer from the bloodstream of the mice. The fractional clearance rate of the LDL tracer was reduced by 50% in the VLF Scd1−/− mice. The magnitude of the decrease was approximately commensurate with the increase in LDL, indicating that reduced clearance was the major mechanism underlying the increase in LDL. We also estimated triglyceride secretion by measuring the increment in serum triglyceride following the injection of poloxamer 407 to inhibit lipoprotein lipase (Millar et al 2005). Although liver triglyceride levels were reduced by 82% in the VLF Scd1−/− mice, there was no difference in triglyceride secretion between these mice and the wild-type mice on the same diet. In addition to the lipid disorders, the VLF Scd1−/− mice had a 50-fold elevation in serum bile acids and a sixfold increase in bilirubin, mostly conjugated, all consistent with cholestasis. In addition, the presence of lipoprotein-X is symptomatic of cholestasis (Havel 1971). Despite cholestasis, there was no defect in bile flow or bile salt secretion. However, biliary phospholipids were decreased substantially (by 44%). Both the lipoprotein and the cholestasis phenotypes were reversible by supplementing the VLF diet with canola oil, but not with coconut oil, suggesting that mono- and/or polyunsaturated fat deficiency was an underlying problem in the Scd1−/− mice on the VLF diet. To better understand which lipid source was critical
50
ATTIE ET AL
for this correction, we supplemented VLF mice with triolein in order to provide a pure source of MUFA. The triolein supplementation was able to abolish cholestasis, improve hepatic function, and restore HDL cholesterol to near normal levels. However, triolein did not reduce the very high levels of LDL in the VLF Scd1−/− mice. The latter defect could only be reversed by polyunsaturated fat. The increased cholesterol in the bloodstream was primarily in the form of free cholesterol, consistent with the accumulation of lipoprotein-X. This is also the case in animals with a deficiency of lecithin : cholesterol acyltransferase (LCAT). The Scd1−/− mice on the VLF diet had a 43% reduction in LCAT activity. Although this helps to account for an increase in free cholesterol, it does not explain the complete loss of HDL because animals heterozygous for a null allele at Lcat have half the normal level of HDL (Sakai et al 1997). We carried out microarray analysis of gene expression in the liver to investigate which genes are dysregulated under the conditions of Scd1 deficiency and the VLF diet. There were many genes whose expression changed as predicted. But, unexpectedly, the expression of many of the genes involved in lipid transport across the canalicular membrane was not significantly changed. The bile salt export pump (ABCB11/BSEP) mRNA abundance was reduced by one-half in the Scd1−/− mice on the VLF diet. It is possible that the balance of mono- and polyunsaturated fats has a greater effect on the activity of lipid transporters than it does on their level of expression.
Scd1 and diabetes susceptibility in leptin-deficient obese mice We carried out hyperinsulinaemic–euglycaemic clamp studies to quantitate the level of insulin sensitivity in lean Scd1−/− mice. In this procedure, insulin levels are raised and glucose is infused to maintain euglycaemia. The glucose infusion rate required to maintain euglycaemia is a measure of insulin sensitivity. The Scd1−/− mice had a striking threefold increase in the glucose infusion rate, indicating a large increase in whole-body insulin sensitivity (Fig. 1). By measuring the accumulation of [14C]deoxyglucose, we were able to quantitate glucose uptake into muscle and adipose tissue. Heart and soleus muscle had an increase of about 2.5-fold in glucose uptake, but overall, the uptake of glucose into the heart became quite dominant over all other insulin sensitive tissues. The liver also showed an increased level of insulin sensitivity as evidenced by an almost total suppression of hepatic glucose output under the hyperinsulinaemic conditions. We introgressed the Scdnull allele into BTBR leptinob/ob mice, which normally develop diabetes between 6 and 10 weeks of age (Fig. 1). The effect of the Scd1 mutation was to increase the severity of the hyperglycaemia. At 6 weeks of age, mean glucose levels in obese male mice was 474 in the Scd1-deficient mice vs. 272 mg/dl in the Scd1wt/wt mice (P < 0.0001). In females, the corresponding values
SCD1 DEFICIENCY
51
FIG. 1. (Upper) Consequences of Scd1 deficiency in lean mice. The loss of Scd1 leads to increased insulin sensitivity. This in turn causes increased muscle glucose uptake and decreased hepatic glucose output. In addition, the mice have a profound loss of adipose tissue mass. (Lower) Consequences of Scd1 deficiency in leptinob/ob mice. A large fraction of the pancreatic islets in these mice have an abnormal morphology. These islets have a dramatic induction in Lpl and Cd36. Thus, we speculate that increased uptake of saturated fatty acids leads to lipotoxicity of the β -cells.
were 344 and 214 mg/dl (P < 0.002). Insulin levels dropped by about 35% in both sexes. Although a reduction in fasting insulin can be a sign of increased insulin sensitivity, in this case, because of the rise in glucose, it appeared to be more likely an indication of reduced insulin secretion. Indeed, when challenged with an intraperitoneal glucose bolus, the mice secreted less than one-half as much insulin as did BTBR leptinob/ob Scd1wt/wt mice.
52
ATTIE ET AL
Insufficient insulin secretion can be a consequence of an intrinsic defect in the process by which β -cells sense glucose and secrete the appropriate amount of insulin. Alternatively, it can be caused by a reduction in β -cell mass. When we isolated islets from the mice, we noted that about 50% of the islets in the BTBR leptinob/ob Scd1−/− mice were small and had an opaque circle in the centre, giving them a ‘fried egg’ appearance. These islets had a >80% reduction in insulin content. Thus, the insulinopaenia of the BTBR leptinob/ob Scd1−/− mice is due to the presence of the abnormal islets and their loss of insulin. Monounsaturated fatty acyl-CoAs are the preferred substrates for incorporation of fatty acids into the sn-2 position of glycerolipids and cholesterol esters. Thus, the production of monounsaturated fatty acids can be limiting for the amount and type of these lipids. In addition, to the extent that palmitate might affect the production of ceramide and sphingolipids, desaturation of palmitate can play a role in the flux through the sphingolipid pathway. As expected, when an animal has a reduction in MUFA synthesis, tissues and serum showed an increase in palmitate and stearate relative to palmitoleate and oleate. The abnormal islets in the BTBR leptinob/ob Scd1−/− mice had threefold more palmitate and stearate as free fatty acid and fourfold more palmitate in triglyceride than the normal-looking islets. We surveyed gene expression by quantitative RT-PCR. The most striking changes in gene expression in the abnormal vs. the normal-looking islets were in Lpl and Cd36. Lpl was up-regulated about 10-fold and Cd36 about 167-fold in the abnormal islets. Both of these changes would be expected to increase the lipid load on the β -cells, through enhanced lipolysis of lipoprotein triglyceride and through enhanced fatty acid transport into the cells. In vitro, chronic exposure of β -cells to fatty acids alters secretory function and induces apoptosis (Shimabukuro et al 1998, Lambert et al 2001, Maedler et al 2003, Maestre et al 2003, Piro et al 2002, El-Assaad et al 2003). Thus, the combination of increased saturated fatty acid availability and increased fatty acid transport can explain a potential lipotoxic effect of Scd1 deficiency on the islets.
Acknowledgements The work was supported by NIDDK grants DK58037 and DK66369, and NHLBI grant HL56593.
References Attie AD, Krauss RM, Gray-Keller MP et al 2002 Relationship between stearoyl-CoA desaturase activity and plasma triglycerides in human and mouse hypertriglyceridemia. J Lipid Res 43:1899–1907
SCD1 DEFICIENCY
53
Cohen P, Miyazaki M, Socci ND et al 2002 Role for Stearoyl-CoA desaturase-1 in leptinmediated weight loss. Science 297:240–243 Coleman DL 1978 Obese and diabetes: two mutant genes causing diabetes-obesity syndromes in mice. Diabetologia 14:141–148 Colombo C, Haluzik M, Cutson JJ et al 2003 Opposite effects of background genotype on muscle and liver insulin sensitivity of lipoatrophic mice. Role of triglyceride clearance. J Biol Chem 278:3992–3999 El-Assaad W, Buteau J, Peyot ML et al 2003 Saturated fatty acids synergize with elevated glucose to cause pancreatic beta-cell death. Endocrinology 144:4154–4163 Havel RJ 1971 The abnormal lipoprotein of cholestasis. N Engl J Med 285:578–579 Lambert G, Sakai N, Vaisman BL et al 2001 Analysis of glomerulosclerosis and atherosclerosis in lecithin cholesterol acyltransferase-deficient mice. J Biol Chem 276:15090– 15098 Lan H, Rabaglia ME, Stoehr JP et al 2003 Gene expression profi les of nondiabetic and diabetic obese mice suggest a role of hepatic lipogenic capacity in diabetes susceptibility. Diabetes 52:688–700 Maedler K, Oberholzer J, Bucher P, Spinas GA, Donath MY 2003 Monounsaturated fatty acids prevent the deleterious effects of palmitate and high glucose on human pancreatic beta-cell turnover and function. Diabetes 52:726–733 Maestre I, Jordan J, Calvo S et al 2003 Mitochondrial dysfunction is involved in apoptosis induced by serum withdrawal and fatty acids in the beta-cell line INS-1. Endocrinology 144:335–345 Millar JS, Cromley DA, McCoy MG, Rader DJ, Billheimer JT 2005 Determining hepatic triglyceride production in mice: comparison of poloxamer 407 with Triton WR-1339. J Lipid Res 46:2023–2028 Miyazaki M, Kim YC, Ntambi JM 2001 A lipogenic diet in mice with a disruption of the stearoyl-CoA desaturase 1 gene reveals a stringent requirement of endogenous monounsaturated fatty acids for triglyceride synthesis. J Lipid Res 42:1018–1024 Miyazaki M, Dobrzyn A, Man WC et al 2004 Stearoyl-CoA desaturase 1 gene expression is necessary for fructose-mediated induction of lipogenic gene expression by sterol regulatory element-binding protein-1c-dependent and -independent mechanisms. J Biol Chem 279:25164–25171 Nadler ST, Attie AD 2001 Please pass the chips: genomic insights into obesity and diabetes. J Nutr 131:2078–2081 Ntambi JM, Miyazaki M 2003 Recent insights into stearoyl-CoA desaturase-1. Curr Opin Lipidol 14:255–261 Ntambi JM, Miyazaki M, Stoehr JP et al 2002 Loss of stearoyl-CoA desaturase-1 function protects mice against adiposity. Proc Natl Acad Sci USA 99:11482–11486 Piro S, Anello M, Di Pietro C et al 2002 Chronic exposure to free fatty acids or high glucose induces apoptosis in rat pancreatic islets: possible role of oxidative stress. Metabolism 51:1340–1347 Rahman SM, Dobrzyn A, Dobrzyn P et al 2003 Stearoyl-CoA desaturase 1 deficiency elevates insulin-signaling components and down-regulates protein-tyrosine phosphatase 1B in muscle. Proc Natl Acad Sci USA 100:11110–11115 Sakai N, Vaisman BL, Koch CA et al 1997 Targeted disruption of the mouse lecithin:cholesterol acyltransferase (LCAT) gene. Generation of a new animal model for human LCAT deficiency. J Biol Chem 272:7506–7510 Shimabukuro M, Zhou YT, Levi M, Unger RH 1998 Fatty acid-induced beta cell apoptosis: a link between obesity and diabetes. Proc Natl Acad Sci USA 95:2498–2502 Stoehr JP, Nadler ST, Schueler KL et al 2000 Genetic obesity unmasks nonlinear interactions between murine type 2 diabetes susceptibility loci. Diabetes 49:1946–1954
54
ATTIE ET AL
DISCUSSION Spiegelman: In terms of the expression profi ling correlating with genetics, a concern is that these things could be rather direct or extremely indirect. For example, if insulin was elevated in some but not all of those things, having some alterations in the insulin pathway would play out in some of these regulatory schemes as well. Do you have any clever way of crossing the patterns of some genes against other genes, to weed out different strain and chromosomal influences on broad phenomenon such as hyperinsulinaemia, as opposed to what I assume you are looking for—something more selective in the Scd1 pathway? Attie: As complicated a question that is, it is even more complicated a question than you think it is! It raises some important statistical questions. We think about this quite a bit. It also is related to the issue of directionality. You can construct networks on the basis of a correlation matrix, where various phenotypes are correlated, but it doesn’t tell you what the directionality is. You get a bit of help when you start with genetic variation, because at the very least the first arrow from genetic variation can only go in one direction. This is one of our ground points. It gets more difficult if you have a metabolite and an mRNA: the mRNA could be regulated by the metabolite or vice versa. We thus have to set up networks that propose testable experiments. The other thing we can do is to isolate the gene loci with congenic strains. Those are valuable tools. Hotamisligil: The cross of Ob/Ob and Scd1-deficiency in the B6 background is reported to be highly beneficial for all of the measured parameters, including obesity, hepatosteosis and insulin sensitivity. If you suppress Scd1 in the liver by RNAi, this also results in a beneficial phenotype in high fat-fed mice. Are the two different phenotypes that you see resulting from Scd1 deficiency a function of where Scd1 is active primarily in those two strains? I am cross-referencing here. When you bred Scd1 deficiency into your diabetic Ob model, this made things worse, whereas everything else seems to be making things much better. Attie: In lean mice it made things better, but in the obese mice it made things worse. Hotamisligil: In the Ob/Ob cross that Jeff Friedman published (Cohen et al 2002), things appeared to be much better. Attie: He didn’t publish any glucose data in that paper. They were leaner. The RNAi knockdown was also with lean mice. It was using a high fat diet, but the mice weren’t obese. The purpose of that study was primarily to look at insulin sensitivity, and it improved, just as you’d expect. Hotamisligil: That result is understandable. There are other activities of Scd1 elsewhere in the body that might be important for systemic metabolism. Attie: The point is that the experiments you are referring to with the knockdown did replicate those phenotypes. There is something we don’t understand here.
SCD1 DEFICIENCY
55
O’Rahilly: I have a question regarding the marked increase in energy expenditure in the asebia Scd1 knockout. How much is the dermatological phenotype contributing to heat loss? Attie: We were concerned about the heat loss. The graduate student in my lab who performed this experiment, Jonathan Stoehr, made sure he kept the animals warm enough. They were above ambient temperature, so I don’t think it is due to the ambient temperature. O’Rahilly: I was interested in seeing the bile acid story in relation to Johan Auwerx’s recent work showing bile acids being thermogenic in peripheral tissues, via thyroid hormone de-iodination. Is there any play into that story? Attie: That’s an interesting idea: I haven’t thought about that connection. Yang: Have you measured DAG or ceramide in the Scd1 knockout? Attie: We have measured the expression of serine palmitoyl transferase and it is down-regulated, not up-regulated. Thus we wouldn’t expect an increase in ceramide. Yang: Is there any similarity in genotype between your knockout mice and lipodystrophy mice? Attie: It is the dramatic difference that is interesting here. They have a dramatic loss of adipose tissue, their leptin levels are down and they are hyperphagic just like lipodystrophic animals. These animals eat a tremendous amount, perhaps three times more than a normal mouse. Yet, they are insulin sensitive. Spiegelman: Are they lipodystrophic or just extremely lean? O’Rahilly: Almost by definition they are lean. Spiegelman: They sound like a lean animal, not a lipodystrophic one. Lipodystrophic animals usually have some fundamental disruption of fat cells or fat cell biology, whereas you can be extremely lean but not lipodystrophic. Attie: At a cellular level I don’t know how to make the distinction. You also talk about heterozygotes for lipodystrophy, don’t you? They are not totally devoid of adipose tissue but they are still called lipodystrophic. Spiegelman: Lipodystrophic animals almost always have hepatomegaly or some disruption of the liver function, so that the level of fat doesn’t match the energy balance. If the level of fat matches the energy balance, I would call that a lean mass. O’Rahilly: I would agree that it is universal in all lipodystrophic cases where we know the primary cause. However, we are occasionally faced with humans where it is difficult to know whether or not they are lean. Once you know where you start from, then you can say this is what happens with lipodystrophy. Sometimes, you see an animal or a human and it is not that easy to figure out. Fat cells that don’t have fat in them look a lot like fibroblasts. Adiponectin levels can be a clue, because the smaller the fat cell gets the more adiponectin it makes, by and large. Do you know the adiponectin levels in your animals?
56
ATTIE ET AL
Attie: No. O’Rahilly: Lipodystrophic animals tend to have low adiponectin. Spiegelman: But if there is a disruption between the amount of fat that is supposed to be stored and the amount that is stored, because of an insufficient fat mass, almost invariably the liver is going to be in the loop. I am not a clinician, but I have heard enough lectures on lipodystrophy to know that hepatomegaly and hepatosteatosis are a signature for lipodystrophy. O’Rahilly: It is conceivable that an animal could have an inability to make fat and also a bizarre metabolic dysregulation that gave it a high energy output too. They are independent variables so it is possible that by screwing up lipid synthesis rather than adipocyte formation it is more likely to be leanness rather than lipodystrophy. Spiegelman: There is no insulin resistance, there is no hepatomegaly and no hepatosteatosis. Attie: If that is the definition of lipodystrophy then we can’t call it that. Someone mentioned earlier the concept that a depot is needed for the excess fat; if not, it will spill over somewhere, either in the form of steatosis or lipotoxicity. In the case of the lean mice, we don’t have this: we see loss of adipose tissue (not a total loss), and yet the animals are more insulin sensitive and they have less hepatic triglyceride. Spiegelman: I have to go back to Howard Green’s earliest papers on fat cell differentiation. He showed in 1976 that you could have perfectly normal fat cell differentiation with no fat accumulation. They are independent variables. There could be differentiated adipose tissue with no fat in it. Morphologically you wouldn’t know it is fat. Fat cell differentiation doesn’t require lipid synthesis or accumulation. Attie: Then how do you explain the convergence of the lipodystrophic syndromes that are apparently a consequence of the loss of lipogenic enzymes such as monoacyl glyceroltransferase? Would you call the DGAT mouse lyphodystrophic? Spiegelman: I am not sure I would. Hotamisligil: I would not either. It seems like a fine animal. O’Rahilly: AGPAT2 deficiency in humans causes lipodystrophy. Lipodystrophy happens whenever there is an impaired ability to store triacylglycerol normally in adipocytes under circumstances where energy intake habitually exceeds energy expenditure even by a small amount Spiegelman: I don’t doubt that if you screw up enzymes in fat metabolism you could end up with fat cell apoptosis, or make toxic intermediate products. After all, if you have a deficiency in haem biosynthesis you can get death of liver cells. This doesn’t mean that haem is required for liver differentiation, but it can mess it up.
SCD1 DEFICIENCY
57
Attie: A premise in this discussion is that the enzymes have one function. I think it is questionable: we can’t assume that they only have one function. Hotamisligil: I want to ask about the two types of islets. Could there be a genetic basis for this, or is it just the penetrance of the phenotype? Attie: We do a lot of work on heterogeneity of islets. We have been looking at the massive induction of what we have now discovered as a growth factor in islets, cholecystokinin. The expression is non-uniform—it is not in all islets. We see lots of evidence for heterogeneity within islets. Their orientation with respect to the microvasculature is non-uniform, so there are many explanations for how there could be heterogeneity in islets that is not necessarily genetic. Shi: If you give patients SCD1 inhibitors, would this exacerbate the dyslipidaemia? Attie: The lean mice that don’t have other problems seem to be better off when they don’t have Scd1, so lots of drug companies are working on this as a target. But there is a lot of genetic variation in humans. Some of the extreme cases I showed could occur in some people. I’d be concerned about those individuals. Zhang: For type A or type 2 islets, you used Ob/Ob islets as a control. These are not normal. Before 40 weeks of age they have a β cell dysfunction. They show impaired glucose-stimulated insulin secretion. Attie: It depends what strain you are studying. The B6 has, as a strain, a defect in insulin secretion. It is the nicotinamide nucleotide transhydrogenase gene that is affected (Freeman et al 2006). Apart from that they are extremely diabetes resistant because they are able to stimulate β -cell proliferation and greatly expand their β -cell mass. If you use that as the control, the obese mouse doesn’t have a loss of that function in the B6 strain. It is quite robust. Zhang: If you compare what is going on in type B islets with the lean mice, what do you see? Attie: The lean islets were somewhat similar to the Ob. They don’t have the induction of CD36 and don’t show the changes in many other genes that we studied. This was unique to the loss of Scd1 and in the subpopulation of the islets with the loss Scd1. References Cohen P, Miyazaki M, Socci ND et al 2002 Role for stearoyl-CoA desaturase-1 in leptinmediated weight loss. Science 297:240–243 Freeman H, Shimomura K, Horner E et al 2006 Nicotinamide nucleotide transhydrogenase: a key role in insulin secretion. Cell Metab 3:35–45
Fatty Acids and Lipotoxicity in Obesity and Diabetes: Novartis Foundation Symposium 286, Volume 286. Edited by Gregory Bock and Jamie Goode Copyright Novartis Foundation 2007. ISBN: 978-0-470-05764-3
The role of lipin 1 in adipogenesis and lipid metabolism Karen Reue Department of Human Genetics, David Geffen School of Medicine, University of California, Los Angeles, CA 90095, USA
Abstract. In conditions such as obesity and lipodystrophy, insulin resistance and diabetes are associated with the dysregulation of lipid metabolism in adipose tissue, skeletal muscle and liver. One factor that influences lipid homeostasis in all three of these tissues is lipin 1. Studies in mouse models and humans have established a relationship between lipin 1 levels and both triglyceride storage in adipose tissue and insulin sensitivity. Thus, lipin 1 deficiency in the mouse results in lipodystrophy and insulin resistance. In contrast, fat-specific expression of a lipin 1 transgene results in increased triglyceride storage in adipose tissue and insulin sensitivity, despite the development of obesity. In humans, lipin 1 expression levels in adipose tissue are positively correlated with insulin sensitivity, and inversely correlated with inflammatory cytokine expression and intramyocellular lipid, a key marker of insulin resistance. These data from mouse and human studies suggest a role for lipin 1 in directing lipid to the appropriate storage site in adipose tissue, thus potentially reducing lipid accumulation in tissues such as muscle and liver. The mechanism of lipin 1 action appears to be complex, with evidence for roles in triglyceride biosynthesis and in the regulation of gene expression. 2007 Fatty acids and lipotoxicity in obesity and diabetes. Wiley, Chichester (Novartis Foundation Symposium 286) p 58–71
Obesity, insulin resistance, and associated pathologies have become increasingly prevalent in many populations throughout the world. It is therefore a priority to understand the physiological, biochemical and molecular processes that lead to altered metabolism in these conditions. Dysregulation of lipid metabolism in tissues such as adipose tissue, skeletal muscle, liver and pancreatic β cells are one contributing mechanism. Previous work has suggested that adipose tissue is a ‘safe’ site for fat storage, and that sequestration of fat specifically in adipose tissue may protect other tissues from toxic reactive lipid species that may impair cellular function or lead to apoptosis (reviewed in Unger 2003, Eldor & Raz 2006, Slawik & Vidal-Puig 2006). The importance of adipose tissue in metabolic homeostasis is underscored by the development of insulin resistance and diabetes when adipose tissue lipid storage is compromised. This occurs in obesity, where fat cell storage 58
LIPIN 1
59
capacity is exceeded, and in lipodystrophy, due to the inability of adipocytes to store fat (Garg & Misra 2004, Hegele & Pollex 2005, Petersen & Shulman 2006, Reue & Phan 2006). One factor that influences fat storage in adipose tissue is lipin 1. Indeed, alterations in lipin 1 expression levels in the mouse can produce extreme changes in fat storage ranging from lipodystrophy to obesity (reviewed in Reue & Donkor 2006). Here we will describe studies in both mouse and human that support a role for lipin 1 in the regulation of lipid storage in adipose tissue, and the corresponding effects on metabolic homeostasis. Lipin 1 is required for fat storage in adipose tissue Lipin 1 was originally identified in the fatty liver dystrophy ( fld ) mutant mouse strain. The fld mouse is a model of generalized lipodystrophy characterized by the absence of normal adipose tissue from birth, and the development of a peripheral neuropathy and insulin resistance as adults (Langner et al 1989, Reue et al 2000). Similar to humans and other lipodystrophic mouse models, the fld mouse initially develops a fatty liver. However, unique to this model, the fatty liver spontaneously resolves by three weeks of age. Using a positional cloning approach, we determined that a null mutation in the Lpin1 gene is responsible for the complex array of metabolic defects observed in the fld mouse (Peterfy et al 2001). We identified strong similarity between lipin 1 and two additional predicted protein sequences, lipin 2 and lipin 3. A single lipin homologue is also present in invertebrates and yeast, suggesting that lipin functions in a fundamental cellular process. All members of the lipin family possess a nuclear localization signal and have high sequence conservation for extended stretches at the N- and C-terminal ends (Peterfy et al 2001). Through alternative mRNA splicing, the Lpin1 gene gives rise to two abundant protein isoforms, lipin 1α and lipin 1β (Huffman et al 2002, Peterfy et al 2005). The lipin 1β isoform contains an additional 33 amino acids not present in the lipin 1α form, but is otherwise identical in sequence (Fig. 1). However, as described below, the two lipin 1 isoforms appear to have distinct roles in adipocyte development, and perhaps other processes. The lipodystrophic phenotype of the fld mouse and high expression levels of lipin in normal adipose tissue suggested that lipin 1 has a critical role in the adipocyte. Adipose tissue from the fld mouse is comprised of cells that fail to accumulate triglyceride and maintain the appearance of immature adipocyte precursors (Reue et al 2000). In vitro, lipin-deficient preadipocytes cannot be induced to differentiate in response to hormonal signals. This is associated with failure to express key adipogenic transcription factors (e.g. PPARγ and C/EBPα ), and inability to synthesize triglyceride (Phan et al 2004). The lipin 1α and lipin 1β isoforms appear to have distinct roles in adipogenesis, as evidenced by differences in expression dynamics, subcellular localization, and effects on gene expression. Expression of
60
REUE Lipin-1α
(a)
Lipin-1β 924 aa
891 aa NLS 33 aa
NLS NLIP
CLIP
CLIP
NLIP
(b) Nuclear localization
Cytoplasmic localization
Lipin-1α and DNA co-localize
(c)
Lipin-1α
Adipogenic gene expression
Lipin-1β
Lipin-1β
DNA
Lipogenic gene expression
FIG. 1. Lipin 1α and β protein isoforms exhibit distinct functional characteristics. (a) Lipin 1α and 1β isoforms, derived from alternative splicing of the Lpin1 gene, are identical except for an additional 33 amino acids in the 1β isoform. NLS, nuclear localization signal; NLIP and CLIP, evolutionarily conserved N- and C-terminal lipin domains. (b) Confocal microscopy images of nuclear and cytoplasmic staining of 3T3-L1 adipocytes transfected either with lipin 1α or lipin 1β. Lipin 1α co-localizes with DNA in the nucleus, whereas lipin 1β is present predominantly in the cytoplasm. Dark, unstained areas are lipid droplets. (c) Lipin 1 isoforms promote distinct patterns of gene expression upon transfection of lipin-deficient preadipocytes. Lipin 1α induces expression of adipogenic genes, including PPARγ and C/EBPα , whereas lipin 1β induces lipogenic genes, such as fatty acid synthase, stearoyl CoA desaturase and diacylglycerol acyltransferase. Adapted from Reue & Donkor (2006) with permission.
both lipin 1 isoforms is dramatically induced during differentiation of 3T3-L1 preadipocytes, but the lipin 1α isoform predominates at 2 days after induction, while lipin 1β is the major form at day 6 (Peterfy et al 2005). Despite the fact that both lipin 1 isoforms possess a nuclear localization signal, the two exhibit distinct subcellular localization, with lipin 1α and lipin 1β more likely to be found in the nucleus and cytoplasm, respectively. Most importantly, there is evidence for functional differences between the isoforms. Selective replacement of lipin 1α or lipin 1β into lipin-deficient cells revealed that lipin 1α expression leads to a dramatic induction of PPARγ and C/EBPα gene expression, suggesting that this isoform enhances adipocyte differentiation (Peterfy et al 2005). On the other hand, lipin 1β expression has no effect on adipogenic gene expression, but induces expression of genes involved in fatty acid and triglyceride synthesis (e.g. fatty acid synthase, stearoyl CoA desaturase 1, diacylglycerol transferase). Thus, the two lipin 1 isoforms appear to serve complementary functions in adipocyte development,
LIPIN 1
61
with lipin 1α promoting adipogenic and lipin 1β promoting lipogenic gene expression. Rare forms of congenital lipodystrophy in humans have been attributed to mutations in a variety of genes, including LMNA, which encodes nuclear envelope proteins, and AGPAT2, a triglyceride biosynthetic enzyme (Garg 2004). Thus far, no mutations in human LPIN1 have been identified in congenital lipodystrophies (Cao & Hegele 2002). However, we have examined the potential role of altered lipin expression levels in acquired lipodystrophy that results from antiretroviral therapy in HIV-1 infected patients. Lipin 1 mRNA levels in abdominal and femoral subcutaneous adipose tissue were determined in three groups of individuals— HIV-1 infected individuals with lipodystrophy, HIV-1 infected individuals without lipodystrophy, and non-infected controls. Consistent with the mouse studies, lipin mRNA levels were correlated with limb fat mass in HIV patients, with lower lipin levels in those with lipodystrophy compared to non-lipodystrophic individuals (Lindegaard et al 2006). However, this was not a result of reduced lipin levels in patients with lipodystrophy compared to control subjects, but rather to an increase in lipin 1β expression in the HIV infected subjects that had not developed lipodystrophy compared to both controls and lipodystrophic individuals. The increased lipin 1β expression in HIV-1 infected subjects without lipodystrophy may represent an inherent difference in a subset of HIV-1 infected individuals that allows them to resist development of lipodystrophy, or may be a temporary compensatory response aimed at maintaining fat mass in response to HIV infection or the antiretroviral drugs. There was also an inverse correlation between lipin levels and adipose tissue expression of inflammatory cytokines, including interleukins (IL)6, IL8 and IL18 (Lindegaard et al 2006). Thus, increased lipin 1β expression levels in HIV-1 infected subjects are associated with maintenance of higher limb fat mass and reduced inflammatory cytokine production, leading to a more favourable metabolic status. Adipose tissue lipin 1 is a determinant of insulin sensitivity Whereas lipin 1 deficiency leads to a lack of adipose tissue lipid storage, enhanced lipin 1 expression directly promotes lipid storage in adipose tissue. Transgenic mice expressing lipin 1β at threefold normal levels in adipose tissue exhibit normal body weight when fed a chow diet, but when challenged with a high fat diet, the transgenic mice gain weight at twice the rate of their wild-type littermates (Phan & Reue 2005). This is associated with normal fat cell number, but increased fat mass due to increased lipid storage per adipocyte (Peterfy et al 2005). The transgenic mice also exhibit enhanced expression of lipogenic genes, such as fatty acid synthase, acetyl CoA carboxylase, and diacylglycerol acyltransferase. Despite obesity, increased lipin 1β expression in mature adipocytes appears to protect against
62
REUE
diet-induced insulin resistance (Phan & Reue 2005). Possible explanations include the lack of lipid accumulation in tissues such as liver and muscle of the transgenic mice, and production of higher levels of insulin-sensitizing adipokines. These results are in stark contrast to the effects of lipin 1β overexpression specifically in skeletal muscle, which causes obesity associated with reduced energy expenditure, and insulin resistance (Phan & Reue 2005). Overall, the results of studies in lipin 1-deficient and lipin 1β transgenic mice reveal a direct correlation between the amount of lipin 1 expression in adipose tissue and both triglyceride accumulation in adipocytes and insulin sensitivity. To determine whether the relationship between lipin 1 expression and insulin sensitivity is maintained over a range of expression levels, we evaluated these parameters in groups of genetically diverse mice and humans. In both mouse and man, there was a strong negative correlation between total lipin 1 mRNA levels in adipose tissue and fasting glucose, insulin, and HOMA-IR values (Suviolahti et al 2006). Furthermore, significant associations were identified between LPIN1 allelic haplotypes in an analysis of more than 1000 members of dyslipidaemic families (Suviolahti et al 2006). These data extend the relationship between lipin expression and glucose metabolism parameters to a continuous range of lipin 1 mRNA levels, and suggest that genetic variations in the LPIN1 gene sequence may underlie differences in lipin 1 levels and/or activity. We directly tested the relationship between lipin 1 expression and insulin sensitivity in human subjects who had been characterized via a frequently sampled intravenous glucose tolerance test. Consistent with the previous results in mouse and humans, we observed a positive correlation between lipin 1β expression in adipose tissue and insulin sensitivity (Fig. 2a) (Yao-Borengasser et al 2006). To eliminate potential confounding effects of obesity, a subset of individuals were selected and matched for body mass index (BMI), sex and age, and were classified into one of two groups based on whether they had normal or impaired glucose tolerance. Total lipin 1, lipin 1α and lipin 1β mRNA levels in adipose tissue were each found to be approximately twofold higher in the normal glucose tolerant subjects compared to those with impaired glucose tolerance (Yao-Borengasser et al 2006). The higher expression levels of both lipin 1 isoforms in insulin sensitive subjects may reflect adipose tissue that is primed for both adipogenesis (lipin 1α ) and lipogenesis (lipin 1β ). The results described above in mouse and humans are consistent with the hypothesis that increased lipin 1 expression in adipose tissue might enhance insulin sensitivity by partitioning lipids into adipose tissue, thus protecting muscle from lipid accumulation. To test this directly, we examined the relationship between adipose tissue lipin 1 expression and intramyocellular lipid (IMCL) in human muscle. In support of the hypothesis, we found that lipin 1α expression in adipose tissue was inversely correlated with IMCL in type 2 muscle fibres (Fig. 2b)
LIPIN 1
63
Publisher's Note: Permission to reproduce this image online was not granted by the copyright holder. Readers are kindly requested to refer to the printed version of this chapter.
FIG. 2. Relationship between lipin 1 expression levels in human adipose tissue and insulin sensitivity, intramyocellular (IMCL) lipid content, and response to pioglitazone. Lipin 1 mRNA levels in human subcutaneous adipose tissue were determined by real-time RT-PCR using primers specific for total lipin, lipin 1α or lipin 1β, as indicated, and normalized to 18S RNA. (a) Correlation of lipin 1β mRNA levels with insulin sensitivity determined via frequently sampled intravenous glucose tolerance test. Data are plotted on a log-log scale. (b) Inverse correlation between lipin 1α and IMCL. Data are plotted on a log-log scale. (c) Induction of lipin 1β expression after 10 weeks of pioglitazone, but not metformin, treatment. *, P < 0.05 vs. baseline; **, P < 0.005 vs. baseline. Data are adapted from Yao-Borengasser et al (2006) with permission.
64
REUE
(Yao-Borengasser et al 2006). Importantly, this effect was independent of obesity and insulin sensitivity, both of which are also associated with increased IMCL. The variation in IMCL was not associated with lipin 1 expression levels in muscle itself, nor with oxidative capacity of muscle (Yao-Borengasser et al 2006). The results described above are consistent with a role for adipose tissue lipin 1 in insulin sensitivity through partitioning lipids into adipose tissue and away from muscle and other sites. The same mechanism has been linked to the effect of thiazoladinedione drugs, such as pioglitazone, which act as PPARγ agonists. Pioglitazone treatment has been shown to improve insulin sensitivity in association with increased subcutaneous adipose tissue volume and decreased IMCL (Rasouli et al 2005). We therefore tested whether treatment with pioglitazone for 10 weeks leads to increased lipin 1 expression in human adipose tissue. Glucose intolerant subjects were treated either with pioglitazone or metformin, an insulin-sensitizing drug that acts to reduce hepatic glucose production. In adipose samples taken before and after treatment, lipin 1 expression was not altered by metformin, but lipin 1β expression was doubled in response to pioglitazone (Fig. 2c) (Yao-Borengasser et al 2006). The same effect on lipin 1β expression was observed in cultured 3T3L1 cells after thiazoladinedione treatment. Given that lipin 1β stimulates lipogenic gene expression (Peterfy et al 2005), induction of lipin 1β by thiazolidinediones may mediate the lipid accumulation that occurs in response to these drugs. Lipin and lipid metabolism in muscle and liver The studies described thus far focus primarily on the role of lipin 1 in adipose tissue lipid metabolism. In fact, lipin 1 is also expressed at substantial levels in other key metabolic tissues including muscle, liver and pancreatic β cells, and may exert important tissue specific effects (Peterfy et al 2001, Lan et al 2003). Studies in lipin 1-deficient and muscle-specific lipin 1β transgenic mice have established that lipin 1 levels in muscle are a determinant of whole body energy expenditure, and of glucose versus fatty acid utilization in muscle (Phan & Reue 2005). Lipin deficiency promotes increased energy expenditure, reduced respiratory quotient (signifying greater reliance on fatty acid fuels), and elevated expression of fatty acid oxidation genes in muscle. Lipin 1β overexpression exclusively in muscle produces the opposite effect, with reduced energy expenditure, elevated respiratory quotient, and reduced fatty acid oxidation gene expression, all of which presumably contribute to the obesity in these animals. In contrast to adiposespecific transgenic mice, the muscle-specific lipin 1β transgenic mice are insulin resistant, pointing to an adipose-specific effect of lipin 1β on enhanced insulin sensitivity. Recent studies have also revealed a fundamental role for lipin 1 in the coordination of lipid metabolism by liver and muscle during the diurnal cycle. We determined that the fld mouse exhibits an unusual metabolic adaptation in response to
LIPIN 1
65
the inability to store triglycerides in adipose tissue. In place of triglycerides, these animals store excess amounts of glycogen in liver and muscle during the fed state, and oxidize a higher proportion of fatty acids (Xu et al 2006). During the fasting state, fld mice utilize the stored glycogen to carry them through the fasting period in place of fatty acids from adipose tissue. This impaired switch between glucose and fat oxidation during transitions between fasted and fed states is reminiscent of the ‘metabolic inflexibility’ observed in insulin resistance and type 2 diabetes in humans (Storlien et al 2004). The heightened demand for fatty acids in muscle of the fld mouse during the fed state is met by fatty acids derived from the liver, and suggests a mechanism for the amelioration of the fatty liver in adult lipindeficient mice (Xu et al 2006). The fact that this does not occur in other types of lipodystrophy indicates a specific effect of lipin deficiency in tissues such as liver and muscle. Mechanisms of lipin 1 action As described above, the two lipin 1 isoforms appear to have distinct functional properties. While its molecular function is not yet well understood, evidence suggests that lipin 1 may have multiple activities, potentially including separate activities in the nucleus versus cytoplasm. Very recently, it was discovered that the lipin homologue in the budding yeast, Saccharomyces cerevisiae, is a lipid biosynthetic enzyme, phosphatidic acid phosphatase-1 (PAP1) (Han et al 2006). PAP1 activity has been studied for decades, but the protein responsible had remained elusive. PAP1 regulates the conversion of phosphatidic acid to diacylglycerol, which is required for triacylglycerol and phospholipid synthesis (Bell & Coleman 1980, Brindley & Waggoner 1998). A key sequence motif required for PAP1 activity (DxDxT) is evolutionarily conserved in every member of the lipin protein family in every species ranging from mammals to yeast. Consistent with this, all known mammalian lipin proteins have been shown to have PAP1 activity, including lipin 1α , lipin 1β, lipin 2 and lipin 3 (Donker et al 2007, Harris et al 2007). Using tissues from fld mice, it was also determined that lipin-1 accounts for all PAP1 activity in white and brown adipose tissue and skeletal muscle; however, PAP1 activity in the liver of fld mice was essentially normal, indicating that other members of the lipin protein family may act in this tissue (Donker et al 2007). The role of lipin-1 as the sole PAP1 in adipose tissue may explain the failure of lipin-1 deficient adipocytes to accumulate triglycerides, as well as the increased triglyceride storage in adipose tissue of mice with enhanced lipin 1β expression (Reue et al 2000, Phan & Reue 2005). It is also consistent with the known effects of thiazoladinedione treatment, which both induces lipin 1β expression and increases triglyceride storage in adipose tissue. Previous studies of PAP1 activity have shown that activity is regulated by several hormones, including insulin, which has also been shown to phosphorylate lipin 1 (Huffman et al 2002).
66
REUE
Evidence also exists for a specific nuclear role for lipin. Thus, the yeast lipin regulates nuclear membrane growth during the cell cycle, through transcriptional regulation of phospholipid biosynthesis (Santos-Rosa et al 2005). Furthermore, yeast lipin was found to physically associate with the promoters of phospholipid biosynthetic genes. Since lipin does not contain a transcriptional activation domain, the most likely scenario is a role as a co-activator or co-repressor that interacts with other factors to influence transcription. The elucidation of the potential roles of lipin 1 isoforms in transcriptional regulation and PAP1 activity should enhance our understanding of the mechanisms underlying the observed relationship between lipin 1 levels and insulin sensitivity. Summary and prospects The studies described here in mouse models and humans suggest a role for lipin 1 in lipid and glucose homeostasis. Although the function of lipin 1 is not fully understood, the available data implicate lipin 1 in the process of triglyceride storage
lipin1 expression
triglyceride per adipocyte
intramyocellular lipid
inflammatory cytokines
insulin-sensitizing adipokines
insulin sensitivity FIG. 3. Model of the effect of enhanced adipose tissue lipin 1 expression on insulin sensitivity. Evidence from mouse and human studies described herein indicate that enhanced lipin 1 expression in adipose tissue is associated with improved insulin sensitivity. Potential mechanisms include increased triglyceride storage per adipocyte, reduced lipid accumulation in muscle, reduced inflammatory cytokine expression in adipose tissue, and enhanced expression of insulin-sensitizing adipokines.
LIPIN 1
67
in adipose tissue with subsequent effects on fat mass, fat accumulation in muscle, inflammatory cytokine levels, adipokine levels, and ultimately, insulin sensitivity (Fig. 3). Many questions remain to be answered. These include questions concerning the molecular function of lipin 1α and 1β isoforms, their physiological roles in muscle, liver and β cells, the regulation of lipin 1 expression, mRNA splicing and activity, and the potential role of lipin 1 as an effector of insulin-sensitizing thiazolidinediones.
Acknowledgements This work was supported by the National Institutes of Health grant HL 28481.
References Bell RM, Coleman RA 1980 Enzymes of glycerolipid synthesis in eukaryotes. Annu Rev Biochem 49:459–487 Brindley DN, Waggoner DW 1998 Mammalian lipid phosphate phosphohydrolases. J Biol Chem 273:24281–24284 Cao H, Hegele RA 2002 Identification of single-nucleotide polymorphisms in the human LPIN1 gene. J Hum Genet 47:370–372 Donker J, Sariahmetoglu M, Dewald J, Brindley DN, Reue K 2007 Three mammalian lipins act as phosphatidate phosphatases with distinct tissue expression patterns. J Biol Chem 282:3450–3457 Eldor R, Raz I 2006 Lipotoxicity versus adipotoxicity—the deleterious effects of adipose tissue on beta cells in the pathogenesis of type 2 diabetes. Diabetes Res Clin Pract 74:S3–S8 Garg A 2004 Acquired and inherited lipodystrophies. N Engl J Med 350:1220–1234 Garg A, Misra A 2004 Lipodystrophies: rare disorders causing metabolic syndrome. Endocrinol Metab Clin North Am 33:305–331 Han GS, Wu WI, Carman GM 2006 The Saccharomyces cerevisiae Lipin homolog is a Mg2+-dependent phosphatidate phosphatase enzyme. J Biol Chem 281:9210–9218 Harris TE, Huffman TA, Chi A et al 2007 Insulin controls subcellular localization and multisite phosphorylation of the phosphatidic acid phosphatase, lipin 1. J Biol Chem 282:277–286 Hegele RA, Pollex RL 2005 Genetic and physiological insights into the metabolic syndrome. Am J Physiol Regul Integr Comp Physiol 289:R663–R669 Huffman TA, Mothe-Satney I, Lawrence JC Jr 2002 Insulin-stimulated phosphorylation of lipin mediated by the mammalian target of rapamycin. Proc Natl Acad Sci USA 99:1047–1052 Lan H, Rabaglia ME, Stoehr JP et al 2003 Gene expression profi les of nondiabetic and diabetic obese mice suggest a role of hepatic lipogenic capacity in diabetes susceptibility. Diabetes 52:688–700 Langner CA, Birkenmeier EH, Ben-Zeev O et al 1989 The fatty liver dystrophy ( fl d ) mutation. A new mutant mouse with a developmental abnormality in triglyceride metabolism and associated tissue-specific defects in lipoprotein lipase and hepatic lipase activities. J Biol Chem 264:7994–8003 Lindegaard B, Larsen LF, Hansen AB, Gerstoft J, Pedersen BK, Reue K 2006 Adipose tissue lipin expression levels distinguish HIV patients with and without lipodystrophy. Int J Obes (Lond) 31:449–456
68
REUE
Peterfy M, Phan J, Xu P, Reue K 2001 Lipodystrophy in the fl d mouse results from mutation of a new gene encoding a nuclear protein, lipin. Nat Genet 27:121–124 Peterfy M, Phan J, Reue K 2005 Alternatively spliced lipin isoforms exhibit distinct expression pattern, subcellular localization, and role in adipogenesis. J Biol Chem 280:32883–32889 Petersen KF, Shulman GI 2006 Etiology of insulin resistance. Am J Med 119:S10–S16 Phan J, Reue K 2005 Lipin, a lipodystrophy and obesity gene. Cell Metab 1:73–83 Phan J, Peterfy M, Reue K 2004 Lipin expression preceding peroxisome proliferator-activated receptor-gamma is critical for adipogenesis in vivo and in vitro. J Biol Chem 279:29558–29564 Rasouli N, Raue U, Miles LM et al 2005 Pioglitazone improves insulin sensitivity through reduction in muscle lipid and redistribution of lipid into adipose tissue. Am J Physiol Endocrinol Metab 288:E930–E934 Reue K, Donkor J 2006 Lipin: a determinant of adiposity, insulin sensitivity and energy balance. Future Lipidol 1:91–101 Reue K, Phan J 2006 Metabolic consequences of lipodystrophy in mouse models. Curr Opin Clin Nutr Metab Care 9:436–441 Reue K, Xu P, Wang XP, Slavin BG 2000 Adipose tissue deficiency, glucose intolerance, and increased atherosclerosis result from mutation in the mouse fatty liver dystrophy ( fl d ) gene. J Lipid Res 41:1067–1076 Santos-Rosa H, Leung J, Grimsey N, Peak-Chew S, Siniossoglou S 2005 The yeast lipin Smp2 couples phospholipid biosynthesis to nuclear membrane growth. EMBO J 24:1931–1941 Slawik M, Vidal-Puig AJ 2006 Lipotoxicity, overnutrition and energy metabolism in aging. Ageing Res Rev 5:144–164 Storlien L, Oakes ND, Kelley DE 2004 Metabolic flexibility. Proc Nutr Soc 63:363–368 Suviolahti E, Reue K, Cantor RM et al 2006 Cross-species analyses implicate Lipin 1 involvement in human glucose metabolism. Hum Mol Genet 15:377–386 Unger RH 2003 Minireview: weapons of lean body mass destruction: the role of ectopic lipids in the metabolic syndrome. Endocrinology 144:5159–5165 Xu J, Lee WNP, Phan J, Saad M, Reue K, Kurland IJ 2006 Lipin deficiency impairs diurnal metabolic fuel switching. Diabetes 55:3429–3438 Yao-Borengasser A, Rasouli N, Varma V et al 2006 Lipin expression is attenuated in adipose tissue of insulin resistant human subjects and increases with PPARγ activation. Diabetes 55:2811–2818
DISCUSSION Daum: With reference to the subcellular localization of lipin, one portion is in the nucleus and the rest in the cytosol. Do you assume that the portion in the nucleus is the transcriptional regulating portion? Reue: That is what we assume, although we have not yet shown it experimentally. We do find that lipin 1 shuttles between cytoplasm and nucleus, raising the possibility that subcellular localization is a regulatory mechanism. Daum: I would assume that when lipin is acting as a phosphatidic acid phosphohydrolase it would have to act on a component that is in the membrane. Reue: It translocates from cytoplasm to the ER membrane. Daum: Would you assume that phosphorylation of this lipin has something to do with the membrane association? It needs to be there.
LIPIN 1
69
Reue: That’s a good bet. We are working on this. Spiegelman: Have you turned the experiment round: if you have PGC1α , what form does it capture? This would be an easy experiment. Reue: That’s a good experiment. Kim: When you say that lipin 1 has a phosphatase activity, do you think that PAP1 activity is crucial for all of these phenotypes? Are they working with scaffold proteins to interact with other proteins to show all the different insulin sensitivity or differentiation activities? Reue: That’s a good question, to which we don’t yet have an answer. It is likely that the inability to accumulate lipid in adipocytes is due to the lack of PAP1 activity. But I think that the role of lipin 1 in early stages of adipocyte differentiation may be related to its coactivator function. One way we can approach this possibility is with mutants. We can make a mutant that knocks out the PAP1 enzyme activity but not the transcriptional coactivator function, and determine the effect on adipocyte differentiation, insulin sensitivity, etc. Yang: Are there any similarities between the DGAT1 knockout mouse and your lipin knockout mice? Reue: DGAT1 is the enzyme just downstream of lipin in the triglyceride biosynthetic pathway. I would call the DGAT1 mutant a lean animal, rather than a lipodystrophic animal. They still have plenty of adipose tissue but they have increased energy expenditure. When they are crossed to ob/ob mice and AY agouti mice, they show improved insulin sensitivity and reduced adiposity. However, the knockout in DGAT1 is not nearly as severe as the lipin 1 knockout in terms of the adipose tissue deficiency. Both lipin 1 and DGAT catalyse steps in triglyceride biosynthesis. The lipin 1 knockout may be more severe, in part, because of its additional role as a transcriptional regulator. Spiegelman: What if the roles in the nucleus and the cytoplasm are closely related in these sense that coactivator proteins are enzymes or they dock enzymes. We tend to think in terms of histone acetyltransferases, ATP-depdendent DNA unwinding, methylation and demethylation: but what if it is not a mistake that nature is using it in these two places, but rather there is a role for the dephosphorylation of lipids in chromatin function? Perhaps it is working in the coactivator complexes by phosphatase activity on something else or on the lipids involved somehow in chromatin. Reue: I think that this is a great idea. We are looking into it. Spiegelman: So PGC1α is capturing it in the same way that it captures SirT1 or CBP. It is just capturing enzymes and using them. Muoio: Along the same lines there is recent evidence that intermediates in the glycerol lipid pathway can function as ligands for some of the nuclear receptors. Might lipin be functioning in this capacity as well?
70
REUE
Reue: We haven’t done any work on that but it makes a lot of sense. Lipin 1 converts PA to DAG. Both are signalling molecules, so one can imagine that the concentrations of those is going to be critical, potentially having effects on nuclear receptor activity. Tabas: In the knockout and transgenic mice, what happens to adiponectin levels? Reue: In lipin 1-deficient mice, adiponectin is barely detectable, owing to a lack of normal adipose tissue and failure to express genes such as adiponectin. In the adipose-specific lipin 1 transgenic strain, there is a 50% increase in adiponectin levels compared to wild-type mice, which could be metabolically important. Hotamisligl: In the transgenic model, which form of lipin was expressed? Reue: The transgenic strain overexpresses the lipin 1β isoform, and maintains wild-type levels of lipin 1α . When we generated those mice, we didn’t have the information regarding the functional differences between the two isoforms. We are now making lipin 1α transgenic mice to determine the physiological effects of this isoform as well. Hotamisligl: So a non-nuclear lipin is creating this phenotype. Reue: Not strictly non-nuclear. While the majority of lipin 1β resides in the cytoplasm of adipocytes, it appears to shuttle between cellular compartments, and a proportion of cells appear to have lipin 1β in the nucleus at a given time. Spiegelman: Going back to your adipogenesis story, it is an attractive idea that you are modifying a lipid in the nucleus and it is working as a ligand for something—perhaps a nuclear receptor. Can you rescue the lipin knockout in adipogenesis by giving it a PPARγ or LXR agonist? Reue: We cannot rescue adipogenesis in lipin 1-deficient cells simply by providing PPARγ ligand. However, keep in mind that these cells do not express PPARγ. When we provide both PPARγ and ligand, we partially rescue the differentiation phenotype. We have not tested LXR agonist. Spiegelman: That is misleading because PPARγ ligand turns on PPARγ, so the fact you don’t turn it on just means that you didn’t start the pathway. Reue: All our differentiation experiments are done with the addition of the PPARγ agonist, rosiglitazone, and the lipin 1-deficient cells still cannot differentiate. Spiegelman: That answers the question: it can’t be a deficiency in the PPARγ agonist. Kim: What about the splice form in different species? Can you fi nd this in yeast? Reue: We have only identified the lipin 1 splice variant in mammals. Yeast is quite different from mammals in that it has a single lipin gene, whereas in mammals there is a family of three genes. In fact, some of the early findings in yeast have been misleading. Yeast databases indicate that Smp2, the lipin homologue, local-
LIPIN 1
71
izes to the mitochondria. This is definitely not the case for mammalian lipin 1. Despite the differences that may exist, the yeast lipin exhibits PAP1 activity, and was instrumental in identifying this role of mammalian lipin proteins. Shi: The muscle-specific transgenics are very obese. What are your thoughts on this? Reue: We probably haven’t characterized them as much as we should have. The main explanation for this phenotype that we have uncovered is that they have dramatically reduced energy expenditure and reduced fatty acid oxidation in muscle. Spiegelman: How is their movement? Reue: Their movement is normal. They are just not expending energy. We have quantified the amount of triglyceride in the muscle, and they show increased storage. This could be due to the direct action of lipin in the muscle, or because they are obese and are accumulating lipid there. They are insulin resistant.
Fatty Acids and Lipotoxicity in Obesity and Diabetes: Novartis Foundation Symposium 286, Volume 286. Edited by Gregory Bock and Jamie Goode Copyright Novartis Foundation 2007. ISBN: 978-0-470-05764-3
The role of the AMP-activated protein kinase in the regulation of energy homeostasis David Carling Cellular Stress Group, MRC Clinical Sciences Centre, Imperial College, Hammersmith Hospital, Du Cane Road, London W12 0NN, UK
Abstract. AMP-activated protein kinase (AMPK) is the downstream component of a protein kinase cascade that plays a major role in maintaining energy homeostasis. Within individual cells, AMPK is activated by a rise in the AMP:ATP ratio that occurs following a fall in ATP levels. AMPK is also regulated by the adipokines, adiponectin and leptin, hormones that are secreted from adipocytes. Activation of AMPK requires phosphorylation of threonine 172 within the catalytic subunit by either LKB1 or calcium/calmodulin dependent protein kinase kinase β (CaMKK β ). AMPK regulates a wide range of metabolic pathways, including fatty acid oxidation, fatty acid synthesis, glycolysis and gluconeogenesis. In peripheral tissues, activation of AMPK leads to responses that are beneficial in counteracting the deleterious effects that arise in the metabolic syndrome. Recent studies have demonstrated that modulation of AMPK activity in the hypothalamus plays a role in feeding. A decrease in hypothalamic AMPK activity is associated with decreased feeding, whereas activation of AMPK leads to increased food intake. Furthermore, signalling pathways in the hypothalamus lead to changes in AMPK activity in peripheral tissues, such as skeletal muscle, via the sympathetic nervous system (SNS). AMPK, therefore, provides a mechanism for monitoring changes in energy metabolism within individual cells and at the level of the whole body. 2007 Fatty acids and lipotoxicity in obesity and diabetes. Wiley, Chichester (Novartis Foundation Symposium 286) p 72–85
There is a growing body of evidence that indicates that the AMP-activated protein kinase (AMPK) cascade plays a major role in the regulation of energy homeostasis. The aim of this article is to present an overview of the regulation of AMPK and discuss how this fits with the concept of AMPK as a cellular fuel gauge, a term that was coined nearly 10 years ago (Hardie & Carling 1997). More recent evidence has emerged that suggests that AMPK also plays a role in the regulation of whole body energy metabolism. In peripheral tissues, AMPK is activated by the adipokines, leptin (Minokoshi et al 2002) and adiponectin (Yamauchi et al 2002), 72
AMPK AND ENERGY HOMEOSTASIS
73
leading to increases in fatty acid and glucose oxidation. In the hypothalamus, changes in AMPK activity have been shown to alter food intake. Activation of AMPK increases feeding whereas inhibition leads to decreased food intake (Andersson et al 2004, Kim et al 2004a, Minokoshi et al 2004). The combined actions of AMPK place it at the convergence of energy homeostasis within the individual cell level as well as on a whole body level, making it a key point for regulation of energy balance within an organism. Elucidating the regulation and role of AMPK is an important aspect in our overall understanding of energy metabolism and has a significant impact on the central issues of the current symposium. Regulation of AMPK AMPK is activated by phosphorylation of threonine 172 within the catalytic α subunit. Recent studies have identified two kinases capable of phosphorylating and activating AMPK, namely LKB1 (Hawley et al 2003, Woods et al 2003) and calcium/calmodulin dependent protein kinase kinase β (CaMKK β ) (Hawley et al 2005, Hurley et al 2005, Woods et al 2005). More recently, TAK1, a member of the MAPKKK family, has been identified as a third kinase that activates AMPK in vitro (Momcilovic et al 2006). Although the physiological role of the different upstream kinases is still under investigation there is evidence that LKB1 is required for activation of AMPK in response to signalling via AMP. In HeLa cells, which lack LKB1, AMPK is not activated in response to ATP depletion and the subsequent rise in the AMP:ATP ratio, but expression of LKB1 restores AMPK activation (Hawley et al 2003). CaMKK β is activated by Ca2+ and is likely to be involved in the activation of AMPK in response to agents that increase intracellular Ca2+ . Consistent with this hypothesis, a recent study demonstrated that CaMKK β is required for activation of AMPK by thrombin, which causes a rise in intracellular Ca2+ , in human endothelial cells (Stahmann et al 2006). Another study showed that CaMKK β is upstream of AMPK in T lymphocytes and is required for activation of AMPK following T cell antigen receptor activation (Tamas et al 2006). These results demonstrate that AMPK can be activated by different signals in different cell types, and these pathways involve distinct upstream kinases. In addition to phosphorylation, AMPK is activated allosterically by AMP. The γ subunit isoforms of AMPK contain four CBS (cystathionine- β -synthase) domains and several lines of evidence indicate that these domains are involved in binding AMP. Part of this evidence comes from studies examining the role of naturally occurring mutations in γ 2 that cause severe cardiac defects in humans (Arad et al 2002, Blair et al 2001, Gollob et al 2001), and a naturally occurring mutation in γ3 in pig that leads to glycogen accumulation in skeletal muscle (Milan et al 2000). Most of these mutations lie within the CBS domains and interfere with the activation of AMPK by AMP (Daniel & Carling 2003, Barnes et al 2004). Furthermore,
74
CARLING
AMP has been shown to bind directly to the CBS domains of the γ subunit isoforms expressed in bacteria (Scott et al 2004). As well as allosteric activation, AMP has been proposed to play a role in the phosphorylation of AMPK. Originally, three separate mechanisms were proposed whereby AMP would promote phosphorylation of AMPK. The first mechanism was direct activation of the upstream kinases by AMP (Hawley et al 1995). However, the evidence supporting this mechanism was based on results obtained using partially purified preparations of the upstream kinase from rat liver, which was subsequently shown to be LKB1. More recent studies using highly purified, recombinant preparations of LKB1 have shown that AMP does not directly activate LKB1 (Woods et al 2003). Similarly, CaMKKβ is not directly activated by AMP (Woods et al 2005, Hawley et al 2005). Two further mechanisms for promotion of phosphorylation by AMP were suggested to be substrate mediated. Binding of AMP to AMPK was proposed to render the kinase a better substrate for phosphorylation by upstream kinases (Hawley et al 1995), whilst making it a worse substrate for dephosphorylation by protein phosphatases (Davies et al 1995). Although these mechanisms provide an attractive model for AMP activation of AMPK, there is relatively little direct evidence supporting them, and for these studies partially purified preparations of AMPK and the upstream kinases were used. Moreover, recent studies have reported that AMP does not promote phosphorylation of AMPK by CaMKK β (Hawley et al 2005, Woods et al 2005). Since both CaMKK β and LKB1 activate AMPK by phosphorylating the same residue within AMPK it is difficult to envisage a mechanism that would account for a substrate-mediated effect of AMP that is specific for LKB1. Moreover, Suter et al (2006) recently reported that AMP does not promote phosphorylation of AMPK using a recombinant preparation of LKB1. Recent results from our laboratory (Sanders et al 2007) offer a plausible explanation for the apparently discrepant observations discussed above. We showed that AMP allosterically activates AMPK and inhibits dephosphorylation of T172 and that both these effects involve the γ subunit. However, AMP has no effect on phosphorylation of AMPK by either LKB1 or CaMKK β. Instead, we provided evidence to suggest that previous results indicating that AMP stimulates phosphorylation of AMPK by LKB1 were confounded by the presence of endogenous protein phosphatase 2Cα (PP2Cα ) in the preparations of the kinases used in the earlier studies. Taken together, these new fi ndings allow a simple model to be drawn for AMPK activation in response to either a rise in the AMP:ATP ratio or a rise in Ca2+ (Fig. 1). Activation of AMPK can occur either by increased phosphorylation, or decreased dephosphorylation, of T172. LKB1 appears to be constitutively active and its activity is not changed by stimuli that activate AMPK (Woods et al 2003, Sakamoto et al 2004). According to the new model, increased T172 phosphorylation by LKB1 would occur in response to decreased dephos-
AMPK AND ENERGY HOMEOSTASIS
A
LKB1 α
γ
α β
γ PPase
B
75
Ca2+ α
γ
T172-PO4
β
AMP
CaMKKβ α
β
γ
T172-PO4
β
PPase FIG. 1. Model for the regulation of AMPK. (A) AMP-dependent activation: under conditions that lead to an increase in AMP, dephosphorylation of AMPK by protein phosphatase (PPase) is inhibited. Since LKB1 is constitutively active, inhibition of the dephosphorylation reaction leads to an increase in T172 phosphorylation and activation of AMPK. In addition to increasing T172 phosphorylation, AMP allosterically activates AMPK, via binding to the γ subunit. (B) Ca2+ -dependent activation: signals that increase Ca2+ activate CaMKK β, increasing T172 phosphorylation and activation of AMPK, and this can occur without an increase in AMP. However, it is possible that in some situations both Ca 2+ and AMP may increase in parallel, and under these conditions AMP will allosterically activate AMPK and inhibit dephosphorylation.
phorylation following a rise in AMP. In parallel, AMP would allosterically activate AMPK (Fig. 1A). Dephosphorylation and inactivation of AMPK would occur when the concentration of AMP returned to basal levels. Unlike LKB1, the activity of CaMKK β is subject to regulation within the cell, and is increased in response to signals that raise intracellular Ca2+ (Tokumitsu et al 2001, Anderson et al 1998). Therefore, signals that increase Ca2+ activate AMPK as a consequence of increased T172 phosphorylation via increased CaMKK β activity. Under these conditions there is no requirement for increased AMP levels (Fig. 1B). Consistent with this model, recent studies have reported activation of AMPK by CaMKK β -mediated signalling without detectable changes in AMP (Stahmann et al 2006, Tamas et al 2006, Hawley et al 2005). A further prediction of this model is that AMPK would be rapidly dephosphorylated following a fall in intracellular Ca2+ , and this may account for the transient nature of AMPK activation observed in CaMKK β mediated responses (Stahmann et al 2006, Tamas et al 2006, Hawley et al 2005). It is important to note that the AMP- and Ca2+ -signalling pathways for activation of AMPK predicted by this model are not mutually exclusive. It is possible that
76
CARLING
under some circumstances, where AMP and Ca2+ rise in parallel, both pathways could operate and under these conditions both CaMKK β and LKB1 would contribute to AMPK activation. Downstream targets of AMPK Once activated, AMPK acts to switch on ATP-generating pathways and to switch off ATP-consuming pathways, protecting the cell from depletion of ATP. In the short-term, for instance, AMPK activation leads to increased glucose uptake in muscle, increased fatty acid oxidation in liver and muscle, decreased fatty acid synthesis and decreased protein synthesis (reviewed in Carling 2004, Hardie 2003, Kahn et al 2005). In the longer term, AMPK has effects on gene expression, although our current understanding of this aspect of AMPK function is relatively poor. However, one important mechanism by which AMPK could control gene expression is via peroxisome proliferator-activated receptor γ co-activator 1α (PGC1α ), since AMPK signalling is required for increased expression of PGC1α in response to chronic energy deprivation (Zong et al 2002). An exciting new development in the role of AMPK came from initial studies investigating the regulation of AMPK by leptin in skeletal muscle. Leptin was shown to activate AMPK in muscle in a biphasic manner. A rapid activation (within 15 minutes) of AMPK is presumed to occur by direct binding of leptin to its receptor present on muscle cells, whereas a slower activation (6 hours) of AMPK occurs via the sympathetic nervous system (SNS) and α -adrenergic stimulation of muscle cells (Minokoshi et al 2002). The fi nding that the long term effect of leptin on AMPK activation in muscle occurred via a mechanism that involved hypothalamic signalling prompted several groups to investigate directly the effect of leptin on AMPK in the hypothalamus. In marked contrast to its effect in muscle, it was found that leptin inhibited AMPK in the hypothalamus (Andersson et al 2004, Minokoshi et al 2004). In addition to leptin, AMPK activity was decreased in the fed state (Minokoshi et al 2004), or following treatment with C75, a fatty acid synthase inhibitor (Kim et al 2004a). Conversely, ghrelin, an orexigenic hormone synthesized in the stomach, activates AMPK in the hypothalamus (Andersson et al 2004). Further studies revealed that signals associated with decreased food intake led to inhibition of hypothalamic AMPK whereas signals associated with increased food intake, including fasting, led to AMPK activation (Andersson et al 2004, Minokoshi et al 2004, Kim et al 2004a, b). Consistent with these fi ndings, pharmacological activation of AMPK in the hypothalamus using 5-amino-4-imidazole carboxamide (AICAR) or following infection of an adenovirus harbouring a mutant form of AMPK with increased basal activity in the hypothalamus, led to increased food intake in rodents. Taken together, these findings indicate that AMPK plays a role in co-ordinating energy homeostasis via effects in the central nervous system (CNS) and in peripheral tissues. Modulation of AMPK activity in
AMPK AND ENERGY HOMEOSTASIS
77
Peripheral tissues e.g. skeletal muscle Leptin Adiponectin Metformin AICAR
+
AMPK +
Increased energy production e.g. fatty acid oxidation, glucose oxidation
SNS
Hypothalamus Leptin Fed C75
-
Ghrelin Fasting AICAR
+
AMPK
NPY
Decreased food intake Increased energy expenditure
AMPK
NPY
Increased food intake Decreased energy expenditure
FIG. 2. Role of AMPK in the regulation of whole body energy metabolism. In peripheral tissues, e.g. skeletal muscle, activation of AMPK by hormones (leptin and adiponectin) or pharmacological agents (metformin and AICAR) leads to increased energy production. In the hypothalamus, leptin inhibits AMPK. In addition, AMPK activity is decreased in the fed state and by C75, a fatty acid synthase inhibitor. Decreased AMPK activity leads to a reduction in the expression of neuropeptide Y (NPY) and a decrease in food intake coupled with an increase in energy expenditure. In contrast, activation of AMPK by ghrelin, fasting or AICAR, increases NPY expression and leads to increased food intake and decreased energy expenditure. Leptin, acting within the hypothalamus, leads to activation of AMPK in skeletal muscle via the SNS and α -adrenergic signalling.
the hypothalamus has a direct effect on food intake as well as an indirect effect on energy expenditure in peripheral tissues, such as skeletal muscle, via the SNS. Furthermore, direct activation of AMPK in peripheral tissues leads to changes in energy metabolism, and these can be both short- and long-term. A model summarizing the effects of AMPK on energy homeostasis is shown in Fig. 2. LKB1 acts upstream of AMPK in liver and is important for glucose homeostasis In a recent study by Shaw et al (2005), LKB1 expression in the liver was abolished by injecting mice homozygous for a conditional floxed allele of LKB1 with adenovirus harbouring the Cre recombinase enzyme. This approach allows LKB1 expression to be abolished in adult mice, eliminating any complications that might arise during early development of the liver. Loss of LKB1 substantially reduced AMPK phosphorylation and activity in the liver, indicating that in this tissue
78
CARLING
LKB1 is the predominant upstream kinase in the AMPK cascade. Deletion of LKB1 led to a marked increase in fasting blood glucose together with impaired glucose tolerance. Notably, however, mice lacking LKB1 showed a normal response to insulin, indicating that peripheral glucose uptake was not affected. A number of genes involved in hepatic gluconeogenesis and lipogenesis were significantly increased in the LKB1-deleted mice. Importantly, the expression of PGC1α was also increased. PGC1α acts upstream of many target genes, including those involved in gluconeogenesis (Yoon et al 2001) and is itself under transcriptional control by the cAMP-response element binding (CREB) protein (Herzig et al 2001). CREB is regulated by a coactivator, transducer of regulated CREB activity 2 (TORC2) (Screaton et al 2004). TORC2 is regulated by phosphorylation and its nuclear localization is dependent on the phosphorylation state of serine 171 (Screaton et al 2004). AMPK and two members of the AMPK-related family of kinases, salt-inducible kinase (SIK) 1 and 2, have been shown to phosphorylate serine 171 within TORC2 promoting binding of the phospho-protein binding adaptor protein, 14-3-3 and sequestration of TORC2 within the cytoplasm (Screaton et al 2004, Koo et al 2005). The level of TORC2 phosphorylation in livers lacking LKB1 was reduced relative to the level in control livers. Consistent with this, the proportion of TORC2 in the nucleus was increased in LKB1-deficient hepatocytes providing strong evidence that LKB1 is required for the regulation of TORC2, and the subsequent regulation of target genes downstream of CREB. Metformin is one of the most widely used drugs for the treatment of type 2 diabetes and has been in use clinically since the 1970s. Metformin acts primarily by decreasing hepatic glucose production, largely by inhibiting gluconeogenesis (Stumvoll et al 1995). In spite of the fact that metformin has been used extensively as a frontline treatment for type 2 diabetes, the molecular mechanism by which metformin acts to lower blood glucose remained enigmatic until very recently. A key breakthrough in uncovering the molecular basis for the effect of metformin came from studies that showed that it activates AMPK by increasing phosphorylation of T172 (Zhou et al 2001, Fryer et al 2002, Hawley et al 2002). Crucially, metformin has no effect on lowering blood glucose in mice lacking hepatic expression of LKB1. The simplest interpretation of this finding is that LKB1 is required for activation of AMPK by metformin, and that in turn this is required for decreased expression of gluconeogenic genes, via phosphorylation of TORC2, and decreased glucose output. Conclusions AMPK plays a key role in monitoring the energy level within individual cells. Following a decrease in ATP, AMPK is activated and responds by increasing ATP supply and decreasing ATP-utilization, thereby acting to restore energy levels.
AMPK AND ENERGY HOMEOSTASIS
79
AMPK also plays a role in the regulation of whole body energy homeostasis. Modulation of AMPK activity in the hypothalamus is involved in controlling food intake and in regulating peripheral energy expenditure. AMPK activity in the CNS and peripheral tissues is modulated by hormones, such as leptin, providing a mechanism that allows co-ordinate regulation of energy balance at the level of the whole body. Recent studies on the regulation of AMPK by upstream kinases suggests that LKB1 is required for activation of AMPK in response to energy depletion, via AMP signalling pathways. LKB1 in the liver is essential for the glucose-lowering effect of metformin, through decreased gluconeogenesis. AMPK is also activated by stimuli that increase intracellular Ca2+ and in this case CaMKK β is an essential upstream component. A simple model accounting for the regulation of AMPK by AMP and Ca2+ , acting via either LKB1 or CaMKK β is proposed. Acknowledgements Work in the author’s laboratory was supported by the Medical Research Council UK, the European Commission (LSHM-CT-2004-005272 and LSHG-CT-2005-518181) and the Wellcome Trust.
References Anderson KA, Means RL, Huang QH et al 1998 Components of a calmodulin-dependent protein kinase cascade. Molecular cloning, functional characterization and cellular localization of Ca2+/calmodulin-dependent protein kinase kinase beta. J Biol Chem 273: 31880–31889 Andersson U, Filipsson K, Abbott CR et al 2004 AMP-activated protein kinase plays a role in the control of food intake. J Biol Chem. 279:12005–12008 Arad M, Benson DW, Perez-Atayde AR et al 2002 Constitutively active AMP kinase mutations cause glycogen storage disease mimicking hypertrophic cardiomyopathy. J Clin Invest 109:357–362 Barnes BR, Marklund S, Steiler TL et al 2004 The 5′-AMP-activated protein kinase gamma3 isoform has a key role in carbohydrate and lipid metabolism in glycolytic skeletal muscle. J Biol Chem 279:38441–38447 Blair E, Redwood C, Ashrafian H, Ostman-Smith I, Watkins H 2001 Mutations in the γ 2 subunit of AMP-activated protein kinase cause familial hypertrophic cardiomyopathy: evidence for the central role of energy compromise in disease pathogenesis. Hum Mol Genet 10:1215–1220 Carling D 2004 The AMP-activated protein kinase cascade—a unifying system for energy control. Trends Biochem Sci 29:18–24 Daniel T, Carling D 2003 Functional analysis of mutations in the gamma2 subunit of AMPactivated protein kinase associated with cardiac hypertrophy and Wolff-Parkinson-White Syndrome. J Biol Chem 277:51017–51024 Davies SP, Helps NR, Cohen PTW, Hardie DG 1995 5′-AMP inhibits dephosphorylation, as well as promoting phosphorylation, of the AMP-activated protein kinase—studies using bacterially expressed human protein phosphatase-2C α and native bovine protein phosphatase-2A(C). FEBS Lett 377:421–425
80
CARLING
Fryer LG, Parbu-Patel A, Carling D 2002 The anti-diabetic drugs rosiglitazone and metformin stimulate AMP-activated protein kinase through distinct signaling pathways. J Biol Chem 277:25226–25232 Gollob MH, Green MS, Tang ASL et al 2001 Identification of a gene responsible for familial Wolff-Parkinson-White syndrome. New Engl J Med 344:1823–1831 Hardie DG 2003 Minireview: the AMP-activated protein kinase cascade: the key sensor of cellular energy status. Endocrinology 144:5179–83 Hardie DG, Carling D 1997 The AMP-activated protein kinase: fuel gauge of the mammalian cell. Eur J Biochem 246:259–273 Hawley SA, Boudeau J, Reid JL et al 2003 Complexes between the LKB1 tumor suppressor, STRAD alpha/beta and MO25 alpha/beta are upstream kinases in the AMP-activated protein kinase cascade. J Biol 2:28 Hawley SA, Selbert MA, Goldstein EG et al 1995 5’-AMP activates the AMP-activated protein kinase cascade and Ca 2+/calmodulin activates the calmodulin-dependent protein kinase-I cascade, via 3 independent mechanisms. J Biol Chem 270:27186–27191 Hawley SA, Gadalla AE, Olsen GS, Hardie DG 2002 The antidiabetic drug metformin activates the AMP-activated protein kinase cascade via an adenine nucleotide-independent mechanism. Diabetes 51:2420–2425 Hawley SA, Pan DA, Mustard KJ et al 2005 Calmodulin-dependent protein kinase kinase-beta is an alternative upstream kinase for AMP-activated protein kinase. Cell Metab 2:9–19 Herzig S, Long F, Jhala US et al 2001 CREB regulates hepatic gluconeogenesis through the coactivator PGC-1. Nature 413:179–183 Hurley RL, Anderson KA, Franzone JM et al 2005 The Ca2+/calmodulin-dependent protein kinase kinases are AMP-activated protein kinase kinases. J Biol Chem 280:29060–29066 Kahn BB, Alquier T, Carling D, Hardie DG 2005 AMP-activated protein kinase: ancient energy gauge provides clues to modern understanding of metabolism. Cell Metab 1:15–25 Kim EK, Miller I, Aja S et al 2004a C75, a fatty acid synthase inhibitor, reduces food intake via hypothalamic AMP-activated protein kinase. J Biol Chem 279:19970–19976 Kim MS, Park JY, Namkoong C et al 2004b Anti-obesity effects of alpha-lipoic acid mediated by suppression of hypothalamic AMP-activated protein kinase. Nat Med 10:727–733 Koo SH, Flechner L, Qi L et al 2005 The CREB coactivator TORC2 is a key regulator of fasting glucose metabolism. Nature 437:1109–1111 Milan D, Jeon JT, Looft C et al 2000 A mutation in PRKAG3 associated with excess glycogen content in pig skeletal muscle. Science 288:1248–1251 Minokoshi Y, Kim YB, Peroni OD et al 2002 Leptin stimulates fatty-acid oxidation by activating AMP-activated protein kinase. Nature 415:339–343 Minokoshi Y, Alquier T, Furukawa N et al 2004 AMP-kinase regulates food intake by responding to hormonal and nutrient signals in the hypothalamus. Nature 428:569–574 Momcilovic M, Hong SP, Carlson M 2006 Mammalian TAK1 activates Snf1 protein kinase in yeast and phosphorylates AMP-activated protein kinase in vitro. J Biol Chem 281:25336–43 Sakamoto K, Goransson O, Hardie DG, Alessi DR 2004 Activity of LKB1 and AMPK-related kinases in skeletal muscle: effects of contraction, phenformin, and AICAR. Am J Physiol Endocrinol Metab 287:E310–E317 Sanders MJ, Grondin PO, Hegarty BD, Snowden MA, Carling D 2007 Investigating the mechanism for AMP activation of the AMP-activated protein kinase cascade. Biochem J 403:139–148 Scott JW, Hawley SA, Green KA et al 2004 CBS domains form energy-sensing modules whose binding of adenosine ligands is disrupted by disease mutations. J Clin Invest 113:274–284 Screaton RA, Conkright MD, Katoh Y et al 2004 The CREB coactivator TORC2 functions as a calcium- and cAMP-sensitive coincidence detector. Cell 119:61–74
AMPK AND ENERGY HOMEOSTASIS
81
Shaw RJ, Lamia KA, Vasquez D et al 2005 The kinase LKB1 mediates glucose homeostasis in liver and therapeutic effects of metformin. Science 310:1642–1646 Stahmann N, Woods A, Carling D, Heller R 2006 Thrombin activates AMP-activated protein kinase in endothelial cells via a pathway involving Ca2+/calmodulin-dependent protein kinase kinase beta. Mol Cell Biol 26:5933–5945 Stumvoll M, Nurjhan N, Perriello G, Dailey G, Gerich JE 1995 Metabolic effects of metformin in non-insulin-dependent diabetes mellitus. New Engl J Med 333:550–554 Suter M, Riek U, Tuerk R et al 2006 Dissecting the role of AMP for allosteric stimulation. activation and deactivation of AMP-activated protein kinase. J Biol Chem 281:32207– 32216 Tamas P, Hawley SA, Clarke RG et al 2006 Regulation of the energy sensor AMP-activated protein kinase by antigen receptor and Ca2+ in T lymphocytes. J Exp Med 203:1665–1670 Tokumitsu H, Iwabu M, Ishikawa Y, Kobayashi R 2001 Differential regulatory mechanism of Ca2+/calmodulin-dependent protein kinase kinase isoforms. Biochemistry 40:13925– 13932 Woods A, Johnstone SR, Dickerson K et al 2003 LKB1 is the upstream kinase in the AMPactivated protein kinase cascade. Curr Biol 13:2004–2008 Woods A, Dickerson K, Heath R et al 2005 Ca2+/calmodulin-dependent protein kinase kinase-beta acts upstream of AMP-activated protein kinase in mammalian cells. Cell Metab 2:21–33 Yamauchi T, Kamon J, Minokoshi Y et al 2002 Adiponectin stimulates glucose utilization and fatty-acid oxidation by activating AMP-activated protein kinase. Nat Med 8:1288–1295 Yoon JC, Puigserver P, Chen G et al 2001 Control of hepatic gluconeogenesis through the transcriptional coactivator PGC-1. Nature 413:131–138 Zhou G, Myers R, Li Y et al 2001 Role of AMP-activated protein kinase in mechanism of metformin action. J Clin Invest 108:1167–1174 Zong H, Ren JM, Young LH et al 2002 AMP kinase is required for mitochondrial biogenesis in skeletal muscle in response to chronic energy deprivation. Proc Natl Acad Sci USA 99:15983–15987
DISCUSSION P Li: One of the question marks surrounding LKB1 as an upstream kinase for AMPK is that the phosphorylation state of LKB1 is not regulated by various signals. You said that phosphorylation is crucial for the activation of AMPK. Here you have a kinase that is always active, so how does this regulate the phosphorylation and activity of AMPK? Carling: To date, the evidence strongly suggests that LKB1 is constitutively active. This is presumably because it does not require phosphorylation within the T-loop region, unlike virtually all other kinases. In addition to AMPK, Dario Alessi and colleagues have shown that LKB1 is upstream of 12 other AMPKrelated kinases (Lizcano et al 2004). LKB1 is therefore an upstream kinase for a whole family of other kinases, not unlike PDK1, which also acts upstream of a range of other kinases. In these cases, I think it is possible that the systems have evolved so that the upstream kinases are constitutively active and that the regulation of phosphorylation of the downstream kinases is substrate-mediated.
82
CARLING
In this way it is possible to activate individual kinases rather than switching on all of the downstream kinases at the same time. The most likely mechanism for this in the AMPK/LKB1 cascade is that phosphorylation of AMPK by LKB1 is regulated by binding of AMP to AMPK following a change in the AMP:ATP ratio within the cell. LKB1 is constitutively active, but under basal conditions when ATP is high and AMP is low, the protein phosphatase(s) acting on AMPK is much more active than the LKB1, which maintains AMPK in the unphosphorylated, inactive form. An increase in AMP leads to inhibition of the phosphatase(s). Under these conditions the net effect will be to increase phosphorylation of AMPK, even though the activity of LKB1 is unchanged. However, we still do not understand the mechanisms that regulate phosphorylation and activation of the AMPK-related kinases by LKB1, as these kinases do not appear to be regulated by AMP. Hotamisligil: You said this is dependent on AMP still. Is activation of AMPK by adiponectin AMP-independent? Carling: This is an interesting point. In our hands we do not see a significant change in the AMP:ATP ratio in response to activation by adiponectin. However, the way that we monitor this is to measure total cellular nucleotides i.e. we lyse the cells and measure the total intracellular pools of ATP, ADP and AMP. It is possible that adiponectin could cause a small, transient increase in AMP, which could even be localized to a specific region of the cell. If this were the case it is very possible that we might not detect such a change. Unfortunately, at the moment the techniques that would allow us to monitor such changes are not available and so we are unable to address this question. Although we have not been able to measure changes in AMP in response to adiponectin, we do know that LKB1 is required. From many other studies we know that signals that activate AMPK by increasing AMP act via LKB1, by inhibiting dephosphorylation, as I mentioned earlier. It is tempting to speculate that since adiponectin activates AMPK in an LKB1dependent manner, it will also signal through an increase in AMP. Alternatively, of course, it could be that there is some signalling molecule, other than AMP, that activates AMPK in response to adiponectin, although no-one has described anything like this to date. Hotamisligil: I can make a simple suggestion. If you take AMPK as an example, you have a hard time thinking of its activation in the absence of AMP regulation, but it is responding to a lot of hormones and regulation of activity depends on a kinase. It might well be in the signal transduction pathway on one of the receptors for these hormones. The fact that you see AMPK regulation in opposite directions in the brain and in the periphery argues strongly that there is an AMP-independent regulation of this pathway. Carling: There is certainly an AMP-independent mechanism. We and others recently showed that Ca2+ acts on the AMPK cascade (Hawley et al 2005, Hurley et al 2005, Woods et al 2005). Signals which activate AMPK by increasing Ca2+
AMPK AND ENERGY HOMEOSTASIS
83
and activating CaMKK β can occur independently of changes in AMP. However, when it comes to LKB1-dependent activation of AMPK there is a problem, since LKB1 is constitutively active. So in this case, it can’t be a simple signal transduction in the way we would normally think of it, because the upstream kinase is switched on. Hotamisligil: Not really. You can have a constitutive kinase that is recruited to different complexes. Carling: Yes, and this is related to the point I was trying to illustrate with my analogy to the PDK1 system. So, for example, although PDK1 is constitutively active, phsophorylation and activation of PKB occurs via a membrane localization mediated by phosphoinositide. I think a similar type of model could explain the activation of AMPK by LKB1, but in this case AMP regulates the dephosphorylation rather than causing relocalization or recruitment of the complexes. Hotamisligil: Has anyone done the experiment of looking at whether some of the subunits of AMPK associate with leptin or adiponectin receptors? Carling: I’m not aware of anyone reporting a direct interaction of AMPK with either of these receptors. A recent study reported the identification of a protein, termed APPL1, that interacts with the cytoplasmic tail of the adiponectin receptor (Mao et al 2006). One possibility, therefore, would be that APPL1 could bind to LKB1 or AMPK, acting as a scaffold with the adiponectin receptor. This could provide a basis for a potential mechanism explaining activation of AMPK by adiponectin, although this is very speculative. We have no direct evidence that AMPK associates with plasma membranes, so I’m not particularly keen on this putative mechanism. Bernlohr: Much of the discussion on AMPK is focused around glucose metabolism and ATP:ADP ratios. But there are many metabolic processes that change nucleotide levels. Do these also play a role in regulating AMPK? I’m thinking of protein degradation and requirement for ubiquitination: utilization of ATP and the production of AMP is part of that reaction. Is this a physiological regulator of the AMPK system? Carling: In whatever cellular system that an increase in AMP can be detected, AMPK is activated. So, if this process leads to an increase in AMP levels, I suspect AMPK would be activated. Bernlohr: What about transient changes in cyclic nucleotide hydrolysis producing AMP? Are these feasible regulators locally or transiently of AMPK? Carling: It would be interesting to determine whether breakdown of cAMP to AMP would produce enough AMP to lead to activation. One of the original observations in this field was that glucagon in the liver decreases lipogenesis by causing an increase in phosphorylation of acetyl-CoA carboxylase (ACC). It turns out that glucagon does this by causing increased phosphorylation of ACC via AMPK, although the precise molecular mechanism has not been elucidated. This particular system is complicated because it has recently been shown that PKA can
84
CARLING
directly phosphorylate AMPK. However, in a recent study (Hurley et al 2006), Lee Witters and colleagues have suggested that raising cAMP levels in intact cells leads to inhibition of AMPK i.e. PKA phosphorylation of AMPK is antagonistic to its activation. Attie: Along the same lines as David Bernlohr’s question, Uyeda published a paper (Kawaguchi et al 2002) arguing something similar, but he was talking about fatty acid activation being a major source of AMP production. The capacity of cells for fatty acid uptake and activation is extremely high. This would be a large amount, but the numbers in the paper didn’t seem right. Carling: I was involved in a collaboration with David Saggerson where we looked at the question of whether the AMP produced following esterification of fatty acids could activate AMPK. David was interested in determining what mechanism could be responsible for a ‘feed-forward’ activation of fatty acid oxidation. One potential mechanism would be via an increase in fatty acid transport into the cell and its subsequent esterification, which would in turn cause an increase in AMP and activation of AMPK, driving fatty acid oxidation. However, in the system we were studying, which was the perfused rat heart, we did not detect an increase in AMP, despite seeing activation of AMPK (Clark et al 2004). Attie: Did this involve adding fatty acids to cells? Carling: Yes. This was in response to perfusion with physiological concentrations of palmitate or oleate. Of course we are faced with the same problem I alluded to earlier—namely that it is possible that these conditions do lead to small changes in AMP that we are unable to detect by the methods we have available at present. A solution to this would be to develop a method for determining AMP levels in specific regions within the cell and this is something we are very keen to explore. Kim: Several researchers have suggested that the metformin cases don’t change the ATP:AMP ratio, but they do change the mitochondrial function. This is how they can modulate the AMPK activity. Have you thought about this? Spiegelman: Mitochondria are going to play into all these pathways quite directly. Carling: Perhaps, but whether this is the whole story remains to be elucidated. Certainly, if the AMP:ATP ratio increases within the cell, AMPK is activated—as long as LKB1 is present. Again, with metformin we have not been able to measure changes in AMP, even though AMPK is activated. I think it is still a possibility that there is another signal, not AMP, that could mediate activation. It would be tempting to speculate that this is linked to mitochondrial function, but that is just speculation. Spiegelman: Furthermore, mitochondria act as a sink for Ca2+ and this storage depends on the membrane potential. Things that lower the membrane potential cause Ca2+ release from mitochondria.
AMPK AND ENERGY HOMEOSTASIS
85
References Clark H, Carling D, Saggerson D 2004 Covalent activation of heart AMP-activated protein kinase in response to physiological concentrations of long-chain fatty acids. Eur J Biochem 271:2215–2224 Hawley SA, Pan DA, Mustard KJ et al 2005 Calmodulin-dependent protein kinase kinase-beta is an alternative upstream kinase for AMP-activated protein kinase. Cell Metab 2:9–19 Hurley RL, Anderson KA, Franzone JM, Kemp BE, Means AR, Witters LA 2005 The Ca2+/ calmodulin-dependent protein kinase kinases are AMP-activated protein kinase kinases. J Biol Chem 280:29060–29066 Hurley RL, Barre LK, Wood SD et al 2006 Regulation of AMP-activated protein kinase by multisite phosphorylation in response to agents that elevate cellular cAMP. J Biol Chem 281:36662–36672 Kawaguchi, T, Osatomi K, Yamashita H et al 2002 Mechanism for fatty acid ‘sparing’ effect on glucose-induced transcription: regulation of carbohydrate-responsive element-binding protein by AMP-activated protein kinase. J Biol Chem 277:3829–3835 Lizcano JM, Goransson O, Toth R et al 2004 LKB1 is a master kinase that activates 13 kinases of the AMPK subfamily, including MARK/PAR-1. EMBO J 23:833–843 Mao X, Kikani CK, Riojas RA et al 2006 APPL1 binds to adiponectin receptors and mediates adiponectin signalling and function. Nat Cell Biol 8:516–523 Woods A, Dickerson K, Heath R et al 2005 Ca2+/calmodulin-dependent protein kinase kinasebeta acts upstream of AMP-activated protein kinase in mammalian cells. Cell Metab 2:21–33
Fatty Acids and Lipotoxicity in Obesity and Diabetes: Novartis Foundation Symposium 286, Volume 286. Edited by Gregory Bock and Jamie Goode Copyright Novartis Foundation 2007. ISBN: 978-0-470-05764-3
Endoplasmic reticulum stress and inflammation in obesity and type 2 diabetes Gökhan S. Hotamisligil Department of Genetics and Complex Diseases, Harvard School of Public Health, Boston, MA 02115, USA
Abstract. Metabolic and immune systems are among the most fundamental requirements for survival, and mechanisms integrating and co-ordinating the activities of these responses have been evolutionarily highly conserved throughout the species. Disruption of the interface between inflammatory and metabolic pathways, most notably insulin action, is central to the pathogenesis of a cluster of chronic metabolic diseases, particularly obesity, insulin resistance, type 2 diabetes and cardiovascular disease, which collectively constitute the greatest threat to the global human health and welfare. The c-Jun N-terminal kinase ( JNK) is a critical mediator linking inflammatory signals to insulin resistance in obesity. In recent years, we have demonstrated that endoplasmic reticulum (ER) dysfunction and the integrated stress responses are important in the emergence of abnormal JNK activity, inflammatory responses, and insulin resistance in obesity. Blocking JNK activity through chemical of genetic means or targeting ER function through chemical chaperones or by genetics leads to marked metabolic improvement and normalization of glucose metabolism in mice models of obesity. 2007 Fatty acids and lipotoxicity in obesity and diabetes. Wiley, Chichester (Novartis Foundation Symposium 286) p 86–98
Obesity and the associated metabolic diseases now constitute a major threat to global human health regardless of geographical or socio-economical boundaries. Hence, understanding of the pathogenesis and molecular mechanisms underlying these diseases are of critical importance for the development of effective and novel preventive and/or therapeutic measures. Obesity, insulin resistance and type 2 diabetes are closely associated with chronic low-grade inflammation (Hotamisligil 2006). Mounting experimental, epidemiological and clinical evidence has linked inflammation, and the molecules and networks integral to inflammatory responses, to the development of these metabolic diseases, particularly in the context of obesity and type 2 diabetes. Interestingly, a central site for the inflammatory alterations in obesity is the adipose tissue (Hotamisligil et al 1993). At this site, numerous inflammatory cytokines, chemok86
ER STRESS IN OBESITY AND DIABETES
87
ines and acute-phase reactants are produced in an obesity-dependent manner (Shoelson 2006, Hotamisligil et al 1993, Wellen & Hotamisligil 2005). Adipose tissue also recruits inflammatory cells such as macrophages during the course of obesity and the presence of these cells contributes to chronic inflammation and deterioration of metabolic homeostasis. Obesity-induced inflammation is causal to development of insulin resistance as suppression of these pathways through pharmacological or genetic means result in enhanced insulin action and metabolic control in mice (Hotamisligil 2006, Uysal et al 1997, Ventre et al 1997, Hirosumi et al 2002, Yuan et al 2001). Since obesity results in the production of an array of inflammatory mediators, mechanisms controlling the expression and production of these integrated inflammatory and stress responses, as well as their impact on insulin action, are likely to be of major importance to the development type 2 diabetes and other complications of obesity. Insulin generates its biological functions through a complex network of signalling events triggered upon its binding to the insulin receptor (Taniguchi et al 2006). Among these, a critical event mediating insulin action is the stimulation of tyrosine phosphorylation of IRS proteins. In the overwhelming majority of cases of insulin resistance, this and further downstream steps of insulin receptor signalling are defective in experimental models and in humans (Wellen & Hotamisligil 2005, Saltiel & Kahn 2001). Tumour necrosis factor (TNF) β and other cytokines, lipid mediators and other stress pathways also target this element of insulin receptor signalling through an inhibitory serine phosphorylation of IRS1 (Hotamisligil 2006). Serine phosphorylation of IRS1 and inhibition of insulin receptor signalling is universally observed in insulin resistant cells and tissues, animal models and humans with type 2 diabetes (Taniguchi et al 2006, Aguirre et al 2000, Paz et al 1997). It is now established that IRS1 is phosphorylated by a variety of kinases, which interferes with the ability of this protein to engage in insulin receptor signalling and results in alterations in insulin action (Zick 2005). Among these IRS-modifying enzymes, mounting evidence indicate that activation of c-Jun N-terminal kinase (JNK), inhibitor of NF-κ B kinase (IKK), and protein kinase C (PKC) are central to mediating insulin resistance in response to a variety of stresses that occur in obesity, and other conditions of insulin resistance. They are all reported to be able to inhibit insulin action by serine phosphorylation of IRS1 (Hirosumi et al 2002, Aguirre et al 2000, Gao et al 2003, Griffin et al 1999), although the activity of IKK in this regard is not yet well established under physiological conditions. IRS1 serine phosphorylation disrupts insulin receptor signalling via several distinct mechanisms and blocks insulin action (Zick 2005, Hotamisligil et al 1996, Gao et al 2002). These kinases also exert powerful effects on gene expression, including promoting further inflammatory mediators through transcriptional activation of activator protein 1 (AP1) complexes and NF-κ B (Baud
88
HOTAMISLIGIL
& Karin 2001). The transcriptional networks that also impact insulin action and metabolic pathways through these signalling molecules have not yet been characterized in metabolic disease context. Most importantly, JNK activity is significantly increased in obesity in several critical metabolic sites such as adipose, muscle and liver tissues (Hirosumi et al 2002). JNK can incorporate many stress signals and is activated upon exposure to cytokines such as TNFβ, as well as by free fatty acids (Aguirre et al 2000, Gao et al 2002). JNK is also activated during ER stress inducing conditions in vitro (Urano et al 2000) and recent studies in our lab demonstrated that ER stress might underlie the obesity-induced increase in JNK activity and consequently, inflammatory responses and insulin resistance (see below). Genetic JNK1-deficiency protects mice from obesity induced JNK activation, IRS1 serine phosphorylation, and consequently insulin resistance, fatty liver and diabetes (Hirosumi et al 2002, Tuncman et al 2006). Recent studies examining the functional interactions between JNK isoforms revealed that JNK2 also participates in metabolic regulation, but that this function is normally masked due to compensation by JNK1 in the absence of JNK2 (Tuncman et al 2006). In JNK2 –/– mice, JNK1-haploinsufficiency produces a phenotype essentially identical to total JNK1-deficiency. In addition, JNK2 activity has been implicated in the pathogenesis of type 1 diabetes and atherosclerosis (Jaeschke et al 2005, Ricci et al 2004). No metabolic action has yet been described for the JNK3 isoform. In the pathological states mentioned above, it is likely that total JNK activity level and duration are the critical determinants of pathology, since a variety of interventions to block JNK activity in established models of obesity and diabetes improved systemic glucose homeostasis and insulin sensitivity, as well as atherosclerosis (Ricci et al 2004, Kaneto et al 2004, Liu & Rondinone 2005). It is important to note that mutations in JNK binding protein (JIP) also cause abnormal JNK activity and type 2 diabetes in humans (Waeber et al 2000). Taken together, these findings indicate that JNK activation is universal in obesity and diabetes and its inhibition may be a promising therapeutic avenue for diabetes in humans. IKK β is another inflammatory kinase that is critical in the development of insulin resistance and metabolic dysfunction (Shoelson et al 2006). In cultured cells, blocking IKK β function renders cells resistant to TNF-induced insulin resistance. Mice heterozygous for a null mutation in IKK β are partially protected from obesity-induced insulin resistance, and inhibition of IKKβ by administration of high dose salicylates improves insulin action in experimental models and humans (Yuan et al 2001, Hundal et al 2002). Studies involving tissue- or cell-type specific modulation of IKK β reveal that its activity in liver impacts systemic metabolism, but that its role in muscle is probably less important in terms of metabolic regulation (Cai et al 2005, Rohl et al 2004). Similar to JNK, activation of IKK β in liver alone appears to be sufficient to generate severe systemic insulin resistance (Cai
ER STRESS IN OBESITY AND DIABETES
89
et al 2005). Interestingly, myeloid-specific deletion of IKK β also results in partial protection against obesity- or LPS-induced insulin resistance (Arkan et al 2005), further demonstrating the potential role of inflammatory cells in the deterioration of metabolic homeostasis in obesity. Currently, clinical trials are underway to address the therapeutic profi le of high dose salicylate, which can also block IKK β and JNK, in diabetic humans (Shoelson et al 2006). Other molecules that are also critical in the inflammatory-metabolic cross-talk include protein kinase C (Yu et al 2002, Kim et al 2004, Boden et al 2005), the suppressor of cytokine signalling (SOCS) molecules, SOCS1, 3 and 6 (Emanuelli et al 2000, 2001, Mooney et al 2001, Shi et al 2006), toll-like receptors, and nuclear hormone receptors expressed in macrophages and other metabolically important cell types (Shi et al 2006, Song et al 2006, Lin et al 2000, Bookout et al 2006, Ogawa et al 2005). Like JNK and IKK, most of these molecules also appear to inhibit insulin action at the level of insulin receptor substrates either by blocking signalling or promoting degradation of insulin receptor substrates (Hotamisligil 2006, Zick 2005). Due to limitations of space, these pathways are not described in detail here. Studies in our lab attempting to address the mechanisms involved in sensing the stresses associated with energy surplus and obesity and leading to activation of inflammatory pathways resulted in the discovery of endoplasmic reticulum (ER) dysfunction as a critical feature in obesity and type 2 diabetes (Ozcan et al 2004). ER is a vast network of membranes where all the secretory and membrane proteins are assembled into their secondary and tertiary structures. Proper folding, maturation, storage and transport of these proteins takes place in this organelle. Unfolded or misfolded proteins are detected and removed from the ER through the 26S proteosome system. Accumulation of unfolded proteins, energy and nutrient fluctuations, hypoxia, toxins, viral infections, and increased demand on the synthetic machinery give rise to perturbations in the ER lumen and create ER stress. Under these conditions, ER activates a complex response system called UPR (unfolded protein response) to restore the functional integrity of the organelle. The details of UPR sensing and signalling systems are not covered here (Marciniak & Ron 2006, Zhao & Ackerman 2006, Zhang & Kaufman 2006). However, it is important to note that the two critical kinases involved in inhibition of insulin action, JNK and IKK β, are both closely linked to ER stress and UPR. Since ER is highly responsive to nutrient and energy fluxes and its stress is linked to JNK and IKK activation, it is possible to postulate that ER is the site for the sensing of metabolic stress and translation of that stress into broader inflammatory signalling and responses. In fact, ER could be considered an essential and ancient site of integration between nutrient and pathogen responses as it is exquisitely sensitive to glucose and energy availability, lipids, pathogens and pathogenassociated components. In addition to increased demand on the synthetic capacity
90
HOTAMISLIGIL
at several organs such as liver and adipose tissue, increased adiposity in obesity produces an environment that further challenges ER function and capacity with architectural constraints limiting ER expansion as well as altered energy and nutrient availability. Obesity provides many of these conditions that could lead to ER stress (Hotamisligil 2005). In fact, work in our laboratory has recently demonstrated that ER stress is increased in both dietary and genetic obesity in adipose and liver tissues (Ozcan et al 2004, Hotamisligil 2005). In cellular systems, induction of ER stress leads to insulin resistance, at least in part through IRE-1-dependent activation of JNK. Modulation of ER folding capacity through gain- and loss-of function studies with XBP1 demonstrated a close link between ER function and insulin action in cells and animal models. In mice, XBP1 haploinsufficiency results in predisposition to diet induced insulin resistance and type 2 diabetes (Ozcan et al 2004). These initial observations were followed by additional independent studies linking ER function to insulin sensitivity. For example, decreasing the expression of an ER resident chaperone, oxygen regulated protein 150 (ORP150) leads to insulin resistance (Ozawa et al 2005). On the other hand, whole body or liver-specific overexpression of the chaperone ORP150 in a mouse model of diabetes improves insulin sensitivity and glucose tolerance (Ozawa et al 2005, Nakatani et al 2005). Together, these results indicate that ER stress plays an important role in mediating JNK activity and insulin resistance in obesity and that increasing cellular folding capacity may be a promising therapeutic approach. Interestingly, genetic variations in several genes that are critical in ER function have recently been linked to insulin resistance in humans (Kovacs et al 2002, Chu et al 2007). While ER stress has been causally linked to islet survival and function in both experimental models and humans (Marciniak & Ron 2006), further studies will be necessary to defi nitively determine the role of ER stress in insulin action and metabolic regulation in humans (Zhao & Ackerman 2006, Zhang & Kaufman 2006) and link defects in insulin and secretion via this mechanism. In preclinical studies, we have recently demonstrated that small molecule chemical chaperones are extremely effective in alleviating obesity-induced ER stress and treating insulin resistance and diabetes in mice (Ozcan et al 2006). Among the chemical chaperones tested, 4-phenyl butyric acid (PBA) and taurine-conjugated ursodeoxycholic acid (TUDCA) demonstrated strong capacity to alleviate ER stress in both cultured cells and in whole animals. Administration of these compounds to a severe genetic model of obesity and insulin resistance (ob/ob, leptindeficient model) resulted in normalization of blood glucose levels and systemic insulin resistance within 4–6 days of treatment. Concomitant with the correction of biochemical signs of ER stress in liver and adipose tissue, chemical chaperone administration resulted in complete reversal of the obesity-induced JNK activation and IRS1 serine phosphorylation and consequently recovered insulin receptor signalling (Ozcan et al 2006). Since both PBA and TUDCA have been safely used
ER STRESS IN OBESITY AND DIABETES
91
in humans they could be tested for potential proof of principle for efficacy in type 2 diabetes. ER stress also provides multiple links with the emergence of inflammatory responses. First, and as stated above, ER stress leads to activation of both JNK and IKK (Urano et al 2000, Deng et al 2004, Hu et al 2006). Second, inflammatory mediators can trigger ER stress and lead to propagation of general stress responses (Zhang et al 2006). Third, ER stress leads to activation of CREB-H, which then plays a significant role in inflammatory and acute phase responses in liver (Zhang et al 2006). Fourth, ER is a major source of reactive oxygen species, and consequently oxidative stress in the cells (Cullinan & Diehl 2006, Xue et al 2005). Oxidative stress is emerging as a feature of obesity and important player in the development of insulin resistance in obesity (Koshkin et al 2003, Furukawa 2004, Houstis et al 2006). Taken together, the effector arms of ER stress and the associated stress responses are tightly linked to inflammatory pathways at many levels that are all critical for insulin action and metabolic homeostasis. It remains to be tested whether the metabolic regulation through ER is causally linked to these pathways in vivo. In summary, discovery of the role of ER dysfunction in obesity and type 2 diabetes is likely to open new avenues for mechanistic exploration of metabolic diseases and development of new therapeutic opportunities for these devastating disorders. Acknowledgements I thank members of my laboratory for their contributions with support from the National Institutes of Health, American Diabetes Foundation, Pew Foundation, Sandler Foundation, and Harvard School of Public Health. I regret the inadvertent omission of many references due to limitations in space.
References Aguirre V, Uchida T, Yenush L, Davis R, White MF 2000 The c-Jun NH(2)-terminal kinase promotes insulin resistance during association with insulin receptor substrate-1 and phosphorylation of Ser(307). J Biol Chem 275:9047–9054 Arkan MC, Hevener AL, Greten FR et al 2005 IKK-beta links inflammation to obesity-induced insulin resistance. Nat Med 11:191–198 Baud V, Karin M 2001 Signal transduction by tumor necrosis factor and its relatives. Trends Cell Biol 11:372–377 Boden G, She P, Mozzoli M et al 2005 Free fatty acids produce insulin resistance and activate the proinflammatory nuclear factor-kappaB pathway in rat liver. Diabetes 54: 3458–3465 Bookout AL, Jeong Y, Downes M, Yu RT, Evans RM, Mangelsdorf DJ 2006 Anatomical profi ling of nuclear receptor expression reveals a hierarchical transcriptional network. Cell 126:789–799 Cai D, Yuan M, Frantz DF et al 2005 Local and systemic insulin resistance resulting from hepatic activation of IKK-beta and NF-kappaB. Nat Med 11:183–190
92
HOTAMISLIGIL
Chu WS, Das SK, Wang H 2007 Activating transcription factor 6 (ATF6) sequence polymorphisms in type 2 diabetes and pre-diabetic traits. Diabetes 56:856–862 Cullinan SB, Diehl JA 2006 Coordination of ER and oxidative stress signaling: the PERK/ Nrf2 signaling pathway. Int J Biochem Cell Biol 38:317–332 Deng J, Lu PD, Zhang Y et al 2004 Translational repression mediates activation of nuclear factor kappa B by phosphorylated translation initiation factor 2. Mol Cell Biol 24:10161– 10168 Emanuelli B, Peraldi P, Filloux C et al 2000 SOCS-3 is an insulin-induced negative regulator of insulin signaling. J Biol Chem 275:15985–15991 Emanuelli B, Peraldi P, Filloux C et al 2001 SOCS-3 inhibits insulin signaling and is upregulated in response to tumor necrosis factor-alpha in the adipose tissue of obese mice. J Biol Chem 276:47944–47949 Furukawa S, Fujita T, Shimabukuro M et al 2004 Increased oxidative stress in obesity and its impact on metabolic syndrome. J Clin Invest 114:1752–1761 Gao Z, Hwang D, Bataille F et al 2002 Serine phosphorylation of insulin receptor substrate 1 by inhibitor kappa B kinase complex. J Biol Chem 277:48115–48121 Gao Z, Zuberi A, Quon MJ, Dong Z, Ye J 2003 Aspirin inhibits serine phosphorylation of insulin receptor substrate 1 in tumor necrosis factor-treated cells through targeting multiple serine kinases. J Biol Chem 278:24944–24950 Griffi n ME, Marcucci MJ, Cline GW et al 1999 Free fatty acid-induced insulin resistance is associated with activation of protein kinase C theta and alterations in the insulin signaling cascade. Diabetes 48:1270–1274 Hotamisligil GS 2005 Role of endoplamic reticulum stress and c-Jun NH2-terminal kinase pathways in inflammation and origin of obesity and diabetes. Diabetes 54 (Suppl 2): S73–78 Hotamisligil GS 2006 Inflammation and metabolic disorders. Nature 444:860–867 Hotamisligil GS, Shargill NS, Spiegelman BM 1993 Adipose expression of tumor necrosis factor-alpha: direct role in obesity-linked insulin resistance. Science 259:87–91 Hotamisligil GS, Peraldi P, Budavari A, Ellis R, White MF, Spiegelman BM 1996 IRS-1mediated inhibition of insulin receptor tyrosine kinase activity in TNF-alpha- and obesityinduced insulin resistance. Science 271:665–668 Hirosumi J, Tuncman G, Chang L et al 2002 A central role for JNK in obesity and insulin resistance. Nature 420:333–336 Houstis N, Rosen ED, Lander ES 2006 Reactive oxygen species have a causal role in multiple forms of insulin resistance. Nature 440:944–948 Hu P, Han Z, Couvillon AD, Kaufman RJ, Exton JH 2006 Autocrine tumor necrosis factor alpha links endoplasmic reticulum stress to the membrane death receptor pathway through IRE1alpha-mediated NF-kappaB activation and down-regulation of TRAF2 expression. Mol Cell Biol 26:3071–3084 Hundal RS, Petersen KF, Mayerson AB et al 2002 Mechanism by which high-dose aspirin improves glucose metabolism in type 2 diabetes. J Clin Invest 109:1321–1326 Jaeschke A, Rincon M, Doran B et al 2005 Disruption of the Jnk2 (Mapk9) gene reduces destructive insulitis and diabetes in a mouse model of type I diabetes. Proc Natl Acad Sci USA 102:6931–6935 Kaneto HNY, Miyatsuka T, Kawamori D et al 2004 Possible novel therapy for diabetes with cell-permeable JNK-inhibitory peptide. Nat Med 10:1128–1132 Kim JK, Fillmore JJ, Sunshine MJ et al 2004 PKC-theta knockout mice are protected from fat-induced insulin resistance. J Clin Invest 114:823–827 Koshkin V, Wang X, Scherer PE, Chan CB, Wheeler MB 2003 Mitochondrial functional state in clonal pancreatic beta-cells exposed to free fatty acids. J Biol Chem 278:19709– 19715
ER STRESS IN OBESITY AND DIABETES
93
Kovacs P, Yang X, Permana PA, Bogardus C, Baier LJ 2002 Polymorphisms in the oxygenregulated protein 150 gene (ORP150) are associated with insulin resistance in Pima Indians. Diabetes 51:1618–1621 Lin Y, Lee H, Berg AH, Lisanti MP, Shapiro LS, Scherer PE 2000 The lipopolysaccharideactivated toll-like receptor (TLR)-4 induces synthesis of the closely related receptor TLR-2 in adipocytes. J Biol Chem 275:24255–24263 Liu G, Rondinone CM 2005 JNK: bridging the insulin signaling and inflammatory pathway. Curr Opin Investig Drugs 6:979–987 Marciniak SJ, Ron D 2006 Endoplasmic reticulum stress signaling in disease. Physiol Rev 86:1133–1149 Mooney RA, Senn J, Cameron S et al 2001 Suppressors of cytokine signaling-1 and -6 associate with and inhibit the insulin receptor. A potential mechanism for cytokine-mediated insulin resistance. J Biol Chem 276:25889–25893 Ogawa S, Lozach J, Benner C et al 2005 Molecular determinants of crosstalk between nuclear receptors and toll-like receptors. Cell 122:707–721 Ozawa K, Miyazaki M, Matsuhisa M 2005 The endoplasmic reticulum chaperone improves insulin resistance in type 2 diabetes. Diabetes 54:657–663 Ozcan U, Cao Q, Yilmaz E et al 2004 Endoplasmic reticulum stress links obesity, insulin action, and type 2 diabetes. Science 306:457–461 Ozcan U, Yilmaz E, Ozcan L et al 2006 Chemical chaperones reduce ER stress and restore glucose homeostasis in a mouse model of type 2 diabetes. Science 313:1137–1140 Nakatani Y, Kaneto H, Kawamori D et al 2005 Involvement of endoplasmic reticulum stress in insulin resistance and diabetes. J Biol Chem 280:847–851 Paz K, Hemi R, LeRoith D et al 1997 A molecular basis for insulin resistance. Elevated serine/ threonine phosphorylation of IRS-1 and IRS-2 inhibits their binding to the juxtamembrane region of the insulin receptor and impairs their ability to undergo insulin-induced tyrosine phosphorylation. J Biol Chem 272:29911–29918 Ricci R, Sumara G, Sumara I et al 2004 Requirement of JNK2 for scavenger receptor Amediated foam cell formation in atherogenesis. Science 306:1558–1561 Rohl M, Pasparakis M, Baudler S et al 2004 Conditional disruption of IkappaB kinase 2 fails to prevent obesity-induced insulin resistance. J Clin Invest 113:474–481 Saltiel AR, Kahn CR 2001 Insulin signaling and the regulation of glucose and lipid metabolism. Nature 414:799–806 Shi H, Kokoeva MV, Inouye K, Tzameli I, Yin H, Flier JS 2006 TLR4 links innate immunity and fatty acid-induced insulin resistance. J Clin Invest 116:3015–3025 Shoelson SE, Lee J, Goldfi ne AB 2006 Inflammation and insulin resistance. J Clin Invest 116:1793–1801 Song MJ, Kim KH, Yoon JM, Kim JB 2006 Activation of Toll-like receptor 4 is associated with insulin resistance in adipocytes. Biochem Biophys Res Commun 346:739–745 Taniguchi CM, Emanuelli B, Kahn CR 2006 Critical nodes in signalling pathways: insights into insulin action. Nat Rev Mol Cell Biol 7:85–96 Tuncman G, Hirosumi J, Solinas G, Chang L, Karin M, Hotamisligil GS 2006 Functional in vivo interactions between JNK1 and JNK2 isoforms in obesity and insulin resistance. Proc Natl Acad Sci USA 103:10741–10746 Urano F, Wang X, Bertolotti A et al 2000 Coupling of stress in the ER to activation of JNK protein kinases by transmembrane protein kinase IRE1. Science 287:664–666 Uysal KT, Wiesbrock SM, Marino MW, Hotamisligil GS 1997 Protection from obesity-induced insulin resistance in mice lacking TNF-alpha function. Nature 389:610–614 Ventre J, Doebber T, Wu M et al 1997 Targeted disruption of the tumor necrosis factoralpha gene—metabolic consequences in obese and nonobese mice. Diabetes 46:1526– 1531
94
HOTAMISLIGIL
Waeber G, Delplanque J, Bonny C et al 2000 The gene MAPK8IP1, encoding islet-brain-1, is a candidate for type 2 diabetes. Nat Genet 24:291–295 Wellen KE, Hotamisligil GS 2005 Inflammation, stress, and diabetes. J Clin Invest 115:1111–1119 Yu C, Chen Y, Cline GW et al 2002 Mechanism by which fatty acids inhibit insulin activation of insulin receptor substrate-1 (IRS-1)-associated phosphatidylinositol 3-kinase activity in muscle. J Biol Chem 277:50230–50236 Yuan M, Konstantopoulos N, Lee J et al 2001 Reversal of obesity- and diet-induced insulin resistance with salicylates or targeted distruption of Ikkbeta. Science 293:1673–1677 Zick Y 2005 Ser/Thr phosphorylation of IRS proteins: a molecular basis for insulin resistance. Sci STKE 2005(268):pe4 Zhang K, Kaufman RJ 2006 The unfolded protein response: a stress signaling pathway critical for health and disease. Neurology 66(2 Suppl 1):S102–109 Zhang K, Shen X, Wu J et al 2006 Endoplasmic reticulum stress activates cleavage of CREBH to induce a systemic inflammatory response. Cell 124:587–599 Zhao L, Ackerman SL 2006 Endoplasmic reticulum stress in health and disease. Curr Opin Cell Biol 18:444–452 Xue X, Piao J, Nakajima A et al 2005 Tumor necrosis factor alpha (TNFalpha) induces the unfolded protein response (UPR) in a reactive oxygen species (ROS)-dependent fashion, and the UPR counteracts ROS accumulation by TNFalpha. J Biol Chem 280:33917–33925
DISCUSSION Spiegelman: The effects you have seen on insulin sensitivity in diabetes are remarkable. Are you impacting food intake, obesity and energy balance at all? Are these remarkable effects independent of the effects of food intake and body composition? Hotamisligil: The effects of chemical chaperones are largely independent of all other metabolic changes. With PBA there is no difference in metabolic rates, food intake and physical activity. With TUDCA there is some effect on body weight, but this is only 5–10% difference. I would say that the insulin-sensitizing effects are largely independent of obesity and do not impact food intake or physical activity. Attie: Are you convinced that the effect you see is purely a chaperone effect? Have you thought about using a stereoisomer? Hotamisligil: There is no question that these compounds act as chemical chaperones to modulate ER. It is very difficult to achieve this anywhere outside the ER. We see the same effects within 10 d of administration in the animals. So, it is clear that these agents act to regulate ER. However, I can not state that it is the only reason they produce this strong antidiabetic activity. If you are giving a lot of drug it is possible that you get additional hits. O’Rahilly: You mentioned that there was a 10% difference in body weight. Our experience with the Ob/Ob mouse is that this is enough to get quite a big change in insulin sensitivity. It is easy to underestimate the effect that a small decrement in body weight gained can have.
ER STRESS IN OBESITY AND DIABETES
95
Muoio: Can you comment on how systemic lipids are affected by the treatment? Also, what is the involvement of ER stress as a mechanism that is universally operative in different tissues? One of your first papers suggested that ER stress was not noticeable in skeletal muscle, yet here you show a robust effect of deoxycholic acid on the muscle tissue. Hotamisligil: In our original work, other than JNK activation we couldn’t detect other signs of ER stress in the muscle tissue in obese mice. Others have since reported regulation of chaperones themselves in the muscle tissue. I am not sure that muscle is the principle target for this mechanism. Having said that, I must mention that the reagents used to examine ER stress, especially in vivo are far from perfect. It is difficult to detect an increase unless it is a robust one. XBP splicing is a dynamic event, because it indicates both the presence of stress and also compensation for the stress. So it is tough to get a clear readout by just monitoring XPP splicing. We can’t rule out the possibility of ER stress in other sites, but we didn’t see it in the original study. Here we see an increase in muscle insulin sensitivity, but again this could be secondary—we can’t ascribe it to direct effects. With respect to lipids, we don’t see much effect on systemic lipid levels. The only strong lipid effect seen is hepatic triglyceride accumulation. Bernlohr: What are the initiating events in obesity that lead to the stress itself? It is unequivocal that the stress is the major phenomenon, but what are some of the molecular bases for how the stress actually happens? Hotamisligil: In all these pathways there is a chicken and egg problem. I can envisage a scenario where there is excess entry of lipids into the ER. In liver, obesity could cause an increase in protein synthesis and stress the ER, although some people think that this isn’t possible. Liver synthesizes 90 million protein molecules per minute. In obesity, liver extracts have twice the amount of protein, so there is enormous demand on protein synthesis in liver. Lipid load could cause an increase in protein synthesis. TNF could be activating ER stress, and ER stress could be producing TNF. ROS activate ER stress, but can also be produced as a result of ER stress. My gut reaction is that the initiating event is metabolic in origin, whether it is a change in synthetic demand on liver, on fat and on islets: places where there is a significant increase in protein load and lipid trafficking. On top of that, the events we discussed can all trigger all of the pathways I mentioned, which feed on each other. Bernlohr: You showed Ca2+ fluxes. Do these change across the ER in lean and obese situations? Hotamisligil: I don’t have a clear answer for this. There isn’t clear evidence yet that Ca2+ fluxes are disrupted in the course of obesity. Tabas: Increased client load—such as a β cell making lots of insulin or a plasma cell making a lot of IgG—results in UPAR being activated. If it wasn’t, you’d be
96
HOTAMISLIGIL
in trouble. This is the concept. One nice tool is that a little bit of cycloheximide under cases of increased client load corrects the situation. You can relieve ER stress with cycloheximide. It makes the cells a lot less stressful. In your cell culture models, the question would be can you reverse these effects by adding cycloheximide? Hotamisligil: We haven’t done this. Here we are using tunicamycin and thapsigargin to induce ER stress. I have to admit I am not a great fan of these models, but there aren’t many alternatives. Tabas: In vitro, do you have any model that you think mimics the in vivo situation? Hotamisligil: We have one, which is a defective protein synthesis model. In such a model we get ER stress spontaneously. In this system, we can rescue insulin action by using chemical chaperones but these are early stages. Tabas: If you have a hepatocyte model, with increased fatty acid flux, that would be interesting. The Ca2+ idea you brought up is fascinating. This is something else that could be examined experimentally. The causation experiment is to add Ca2+ buffers such as BAPTA to the cell. For the processes that we have studied where the trigger is Ca2+ , we can reverse it all by BAPTA. Redox state is another example. For each of these ideas, it seems overwhelming when you are approaching them, but if you take each on its own and design the proper causation experiment, it becomes malleable. Hotamisligil: In culture this can be done. For example, if you treat liver cells with palmitate, you can induce severe ER stress. But it is hard to say that this is the trigger in the whole organism. It is challenging in vivo to test individual triggers but perhaps doable with careful designs for select pathways. Bernlohr: My problem is that when I try to draw these out, there are some inducers that go clockwise, and others that go counter-clockwise in the cycle. Spiegelman: What about the role of ROS? In principle you could design experiments to neutralize ROS either enzymatically or chemically and ask whether it is lipid overload hitting the liver cell in disproportion to the ATP requirements of the cell as a triggering mechanism. Where would you guess that ROS fits in this scheme: up or downstream? Hotamisligil: I would say that ROS is a cumulative result of all of the dysfunctional organelles, principally mitochondria and ER. It is probably a downstream signal after the production of ER stress, related or unrelated to mitochondrial dysfunction. Of course, it could also be downstream of inflammatory pathways. Spiegelman: Could you rescue ER stress by over-production of SOD2? If it is downstream, as in your model, you wouldn’t. Hotamisligil: The fact that ER stress can go directly to JNK and IKK pathways indicates that these pathways could be activated even in the absence of ROS. This
ER STRESS IN OBESITY AND DIABETES
97
may not be sufficient enough intervention but an interesting experiment to try nevertheless. Shi: The increased lipids synthesized in the ER in the obese state are presumably causing ER stress and insulin resistance. Has anyone looked at the DGAT1 knockout mouse to see whether it has improved insulin sensitivity through reduced ER stress? Also, obesity causes changes in mitochondrial morphology. How about the ER morphology? Hotamisligil: We are doing some of this work right now. In the course of obesity there is ER distension in liver. But these aren’t particularly easy experiments. At the moment we are looking at ER morphology changes during the course of obesity: if they do happen, when do they happen? There is a whole literature on ER stress, islet function and insulin production. This includes mutations in humans. One mutation causes islet death early on and thus juvenile diabetes. Most of the islet-related work has been in the context of type 1 diabetes or juvenile diabetes where there is clear distension of the ER as the islets morph into pathological states. If you initiate things with ER stress you can get morphological changes in the mitochondria. We are also looking into this. Shi: Most of the data you presented are from peripheral tissues except the islets. As you know that ER stress contributes to islet beta cell apoptosis and neonatal diabetes in PERK knockout mice and in Wolcott-Rallison Syndrome, how much of the phenotype is caused by improvement in islet function? Hotamisligl: So far, in vivo the only model that has provided information on this is the XBP1 haploinsufficiency. In this model we have not seen a major islet phenotype. The fact that they don’t have sufficient compensation and hypertrophy in the presence of peripheral insulin resistance might be an indication of some islet involvement. In fact, we have seen a small increase in apoptosis in the islets of XBP1 haploinsufficiency. It is not a major component, but there is no question that the islets will be part of the ER pathophysiology. In my view, ER stress could be an integrating mechanism for defective islet function and impaired peripheral insulin sensitivity. Spiegelman: What is the status of human studies in terms of detection of ER stress in various human tissues, with obesity, glucose intolerance and diabetes? Hotamisligil: There has been very little work done. Most people look at chaperone expression which is dysregulated in obesity. This is an important question and there is a need for additional reagents to be able to do this. O’Rahilly: What about genetic conditions where there is a misfolded protein, and therefore you’d expect the system to be under continual life-long stress, like the abnormal Z phenotype in α1 antitrypsin. If your hypothesis is correct you’d expect a higher prevalence of insulin-resistant conditions in these. As far as I know, this isn’t the case for things like α1 antitrypsin deficiency. This might be a nice model.
98
HOTAMISLIGIL
Hotamisligl: There are data in α1 antitrypsin deficiency, but they are somewhat confusing because this condition itself can disrupt islet function. There are interesting data from cystic fibrosis patients, who now can live to be 50 or 60. Over the age of 40, glucose intolerance and type 2 diabetes become very prevalent among cystic fibrosis patients. These individuals experience frequent infections so it is hard to draw clear conclusions. There is anecdotal evidence that in the early trials where CF patients were treated with PBA at high doses, they came off their diabetic treatments and achieved better glucose control. Spiegelman: Where will the answer to this hypothesis in humans come from? Pharmacology? Hotamisligil: Yes. These two drugs are readily applicable to humans. TUDCA has not been approved for human use in the US but it has been used in clinical trials. UDCA is safe and has been used widely in humans. PBA could also be used in humans. Han: In which life period can you see ER stress? Can only leptin mutations induce ER stress? Hotamisligil: There is comparable ER stress in the high fat diet model in mice. We haven’t looked at the other models. We compared Ob/Ob and high fat diet and they seemed to be of equal magnitude. Spiegelman: Do you see the molecular changes in adipose tissue? Hotamisligil: Yes. Spiegelman: That is easy to get from patients. Hotamisligl: I agree, materials are there but the markers are not easy to study for a clear picture yet. The only thing known in humans is increased JNK activity. Muoio: In the high fat diet model, how soon is this programme induced? Hotamisligil: We haven’t looked at the time course. We are now doing an experiment that we hope will give us a much better idea about this.
Fatty Acids and Lipotoxicity in Obesity and Diabetes: Novartis Foundation Symposium 286, Volume 286. Edited by Gregory Bock and Jamie Goode Copyright Novartis Foundation 2007. ISBN: 978-0-470-05764-3
The impact of insulin resistance on macrophage death pathways in advanced atherosclerosis Ira Tabas, Tracie Seimon, Jerry Arellano, Yankun Li, Fabien Forcheron, Dongying Cui, Seongah Han, Chien-Ping Liang, Alan Tall and Domenico Accili Department of Medicine, 630 West 168th Street, Columbia University, New York, NY 10032, USA
Abstract. Macrophage death in advanced atherosclerosis causes plaque necrosis, which promotes plaque rupture and acute atherothrombotic vascular events. Of interest, plaque necrosis and atherothrombotic disease are markedly increased in diabetes and metabolic syndrome. We discovered a novel ‘multi-hit’ macrophage apoptosis pathway that appears to be highly relevant to advanced atherosclerosis. The elements of the pathway include: (a) activation of the unfolded protein response (UPR) by cholesterol overloading of the endoplasmic reticulum or by other UPR activators known to exist in atheromata; and (b) pro-apoptotic signalling involving the type A scavenger receptor (SRA). The downstream apoptosis effectors include CHOP (GADD153) for the UPR and JNK for SRA signalling. Remarkably, components of this pathway are enhanced in macrophages with defective insulin signalling, including UPR activation and SRA expression. As a result, insulin-resistant macrophages show increased susceptibility to apoptosis when exposed to UPR activators and SRA ligands. Moreover, the advanced lesions of atherosclerosisprone mice reconstituted with insulin-resistant macrophages show increased macrophage apoptosis and plaque necrosis. Based on these fi ndings, we propose that one mechanism of increased plaque necrosis and atherothrombotic vascular disease in insulin resistant syndromes is up-regulation of a two-hit signal transduction pathway involved in advanced lesional macrophage death. 2007 Fatty acids and lipotoxicity in obesity and diabetes. Wiley, Chichester (Novartis Foundation Symposium 286) p 99–112
Diet and lifestyle habits of industrialized societies have led to a situation where almost all adult individuals in these societies have variable numbers of subendothelial deposits of cholesterol, inflammatory cells and extracellular matrix in focal areas of the arterial tree (Braunwald 1997). The vast majority of these focal subendothelial deposits, called atherosclerotic lesions or plaques, are relatively small, stable and asymptomatic, and will remain so during the lifespan of the individual. However, in approximately 50% of the population, a small number of these 99
100
TABAS ET AL
atherosclerotic lesions will undergo a gradual transformation that can prove deadly. This transformation is characterized by changes in the morphology of the lesions that render them susceptible to erosion or rupture (Aikawa & Libby 2004, Kolodgie et al 2004, Libby 2000). If erosion or rupture does eventually occur in these susceptible, or ‘vulnerable,’ plaques, the exposure of thrombogenic plaque material to the bloodstream can lead to acute thrombotic vascular occlusion and infarction of the tissue supplied by the affected artery. In the heart, this scenario results in sudden cardiac death and/or myocardial infarction, and in the brain, this process can lead to certain types of stroke (Aikawa & Libby 2004, Kolodgie et al 2004, Libby 2000). Understanding the mechanisms of stable-to-vulnerable plaque transformation at a cellular and molecular level will almost certainly suggest new therapeutic strategies to combat atherothrombotic vascular disease. Because both atherogenesis itself and the transformation to plaque vulnerability are complex processes, a reductionist approach is necessary to elucidate molecular/cellular mechanisms. In this context, we have focused on one of the more prominent cellular events associated with plaque vulnerability, namely, advanced lesional macrophage death. In early life, the arterial subendothelium is largely devoid of cells and extracellular lipids. However, by the early teen years, areas of the arterial tree characterized by disturbed blood flow begin to accumulate circulating lipoproteins in the subendothelial space (Williams & Tabas 1995, 1998). The most important of these lipoproteins are low density lipoproteins (LDL), derived from hepatic secretion of VLDL, and so-called remnant lipoproteins, derived from partial catabolism of intestinally derived chylomicrons and hepatic-derived VLDL (Williams & Tabas 1995, 1998). The key initiating event of subendothelial lipoprotein retention, which is influenced by the plasma level of these ‘atherogenic’ lipoproteins and poorly understood genetic factors at the level of the arterial wall, triggers a series of biological responses (Williams & Tabas 1995, 1998). The most prominent response to subendothelial lipoprotein retention is monocyte infi ltration. The monocytes differentiate into macrophages in the subendothelial space, and the macrophages then ingest the retained lipoproteins by a variety of processes including endocytosis, pinocytosis and phagocytosis (Tabas 2002). Lipoprotein uptake by macrophages results in a large delivery of lipoprotein-derived lipids, particularly cholesterol. Although excess cholesterol can be toxic to cells, early lesional macrophages remain healthy by effluxing the cholesterol and by storing it in a relatively harmless form called cholesteryl fatty acid esters (Glass & Witztum 2001). This latter process, which gives rise to cytoplasmic lipid droplets and thus a ‘foamy’ appearance to the macrophages, is catalysed by an endoplasmic reticulum (ER) enzyme called acyl-CoA:cholesterol acyltransferase (ACAT) (Chang et al 2001). Early foam cell lesions tend be non-occlusive and stable against erosion or rupture,
MACROPHAGE DEATH IN ATHEROSCLEROSIS
101
which is in part due to a ‘protective’ collagenous ‘cap’ that separates the lesion from the bloodstream (Aikawa & Libby 2004, Kolodgie et al 2004).
Results and discussion The importance of macrophage death in advanced atherosclerotic lesions As alluded to above, the vast majority of these foam cell lesions remain nonocclusive and stable throughout the life of the individual. Indeed, reversal of atherosclerotic risk factors, especially lowering of plasma LDL by drugs and dietary changes, can cause at least partial regression of these lesions (Williams & Tabas 2005). However, a few of the lesions, particularly in the setting of persistent risk factors such as high plasma LDL, cigarette smoking and diabetes, can progress to the aforementioned vulnerable plaque stage. One of the key features of vulnerable plaques are vast areas of macrophage debris, which result from death of lesional macrophages (Tabas 2005). These areas of dying macrophages and macrophage debris, often referred to as ‘necrotic cores’ or ‘lipid cores,’ almost certainly contribute to plaque erosion or rupture by promoting inflammation, physical stress on the fibrous cap, and thrombosis (Libby & Clinton 1993, Tabas 2005). It is likely that these processes promoted by dead macrophages complement those triggered by living macrophages, such as secretion of matrix proteases and inflammatory cytokines, to induce plaque breakdown and acute vascular thrombosis (Libby & Clinton 1993, Tabas 2005). According to the above scenario, advanced lesional macrophage death may be a key cellular event in stable-to-vulnerable plaque transformation. However, simply relating macrophage death to necrotic core formation and worsening of plaques can be misleading (Fig. 1). Macrophages in all stages of atherosclerosis undergo a certain basal rate of turnover by the caspase-dependent death process known as apoptosis (Tabas 2005). In early lesions, these apoptotic macrophages are rarely seen because they are rapidly and efficiently phagocytosed by neighbouring macrophages (Liu et al 2005, Tabas 2005). This process of apoptotic cell clearance via phagocytosis, often referred to as ‘efferocytosis,’ is a normal physiological response to apoptosis, is non-inflammatory and prevents cellular necrosis (Henson et al 2001). Indeed, promoting early lesional macrophage apoptosis by genetic manipulation in mouse models of atherosclerosis results in a decrease in lesion cellularity and size, because the lesional macrophages are safely decreased in number (Liu et al 2005). In advanced lesions, however, there is evidence that efferocytosis of apoptotic macrophages is less than completely efficient (Schrijvers et al 2005, Tabas 2005). Apoptotic cells that are not rapidly cleared by efferocytosis become leaky and trigger an inflammatory response—a process referred to as post-
102
TABAS ET AL
Early Lesions
Advanced Lesions
Mφ apoptosis ↓ Rapid clearance of apoptotic Mφs (efferocytosis) ↓ Decreased lesion cellularity ↓ Decreased lesion progression
Mφ apoptosis ↓ Inefficient clearance of apoptotic Mφs (defective efferocytosis) ↓ Post-apoptotic Mφ necrosis ↓ Plaque necrosis (‘necrotic core’) ↓ Plaque disruption ↓ Acute vascular events
FIG. 1. Model of the opposing roles of macrophage (Mφ ) death in atherosclerosis. In early lesions, apoptotic macrophages are rapidly cleared by efferocytosis, leading to reduced lesion cellularity and decreased lesion progression. In advanced lesions, however, efferocytosis is not as efficient, and so some of the apoptotic macrophages become secondarily necrotic. These necrotic macrophages gradually coalesce, forming the necrotic core. The combination of plaque necrosis, i.e. dead macrophages, and residual living macrophages (not shown) promotes plaque disruption and acute lumenal thrombosis. See text and Tabas (2005) for details.
apoptotic, or secondary, necrosis (Fiers et al 1999, Tabas 2005). By this manner, the necrotic core of advanced atherosclerotic lesions gradually forms. The UPR–SRA model of advanced lesional macrophage apoptosis Based on the above description of plaque progression, we feel that understanding the cellular/molecular basis of two key processes in advanced lesional macrophages—apoptosis and efferocytosis—is likely to shed new light on how stable atherosclerotic lesions transform into vulnerable plaques. For purposes of focus, this chapter will address the issue of how macrophages might undergo apoptosis in advanced atherosclerotic lesions. Our studies in this area began with a single model based on observations in human vulnerable plaques, but ensuing mechanistic studies of this model uncovered a much broader array of possible triggers that are likely to be relevant to atherosclerosis. The initial model was based on observations that the macrophages in vulnerable human plaques contain more unesterified, or ‘free,’ cholesterol (FC) than is typically observed in earlier lesional macrophage foam cells (see above) (Aikawa & Libby 2004, Burke et al 2003, Guyton & Klemp 1994, Kolodgie et al 2004, Kruth 1984, Lundberg 1985, Small 1988). Although the mechanism of FC accumulation is not known, it is likely promoted by dysfunctions of ACAT-mediated cholesterol esterification and cellular cholesterol efflux (see above).
MACROPHAGE DEATH IN ATHEROSCLEROSIS
103
Based on this observation and the known cytotoxic effects of excess intracellular FC (Warner et al 1995, Yao & Tabas 2000, 2001), we sought to understand how FC accumulation would effect macrophages. The model we chose was one in which primary tissue macrophages (murine peritoneal macrophages) were exposed in culture to atherogenic lipoproteins in the setting of pharmacological or genetic ACAT dysfunction. We found that FC accumulation was a trigger for caspasedependent macrophage apoptosis by pathways involving both the Fas death receptor and well-described mitochondrial apoptotic mechanisms (Yao & Tabas 2000, 2001). Initially, we imagined that excess FC in the plasma membrane and/or mitochondria might somehow trigger these events. However, our studies revealed that the key organelle was neither of the above but rather the ER (Feng et al 2003a, 2003b). In retrospect, the ER would have been a logical candidate, because the ER membrane bilayer normally has a relatively low cholesterol : phospholipid ratio. This property is responsible for the fluid nature of the ER membrane, which is necessary for its proper function (Davis & Poznansky 1987). When the cholesterol : phospholipid ratio of the ER membrane is increased, such as occurs during FC enrichment of macrophages, the membrane undergoes a phase transition to a more ordered state (Li et al 2004). This abnormal state leads to dysfunction of critical ER membrane proteins, including a protein called SERCA, which controls calcium levels in the ER (Li et al 2004). The macrophage responds to this state of ER ‘stress’ by activating a coordinated signal transduction pathway known as the unfolded protein response (UPR) (Feng et al 2003a). The UPR is triggered by a wide variety of ER stressors, and its major function is to reverse the stress and protect the ER while carrying out this repair function (Ma & Hendershot 2001, Ron 2002, Welihinda et al 1999). Thus, for example, protein translation is suppressed, unfolded proteins are degraded, and protein chaperones are induced (Ma & Hendershot 2001, Ron 2002, Welihinda et al 1999). However, if ER stress is prolonged or unable to be repaired, a distal branch of the UPR can promote apoptosis (McCullough et al 2001, Oyadomari et al 2002, Zinszner et al 1998). In many cases, including the situation with FC-enriched macrophages, the apoptotic function of the UPR is affected by the transcription factor called CHOP, or GADD153 (McCullough et al 2001, Oyadomari et al 2002, Zinszner et al 1998). Although the exact mechanism of CHOP-induced apoptosis is still be explored in our and other laboratories, CHOP-deficient macrophages are substantially protected against FC-induced apoptosis (Feng et al 2003a). Moreover, recent unpublished data in our laboratory suggest that advanced lesions of Chop –/– mice on an atherogenic background have less macrophage death and less plaque necrosis than Chop +/+ mice on the same background. The concept that the UPR is an apoptosis trigger is simplistic. In fact, the UPR normally promotes cell survival to enable repair (see above), and apoptosis is
104
TABAS ET AL
induced only when something else goes awry (Ma & Hendershot 2001, Ron 2002, Welihinda et al 1999). Thus, the UPR can, under certain circumstances, be necessary for apoptosis, but it is never sufficient by itself. Rather, the UPR can enable apoptosis in response to another ‘hit.’ This two-hit model of UPR-induced apoptosis is critical to understanding how FC enrichment kills macrophages. In this scenario, the UPR–CHOP pathway is necessary but not sufficient for FC-induced apoptosis. We found that the ‘second hit’ in this model involved engagement of a receptor for atherogenic lipoproteins called the type A scavenger receptor (SRA) (DeVries-Seimon et al 2005) (Fig. 2). This second hit was affected by the method we had chosen to deliver cholesterol to the macrophages, namely, because we had used a typical SRA-binding atherogenic lipoprotein to load the cells with cholesterol (DeVries-Seimon et al 2005). Thus, atherogenic lipoproteins induce death by satisfying both hits of the two-hit model: they deliver cholesterol to the ER to activate the UPR–CHOP pathway, and they engage the SRA to supply the second
Atherogenic ER ‘stressors’ (FC, nitric oxide, hypoxia, oxidant stress, homocysteine, glucosamine)
Atherogenic SRA ligands (modified LPs, AGEs, glycated collagen, β-amyloid)
↓ cell survival signalling
Endoplasmic reticulum
↑ JNK
SRA signalling
UPR-CHOP
Apoptosis Plaque disruption Acute lumenal thrombosis Tissue infarction (MI, stroke)
defective phagocytic clearance
Post-apoptotic Mφ necrosis
FIG. 2. The UPR–SRA two-hit model of macrophage (Mφ ) apoptosis. UPR activation by FC loading or other UPR activators known to be present in advanced atherosclerotic lesions provides the potential for apoptosis through the UPR effector CHOP. However, apoptosis does not occur unless a second hit is present, which in the case of macrophage apoptosis is engagement of the SRA. Ongoing mechanistic studies suggest that the SRA triggers both a pro-apoptotic JNK pathway and inhibits a cell survival pathway.
MACROPHAGE DEATH IN ATHEROSCLEROSIS
105
hit. We now know that any combination of UPR activators and SRA ligands, many of which exist in advanced atherosclerotic lesions (Fig. 2), can trigger macrophage apoptosis independently of FC enrichment and ACAT dysfunction. These findings are very important, because they extend the array of possible triggers of advanced lesional macrophage death beyond the FC enrichment/ACAT dysfunction model. How SRA engagement supplies the second apoptotic hit to ER-stressed macrophages is a key area of current focus in the laboratory. Our current data suggest a fascinating mechanism in which SRA engagement triggers activation of the proapoptotic mitogen-activated protein kinase JNK while selectively silencing a cell survival pathway in the macrophages. The link to insulin resistance Humans with type 2 diabetes have a markedly increased risk of atherothrombotic vascular disease (Plutzky et al 2002). Of significant relevance to the topic of this chapter, diabetic lesions are characterized by increased necrotic cores (Burke et al 2004), suggesting an enhancement of advanced lesional macrophage death. The increase in plaque progression in diabetics is likely due to a combination of insulin resistance and hyperglycaemia. Certainly, insulin resistance by itself is a strong risk factor, because humans with insulin resistance without hyperglycaemia, e.g. in metabolic syndrome, have an increased incidence of coronary artery disease (Grundy 2004). There are a number of possible mechanisms whereby insulin resistance might promote plaque progression, including enhancement of systemic atherosclerotic risk factors, like high-density lipoprotein (HDL) and hypertension, and direct effects on cells of the arterial wall, notably endothelial cells (Plutzky et al 2002). In a recent collaboration with Drs Alan Tall and Domenico Accili of Columbia University, we have been investigating another possible mechanism, namely, proapoptotic effects on macrophage insulin resistance in the context of the UPR–SRA model described above. Although macrophages do not have insulin-regulatable glucose transporters, they do have insulin receptors that respond to insulin through the canonical insulin signalling pathway (Liang et al 2004). Thus, acute treatment of macrophages with physiologic insulin concentrations leads to phosphorylation of the insulin receptor, insulin receptor substrate 2 (IRS2), and Akt (Liang et al 2004). Most relevant to the topic of this discussion, macrophages isolated from insulin-resistant mouse models, such as the hyperinsulinaemic leptin-deficient ob/ob mouse, have down-regulated insulin receptors and depressed insulin signaling (Liang et al 2004). Thus, macrophages, like hepatocytes and skeletal muscle cells, become insulin resistant in the setting of hyperinsulinaemia in vivo. Remarkably, two of the more prominent characteristics of macrophages in the insulin-resistant state are activation of the UPR and up-regulation of the SRA (Han
106
TABAS ET AL
et al 2006, Liang et al 2004). Although the molecular mechanisms behind these two processes are still under investigation, SRA up-regulation appears to be posttranscriptional and causally related to UPR activation (Han et al 2006, Liang et al 2004). These findings directly suggest that insulin-resistant macrophages would be more susceptible to the model of apoptosis described above. Indeed, we found that such macrophages show markedly enhanced apoptosis when exposed to SRAmediated FC enrichment conditions or when treated with the combination of a UPR activator and an SRA ligand (Han et al 2006). Most importantly, we have data supporting this model in advanced atherosclerotic lesions. Specifically, we found that the advanced lesions of Ldlr–/– mice have increased macrophage apoptosis and lesional necrosis when they are reconstituted with bone marrow from insulin-resistant mice (Han et al 2006). Studies are ongoing to determine whether the increased apoptosis observed in insulin-resistant macrophages represents an enhancement of the same UPR–SRA pro-apoptotic signalling pathway used by insulin-sensitive macrophages (described above), as suggested by the UPR activation and SRA up-regulation in these cells (Fig. 3), or whether the insulin-resistant state triggers a second pro-apoptotic
↓ insulin signalling in Mφs
INSULIN RESISTANCE
Alteration in adipocytokines
}
?
UPR
↑ SRA
↑ Chol loading
↑ Apoptosis trigger (JNK, ↓ survival signalling)
↑ Mφ apoptosis in advanced lesions ↑ lesional necrosis ↑ plaque disruption ↑ lumenal thrombosis ↑ tissue infarction (MI, stroke)
FIG. 3. Model of how insulin resistance promotes advanced lesional macrophage (Mφ ) apoptosis. Macrophage insulin resistance directly activates the UPR, which in turn leads to upregulation of the SRA. Each of these processes contributes to apoptosis by the two-hit model outlined in Fig. 2. In addition, the up-regulation of the SRA might promote enhanced internalization of lipoprotein-derived cholesterol, which might further activate the UPR in the setting of dysfunctional ACAT and suppressed cholesterol efflux. We are also exploring the possibility that alterations in adipocytokines in the setting of systemic insulin resistance might affect advanced lesional macrophage apoptosis.
MACROPHAGE DEATH IN ATHEROSCLEROSIS
107
pathway that is additive or synergistic with the ‘basal’ pathway. Furthermore, the laboratory is investigating molecular links between systemic insulin resistance and advanced lesional macrophage apoptosis, such as those that may be mediated by circulating adipocytokines that are altered in insulin-resistant states. Summary and conclusions Advanced lesional macrophage death is a key event in the transformation of asymptomatic atherosclerotic lesions into plaques that have the potential to rupture and cause acute vascular events. While multiple mechanisms are likely responsible for advanced lesional macrophage death, in vivo evidence suggests that a two-hit proapoptotic pathway involving the UPR and the SRA plays an important role. Apoptosis in this two-hit model results from pro-apoptotic CHOP from the UPR branch and pro-apoptotic JNK, plus suppression of cell-survival signalling, from the SRA branch. Remarkably, both of these hits are up-regulated in macrophages that are insulin resistant, and an atherosclerotic mouse model with insulinresistant macrophages shows evidence of increased advanced lesional macrophage death and plaque necrosis. Further studies are needed to defi ne exactly how defective insulin signalling in macrophages leads to UPR activation and SRA upregulation; whether these events are responsible for the advanced lesional macrophage death observed in vivo; and whether other events associated with insulin resistance, such as alterations in adipocytokines, might contribute to this process. Acknowledgements The work described in this chapter was supported by NIH grants HL75662 and HL54591 and USA Medical Research and Material Command Grant PR054352.
References Aikawa M, Libby P 2004 The vulnerable atherosclerotic plaque: pathogenesis and therapeutic approach. Cardiovasc Pathol 13:125–138 Braunwald E 1997 Cardiovascular medicine at the turn of the millennium: triumphs, concerns, and opportunities. N Engl J Med 337:1360–1369 Burke AP, Virmani R, Galis Z, Haudenschild CC, Muller JE 2003 34th Bethesda Conference: Task force #2—What is the pathologic basis for new atherosclerosis imaging techniques? J Am Coll Cardiol 41:1874–1886 Burke AP, Kolodgie FD, Zieske A et al 2004 Morphologic findings of coronary atherosclerotic plaques in diabetics: a postmortem study. Arterioscler Thromb Vasc Biol 24:1266–1271 Chang TY, Chang CC, Lin S, Yu C, Li BL, Miyazaki A 2001 Roles of acyl-coenzyme A:cholesterol acyltransferase-1 and -2. Curr Opin Lipidol 12:289–296 Davis PJ, Poznansky MJ 1987 Modulation of 3-hydroxy-3-methylglutaryl-CoA reductase by changes in microsomal cholesterol content or phospholipid composition. Proc Natl Acad Sci USA 84:118–121
108
TABAS ET AL
DeVries-Seimon T, Li Y, Yao PM et al 2005 Cholesterol-induced macrophage apoptosis requires ER stress pathways and engagement of the type A scavenger receptor. J Cell Biol 171:61– 73 Feng B, Yao PM, Li Y et al 2003a The endoplasmic reticulum is the site of cholesterol-induced cytotoxicity in macrophages. Nat Cell Biol 5:781–792 Feng B, Zhang D, Kuriakose G, Devlin CM, Kockx M, Tabas I 2003b Niemann-Pick C heterozygosity confers resistance to lesional necrosis and macrophage apoptosis in murine atherosclerosis. Proc Natl Acad Sci USA 100:10423–10428 Fiers W, Beyaert R, Declercq W, Vandenabeele P 1999 More than one way to die: apoptosis, necrosis and reactive oxygen damage. Oncogene 18:7719–7730 Glass CK, Witztum JL 2001 Atherosclerosis. the road ahead. Cell 104:503–516 Grundy SM 2004 Obesity, metabolic syndrome, and cardiovascular disease. J Clin Endocrinol Metab 89:2595–2600 Guyton JR, Klemp KF 1994 Development of the atherosclerotic core region. Chemical and ultrastructural analysis of microdissected atherosclerotic lesions from human aorta. Arterioscler Thromb 14:1305–1314 Han S, Liang CP, DeVries-Seimon T et al 2006 Macrophage insulin receptor deficiency increases ER stress-induced apoptosis and necrotic core formation in advanced atherosclerotic lesions. Cell Metab 3:257–266 Henson PM, Bratton DL, Fadok VA 2001 Apoptotic cell removal. Curr Biol 11:R795–805 Kolodgie FD, Virmani R, Burke AP et al 2004 Pathologic assessment of the vulnerable human coronary plaque. Heart 90:1385–1391 Kruth HS 1984 Localization of unesterified cholesterol in human atherosclerotic lesions. Demonstration of fi lipin-positive, oil-red-O-negative particles. Am J Pathol 114:201–208 Li Y, Ge M, Ciani L et al 2004 Enrichment of endoplasmic reticulum with cholesterol inhibits SERCA2b activity in parallel with increased order of membrane lipids. Implications for depletion of ER calcium stores and apoptosis in cholesterol-loaded macrophages. J Biol Chem 279:37030–37039 Liang CP, Han S, Okamoto H et al 2004 Increased CD36 protein as a response to defective insulin signaling in macrophages. J Clin Invest 113:764–773 Libby P 2000 Changing concepts of atherogenesis. J Intern Med 247:349–358 Libby P, Clinton SK 1993 The role of macrophages in atherogenesis. Curr Opin Lipidol 4:355–363 Liu J, Thewke DP, Su YR, Linton MF, Fazio S, Sinensky MS 2005 Reduced macrophage apoptosis is associated with accelerated atherosclerosis in low-density lipoprotein receptor-null mice. Arterioscler Thromb Vasc Biol 25:174–179 Lundberg B 1985 Chemical composition and physical state of lipid deposits in atherosclerosis. Atherosclerosis 56:93–110 Ma Y, Hendershot LM 2001 The unfolding tale of the unfolded protein response. Cell 107:827–830 McCullough KD, Martindale JL, Klotz LO, Aw TY, Holbrook NJ 2001 Gadd153 sensitizes cells to endoplasmic reticulum stress by down-regulating Bcl2 and perturbing the cellular redox state. Mol Cell Biol 21:1249–1259 Oyadomari S, Koizumi A, Takeda K et al 2002 Targeted disruption of the Chop gene delays endoplasmic reticulum stress-mediated diabetes. J Clin Invest 109:525–532 Plutzky J, Viberti G, Haffner S 2002 Atherosclerosis in type 2 diabetes mellitus and insulin resistance: mechanistic links and therapeutic targets. J Diabetes Complications 16:401– 415 Ron D 2002 Translational control in the endoplasmic reticulum stress response. J Clin Invest 110:1383–1388
MACROPHAGE DEATH IN ATHEROSCLEROSIS
109
Schrijvers DM, De Meyer GR, Kockx MM, Herman AG, Martinet W 2005 Phagocytosis of apoptotic cells by macrophages is impaired in atherosclerosis. Arterioscler Thromb Vasc Biol 25:1256–1261 Small DM 1988 George Lyman Duff memorial lecture. Progression and regression of atherosclerotic lesions. Insights from lipid physical biochemistry. Arteriosclerosis 8:103–129 Tabas I 2002 Consequences of cellular cholesterol accumulation: basic concepts and physiological implications. J Clin Invest 110:905–911 Tabas I 2005 Consequences and therapeutic implications of macrophage apoptosis in atherosclerosis: the importance of lesion stage and phagocytic efficiency. Arterioscler Thromb Vasc Biol 25:2255–2264 Warner GJ, Stoudt G, Bamberger M, Johnson WJ, Rothblat GH 1995 Cell toxicity induced by inhibition of acyl coenzyme A:cholesterol acyltransferase and accumulation of unesterified cholesterol. J Biol Chem 270:5772–5778 Welihinda AA, Tirasophon W, Kaufman RJ 1999 The cellular response to protein misfolding in the endoplasmic reticulum. Gene Expr 7:293–300 Williams KJ, Tabas I 1995 The response-to-retention hypothesis of early atherogenesis. Arterioscler Thromb Vasc Biol 15:551–561 Williams KJ, Tabas I 1998 The response-to-retention hypothesis of atherogenesis, reinforced. Curr Opin Lipidol 9:471–474 Williams KJ, Tabas I 2005 Lipoprotein retention—and clues for atheroma regression. Arterioscler Thromb Vasc Biol 25:1536–1540 Yao PM, Tabas I 2000 Free cholesterol loading of macrophages induces apoptosis involving the fas pathway. J Biol Chem 275:23807–23813 Yao PM, Tabas I 2001 Free cholesterol loading of macrophages is associated with widespread mitochondrial dysfunction and activation of the mitochondrial apoptosis pathway. J Biol Chem 276:42468–42476 Zinszner H, Kuroda M, Wang X et al 1998 CHOP is implicated in programmed cell death in response to impaired function of the endoplasmic reticulum. Genes Dev 12:982–995
DISCUSSION O’Rahilly: I was interested in the data generated from insulin receptor null cells. I would advise caution about using them as the unique model. If I had used this as my sole model for insulin resistance I would have got the story of insulin and adiponectin completely wrong. If you are thinking of the macrophages in an insulin resistant milieu, it is hard to know which cells are being exposed to a hyperinsulinaemic environment because they are not primarily insulin resistant themselves. The absolute absence of insulin may not be what they are seeing in vivo. Tabas: The point you are raising is exactly why we designed the experiment that way. It was a reductionist approach to try to get at the bare mechanism of what is going on. You are right, but everything I have shown in vitro has been completely reproduced with Ob/Ob. We are now going back in vivo and doing everything with Ob/Ob. However, the reason we designed the first experiment the way we did is because of the points you raised. This model is unique. It is amazing how many
110
TABAS ET AL
people don’t get this: insulin receptor down-regulation is an inherent part of insulin resistance. Our goal wasn’t to try to attack the whole thing at first, but to look at what would happen if by itself we decrease insulin signalling in macrophages. Our second goal is to put this in a larger picture, as you are suggesting. Spiegelman: I know that Ob/Ob mice have down-regulated insulin receptors, but I was under the impression that this wasn’t generally true in humans with obesity and insulin resistance. O’Rahilly: They do have down-regulated insulin receptors in a variety of cells. We haven’t quantified the pattern precisely. In the average patient with insulin resistance, there are certainly some cells in the body that are seeing too much insulin action. The dermal fibroblast response of acanthosis nigricans, which is commonly seen, is almost certainly a hypertrophic response of cells to insulin, either occurring through its own receptor, hybrids or the insulin-like growth factor (IGF) receptor. Working out which tissues are seeing too much and which are seeing not enough insulin in any individual circumstance of hyperinsulinaemia is hard. Tabas: We have data from human liver. These show a straight-line correlation between the level of insulin resistance and insulin receptors on hepatocytes. In the most insulin resistant states the insulin receptors in the hepatocytes are undetectable by western blot. Hotamisligil: It has been known for a long time that the down-regulation of insulin receptor quantity is insufficient to explain the extent of insulin resistance, which was the original observation that led to the examination of insulin receptor signalling. This is a critical phenomenon that is often overlooked. Insulin resistance is not uniform in the body: some cells are seeing too much, some parts of insulin signalling are more active than others. Tabas: This is something we have known for a long time. Brown and Goldstein (Shimomura et al 2000) brought this up in the liver where their feeling is that the insulin pathway that is involved in fatty acid metabolism and triglyceride synthesis remains hyperactive because of the high levels of circulating insulin. One of the reasons why the knockout is advantageous is because it has zero insulin receptors. But I want to emphasize that the knockout was done at stage 1 in our study: we wanted to take a reductionist approach. Stage 2 is to put this in a larger context. Hotamisligil: In the macrophage-specific insulin receptor knockout model versus the total knockout bone marrow transplantation model, are there differences between those experiments? There was also an IRS2 knockout, and looking at the total lesion area, it looked completely the other way (Han et al 2006, Baumgartl et al 2006). This is a little confusing. Tabas: You shouldn’t be confused at all. If you go back and look at the data in that paper and the data in our paper, the differences are really not the big. We
MACROPHAGE DEATH IN ATHEROSCLEROSIS
111
show a little bit of an increase; in that paper there was somewhat of a decrease. In my mind that is not the story. The story, if there is a story at all, will be in the advanced lesional morphology events. The slight increase in lesion area we got and the moderate decrease in the IRS2 knockout is noise and it is not the most important part of the story. Hotamisligil: I am intrigued by your comment that lesion size is less of a story. Tabas: This is my bias. In humans, even though we now use coronary angiography as our gold standard, there is not a great correlation between lesion size and events. The best thing that angiography is doing is showing us the lesions that are associated with the dangerous lesions, which are known to be much more moderate in size than the larger lesions. The lesson we have learned from humans is that lesion morphology is more important than lesion size. It is on the basis of that knowledge that I have created a bias in using the mouse as a model that we are much more interested in advanced lesional morphology. Spiegelman: The data you showed where adiponectin was suppressing UPR looks extremely impressive. To what extent can you hypothesize that this is the molecular basis of adiponectin action in metabolism? And have you examined which of the known adiponectin receptors is involved in that response? Tabas: Could this be the answer to all the actions of adiponectin? We don’t know. It is robust, and it was revealing for me to get Philipp Scherer’s perspective. I am interested in atherosclerosis and macrophages, and he says, ‘Ira, you have an in vitro assay that can distinguish between the high and low molecular weight forms!’ It is highly discriminatory between the potent and non-potent forms. We have been focused on macrophages so far and have tried to do just a few pilot experiments with some liver cell lines. Spiegelman: You don’t want to use liver cell lines, but primary hepatocytes. Tabas: Could this be important in some of the other effects of adiponectin, particularly in view of what Gökhan Hotamisligil has taught us about how suppression of the UPR can improve insulin resistance? We know that high molecular weight adiponectin has been associated with improvement in insulin resistance, so could it all fit together? One of the reasons I was so excited about being at this meeting is so we could talk about this. Hotamisligil: This is extremely exciting. It could not only explain why adiponectin is increasing insulin sensitivity, but it could also represent a nice loop, with increased UPR because of adiponectin and ER stress itself might be related to why there is less adiponectin. Adiponectin is one of the most complex molecules assembled in the ER. I would bet that if you induce ER stress, this blocks adiponectin synthesis. Tabas: That’s a good point. If adiponectin is made mostly by adipocytes, why do obese people have less adiponectin? Philipp Scherer has two ideas, one of which is exactly what you said. His other idea is related to inflammatory cytokines.
112
TABAS ET AL
Spiegelman: That’s a gene expression change. Tabas: He thinks that two things are going on. He thinks the transcriptional is related to cytokines, and the post-transcriptional may be receptors. Remember, my collaborator is Philipp Scherer. This is the world I live in and the reagents I have available. His single knockouts don’t show much, but the R1 and R2 double knockout is lethal. Professor Kadowaki finds small effects in R1 and R2 knockouts but his double knockout is viable, probably because there is a little leak through. He finds changes in insulin signalling. We have taken the R1 knockout and subjected it to R2 RNAi, getting 80% suppression. There is no diminution of the effect I have shown. We have no evidence at this point that R1 and R2 are involved. I am not sure what adiponectin is doing. We also know that AMPK is not involved. I flew my postdoc to Benoit Viollet in Paris. It turns out that macrophages only express the α1 form of adiponectin. There is no effect at all knocking out α1. As an outsider to this field I’m making no assumptions about the role of adiponectin. It may not even be the molecule itself that is doing something: it has lipid pockets so it could be delivering something. Hotamisligil: Perhaps it is acting as a chaperone, providing it gets into the cell. Tabas: The reason it isn’t acting like a chaperone is that there is no chaperone that would not suppress eIF2 α phosphorylation. Hotamisligil: What if it is inhibiting an eIF2 α phosphatase and acting as a chaperone? Tabas: That’s possible but unlikely in my opinion. References Baumgartl J, Baudler S, Scherner M et al 2006 Myeloid lineage cell-restricted insulin resistance protects apolipoproteinE-deficient mice against atherosclerosis. Cell Metab 3:247–256 (Erratum: 2006 Cell Metab 3:469) Han S, Liang CP, DeVries-Seimon T et al 2006 Macrophage insulin receptor deficiency increases ER stress-induced apoptosis and necrotic core formation in advanced atherosclerotic lesions. Cell Metab 3:257–266 Shimomura I, Matsuda M, Hammer RE, Bashmakov Y, Brown MS, Goldstein JL 2000 Decreased IRS-2 and increased SREBP-1c lead to mixed insulin resistance and sensitivity in livers of lipodystrophic and ob/ob mice. Mol Cell 6:77–86
Fatty Acids and Lipotoxicity in Obesity and Diabetes: Novartis Foundation Symposium 286, Volume 286. Edited by Gregory Bock and Jamie Goode Copyright Novartis Foundation 2007. ISBN: 978-0-470-05764-3
Fatty acid transport in adipocytes and the development of insulin resistance Sandra Lobo and David A. Bernlohr Department of Biochemistry, Molecular Biolog y and Biophysics, The University of Minnesota, Minneapolis, MN 55455, USA
Abstract. Fatty acid influx into adipocytes is a complex multifactoral process driven by biochemical and biophysical processes linking transmembrane flux to the ATPdependent esterification of fatty acids. Adipocyte proteins implicated in free fatty acid (FFA) influx include CD36 functioning as a general lipid receptor, caveolin 1 functioning as a component of an endocytotic/exocytotic vesicular cycle and the acyl CoA synthetases (FATP1, ACSL1) catalysing esterification of lipids producing acyl CoAs. In adipocytes, CD36, ACSL1 and FATP1 translocate from intracellular sites to the plasma membrane in response to insulin thereby positioning these key proteins to facilitate FFA esterification. Lentiviral delivery of shRNA targeting FATP1 in 3T3-L1 adipocytes results in a complete loss of insulin-stimulated FFA uptake, decreased accumulation of TAG/DAG/MAG and potentiated insulin-stimulated 2-deoxyglucose uptake. Increased insulin-stimulated hexose uptake in FATP1 knockdown adipocytes is correlated with increased tyrosine phosphorylation and abundance of IRS1 protein. Evaluation of the lipid activated serine kinases implicated in insulin signalling reveals that S6K and JNK1 were not altered in abundance or phosphorylation in FATP1 knockdown adipocytes but that the phosphorylation of PKCθ and abundance of IKKα/β were significantly reduced. These results suggest lipid droplet pools in the adipocyte play a major role in regulating kinase cascades controlling insulin action. 2007 Fatty acids and lipotoxicity in obesity and diabetes. Wiley, Chichester (Novartis Foundation Symposium 286) p 113–126
Circulating lipoprotein-bound triacylglycerol derived either from dietary fat (chylomicra) or dietary carbohydrate (VLDL), provides the adipose tissue beds with an abundant source of calories to be stored during nutrient excess and liberated during nutrient depletion (Bernlohr & Simpson 1996). Insulin-stimulated secretion of lipoprotein lipase and subsequent anchoring of the enzyme into the luminal side of the endothelial cell wall positions adipose tissue to accept chylomicra and VLDL-derived fatty acids for influx and metabolism. Despite this well-known and well-accepted physiology, the components of the fatty acid influx system are only now beginning to be defined in molecular terms. 113
114
LOBO & BERNLOHR
A complicating factor in defining the biochemical processes germane to fatty acid influx is the variable solubility of the molecules themselves. As opposed to dietary carbohydrates such as glucose and fructose that have very similar chemical properties, the solubility of saturated long and very long chain free fatty acids (FFAs) differ greatly from polyunsaturated fatty acids (Kamp & Hamilton 2006). Moreover, the rate of fatty acid diffusion across biological membranes varies greatly between different molecular species and often confounds conventional analysis methodologies (Hamilton 2003). As such, functional assays involving gain or loss strategies in model organisms have typically been employed to identify molecular components of the fatty acid influx system in adipocytes. From such analyses three classes of proteins have been implicated: fatty acid receptors, proteins that facilitate bilayer movement and acyl CoA synthetases. CD36/fatty acid translocase: a fatty acid receptor Fatty acid translocase (FAT), also known as CD36, was identified as an 88 kDa protein from rat adipocytes that was specifically labelled with a reactive sulfo-Nsuccinimidyl derivative of oleic acid (Abumrad et al 1993). FAT/CD36 belongs to the family of class B scavenger receptors and is universally expressed in most tissues especially in tissue with high metabolic activity such as heart and muscle, and tissues involved in lipid storage and absorption (adipose and intestine). Evidence gathered from experiments in cultured cells and rat and mouse models with altered expression of CD36 have designated a prominent role for CD36 as a fatty acid receptor linked to lipid influx. Knockout of CD36 in mice (CD36 –/– ) resulted in a 60–80% decrease in fatty acid uptake in adipose tissue, which was similar to that observed in CD36-deficient humans (Coburn et al 2001). CD36 in adipose tissue is regulated at the expression level by the nuclear peroxisome proliferatoractivated receptor γ (PPARγ ), tumour necrosis factor α (TNFα ) and interleukin 1 (IL1), and by its translocation to the plasma membrane in response to insulin (Sato et al 2003, Memon et al 1998, Luiken et al 2002). The mechanism by which FAT/CD36 facilitates fatty acid uptake is unknown. The predicted CD36 protein structure is unusual with 10 potential N-linked glycosylation sites in the extracellular domain and the protein is tethered to the membrane by two short transmembrane regions at each terminal. CD36 is preferentially localized to detergent resistant membranes also known as lipid rafts, which are structurally ordered microdomains in the plasma membrane matrix consisting of lateral assemblies of sphingolipids and cholesterol in the outer leaflet of the lipid bilayer connected to phospholipid and cholesterol in the inner leaflet (Pohl et al 2005). The positioning of CD36 to lipid rafts is critical to its function and disruption of the lipid raft structure by depletion of cholesterol results in a reduction of fatty acid uptake (Pohl et al 2004). It is hypothesized that the binding of a fatty
ADIPOCYTE FATTY ACID TRANSPORT
115
acid ligand to CD36 may induce dimerization and facilitate LCFA influx via endocytotic mechanisms or may provide a high local LCFA environment thereby producing a concentration gradient across the plasma membrane.
Caveolin 1, a fatty acid binding protein potentially facilitating bilayer movement Caveolins are a family of proteins that coat the cytoplasmic surface of caveolae, which are ‘flask-shaped’ 50–100 nm vesicular invaginations of the plasma membrane. The polymerization of caveolins is hypothesized to result in the formation of caveolae. There are three caveolin proteins identified in vertebrates: caveolin 1, caveolin 2 and caveolin 3. Adipose tissue expresses caveolin 1 and caveolin 2. The adipose plasma membrane is rife with caveolae that are enriched with cholesterol and are thought to function in vesicle trafficking and cholesterol homeostasis (Pol et al 2005). Caveolin 1 binds cholesterol with high affinity and specificity in vitro, and is thought to be involved in the efflux of newly synthesized cholesterol from the endoplasmic reticulum and the influx of cholesterol esters from the plasma (Uittenbogaard & Smart 2000). Caveolin 1 is a crucial structural component required for formation and assembly of lipid rafts and caveolae. Caveolin 1 has also been shown to be a fatty acid binding protein (Trigatti et al 1999). Experiments in HepG2 cells showed that inhibition of caveolae formation caused a significant reduction in oleate uptake, independent of coated pit mediated uptake (Pohl et al 2002). Since the rate of fatty acid diffusion is proportional to the hydrophobicity of the local medium, caveolin binding of fatty acid and cholesterol in the raft environment may provide a site for fatty acid diffusion from the outer to the inner leaflet of the membrane. Inhibition of caveolar function by either depletion of membrane cholesterol using cyclodextrin, over expression of a dominant negative caveolin 1 mutant (CAV DGV ), or disruption of the caveolae actin structure by actin-depolymerizing agents all significantly reduced fatty acid uptake (Pohl et al 2004). Caveolin 1 null mice lacking caveolae demonstrated a reduced fatty acid influx and adipocyte lipiddroplet size, are resistant to diet-induced obesity and have high post-prandial plasma triglyceride and fatty acid levels (Razani et al 2002).
Acyl CoA synthetases: metabolic enzymes facilitating net fatty acid influx There are two broad classes of enzymes implicated in fatty acid influx in adipocytes, the fatty acid transport proteins and acyl CoA synthetases. Both classes of proteins are ATP-dependent fatty acid CoA ligases producing acyl CoAs, thereby trapping the fatty acid in a form incapable of diffusing out of the cell.
116
LOBO & BERNLOHR
Fatty acid transport protein 1 (FATP1) was discovered in an expression cloning screen designed to identify proteins that augment uptake of long chain fatty acids in adipocytes (Schaffer & Lodish 1994). Based on sequence homology, six isoforms (FATP1–6) with specific tissue expression patterns have been identified in the mammalian genome. Adipose tissue predominantly expresses the FATP1 and FATP4 isoforms. FATP1, which has been studied extensively, is transcriptionally down-regulated by insulin, tumour necrosis factor α (TNFα ), interleukin 1 (IL1) and up-regulated by PPARγ in adipocytes (Frohnert et al 1999, Memon et al 1998, Hui et al 1998). Loss of function/gain of function studies in various model systems that include yeast, cultured cells and mice have provided sufficient evidence for a role of FATP1 (but not FATP4) in adipocyte fatty acid uptake. FATPs were found to share a 20–40% sequence identity with a different class of proteins termed long-chain acyl CoA synthetases (ACSLs) that catalyse the ATP-dependent CoA esterification of long-chain fatty acids. Purification of mouse FATP1 and FATP4 and in vitro enzyme assays confirmed that these proteins exhibit intrinsic long-chain and particularly robust very long-chain acyl CoA synthetase activity (Hall et al 2003, 2005). Immunofluorescence studies reveal that in adipocytes, in response to insulin, FATP1 is translocated to the disordered non-lipid raft domain of the plasma membrane from intracellular membranes (Stahl et al 2002). Knockout of FATP1 in mice as well as RNAi knockdowns of FATP1 in 3T3-L1 adipocytes results in a complete loss of the insulin-stimulated component of fatty acid uptake revealing a significant role for FATP1 in insulin-stimulated fatty acid influx (Wu et al 2006). The fatty acid transport function of FATP1 is tied to its acyl CoA synthetase activity as catalytically inactive mutants of FATP1 are non-functional in catalysing fatty acid uptake (Stuhlsatz-Krouper et al 1998). Hamilton et al have proposed that fatty acid metabolism and not translocation of fatty acids across cellular membranes is rate limiting in fatty acid uptake (Guo et al 2006). Hence, an increase in esterification of fatty acids by FATP1 at the plasma membrane and their subsequent metabolism prevents efflux and produces a fatty acid gradient across the membrane. There are five distinct isoforms of long chain acyl CoA synthetases (ACSL1, 3, 4, 5 and 6) that vary in tissue specificity and partitioning of fatty acids towards diverse cellular metabolic pathways. White adipose tissue expresses predominantly ACSL1, whereas ACSL5 is also expressed in brown fat. Similar to FATP1, ACSL1 is transcriptionally up-regulated by PPARγ agonists in adipocytes and also translocates to the plasma membrane in response to insulin. Characterization of the acyl CoA synthetase activity of ACSL1 indicated that although FATP1 and ACSL1 have similar fatty acid substrate specificities, ACSL1 is a higher velocity enzyme for long chain fatty acids over very long chain lipids (Hall et al 2003). Over expression of ACSL1 in fibroblast cells as well as in vitro reconstitution of purified ACSL1
ADIPOCYTE FATTY ACID TRANSPORT
117
into large unilamellar vesicles led to an increase in ATP-dependent fatty acid influx implying a role for ACSL1 in LCFA import (Gargiulo et al 1999, Schmelter et al 2004). Although epitope-tagged ACSL1 and FATP1 co-localize in the adipocyte membrane (Gargiulo et al 1999), the purified proteins do not physically associate with each other and immunoprecipitation studies do not reveal any complex formation. This seems to be a distinction between mammalian FATP1 and ACSL1 compared to their yeast counterparts where over expression results in physical association between the yeast Fat1p and either Faa1p or Faa4p (Zou et al 2003). Model for fatty acid uptake in adipocytes The discussion detailing the results of molecular and biochemical studies provides compelling evidence for the role of several proteins and, potentially, multiple mechanisms facilitating fatty acid influx. It is likely that multiple proteins may work in concert to achieve the rapid fatty acid uptake observed in adipocytes. Moreover, no protein has been biochemically or biophysically demonstrated to facilitate fatty acid fl ip-flop from the outer to the inner leaflet. These studies underscore the potential importance of diffusion of fatty acids across the membrane bilayer as a key step in the process. As such, fatty acid influx across the biological membrane may involve components of both diffusion and protein transport. Based on the published data, one can propose a general mechanism by which fatty acids are taken up into cells. Firstly, in response to insulin, FAT/CD36 as well as ACSL1 and FATP1 translocate from internal sites to caveolae (FAT/CD36) or to non-raft membranes (ACSL1, FATP1). FAT/CD36 may function as a lipid receptor thereby creating a high local concentration of fatty acid in the outer leaflet of the plasma membrane. Such protonated fatty acids diffuse from the outer to the inner leaflet at a rate dictated by the fatty acid chain-length and/or degree of unsaturation. Fatty acids on the inner leaflet of the plasma membrane laterally diffuse to the disordered glycerophospholipid domains harbouring either ACSL1 or FATP1. The enzymatic activation of the fatty acids by either ACSL1 or FATP1 results in the production of acyl CoAs that could be transported to the ER for incorporation into triglycerides. Alternatively, acyl CoAs could be retained by cycling caveolar membranes thereby providing a delivery mechanism from the plasma membrane to the ER. Regulation of insulin signalling by lipid metabolites Central obesity is often associated with the development of type 2 diabetes characterized by resistance of tissues to the actions of insulin, in addition to dyslipidaemia and hypertension. Enlarged visceral adipocytes release large amounts of FFAs into circulation due to increased lipolysis, and insulin resistance in adipose
118
LOBO & BERNLOHR
tissue causes its desensitization to the anti-lipolytic effect of insulin and coincident overproduction of cytokines such as TNFα and IL6. Excess FFAs in the plasma lead to ectopic triglyceride accumulation in non-adipose tissues such as muscle, liver and pancreatic β cells, resulting in insulin resistance and β cell dysfunction. Lowering of elevated plasma FFA levels normalizes insulin sensitivity in diabetic patients. A surplus of cellular FFA and FFA metabolites such as diacylglycerols (DAGs) have been shown to negatively regulate the insulin signalling pathway, leading to impaired insulin activity (Itani et al 2002, Yu et al 2002). Hence, controlling intracellular pool sizes of FFAs may be critical to the regulation of insulin signalling. In adipocytes, the intracellular free fatty acid and DAG pools can be derived from either lipogenesis or lipolysis. Quantitative analysis of complex lipid synthesis from fatty acids following cellular uptake indicated that fatty acids are quantitatively activated to CoA esters within seconds of their uptake and are rapidly incorporated into triacylglycerol. Hence, lipid influx contributes very little, if at all, to the pool sizes of fatty acids and diglycerides implicated in the development of insulin resistance. In contrast, a variety of animal knockout models (ATGL, HSL) have shown that lipolytically derived fatty acids and DAG are the major contributors to these key lipid pools in the adipocyte. The role of FFAs in the cellular mechanism of insulin has been studied at the molecular level. FFAs and DAG can activate protein kinase C isoforms such as PKCθ that leads to the activation of serine kinases such as inhibitor of κ B kinase (IKK β ) and c-JUN N-terminal kinase (JNK). Activation of IKKβ and JNK leads in turn to serine (Ser307) phosphorylation of the insulin receptor substrate 1 (IRS1) inhibiting its interaction with the insulin receptor and promoting its degradation. Reduction in IRS1 protein levels leads in turn to attenuation of phosphatidylinositol-3-kinase activation, diminished Akt phosphorylation and reduced glucose transport, producing an insulin resistant state (Yu et al 2002). Fatty acid transport plays a significant role in insulin resistance and the metabolic syndrome. The expression levels of CD36 were found to negatively correlate with insulin resistance, probably due to its contribution to disturbed fatty acid metabolism. Recently a TT genotype of a promoter polymorphism in the CD36 gene was related to higher fasting glucose concentrations, insulin resistance and type 2 diabetes (Corpeleijn et al 2006). Evidence for the role of FATP1 in insulin resistance comes from observing the phenotype of FATP1-null mice. Mice depleted of FATP1 expression were protected from high fat diet-induced insulin resistance (Wu et al 2006). Moreover, thiazolidinedione treatment, known to increase insulin sensitivity, is associated with increased expression of CD36 and FATP1, suggesting that these proteins may mediate part of the insulin-sensitizing effects of this class of drugs. Consistent with this, shRNA-mediated reduction in FATP1 expression in 3T3-L1 adipocytes leads to increased tyrosine phosphorylation and abundance
ADIPOCYTE FATTY ACID TRANSPORT
119
FIG. 1. Lipid metabolism in adipocytes and regulation of FFA and DAG production.
of IRS1 and a 30% increase in basal and insulin-stimulated glucose uptake (Lobo & Bernlohr 2007). Expression and phosphorylation of lipid activated serine kinases such as p70S6 and JNK1 were unaltered in FATP1 knockdown adipocytes although the phosphorylation of PKCθ and abundance of IKKα/β were significantly reduced, suggesting a key role of the NF-κ B pathway as an important initiator of insulin resistance in adipocytes (Fig. 1). Acknowledgements Supported by NIH DK 053189 and ADA RA12 to DAB.
References Abumrad NA, el-Maghrabi MR, Amri EZ, Lopez E, Grimaldi PA 1993 Cloning of a rat adipocyte membrane protein implicated in binding or transport of long-chain fatty acids that is induced during preadipocyte differentiation. Homology with human CD36. J Biol Chem 268:17665–17668 Bernlohr DA, Simpson MA 1996 Adipose tissue and lipid metabolism. In: Vance DE, Vance JE (eds) Biochemistry of lipids, lipoproteins and membranes (3rd ed.). Elsevier p 257–281
120
LOBO & BERNLOHR
Coburn CT, Hajri T, Ibrahimi A, Abumrad NA 2001 Role of CD36 in membrane transport and utilization of long-chain fatty acids by different tissues. J Mol Neurosci 16:117–121; discussion 151–157 Corpeleijn E, van der Kallen CJ, Kruijshoop M et al 2006 Direct association of a promoter polymorphism in the CD36/FAT fatty acid transporter gene with type 2 diabetes mellitus and insulin resistance. Diabet Med 23:907–911 Frohnert BI, Hui TY, Bernlohr DA 1999 Identification of a functional peroxisome proliferator-responsive element in the murine fatty acid transport protein gene. J Biol Chem 274:3970–3977 Gargiulo CE, Stuhlsatz-Krouper SM, Schaffer JE 1999 Localization of adipocyte long-chain fatty acyl-CoA synthetase at the plasma membrane. J Lipid Res 40:881–892 Guo W, Huang N, Cai J, Xie W, Hamilton JA 2006 Fatty acid transport and metabolism in HepG2 cells. Am J Physiol Gastrointest Liver Physiol 290:G528–534 Hall AM, Smith AJ, Bernlohr DA 2003 Characterization of the acyl-CoA synthetase activity of purified murine fatty acid transport protein 1. J Biol Chem 278:43008–43013 Hall AM, Wiczer BM, Herrmann T, Stremmel W, Bernlohr DA 2005 Enzymatic properties of purified murine fatty acid transport protein 4 and analysis of acyl-CoA synthetase activities in tissues from FATP4 null mice. J Biol Chem 280:11948–11954 Hamilton JA 2003 Fast fl ip-flop of cholesterol and fatty acids in membranes: implications for membrane transport proteins. Curr Opin Lipidol 14:263–271 Hui TY, Frohnert BI, Smith AJ, Schaffer JE, Bernlohr DA 1998 Characterization of the murine fatty acid transport protein gene and its insulin response sequence. J Biol Chem 273:27420–27429 Itani SI, Ruderman NB, Schmieder F, Boden G 2002 Lipid-induced insulin resistance in human muscle is associated with changes in diacylglycerol, protein kinase C, and IkappaB-alpha. Diabetes 51:2005–2011 Kamp F, Hamilton JA 2006 How fatty acids of different chain length enter and leave cells by free diffusion. Prostaglandins Leukot Essent Fatty Acids 75:149–159 Lobo S, Bernlohr D 2007 Fatty acid metabolism in adipocytes: functional analysis of fatty acid transport proteins 1 and 4. J Lipid Res 48:609–620 Luiken JJ, Bonen A, Glatz JF 2002 Cellular fatty acid uptake is acutely regulated by membraneassociated fatty acid-binding proteins. Prostaglandins Leukot Essent Fatty Acids 67:73– 78 Memon RA, Feingold KR, Moser AH, Fuller J, Grunfeld C 1998 Regulation of fatty acid transport protein and fatty acid translocase mRNA levels by endotoxin and cytokines. Am J Physiol 274:E210–217 Pohl J, Ring A, Stremmel W 2002 Uptake of long-chain fatty acids in HepG2 cells involves caveolae: analysis of a novel pathway. J Lipid Res 43:1390–1399 Pohl J, Ring A, Ehehalt R et al 2004 Long-chain fatty acid uptake into adipocytes depends on lipid raft function. Biochemistry 43:4179–4187 Pohl J, Ring A, Korkmaz U, Ehehalt R, Stremmel W 2005 FAT/CD36-mediated longchain fatty acid uptake in adipocytes requires plasma membrane rafts. Mol Biol Cell 16: 24–31 Pol A, Martin S, Fernandez MA et al 2005 Cholesterol and fatty acids regulate dynamic caveolin trafficking through the Golgi complex and between the cell surface and lipid bodies. Mol Biol Cell 16:2091–2105 Razani B, Combs TP, Wang XB et al 2002 Caveolin-1-deficient mice are lean, resistant to dietinduced obesity, and show hypertriglyceridemia with adipocyte abnormalities. J Biol Chem 277:8635–8647 Sato O, Kuriki C, Fukui Y, Motojima K 2002 Dual promoter structure of mouse and human fatty acid translocase/CD36 genes and unique transcriptional activation by peroxisome proliferator-activated receptor alpha and gamma ligands. J Biol Chem 277:15703–15711
ADIPOCYTE FATTY ACID TRANSPORT
121
Schaffer JE, Lodish HF 1994 Expression cloning and characterization of a novel adipocyte long chain fatty acid transport protein. Cell 79:427–436 Schmelter T, Trigatti BL, Gerber GE, Mangroo D 2004 Biochemical demonstration of the involvement of fatty acyl-CoA synthetase in fatty acid translocation across the plasma membrane. J Biol Chem 279:24163–24170 Stahl A, Evans JG, Pattel S, Hirsch D, Lodish HF 2002 Insulin causes fatty acid transport protein translocation and enhanced fatty acid uptake in adipocytes. Dev Cell 2:477–488 Stuhlsatz-Krouper SM, Bennett NE, Schaffer JE 1998 Substitution of alanine for serine 250 in the murine fatty acid transport protein inhibits long chain fatty acid transport. J Biol Chem 273:28642–28650 Trigatti BL, Anderson RG, Gerber GE 1999 Identification of caveolin-1 as a fatty acid binding protein. Biochem Biophys Res Commun 255:34–39 Uittenbogaard A, Smart EJ 2000 Palmitoylation of caveolin-1 is required for cholesterol binding, chaperone complex formation, and rapid transport of cholesterol to caveolae. J Biol Chem 275:25595–25599 Wu Q, Ortegon AM, Tsang B, Doege H, Feingold KR, Stahl A 2006 FATP1 is an insulinsensitive fatty acid transporter involved in diet-induced obesity. Mol Cell Biol 26:3455–3467 Yu C, Chen Y, Cline GW et al 2002 Mechanism by which fatty acids inhibit insulin activation of insulin receptor substrate-1 (IRS-1)-associated phosphatidylinositol 3-kinase activity in muscle. J Biol Chem 277:50230–50236 Zou Z, Tong F, Faergeman NJ, Borsting C, Black PN, DiRusso CC 2003 Vectorial acylation in Saccharomyces cerevisiae. Fat1p and fatty acyl-CoA synthetase are interacting components of a fatty acid import complex. J Biol Chem 278:16414–16422
DISCUSSION Black: Jean Schaffer has also shown that FATP1 and ACSL1 are functionally associated and interact in the adipocyte (Gargiulo et al 1999, Richards et al 2006). Bernlohr: She showed that they co-localize; she hasn’t shown that they associate physically. Black: They do co-immunoprecipitate (co-IP). Bernlohr: Tag overexpressed proteins do, but my lab and three others have no evidence for any co-IP. If we overexpress Tag versions we can co-IP them, but if we over-express non-tagged versions we can’t. I’d like to believe this story but we don’t have any data. Black: The overexpression worries me. All of the ACSL1 overexpression data lead to aberrant lipid deposition, particularly within triglyceride droplets and membraneous whorls. I think overexpression is screwing up those homeostatic switches rather dramatically with relatively subtle changes in the expression levels of those proteins. Bernlohr: We went into this thinking we would find complex formation. I don’t have any data for this. Shi: A key question about lipid storage in the adipocyte is when they should be stored as a fuel and when they should be used as a signal molecule. Do FATP
122
LOBO & BERNLOHR
proteins play a role in regulating lipid traffic between different cellular compartments? Bernlohr: At this point, our evidence suggests that they are metabolic enzymes whose function is to catalyse the ATP and CoA-dependent trapping. Therefore they funnel into the metabolic fuel site, not so much the signal site. Shi: If you feed the cell with saturated versus polyunsaturated lipids, would they go in a different direction? Bernlohr: No, we don’t see any difference. However, in our studies the fatty acids are buffered with BSA and are delivered under physiological conditions. If you do them in an unbuffered manner in glucose starved cells (so ATP pools are low) you will get activation of other pathways. I think it is all in context. This is an ATPdependent process. If ATP levels are altered this will influence the pathways the lipids will go via. Many of the effects are likely to be reflecting experimental differences. We can get the same kind of activation of JNK with fatty acids, but if we use them buffered and under physiological conditions we don’t see any JNK activation. Attie: Do these proteins bind any insulin signalling proteins? Bernlohr: There are some that might. That’s all we can say. Hotamisligil: Can you put de novo fatty acid synthesis into your picture? The composition in adipose tissue, according to some schools of thought, is a reflection of what percentage de novo fatty acid synthesis is contributing to the eventual fatty acid load. Spiegelman: This is a bit of a species thing. Bernlohr: Yes, in human versus rodents it does a fl ip. But your point is well taken. It is difficult to study, because you would want to do this in the overexpressors and knockouts. However, they clearly change their ability to run de novo lipogenesis. The knockout animals have increased conversion of acetate to triglyceride, they are very insulin sensitive and AMPK phosphorylation is down, so trying to determine the relative fraction of lipid that is deposited from de novo versus uptake in either the animal model or cell culture model is difficult. In humans we largely believe that the dietary uptake pathway contributes a significant fraction. Tabas: You put forward the idea that the free fatty acids come from the hydrolysis not the usual uptake, this is an important point. I don’t think you mentioned much about the lipid droplet proteins. How do these fit into the picture? Bernlohr: The primary lipid droplet protein is perilipin in adipocytes. It is clearly a key component here. Even in the FABP animal models that Bruce and Gökhan have made, perilipin expression is altered. Clearly, the FABP doesn’t seem to have any physical accociation with perilipin, but the lipase has a component that may associate with perilipin on small lipid droplets. There is a dynamic here on or around the large droplet in response to forskolin stimulation being deconvoluted into smaller droplets. The smaller ones may be the most lipolytically active.
ADIPOCYTE FATTY ACID TRANSPORT
123
Spiegelman: I recall from the structural work that the FABPs are like a barrel into which the fatty acids enter. How do you envisage the structural basis of the function of an apo FABPs versus a bound FABP. Bernlohr: In our structural papers we identified a single residue, Phe57. We think of this as a molecular gatekeeper. It sits on a loop, and once a fatty acid goes in it does a 140° shift from the apo to holo form. Structurally this is the only change we see in the bound versus the free form. We are testing whether or not this is sufficient to signal apo or holo in a cellular context by making mutants in this region and then using the FRET assay as a way of assessing physical association. Spiegelman: I guess eventually you could try to crystallize some of these interacting partners. Bernlohr: We have co-crystals of a component of the lipase with a fatty acid binding. Spiegelman: You say you did this by yeast-two-hybrid. How did you do this in light of the notion of apo versus non-apo FABP? Bernlohr: In yeast they bind fatty acid. In the yeast two-hybrid you get no signal in an R126Q mutant. They require fatty acid to be bound. O’Rahilly: You posed the question of how a fat cell knows how fat it is. How will you know when you have got there? In other words, what does the fat cell do with the information, and what is your readout for whether it is intelligently interpreting this information? I am thinking of things such as leptin and adiponectin. Presumably if you interfere with the step that is involved in the sensing of how large the cell is, you would expect to find a big fat cell with low leptin. Bernlohr: Adipokine biology is changed in the FABP4 knockout versus transgenic animals. Göhkan has shown that the TNFα production is significantly changed. Insulin receptor signalling is also altered. This all occurs through a single alteration in the adipocyte FABP. The situation is actually a little more complex because there is a second FABP that is a minor one expressed in adipocytes. This is expressed to a greater extent in macrophages. The most dramatic metabolic phenotype is in the double knockout in which the animals lack both the adipocyte and epithelial fatty acid-binding proteins, which means that the adipose tissue will now be devoid of the FABPs. O’Rahilly: I guess leptin will be most compelling, because we know that intrinsic to the physiology is the process by which as the adipocyte accumulates triglyceride it makes more leptin and is part of a physiological loop signal. If you could dissociate this, then you will have got there. Bernlohr: Yes, the knockout animals have reduced TNFα expression. They also have increased adiponectin expression. O’Rahilly: Leptin is the key physiologically. Bernlohr: We have never seen any changes in leptin. I believe that there are major changes in the double knockout, though.
124
LOBO & BERNLOHR
Hotamisligil: The double knockout animals will become lean. Therefore they have very low leptin levels. Actually, they are disproportionately low. This might raise the possibility of increased leptin sensitivity in FABP-deficient mice. Sabin: Are there any differences between saturated and unsaturated fatty acids in how they bind with FABP? Bernlohr: The volume of a fatty acid is about 300 Å 3 and the volume of the cavity is about 900. Each lipid is bound uniquely within the cavity, but all lipids induce the phenylalanine 57 rotamer shift. This is at least a consistency. Saito: I have a question about FABP in brown adipocytes. There are at least two isoforms of FABP in brown adipocytes: adipose-type AFABP and heart-type HFABP. Do you have any data or ideas about HFABP in brown fat? Bernlohr: There is some controversy in the literature. We don’t find that. The FABP5 and FABP4 are both found in brown fat. Hotamisligil: When we looked at metabolic regulation in the FABP knockouts, although the predominant site of their action is the adipocyte, major metabolic differences turned out to be in muscle and liver. We still don’t understand how this happens. There is some evidence that FABPs regulate peptide signals. Recently we also observed that these proteins, which do not follow the classic secretory pathway and do not have the signatures of any secretory proteins, do make it into circulation. Their levels correlate well with obesity and insulin resistance. In this sense, FABP4 in particular looks similar to the story with RBP4, conceptually. This might change the view that we have about how these proteins work from a distance. Bernlohr: There is a clinical assay for heart FABP, for looking at early stage heart attacks. When damage is done this small protein leaks out readily. This has been known for several years. I am not convinced that the serum forms don’t represent cell death and necrotic action, with some leakage of FABP as opposed to a bonafide biological process. Spiegelman: I have no horse in this race at all. I guess it could be physiologically or pathophysiologically significant, even if it gets there by steady state apoptosis of fat cells. It is such an abundant protein. Hotamisligil: The striking observation is its strong correlation with body mass index (BMI) and insulin sensitivity in humans. Spiegelman: Fat mass will also correlate. I guess one thing you could do to test this would be to develop a neutralizing reagent in the blood. Presumably this wouldn’t affect the stuff in the fat cell. Hotamisligil: These are ongoing experiments. I am convinced that they are secreted in a regulated manner from a non-traditional pathway. We have also done some experiments in fat cells. If we artificially introduce non-secreted molecules to the same abundance and then follow these and FABP, nothing comes out but aP2. Spiegelman: You alluded to RBP4-like results. Do you think this is physiologically significant?
ADIPOCYTE FATTY ACID TRANSPORT
125
Hotamisligil: We don’t have evidence for this yet but we are working on it. The RBP4 story is that it is increased in all the obesity models, so do the FABPs, especially aP2. In the knockouts we see markedly improved insulin sensitivity in all models. Overexpression suppresses insulin action. David Bernlohr has shown that, we also collaborated with him on a model with FABP5 showing the same results. We have a paper in the pipeline showing that if you chemically block FABP lipid binding, you get good insulin sensitising results in vivo. Spiegelman: I wanted to comment on PPAR and PPARγ. There is a common feeling that fatty acids are ligands for the PPARs. The data that they are physiologically significant for PPARα and δ are pretty good, but the data that the common fatty acids are ligands for PPARγ are quite bad. I have an argument at every nuclear receptor meeting about this subject, so I’d be cautious about having those dietary fatty acids heading towards PPARγ. Bernlohr: I think they are substrates for the oxidation pathways that might generate molecules. Glass: I would agree. This remains one of the unsolved problems in PPARγ biology. Many people have worked hard on this and have given up! Spiegelman: I never give up! I don’t succeed, but I never give up. Glass: It is essential to solve this problem if we are to understand the biology. Hotamisligil: If we delete FABPs in macrophages we get clear high activity from PPARγ. Spiegelman: That makes sense: there the levels of free fatty acids are presumably going way up. Bernlohr: No, they are constant. Free unbound is constant. The bound form is changing dramatically. The bulk of the lipid is already in the ER/perinuclear region. Hotamisligil: This doesn’t say whether it is a ligand directly binding, or whether it regulates the synthesis of something else. But it is clear: if you take out FABP4, you get higher PPARγ activity, and it is not because of PPARγ abundance. Shi: This futile recycling process has been studied for a number of years. What is your thought on this? Why does the body invest so much energy to divide the newly absorbed fatty acids and the stored ones? Bernlohr: The trivial answer is that the body considers this a necessary investment. At the same time, lipolysis induces two to three times more fatty acids than are actually utilized, and the triglycerides are recycled back to the adipocytes. Apparently the recycling is important enough that the adipocyte protects it. Daum: I have a question about the lipid droplets. In the transition state from lean to obese, you observe an increase in lipid droplet size. Would this situation create different populations of lipid droplets, not just regarding size but also the lipid and protein composition? Bernlohr: In the animal model, H&E staining reveals simply that the droplets are larger and there are more of them. We don’t see a change in the distribution.
126
LOBO & BERNLOHR
References Gargiulo CE, Stuhlsatz-Krouper SM, Schaffer JE 1999 Localization of adipocyte long-chain fatty acyl-CoA synthetase at the plasma membrane. J Lipid Res 40:881–892 Richards MR, Harp JD, Ory DS, Schaffer JE 2006 Fatty acid transport protein 1 and long-chain acyl coenzyme A synthetase 1 interact in adipocytes. J Lipid Res 47:665–672
Fatty Acids and Lipotoxicity in Obesity and Diabetes: Novartis Foundation Symposium 286, Volume 286. Edited by Gregory Bock and Jamie Goode Copyright Novartis Foundation 2007. ISBN: 978-0-470-05764-3
Vectorial acylation: linking fatty acid transport and activation to metabolic trafficking Paul N. Black and Concetta C. DiRusso Centers for Metabolic Disease, Ordway Research Institute and Cardiovascular Sciences, Albany Medical College, Albany, NY 12208, USA
Abstract. The process of fatty acid transport across the plasma membrane occurs by several mechanisms that involve distinct membrane-bound and membrane-associated proteins and enzymes. Amongst these are the fatty acid transport proteins (FATP) and long-chain acyl CoA synthetases (ACSL). We have shown the yeast orthologues of FATP and ACSL form a physical complex at the plasma membrane and are required for fatty acid transport, which proceeds through a coupled process linking transport with metabolic activation and termed vectorial acylation. At present six isoforms of FATP and five isoforms of ACSL have been identified in mice and human. In addition there are a number of splice variants of different ACSL isoforms; recent work from our laboratory has found at least one splice variant in human FATP2. The different FATP and ACSL isoforms have distinct tissue expression profi les and along with different cellular locations suggest they function in the trafficking of fatty acids into discrete metabolic pools. More specifically, we hypothesize the different FATP and ACLS isoforms function individually and co-ordinately to move distinct classes of fatty acids into these different metabolic pools. The concerted activity of these proteins allows the cell to discriminate different classes of fatty acids and provides the mechanistic basis underpinning the selectivity and specificity of the fatty acid transport process. 2007 Fatty acids and lipotoxicity in obesity and diabetes. Wiley, Chichester (Novartis Foundation Symposium 286) p 127–141
The processes governing the transport of fatty acids across biological membranes is complex and distinct from those involved in the transport of hydrophilic solutes, including sugars and amino acids. The biophysical properties of fatty acids allow them to readily partition into the lipid bilayer and, in a protonated form, to fl ip between the two surfaces of the membrane. Recent studies have shown the fl ip through the membrane is rate-limiting, which supports the hypothesis that specific proteins are required to promote this process (Kampf et al 2006). Our work has addressed the biochemical mechanisms governing the transport of fatty acids 127
128
BLACK & DIRUSSO
across the membrane and into downstream metabolic pools and while initially focused on defining this process in bacteria and yeast, has now been extended to include mammalian systems. Emerging evidence demonstrates that at least one process driving fatty acid transport in eukaryotic cells is vectorial acylation, which requires a fatty acid transport protein (FATP) and a long-chain acyl CoA synthetase (ACSL) that function in the concomitant transport and activation of fatty acids. Other proteins involved in fatty acid transport include FAT/CD36, FABPpm and specific isoforms of FABP, but the focus of this article will be on the FATP and ACSL isoforms and current research on how these proteins are involved in the transport and trafficking of fatty acids into different metabolic pools. The discussion will address the issues of specificity and selectivity in the process of fatty acid transport and trafficking, particularly in the context of diabetes, hyperlipidaemia and obesity. Concept of vectorial acylation The principle behind vectorial acylation is that exogenous fatty acids are trafficked across the plasma membrane concomitant with activation to CoA thioesters. This idea was put forth by Peter Overath in 1969 in studies describing the role of ACSL [referred to as FadD] from Escherichia coli in fatty acid metabolism (Overath et al 1969). In these experiments, intracellular free fatty acids could not be detected following incubation with exogenous long-chain fatty acids. Rather acyl CoA derivatives were detected, which were destined for either complex lipid synthesis or β -oxidation. Bacterial cells lacking this enzyme are also unable to transport exogenous fatty acids further supporting this concept. In the yeast Saccharomyces cerevisiae, the same is true, but with an added layer of complexity. In this system, exogenous fatty acids are converted to acyl CoAs upon transport. Cells defective in the two major ACSLs (Faa1p and Faa4p) are unable to accumulate intracellular oleoyl CoA following incubation with oleate (Faergeman et al 2001). This is also the case for yeast cells defective in FATP [designated Fat1p] (DiRusso et al 2000). Subsequent studies have shown these two proteins interact and function in concert at the plasma membrane, which along with our earlier studies confirm the mechanistic roles of FATP and ACSL in vectorial acylation (Zou et al 2003). The situation in mammalian cells adds yet another layer of complexity as there are at least 6 FATP and 5 ACSL isoforms. Yet in mammalian cells specific isoforms of FATP and ACSL appear to mediate vectorial acylation in a manner analogous to that defined in yeast. When the first FATP was identified using expression cloning, an often-overlooked result was the identification of ACSL1, which when over expressed in Cos7 cells also increased fatty acid transport (Schaffer & Lodish 1994). It is now known that FATP1 and ACSL1 form a complex at the plasma membrane in mouse adipocytes analogous to what we defined in yeast (Richards et al 2006, Zou et al
VECTORIAL ACYLATION OF FATTY ACIDS
129
2003). Figure 1A provides an overview of vectorial acylation stemming from these studies. Some isoforms of FATP function in concert with a cognate ACSL to promote the coupled transport-activation of fatty acids. Some of the FATPs may function alone in vectorial acylation of specific classes of fatty acids. The activities
FIG. 1. Vectorial acylation and acyl CoA metabolism. (A) Concept of vectorial acylation. (1) Fatty acid transport coupled activation as defi ned in E. coli. Fatty acids bind to and fl ip between the membrane surfaces [shaded arrow]; ACSL [shaded oval] becomes membrane-associated and functions to abstract fatty acids from the membrane concomitant with activation to acylCoA [FA-CoA]; (2) fatty acid transport coupled activation requiring both an FATP [shaded rectangle] and a cognate ACSL; in this scenario, FATP is involved in the transmembrane movement of the fatty acids and the ACSL functions in activation; (3) fatty acid transport coupled activation requiring only an FATP; in this case the FATP is required for both transport and activation. (B) Acyl-CoA metabolism within the cell. Fatty acids must be activated to acyl-CoA prior to metabolism. As shown, acyl-CoA pools arise from vectorial acylation [import/activation], lipolysis of triglyceride stores, chain turnover from complex lipids and de novo biosynthesis. Acyl-CoAs are important signalling molecules, incorporated into complex lipids and triglycerides, are substrates for energy production through β -oxidation and substrates for protein N-myristoylation and palmitoylation. Acyl-CoAs are also subject to chain modification, including desaturation and elongation. Acyl chains in complex lipids serve both structural and regulatory roles.
130
BLACK & DIRUSSO
of these proteins are reflected by changes in the acyl-CoA pools. As illustrated in Fig. 1B, acyl-CoA pools arise from four different sources: [1] coupled import/activation [vectorial acylation], [2] de novo synthesis, [3] liberation of stored fatty acids following lipolysis of triglyceride stores and [4] from complex lipid turnover. AcylCoAs are also subject to chain modification, including desaturation and elongation as occurs in the synthesis of monounsaturated and very long-chain highly unsaturated fatty acids. These pools are envisioned to be fairly discrete although one pool may overlap another. We suspect the different FATP and ACSL isoforms function to partition distinct classes of fatty acids into discrete metabolic pools concomitant with activation. Features of the FATPs The first fatty acid transport protein [now designated mmFATP1 and a member of the Slc27 protein family], was identified using expression cloning from a cDNA library prepared from murine 3T3-L1 adipocytes, directly demonstrating a physiological role in the net accumulation of fatty acids across a biological membrane (Schaffer & Lodish 1994). Six isoforms of FATP have subsequently been identified experimentally in mice, rats, and humans [e.g. in mice, mmFATP1, 2, 3, 4, 5 and 6] (Gimeno et al 2003, Hirsch et al 1998, Schaffer & Lodish 1994, Stahl et al 1999). Most recently we have identified two splice variants for human FATP2 suggesting such variants in the other FATP isoforms may also exist (C. DiRusso, P. Watkins, and P. Black, unpublished). Members of the FATP family have also been identified experimentally or by sequence comparisons in non-mammalian systems including Caenorhabditis elegans, Drosophila melanogaster, S. cerevisiae, and Mycobacterium tuberculosis (Hirsch et al 1998, Faergeman et al 1997). The mammalian FATPs have broad tissue distribution, but there is a preference for specific isoforms in different tissues. FATP1 for example is highly expressed in skeletal muscles, but poorly expressed in intestine. FATP4, by comparison, is highly expressed in intestine. The FATPs are highly conserved proteins, which are localized in part to the plasma membrane and are likely to be homodimers with subunits of 650–700 amino acid residues in length (Gimeno et al 2003, Schaffer & Lodish 1994, Stahl et al 1999, Lewis et al 2001, Richards et al 2003). Two domains within all FATPs share significant identity and serve as distinguishing sequence elements or motifs (Fig. 2): The ATP/AMP binding motif [common to all adenylate-forming enzymes, including the ACSLs] and the FATP/ACSVL motif, which is restricted to this group of proteins. The transmembrane domains within the FATPs are amino-terminal proximal; one transmembrane domain has been identified in mmFATP1 (Lewis et al 2001) and two in the yeast Fat1p (T. Obermeyer, C. DiRusso and P. Black, submitted). Both topological studies place the highly conserved ATP/AMP and FATP/VLACS motifs on the cytoplasmic face of the plasma membrane.
VECTORIAL ACYLATION OF FATTY ACIDS
131
Fatty Acid Transport Proteins Core ATP/AMP motif hsFATP1 FYIYTSGTTGLPKAAIVVHSR hsFATP2a LYIYTSGTTGLPKAAMITHQR hsFATP2b LYIYTSGTTG----------hsFATP3 LYIFTSGTTGLPKAARISHLK hsFATP4 FYIYTSGTTGLPKAAIVVHSR hsFATP5 LFIYTSGTTGLPKPAILTHER hsFATP6 LYIFTSGTTGLPKAAVISQLQ
Core FATP/VLACS motif YMYFRDRSGDTFRWRGENVSTTEVE FIYFHDRVGDTFRWKGENVATTEVA FIYFHDRVGDTFRWKGENVATTEVA FLRFHDRTGDTFRWKGENVATTEVA YLYFRDRTGDTFRWKGENVSTTEVE FLYFRDRLGDTFRWKGENVSTHEVE FLYFWDRTGDTFRWKGENVATTEVA
Acyl CoA Synthetases Core ATP/AMP motif hsACLS1 VICFTSGTTGNPKGAMVTHRN hsACLS3 VIMYTSGSTGLPKGVMISHSN hsACSL4 IVMYTSGSTGRPKGVMMHHSN hsACSL5a VICFTSGTTGDPKGAMITHQN hsACSL5b VICFTSGTTGDPKGAMITHQN hsACSL6 IVCFTSGTTGNPKGAMLTHGN
Core FACS motif TGDIGKWLPNGTLKIIDRKKHIFKL TGDIGEFEPDGCLKIIDRKKDLVKL TGDIGEFHPDGCLQIIDRKKDLVKL TGDIGRWLPNGTLKIIDRKKNIFKL TGDIGRWLPNGTLKIIDRKKNIFKL TGDIGKWLPAGTLKIIDRKKHIFKL
FIG. 2. Distinguishing motifs in the FATP and ACSL families. The core ATP/AMP motif for the human FATPs and ACSLs are shown on the left [residues ∼270–310 for the FATPs and residues ∼290–340 for the ACSLs]. As noted, hsFATP2b lacks part of the core ATP/AMP motif, which is required for ATP binding and adenylate formation. The core FATP/VLACS [upper right; residues ∼530–580] and FACS [lower right; residues ∼590–640] motifs are in approximately the same regions on these proteins in terms of linear sequence. The functional aspects of the FACS motif, including its contribution to the fatty acid binding site, has been defi ned in studies using the bacterial enzyme.
Several FATPs also function as very long-chain acyl CoA synthetases [ACSVL] (Coe et al 1999, DiRusso et al 2000). Bernlohr and colleagues have determined the acyl CoA synthetase specificity profi les of purified mmFATP1 and mmFATP4 (Hall et al 2003, 2005). These enzymes have broad substrate specificity for activating long- and very long-chain fatty acids [16–24 carbons in length]. By comparison, tissues from embryonic FATP4 null mice are defective in very long-chain acyl CoA synthetase activity; the long-chain activities are comparable to the wild-type littermates (Hall et al 2005). Unlike several of the ACSLs, the acyl CoA synthetase activity intrinsic to mmFATP1 and mmFATP4 is insensitive to inhibition by triacsin C. Fat1p from yeast also has very long-chain acyl CoA synthetase activity but this activity is distinguishable from transport (Zou et al 2002). As detailed below, there is considerable evidence that several FATP isoforms pair with cognate ACSLs, which provide the long chain activating activity essential to drive transport. In 293 cells stably expressing mmFATP1, exogenous fatty acids are preferentially channelled into the triglyceride biosynthetic pathway (Hatch et al 2002). This is accompanied by down regulation of de novo sphingomyelin and cholesterol biosynthetic pathways in actively growing cells, indicating that the FATP-dependent mechanism governing fatty acid trafficking is also linked to lipid biosynthetic pathways.
132
BLACK & DIRUSSO
The function of both the ATP/AMP and FATP/VLACS motifs noted above have been determined from studies using directed mutagenesis of mmFATP1 and Fat1p (Gargiulo et al 1999, Zou et al 2002). In the studies using mmFATP1 with substitutions within the ATP/AMP motif, fatty acid transport activity and ATP binding were reduced; long- and very long-chain activation activities were not evaluated (Gargiulo et al 1999). One interpretation of these data is that the formation of an acyl-adenylate intermediate is both a part of their reaction mechanism and is required for transport, consistent with vectorial acylation. The most extensive studies describing the functional contributions of the ATP/AMP and FATP/ VLACS motifs come from studies on the yeast Fat1p (Zou et al 2002). Substitutions within the ATP/AMP and FATP/VLACS motifs result in reduced ACSVL activities, which correlates to decreases in long-chain fatty acid transport. Two substitutions within these regions resulted in proteins that were able to function in transport, but not in activation, further supporting the notion that transport can be distinguished from activation (Zou et al 2002). As noted above, we have defined two splice variants for human FATP2. Of interest is a form that lacks part of the ATP/AMP motif, resulting in a protein that differs in molecular weight by ∼5.7 kDa from the longer form. More importantly when fatty acid transport (Li et al 2005) and activation (Zou et al 2002) profi les were determined for each when expressed in yeast defective in transport and activation, distinguishing features were uncovered. The long-form of human FATP2 functions in both fatty acid transport and activation of very long-chain fatty acids while the short form functions in transport; the activation of long-chain fatty acids were essentially the same for both forms and just slightly higher than the vector control (C. DiRusso, P. Watkins, and P. Black, unpublished). Thus in these naturally occurring variants, fatty acid transport can be distinguished from fatty acid activation. In the case of the human FATP2 short form, we suspect it must necessarily function in concert with a cognate ACSL to promote transport and trafficking into downstream metabolism. Features of the ACSLs Long-chain acyl CoA synthetase [ACSL; fatty acid CoA ligase : AMP forming; EC 6.2.1.3] occupies a central position in intermediary metabolism; different isoforms of the enzyme appear to function in the metabolic activation and trafficking of exogenous fatty acids into distinct metabolic pools (Coleman et al 2002, Black & DiRusso 2003). We suspect different ACSLs in concert with distinct FATPs provide both selectivity and specificity of fatty acid transport and trafficking. Members of the ACSL family have been identified in all organisms for which there is genomic sequence information, testifying to their essential nature and to their evolutionary conservation. The formation of acyl-CoA is requisite for subsequent
VECTORIAL ACYLATION OF FATTY ACIDS
133
metabolism of fatty acids including β -oxidation, incorporation into phospholipids, triglyceride synthesis, and protein modification such as N-myristoylation and palmitoylation (Fig. 1B). In addition, there is substantial information showing these compounds function as bioactive metabolites, including acting as effectors of transcriptional regulation (Black & DiRusso 2003). The ACSLs are highly conserved enzymes, usually homodimers with subunits of 600–700 amino acid residues in length. The subunits are characterized by two structural and functional motifs, which specify substrate binding and catalysis (Fig. 2). The ATP/AMP binding motif is shared with the larger family of adenylate forming enzymes, including the FATPs, and is essential for catalysis. Single amino acid substitutions within the E. coli ACSL FadD corresponding to the ATP/AMP motif result in no activity or reduced activity, which affects kcat alone or both kcat and K m. The second sequence motif, consisting of 25 amino acid residues and designated the FACS motif, is common amongst all ACSL family members and contributes to the fatty acid binding site (Black & DiRusso 2003). Proteins with amino acid substitutions within the ACSL FadD corresponding to this motif have altered fatty acid substrate specificity, which correlates to changes in fatty acid binding. The affinity labelled fatty acid 9-p-azidophenoxy nonanoic acid specifically modifies FadD in the region comprising the FACS motif, consistent with the proposal that this region participates in the binding of long chain fatty acids (Black et al 2000). To date, genes encoding five mammalian ACSL isoforms have been identified and characterized (Coleman et al 2002, Kim et al 2001, Lewin et al 2001). This characterization includes the identification of a number of splice variants for each of the different isoforms. These have been designated ACSL1 and ACSL3–ACSL6 [see http://www.gene.ucl.ac.uk/ nomenclature/genefamily/acs.html ]. Rat ACSL1 [rnACSL1] is the major long chain ACSL in liver and adipose tissue where it contributes acylCoA for triglyceride synthesis (Coleman et al 2002, Kim et al 2001, Lewin et al 2001). Cell fractionation studies demonstrate rnACSL1 is abundant in cytosol. However, soluble rnACSL1 only accounts for 2–3% of the total cell associated activity suggesting that like the ACSLs in S. cerevisiae and E. coli [Faa1p and FadD respectively] this enzyme becomes membrane associated in the active form. Rat ACSL6 [rnACSL6; formerly ACS2] and rnACSL5 share approximately 60% identity with rnACSL1, while rnACSL3 and 4 are more divergent sharing only 30% identity with rnACSL1. Clones for rnACSL3 and 6 were isolated from brain libraries and are highly expressed in neuronal tissues (Setnik & Nobrega 2004). Rat ACSL4 is highly expressed in the adrenal gland but is also found in liver and has specificity towards C20 : 4 and C20 : 5. Rat ACSL5 is the predominant isoform in intestinal epithelial cells, where it is likely to contribute to the synthesis of triglycerides destined for incorporation into chylomicrons (Mashek et al 2006). Like rnACSL1 and 4, rnACSL5 is also found in liver. When expressed in yeast defective in acyl
134
BLACK & DIRUSSO
CoA synthetase activity, rnACSL1, 4 and 6 function in fatty acid transport by vectorial acylation (Tong et al 2006). Somewhat less information has been reported on the murine ACSL isoforms. However, since they share at least 97% identity with the rat enzymes, it is presumed they function and are expressed in manner similar to the rat enzymes. The mmACSL1 was identified in the same functional cloning selection scheme that identified mmFATP1 (Schaffer & Lodish 1994). It was further demonstrated that mmACSL1 like mmFATP1 is found at the plasma membrane and together with mmFATP1 synergistically enhance the levels of fatty acid transport (Gargiulo et al 1999). This led in part to the proposal that the mechanism of FATP-dependent fatty acid transport requires ACSL-mediated activation of the imported fatty acid by vectorial acylation. FATP, ACSL and vectorial acylation Expression of either mmFATP1 or mmACSL1 increases the overall rate of fatty acid transport consistent with the idea that fatty acid import is linked to activation (Gargiulo et al 1999). 3T3-L1 cells transfected with retroviral vectors encoding both mmFATP1 and mmACSL1 transport long-chain fatty acids at rates nearly three-times those measured for mmFATP1 or mmACSL1 alone, indicating a synergistic effect between these two proteins (Gargiulo et al 1999). These observations have been extended in yeast showing Fat1p and an ACSL [Faa1p or Faa4p], form a functional complex at the plasma membrane (Zou et al 2003). Evidence of a physical interaction between Fat1p and Faa1p comes from three independent biochemical approaches. First, a C-terminal peptide of Fat1p deficient in fatty acid transport exerts a dominant negative effect against ACSL activity. Second, protein fusions employing the ACSL Faa1p as bait and portions of the FATP Fat1p as trap are active when tested using the yeast two-hybrid system. And third, co-expressed, differentially epitope tagged Fat1p and Faa1p or Faa4p are co-immunoprecipitated. More recent studies have shown mmFATP1 and mmACSL1 form an oligomeric complex in 3T3-L1 adipocytes (Richards et al 2006). The efficiency of this interaction, also defined using co-immunoprecipitation, is unaltered by insulin treatment [stimulates fatty acid uptake] or by treatment with isoproterenol [decreases fatty acid uptake and stimulates lipolysis]. These studies also showed that inhibition of ACSL1 activity in adipocytes is correlated with a decrease in fatty acid transport further supporting the idea that fatty acid activation is essential for fatty acid transport. Collectively, these data support the hypothesis that one mechanism driving fatty acid transport is vectorial esterification, which requires a multiprotein complex. In yeast this consists of Fat1p and Faa1p [or Faa4p] and in 3T3-L1 adipocytes this consists of a productive interaction between mmFATP1 and mmACSL1. Further studies are required to address specific interactions between the other mammalian FATP and ACSL isoforms.
VECTORIAL ACYLATION OF FATTY ACIDS
135
FATPs, ACSLs and diabetes, hyperlipidaemia and obesity There is emerging information, which suggests defects in the FATPs [particularly FATP1 and FATP4] may be involved in lipid-related pathologies, including type II diabetes, hyperlipidaemia and obesity. Of particular note are two studies that show [1] intronic polymorphisms in the FATP1 gene are associated with increased plasma triglyceride levels in a French population and [2] heterozygotes with a F209S polymorphism in the FATP4 gene in a population of Swedish men had lower body mass indices and lower triglyceride concentrations, systolic blood pressure and insulin concentrations (Gertow et al 2003, 2004). The knockout of mmFATP4 results in restrictive dermopathy in newborn pups, which culminates in death shortly after birth (Herrmann et al 2003). These data suggest that at least one FATP may be crucial for development. More recently a knockout of mmFATP1 has been described, which results in alterations of insulin sensitivity (Kim et al 2004). In mmFATP1 knockout animals insulin-stimulated long-chain fatty acid transport is abolished in adipocytes and depressed in skeletal muscle; in both tissues, the basal levels of transport are not affected. Deletion of FATP5 in the mouse results in aberrant bile acid conjugation, which in some manner is linked to the regulation of body weight. Schaffer and colleagues have constructed transgenic mouse lines that over express mmFATP1 in the heart (Chiu et al 2004). These studies show that increased cardiac expression of mmFATP1 is correlated with increases in fatty acid transport and an increase in cardiac fatty acid metabolism, which after three months of age show evidence of diastolic dysfunction indicating that disturbances of lipid homeostasis results in pathologies similar to diabetic cardiomyopathy. The acyl-CoA synthetases have been associated with a number of diseases, including mental retardation, hypertension and cardiomyopathies (Iwai et al 2003). Piccini et al (1998) have shown that when the human ACSL4 gene contains a deletion, this is accompanied by elliptocytosis and mental retardation. Studies using the mouse model have demonstrated that overexpression of ACLS1 results in lipotoxic cardiomyopathy. More recent studies have shown that hyperleptinaemia prevents the pathologies defined in the ACSL1 transgenic mice (Lee et al 2004).
Prospective The precise mechanistic details of how the different FATP and ACSL isoforms function in trafficking fatty acids into different metabolic pools have not been fully worked out and indeed this represents a considerable experimental challenge. Current efforts in our laboratory are combining the use of shRNA and stable isotope technologies to address these questions. These approaches are likely to yield a novel body of information, which will establish how these proteins function individually and in concert in fatty acid trafficking. A more thorough
136
BLACK & DIRUSSO
understanding of how these proteins and enzymes work in vectorial acylation and intracellular fatty acid trafficking is essential to describe the pathophysiology of lipotoxicity and how this is related to metabolic syndrome, obesity and cardiovascular disease. Acknowledgements This work has been supported by grants from the National Institutes of Health (GM056840), the American Heart Association (0151215T), the National Science Foundation (MCB-0212745), and the Charitable Leadership Foundation.
References Black PN, DiRusso CC 2003 Transmembrane movement of exogenous long-chain fatty acids: proteins, enzymes, and vectorial esterification. Microbiol Mol Biol Rev 67:454–472 Black PN, DiRusso CC, Sherin D et al 2000 Affi nity labeling fatty acyl-CoA synthetase with 9-p-azidophenoxy nonanoic acid and the identification of the fatty acid-binding site. J Biol Chem 275:38547–38553 Chiu HC, Kovacs A, Blanton RM et al 2004 Transgenic expression of FATP1 in the heart causes lipotoxic cardiomyopathy. Circ Res 96:225–233 Coe NR, Smith AJ, Frohnert BI, Watkins PA, Bernlohr DA 1999 The fatty acid transport protein (FATP1) is a very long chain acyl-CoA synthetase. J Biol Chem 274:36300–36304 Coleman RA, Lewin TM, Van Horn CG, Gonzalez-Baro MR 2002 Do long-chain acyl-CoA synthetases regulate fatty acid entry into synthetic versus degradative pathways? J Nutr 132:2123–2126 DiRusso CC, Connell EJ, Faergeman NJ et al 2000 Murine FATP alleviates growth and biochemical deficiencies of yeast fat1Delta strains. Eur J Biochem 267:4422–4433 Faergeman NJ, DiRusso CC, Elberger A, Knudsen J, Black PN 1997 Disruption of the Saccharomyces cerevisiae homologue to the murine fatty acid transport protein impairs uptake and growth on long-chain fatty acids. J Biol Chem 272:8531–8538 Faergeman NJ, Black PN, Zhao XD, Knudsen J, DiRusso CC 2001 The Acyl-CoA synthetases encoded within FAA1 and FAA4 in Saccharomyces cerevisiae function as components of the fatty acid transport system linking import, activation, and intracellular utilization. J Biol Chem 276:37051–37059 Gargiulo CE, Stuhlsatz-Krouper SM, Schaffer JE 1999 Localization of adipocyte long-chain fatty acyl-CoA synthetase at the plasma membrane. J Lipid Res 40:881–892 Gertow K, Skoglund-Andersson C, Eriksson P et al 2003 A common polymorphism in the fatty acid transport protein-1 gene associated with elevated post-prandial lipaemia and alterations in LDL particle size distribution. Atherosclerosis 167:265–273 Gertow K, Bellanda M, Eriksson P et al 2004 Genetic and structural evaluation of fatty acid transport protein-4 in relation to markers of the insulin resistance syndrome. J Clin Endocrinol Metab 89:392–399 Gimeno RE, Ortegon AM, Patel S et al 2003 Characterization of a heart-specific fatty acid transport protein. J Biol Chem 278:16039–16044 Hall AM, Smith AJ, Bernlohr DA 2003 Characterization of the Acyl-CoA synthetase activity of purified murine fatty acid transport protein 1. J Biol Chem 278:43008–43013 Hall AM, Wiczer BM, Herrmann T, Stremmel W, Bernlohr DA 2005 Enzymatic properties of purified murine fatty acid transport protein 4 and analysis of acyl-CoA synthetase activities in tissues from FATP4 null mice. J Biol Chem 280:11948–11954
VECTORIAL ACYLATION OF FATTY ACIDS
137
Hatch GM, Smith AJ, Xu FY, Hall AM, Bernlohr DA 2002 FATP1 channels exogenous FA into 1,2,3-triacyl-sn-glycerol and down-regulates sphingomyelin and cholesterol metabolism in growing 293 cells. J Lipid Res 43:1380–1389 Herrmann T, van der Hoeven F, Grone HJ et al 2003 Mice with targeted disruption of the fatty acid transport protein 4 (Fatp 4, Slc27a4) gene show features of lethal restrictive dermopathy. J Cell Biol 161:1105–1115 Hirsch D, Stahl A, Lodish HF 1998 A family of fatty acid transporters conserved from mycobacterium to man. Proc Natl Acad Sci USA 95:8625–8629 Iwai N, Mannami T, Tomoike H, Ono K, Iwanaga Y 2003 An acyl-CoA synthetase gene family in chromosome 16p12 may contribute to multiple risk factors. Hypertension 41:1041– 1046 Kampf JP, Cupp D, Kleinfeld AM 2006 Different mechanisms of free fatty acid fl ip-flop and dissociation revealed by temperature and molecular species dependence of transport across lipid vesicles. J Biol Chem 281:21566–21574 Kim JH, Lewin TM, Coleman RA 2001 Expression and characterization of recombinant rat Acyl-CoA synthetases 1, 4, and 5. Selective inhibition by triacsin C and thiazolidinediones. J Biol Chem 276:24667–24673 Kim JK, Gimeno RE, Higashimori T et al 2004 Inactivation of fatty acid transport protein 1 prevents fat-induced insulin resistance in skeletal muscle. J Clin Invest 113:756–763 Lee Y, Naseem RH, Duplomb L et al 2004 Hyperleptinemia prevents lipotoxic cardiomyopathy in acyl CoA synthase transgenic mice. Proc Natl Acad Sci USA 101:13624–13629 Lewin TM, Kim JH, Granger DA, Vance JE, Coleman RA 2001 Acyl-CoA synthetase isoforms 1, 4, and 5 are present in different subcellular membranes in rat liver and can be inhibited independently. J Biol Chem 276:24674–24679 Lewis SE, Listenberger LL, Ory DS, Schaffer JE 2001 Membrane topology of the murine fatty acid transport protein 1. J Biol Chem 276:37042–37050 Li H, Black PN, DiRusso CC 2005 A live-cell high-throughput screening assay for identification of fatty acid uptake inhibitors. Anal Biochem 336:11–19 Mashek DG, McKenzie MA, Van Horn CG, Coleman RA 2006 Rat long chain acyl-CoA synthetase 5 increases fatty acid uptake and partitioning to cellular triacylglycerol in McArdleRH7777 cells. J Biol Chem 281:945–950 Overath P, Pauli G, Schairer HU 1969 Fatty acid degradation in Escherichia coli. An inducible acyl-CoA synthetase, the mapping of old-mutations, and the isolation of regulatory mutants. Eur J Biochem 7:559–574 Piccini M, Vitelli F, Bruttini M et al 1998 FACL4, a new gene encoding long-chain acyl-CoA synthetase 4, is deleted in a family with Alport syndrome, elliptocytosis, and mental retardation. Genomics 47:350–358 Richards MR, Listenberger LL, Kelly AA et al 2003 Oligomerization of the murine fatty acid transport protein 1. J Biol Chem 278:10477–10483 Richards MR, Harp JD, Ory DS, Schaffer JE 2006 Fatty acid transport protein 1 and long-chain acyl coenzyme A synthetase 1 interact in adipocytes. J Lipid Res 47:665–672 Schaffer JE, Lodish HF 1994 Expression cloning and characterization of a novel adipocyte long chain fatty acid transport protein. Cell 79:427–436 Setnik B, Nobrega JN 2004 Long-chain acyl-CoenzymeA synthetase-2 mRNA: increased cerebral cortex expression in an animal model of depression. Prog Neuropsychopharmacol Biol Psychiatry 28:577–582 Stahl A, Hirsch DJ, Gimeno RE et al 1999 Identification of the major intestinal fatty acid transport protein. Mol Cell 4:299–308 Tong F, Black PN, Coleman RA, DiRusso CC 2006 Fatty acid transport by vectorial acylation in mammals: roles played by different isoforms of rat long-chain acyl-CoA synthetases. Arch Biochem Biophys 447:46–52
138
BLACK & DIRUSSO
Zou Z, DiRusso CC, Ctrnacta V, Black PN 2002 Fatty acid transport in Saccharomyces cerevisiae. Directed mutagenesis of FAT1 distinguishes the biochemical activities associated with Fat1p. J Biol Chem 277:31062–31071 Zou Z, Tong F, Faergeman NJ et al 2003 Vectorial acylation in Saccharomyces cerevisiae. Fat1p and fatty acyl-CoA synthetase are interacting components of a fatty acid import complex. J Biol Chem 278:16414–16422
DISCUSSION Bernlohr: At the protein level do you know the expression of the two splice variants? Black: Not in whole cells. When Paul Watkins originally looked at human FATP2 as a VLCS, the localization studies showed two bands (Steinberg et al 1999). The smaller form may be around 20% of the larger form. What is interesting is that our clone came from human placenta, and his came from liver. I think that in the placenta FATP2b may be a major protein involved in fatty acid influx from the maternal circulation to the fetal circulation, and it might have a specific activity towards polyunsaturated or highly saturated fatty acids. O’Rahilly: What is your therapeutic target? If you found a molecule that did block it, what would you do with it? Black: The target we are blocking is the different FATP isoforms. The idea would be to block fatty acid import, but this would require a high degree of specificity especially as in some cells fatty acid transport is essential to meet metabolic needs. Some pharmaceutical companies might be interested in FATP as a particular target if one of these small molecules could depress the ability to import fatty acids. This might be a nice way of dampening the system so the absorption of dietary fatty acids is reduced. One thing we are cautious of is that we do not want to select inhibitors that will impact fatty acid import in tissues that rely on fatty acids as metabolic fuel. Moreover, we do not want to depress transport of the good (essential) fatty acids; we have to be careful about discriminating the good fats and bad fats. Muoio: In your model you show that the fatty acids are immediately converted to acyl CoAs upon entering the cells. In diabetes, acyl CoAs often correlate with insulin resistance. Where do most of the acyl CoAs reside in cells? Has there been any progress in trying to localize different pools? Black: That is the big question. The concentration of acyl CoA in the cell will be very low, and once the fatty acid is activated it will move into downstream metabolism. This could be protein modification or triglyceride and complex lipid synthesis. The question is what are the mechanisms that discriminate different acyl CoAs into these pathways? At present no one knows the answer to this question. The tools to address those questions are just now coming online. For example, we are now challenging these different cell types with 13C-labelled fatty acids and
VECTORIAL ACYLATION OF FATTY ACIDS
139
are looking at complex lipids using GC-MS and LC-MS/MS to examine their metabolic fate. For example, if the cell type predominantly expresses FATP2 you may see a pattern ‘a’, and if it predominantly expresses FATP1, you see pattern ‘b’. We don’t know what these patterns are yet, so our initial investigations are directed to establishing this baseline information to drive further experiments. Yang: Do you see the acyl CoA synthetases on the lipid droplets, or are they primarily on the plasma membrane? Black: I think they are on the lipid droplets too. What is interesting is that the acyl CoA synthetases FATPs are not just in one location within the cell. FATP1, for example, is responsive to insulin and moves to the plasma membrane from intracellular sites following stimulation. In studies looking at GFP-labelled yeast FATP or acyl CoA synthetase, or labelled FATPs in mammalian cells, they do not always localize at the plasma membrane. Thus it is apparent that they are functioning in fatty acid metabolism outside the movement of exogenous fatty acids. I suspect these proteins have to be involved with modulating fatty acid movement within the cell and modulating some type of fatty acid homeostasis at that level. This is a complicated set of tasks that we attribute to these proteins and we are just starting to tease this apart in terms of mechanism. Shi: Have you ever observed whether FATP would transport the fatty acid out of the adipocytes? Black: No. Bernlohr: We have knockdowns of some of the fatty acid transport proteins. They result in increased lipolysis from fat cells. It is not clear whether this is transport or simply availability. Spiegelman: Because of the linkage to the CoA and the ATP-requiring step, it is a downstream trip energetically. It would be hard to drive this in the reverse direction. Bernlohr: It would. But if you buy the argument that diffusion is what is facilitating trans bilayer movement, or that there are domains on the protein that facilitate bilayer movement, it could be energy neutral to go across any membrane. Spiegelman: It would have to do that independently of the acyltransferase activity, because otherwise you’d have to drive energetically upstream. Black: Once acyl CoA has been formed, it is pretty much destined to go into the triglyceride or complex lipid pool in the adipocyte. Once the fatty acid is released following lipolysis of the triglyceride store, there is a significant increase in free fatty acids. I suspect that adipocyte FABP is buffering the detergent effects of the fatty acids at the membrane, and in essence there is a mass action effect involved in fatty acid efflux. Bernlohr: Particularly if there is albumin on the outside of the cell. Black: I do not think the FATPs and acyl CoA synthetases play any role in fatty acid efflux. This was one of the criticisms early on in discussions with biophysicists.
140
BLACK & DIRUSSO
If these proteins function in fatty acid transport, it must be one way. On the basis of the topology of the yeast FATP that we have done, the business end is sitting on the intracellular side. Spiegelman: Is there room for some unknown proteins involved in actually binding the fatty acids on the extracellular space? Or is it clear that this is both the binder as well as the acyl transferase activity? Black: There are no data to say these proteins are fatty acid binding proteins. If you purify them and monitor fatty acid binding activity you can define these parameters, but I do not think there is enough of a domain on the extracellular side to constitute a bona fide receptor with fatty acid binding activity. In the case of CD36, which has also been suggested to be a fatty acid transporter, it also functions as an oxidized LDL receptor, which may result in high local concentrations of fatty acids. If the FATPs and acyl CoA synthetases are also localized at the plasma membrane and the fatty acids are partitioning into and fl ipping across the membrane, it is the abstraction out that is the governing force moving fatty acids into downstream metabolic pools. This requires the fatty acid to be metabolically activated to acyl CoA via the process we define as vectorial acylation. Spiegelman: These are sort of equivalent to hexokinase in glucose metabolism. Black: Indeed, if you look at the earlier studies from bacteria, Peter Overath identified the acyl CoA synthetase and proposed the idea of vectorial acylation. He compared this mechanism with the pyruvate phosphotransferase system that is also present in bacteria. I think this is the mechanism here. Is it the only mechanism? I don’t think so. In certain cell types the FATPs may function to abstract the fatty acid from the membrane and deliver it to downstream acyl CoA synthetases. In others the FATP may provide the acyl CoA synthetase activity driving this process forward. Spiegelman: I was going to ask whether the FATPs are misnamed, but I guess not: they do catalyse the phenomenon of fatty acid transport. Black: The misnomer in the FATPs comes from the original work describing FATP1 in the murine adipocyte. More recently, Andreas Stahl has published work were the FATPs are designated solute carriers (Slc) 27a 1, 2, 3, 4, 5, and 6. These are identified in the human genome as solute carriers, and this is clearly incorrect. It is tough for those of us in the field who look at these more as enzymes, in addition to their roles as fatty acid transporters. Indeed, Stahl and co-workers have published a review which shows these proteins as pores (Doege & Stahl 2006). This is wrong. Spiegelman: If we could turn the clock back, would you call them FATPs? Black: No. Spiegelman: It is like calling hexokinase a glucose transporter. Bernlohr: Hexokinase doesn’t facilitate any bilayer movement.
VECTORIAL ACYLATION OF FATTY ACIDS
141
Black: It is also true that the biophysical properties of fatty acids and glucose are so distinct from one another, that from a mechanistic standpoint getting the fatty acid across the membrane is going to be different from getting a glucose molecule across. What would we name them? Acyl CoA synthetases. Shi: Your high-throughput screening is very impressive. You mentioned that you would target the fat absorption process. The major player for absorption is FATP2. The knockout is embryonic lethal. Black: FATP2 is not an embryonic lethal; that is FATP4. FATP2 has been knocked out in studies directed at investigating the role of this protein in X-linked adrenodystrophy. They see a diminution of long-chain acyl CoA synthetase activity. In these studies, fatty acid transport into the cell was not monitored, so we are now working with Paul Watkins to look at these two isoforms. Muoio: Aaron Bonen’s lab in Canada has described CD36 translocation in muscle (Campbell et al 2004 Luiken et al 2003). Recently he has published several papers showing stimulation of CD36 translocation to mitochondria. Can you comment? Black: I am aware of this, but I don’t know what to make of it. Why would CD36, which is heavily glycosylated, go to mitochondria? His thinking is that it is involved in the movement of acyl CoAs and fatty acids to the mitochondria for oxidation. There is another paper suggesting CD36 is on the basolateral surface of the intestinal endothelial cell and is involved in chylomicron synthesis (Drover et al 2005). The more recent work on CD36, with the exception of the transgenic and knockout, is pretty confusing. References Campbell SE, Tandon NN, Woldegiorgis G, Luiken JJ, Glatz JF, Bonen A 2004 A novel function for fatty acid translocase (FAT)/CD36: involvement in long chain fatty acid transfer into the mitochondria. J Biol Chem 279:36235–36241 Doege H, Stahl A 2006 Protein-mediated fatty acid uptake: novel insights from in vivo models. Physiology (Bethesda) 21:259–268 Drover VA, Ajmal M, Nassir F et al 2005 CD36 deficiency impairs intestinal lipid secretion and clearance of chylomicrons from the blood. J Clin Invest 115:1290–1297 Luiken JJ, Coort SL, Willems J et al 2003 Contraction-induced fatty acid translocase/CD36 translocation in rat cardiac myocytes is mediated through AMP-activated protein kinase signaling. Diabetes 52:1627–1634 Steinberg SJ, Wang SJ, Kim DG, Mihalik SJ, Watkins PA 1999 Human very-long-chain acylCoA synthetase: cloning, topography, and relevance to branched-chain fatty acid metabolism. Biochem Biophys Res Commun 257:615–621
Fatty Acids and Lipotoxicity in Obesity and Diabetes: Novartis Foundation Symposium 286, Volume 286. Edited by Gregory Bock and Jamie Goode Copyright Novartis Foundation 2007. ISBN: 978-0-470-05764-3
Lipid storage and mobilization pathways in yeast Günther Daum, Andrea Wagner, Tibor Czabany, Karlheinz Grillitsch and Karin Athenstaedt Institute of Biochemistry, Graz University of Technolog y, Petersgasse 12/2, A-8010 Graz, Austria
Abstract. Biochemistry, cell biology and molecular biology of lipids can be properly studied using the yeast Saccharomyces cerevisiae as a model system. We employ this microorganism to investigate pathways of neutral lipid (triacylglycerol, steryl ester) synthesis, storage and mobilization and to identify major gene products involved in these processes. The steryl ester synthases Are1p and Are2p were shown to catalyze steryl ester formation, and Dga1p and Lro1p were identified as major enzymes of triacylglycerol synthesis. Both triacylglycerols and steryl esters are stored in lipid particles, an intracellular compartment that is structurally reminiscent of lipoproteins. Neutral lipid mobilization is initiated by the triacylglycerol lipases Tgl3p, Tgl4p and Tgl5p, and the steryl ester hydrolases Tgl1p, Yeh1p and Yeh2p. The acyltransferases Are1p, Are2p, Lro1p and Dga1p are located in the endoplasmic reticulum, but a substantial amount of Dga1p is also present in lipid particles. The three triacylglycerol lipases as well as Tgl1p and Yeh1p are components of lipid particles, whereas Yeh2p was detected in the plasma membrane. Thus, enzymatic steps of triacylglycerol and steryl ester metabolism are located in different subcellular compartments. Consequently, regulation of neutral lipid metabolism does not only occur at the enzymatic level but also at the organelle level. 2007 Fatty acids and lipotoxicity in obesity and diabetes. Wiley, Chichester (Novartis Foundation Symposium 286) p 142–154
In most types of cells, neutral lipids serve as an energy reserve and a source of building blocks needed for membrane formation. The yeast Saccharomyces cerevisiae synthesizes triacylglycerols (TAG) and steryl esters (STE) as the most prominent storage lipids. These components are stored in so-called lipid particles (LP), a globular intracellular structure that is sequestered from the cytosolic environment by a phospholipid monolayer containing a small set of proteins. Work in our laboratory performed over the last few years has been focused on the elucidation of yeast metabolic pathways leading to the formation of neutral lipids and of hydrolytic reactions catalysing their mobilization. Yeast gene products involved in these pathways were identified at the molecular level and biochemically characterized. A new picture of the LP emerged from these studies, namely that of a metabolically active organelle in addition to its function as a storage compartment. 142
LIPID STORAGE IN YEAST
143
Biosynthesis of triacylglycerols Prerequisite for the synthesis of TAG in the yeast Saccharomyces cerevisiae is formation of the precursor diacylglycerol (DAG). De novo synthesis of DAG starts with acylation of either glycerol-3-phosphate (G-3-P) or dihydroxyacetone phosphate (DHAP) (Athenstaedt & Daum 1999, Sorger & Daum 2003). These reactions are catalysed by the G-3-P acyltransferases (GAT) Gat1p/Gpt2p and Gat2p/Sct1p, which also act as DHAP acyltransferases. These acylation reactions yield 1-acylG-3-P (lyso-phosphatidic acid) or 1-acyl-DHAP, respectively. 1-Acyl-DHAP is reduced to 1-acyl-G-3-P in an NADPH dependent reaction catalysed by the acylDHAP reductase Ayr1p. 1-Acyl-G-3-P is further acylated in the sn-2 position by the 1-acyl-G-3-P acyltransferase Slc1p yielding phosphatidic acid (PA). Alternatively, PA can be formed from glycerophospholipids by the action of a phospholipase D, e.g. Pld1p from S. cerevisiae. Phosphorylation of DAG by DAG kinase could also lead to the formation of PA, but such a reaction has never been demonstrated in yeast. The reverse reaction, i.e. formation of DAG from PA requires dephosphorylation catalysed by phosphatidate phosphatase(s) (PAP). Recently Pah1p was identified as the major PAP of S. cerevisiae (Han et al 2006). Mutants lacking PAH1 accumulated PA and had reduced amounts of DAG and TAG. Dpp1p and Lpp1p, which were originally discovered as DAG pyrophosphate phosphatases, also act as PAPs although with low activity (Oshiro et al 2003, Toke et al 1998). Finally, formation of DAG from phospholipids through the action of a phospholipase C, e.g. Plc1p, has also to be taken into account. The last step in the pathway of TAG synthesis is acylation of DAG. This reaction is catalysed by two different types of enzymes (Fig. 1). First, TAG is formed by the action of Lro1p (LCAT-related open reading frame), a homologue of the human LCAT (lecithin : cholesterol acyltransferase). Lro1p converts DAG to TAG CoA, FFA, ATP FAA1-4 Sterols
P-Lipids DAG
Acyl-CoA
LRO1 DAG
Acyl-CoA
DGA1 (ARE1, ARE2)
ARE1, ARE2 STE TAG
FIG. 1. Pathways of neutral lipid synthesis in the yeast Saccharomyces cerevisiae. DAG, diacylglycerol; FFA, free fatty acid; STE, steryl ester; TAG, triacylglycerol.
144
DAUM ET AL
in an acyl-CoA independent esterification reaction and uses glycerophospholipids, preferentially phosphatidylethanolamine and phosphatidylcholine, as acyl donors (Dahlqvist et al 2000, Oelkers et al 2000). Secondly, the DAG acyltransferase (DGAT) Dga1p catalyses acyl-CoA dependent acylation of DAG (Sorger & Daum 2002, Oelkers et al 2002, Sandager et al 2002). This enzyme was identified by homology to two DGATs from the fungus Mortierella ramanniana (Cases et al 2001, Lardizabal et al 2001). However, deletion of LRO1 and DGA1 yielded cells that contained 5% of the DAG esterification activity compared to wild type (Sorger & Daum 2002, Oelkers et al 2002). It was shown that Are1p and Are2p, which were originally identified as STE synthases (see below), accounted for the residual enzyme acyl-CoA dependent TAG synthase activity in a dga1∆ lro1∆ strain. This view was confirmed by experiments using a dga1∆ lro1∆ are1∆ are2∆ mutant that completely lacked STE and TAG synthesis (Sorger & Daum 2002, Oelkers et al 2002, Sandager et al 2002, Sorger et al 2004). Lro1p and Dga1p not only catalyze different types of reaction but also exhibit distinct expression profi les. Lro1p appears to play a major role in TAG synthesis during logarithmic growth, whereas Dga1p is more active in the stationary phase (DeRisi et al 1997). Moreover, the subcellular distribution of the two enzymes is different. Dga1p is dually located to the ER and LP, whereas the occurrence of Lro1p is restricted to the ER (Sorger & Daum 2002, 2003). Since all TAG synthesizing enzymes including Are1p and Are2p are present in the ER, this subcellular fraction has to be regarded as a major site of TAG formation. Consequently, TAG formed in the ER has to reach the neutral lipid storage compartment, the LP. How this process is accomplished is still a matter of dispute. Biosynthesis of steryl esters The two STE synthases detected in the yeast, Are1p and Are2p (ACAT related enzymes), form STE from sterols and activated fatty acids (see Fig. 1). The two polypeptides are 50% identical to each other and exhibit approximately 24% identity to human ACAT (acyl-CoA : cholesterol acyltransferase) (Yu et al 1996, Yang et al 1996). Deletion of both ARE1 and ARE2 completely abolished sterol esterification. Therefore, it was assumed that these two gene products are the only STE biosynthetic enzymes of the yeast. In contrast to a reduction of the STE level in an are2∆ single mutant to less than 26% of wild-type, hardly any defect was observed in an are1∆ strain. These findings indicated that Are2p was the major STE synthase of the yeast. Besides their different quantitative contribution to the sterol esterification process, Are1p and Are2p exhibit different substrate specificity (Zweytick et al 2000a). Ergosterol is the preferred substrate of Are2p, whereas Are1p esterifies ergosterol and its precursors with nearly equal efficiency exhibiting a slight preference for lanosterol. It was suggested that the role of Are1p may be
LIPID STORAGE IN YEAST
145
limitation of the amount of lanosterol, thereby interrupting the sterol biosynthetic pathway and at the same time favouring storage of ergosterol precursors in the form of STE in LP. Hydrolysis of STE containing sterol precursors and subsequent conversion of these sterol intermediates to ergosterol appears to be faster than de novo sterol synthesis (Jensen-Pergakes et al 2001). Are1p and Are2p are localized to the ER as demonstrated by enzymatic analysis (Zinser et al 1993) and microscopic inspection using GFP fusion proteins (Zweytick et al 2000a). Thus, similar to TAG synthesis (see above), the major site of STE formation is the ER and consequently not identical to the site of STE storage, the LP. This subcellular scenario again requires efficient pathways of STE assembly into LP. As a matter of fact, storage of STE and assembly of TAG into LP are closely related to the biogenesis of LP. Lipid particles, not only a depot for neutral lipids TAG and STE are unable to integrate in large amounts into phospholipid membrane bilayers. Consequently, they cluster and form the hydrophobic core of LP, whose structure resembles that of lipoproteins from mammals (Kostner & Laggner 1989). In yeast, the hydrophobic core of LP is mainly formed from TAG and STE and sequestered from the cytosolic environment by a phospholipid monolayer with a small amount of characteristic proteins embedded (Zweytick et al 2000b). According to a current model of LP biogenesis, proteins involved in neutral lipid metabolism including enzymes of TAG and STE synthesis accumulate in certain regions of the ER. Once neutral lipids are newly formed in these domains, they may cluster in the hydrophobic region between the two leaflets of the ER membrane. During ongoing synthesis of TAG and STE the neutral lipid droplet grows, and after reaching a certain size buds off the ER. Polypeptides in the environment of this nascent LP may be embedded into the newly formed membrane monolayer of the droplet provided that these polypeptides lack transmembrane spanning domains. A number of LP proteins identified so far (Athenstaedt et al 1999) fulfi l these requirements. The dual localization of many LP proteins in this organelle and in the ER, and the fact that in a dga1∆ lro1∆ are1∆ are2∆ strain which lacks LP, all typical LP proteins of wild-type cells are restricted to the ER (Sorger et al 2004) support the model of LP biogenesis described above. Besides their function as storage compartments, LP may also play a role in stabilization of proteins as shown recently for Erg1p by Sorger et al (2004). Erg1p is the yeast squalene epoxidase and thus a key enzyme of sterol biosynthesis. This polypeptide, which is dually located to LP and ER in wild type (Leber et al 1998), was not only restricted to the ER of a dga1∆ lro1∆ are1∆ are2∆ quadruple mutant, but also occurred at low abundance in this strain (Sorger et al 2004). Transcription of ERG1 was unaffected in the quadruple mutant, but the gene product Erg1p
146
DAUM ET AL
became unstable in this strain. Since Erg1p is enzymatically inactive in LP and only active in the ER (Leber et al 1998), translocation of the protein from LP to the ER and activation in the latter compartment may occur upon requirement. Thus, it was concluded that LP stabilize this protein and serve as a ‘parking lot’ for Erg1p. Consequently, regulation of the sterol biosynthetic pathway may not only occur at the enzyme level, but also through the interaction of organelles. Hydrolysis of triacylglycerols When fatty acids and/or sterols are neither synthesized by the cell nor provided through the medium, these components are set free from TAG and STE stored in LP. The first step of this mobilization process requires hydrolytic enzymes. Recently, a family of yeast TAG lipases named Tgl3p, Tgl4p and Tgl5p (Fig. 2) localized to LP was identified (Athenstaedt & Daum 2003, 2005, Kurat et al 2005). Tgl3p, Tgl4p and Tgl5p contain the consensus sequence GXSXG, which is characteristic for lipolytic enzymes such as related lipases from Candida parapsilosis (Neugnot et al 2002), mammalian cells (Zimmermann et al 2004), Drosophila melanogaster (Grönke et al 2005) and Arabidopsis thaliana (Eastmond 2006). Hydrolysis of TAG by the yeast lipases provides fatty acids and DAG for the biosynthesis of complex membrane lipids. As shown by experiments with isolated LP and isolated tagged polypeptides from overexpression strains, Tgl3p is the yeast TAG lipase with the highest activity (Athenstaedt & Daum 2003, 2005). This view was supported by experiments in vivo using cerulenin as an inhibitor of endogenous fatty acid synthesis which creates a situation of cellular fatty acid depletion. Under these conditions, TAG from wild-type cells was readily mobilized within a few hours, whereas TAG from a tgl3∆ strain was hydrolysed only at a slow rate (Fig. 3). This result, however, did not only confi rm the role of Tgl3p as a major yeast TAG lipase, but also indicated that additional TAG lipases, although less active, were present. Indeed, two such enzymes named Tgl4p and Tgl5p were identified (Athenstaedt & Daum 2005, Kurat et al 2005) and characterized. All three lipases exhibited
TAG TGL3 TGL4 TGL5 DAG
STE Fatty acids
YEH1 YEH2
Fatty acids
TGL1 Sterols
FIG. 2. Degradation of neutral lipids in the yeast Saccharomyces cerevisiae. DAG, diacylglycerol; STE, steryl ester; TAG, triacylglycerol.
LIPID STORAGE IN YEAST Wild-type
15
150
Wild-type + cerulenin tgl3D tgl3D + cerulenin
10
TAG/OD (%)
OD (600 nm)
147
5
100 50 0
0 0
1
3 5 time [h]
7
8,5
0
1
3
5
7
8,5
time [h]
FIG. 3. Tgl3p is the major TAG lipase in vivo of the yeast Saccharomyces cerevisiae. Wild-type and a tgl3∆ mutant strain were grown in the absence or presence of cerulenin, an inhibitor of fatty acid synthesis. The optical density (OD 600 nm) and the relative amount of triacylglycerol (TAG) per OD were measured (data from Athenstaedt & Daum 2003).
TABLE 1 Lack of TAG lipases affects the amount of neutral lipids from the yeast Saccharomyces cerevisiae Strain
TAG (mg/mg CDW)
STE (mg/mg CDW)
TAG/STE
BY4741 (wild-type) tgl3∆ tgl4∆ tgl5∆ tgl3∆tgl4∆ tgl3∆tgl5∆ tgl4∆tgl5∆
1.72 4.11 2.97 1.68 4.28 5.38 4.50
1.93 2.28 1.44 2.11 2.39 2.73 2.27
0.89 1.80 2.06 0.80 1.80 1.97 1.98
tgl3∆tgl4∆tgl5∆
4.58
2.70
1.70
CDW, cell dry weight; STE, steryl ester; TAG, triacylglycerol
TAG lipase activity in vitro when purified close to homogeneity. The lipolytic activity of Tgl5p in vivo was marginal but increased in combination with Tgl4p. The activity of these enzymes demonstrated in vivo is in line with the amount of TAG detected in the different deletion strains (Table 1). Although Tgl3p, Tgl4p and Tgl5p are enzymes with overlapping function they differ slightly in their substrate specificities (Athenstaedt & Daum 2005). Fatty acid analysis of TAG from mutants deleted of TGL3, TGL4 and TGL5 indicated that Tgl3p had low substrate specificity regarding the chain length of the fatty acid, whereas TAG from tgl4∆ was enriched in myristic acid (C14 : 0) and palmitic acid (C18 : 0), and TAG from tgl5∆ in hexacosanoic acid (C26 : 0). All three lipases, however, were highly specific for TAG as a substrate compared to STE. How the activity of the three TAG lipases which are present in the same compartment, namely the LP, is regulated has not yet been clarified.
148
DAUM ET AL
Mobilization of steryl esters Similar to TAG, STE can be mobilized from yeast LP upon requirement. Esterification of ergosterol and sterol intermediates on one hand, and hydrolysis of STE on the other hand play an important role in cellular sterol homeostasis. The yeast Saccharomyces cerevisiae harbours three STE hydrolases named Yeh1p, Yeh2p and Tgl1p (see Fig. 2) (Müllner et al 2005, Köffel et al 2005, Jandrositz et al 2005). In aerobically grown cells, the highest STE hydrolase activity was detected in the plasma membrane (Zinser et al 1993). This activity was attributed to Yeh2p, which was identified as the first yeast STE hydrolase at the molecular level (Müllner et al 2005, Köffel et al 2005). Yeh2p is homologous to several known mammalian STE hydrolases. Deletion of the gene led to complete loss of plasma membrane STE hydrolase activity whereas overexpression resulted in a significant elevation of activity. Overexpression of YEH2 resulted in a disturbance of the structure of the plasma membrane indicating that the role of Yeh2p may be related to the assembly of sterols into this compartment. Since bulk STE mobilization occurred in a yeh2∆ deletion strain at a similar rate as in wild type, it was assumed that Yeh2p is not the only STE hydrolase, but that additional enzymes with overlapping function exist in yeast. Homology searches led to the identification of two other members of the STE hydrolase family in Saccharomyces cerevisiae, Yeh1p and Tgl1p (Köffel et al 2005, Jandrositz et al 2005). Both enzymes, which are paralogues of the mammalian acid lipase family, are components of LP. They exhibit only marginal activity in vitro, but their deletion in the yeh2∆ background led to a complete block of STE hydrolysis. TAG was not metabolized by these enzymes indicating that these hydrolases are specific for STE degradation in vivo. Most recently, Köffel and Schneiter (2006) reported about the predominant role of Yeh1p as a STE hydrolase in haem-deficient yeast cells which mimic anaerobic conditions. It was shown that activation of Yeh1p occurred through the action of Rox3p, a regulator of RNA polymerase II activity, although this effect has not yet been studied at the molecular level. Similar to TAG lipases, Yeh1p and Tgl1p are LP proteins. Consequently, they are located in close vicinity to their substrate. STE hydrolysis catalysed by Yeh1p and Tgl1p must be a regulated process to avoid a futile cycle of STE synthesis and degradation. Yeh2p which is present in the plasma membrane also appears to utilize internal STE from LP as substrate (Müllner et al 2005). The question remains how this enzyme gets access to its substrate.
Summary and conclusions Figure 4 summarizes in a schematic way our recently acquired knowledge about enzymes involved in yeast neutral lipid metabolism, their subcellular localization
LIPID STORAGE IN YEAST
149
TAG & STE FORMATION
STE MOBILIZATION Sterol + FFA Yeh1p Tgl1p
Are2p Are1p Lro1p
Yeh2p
Dga1p Tgl3p
Dga1p
TAG + STE
Tgl4p Erg1p
ER
Erg1p
Erg1p PARKING LOT
Tgl5p
LP
FFA
TAG MOBILIZATION
PM
FIG. 4. Neutral lipid metabolism in the yeast Saccharomyces cerevisiae. FFA, free fatty acid; ER, endoplasmic reticulum; LP, lipid particle; PM, plasma membrane; TAG, triacylglycerol; STE, steryl ester.
and function. This summary gives an overview of the progress that had been made in this field during the last decade. It appears that the major players in the yeast neutral lipid game have been identified, but several important questions remain open. As an example, the physiological relevance of enzymes with overlapping function is far from being understood. The interplay of pathways and the coordination of routes of neutral lipid synthesis, storage and degradation in different organelles need to be elucidated. Mechanisms of neutral lipid migration from their site of synthesis to their site of storage, or mechanisms involved in the access of hydrolytic enzymes to components of the storage compartment have yet to be studied. Mechanisms directing degradation products of neutral lipids to their site of utilization are not known. Finally, the role of neutral lipids as regulatory elements, as elements of detoxification, or as protectors against apoptosis (Low et al 2005) will have to be clarified to obtain a broader view of the biological function of this class of lipids. Acknowledgements Work in our laboratory on yeast neutral lipids has been financially supported by FWF projects P15141, P18857 and DK001 to GD, and Herta Firnberg Fellowship T113 to KA.
150
DAUM ET AL
References Athenstaedt K, Daum G 1999 Phosphatidic acid, a key intermediate in lipid metabolism. Eur J Biochem 266:1–16 Athenstaedt K, Daum G 2003 YMR313c/TGL3 encodes a novel triacylglycerol lipase located in lipid particles of Saccharomyces cerevisiae. J Biol Chem 278:23317–23323 Athenstaedt K, Daum G 2005 Tgl4p and Tgl5p, two triacylglycerol lipases of the yeast Saccharomyces cerevisiae are localized to lipid particles. J Biol Chem 280:37301–37309 Athenstaedt K, Zweytick D, Jandrositz A, Kohlwein SD, Daum G 1999 Identification and characterization of major lipid particle proteins of the yeast Saccharomyces cerevisiae. J Bacteriol 181:6441–6448 Cases S, Stone SJ, Zhou P et al 2001 Cloning of DGAT2, a second mammalian diacylglycerol acyltransferase, and related family members. J Biol Chem 276:38870–38876 Dahlqvist A, Stahl U, Lenman M et al 2000 Phospholipid : diacylglycerol acyltransferase: an enzyme that catalyzes the acyl-CoA-independent formation of triacylglycerol in yeast and plants. Proc Natl Acad Sci USA 97:6487–6492 DeRisi JL, Iyer VR, Brown PO 1997 Exploring the metabolic and genetic control of gene expression on a genomic scale. Science 278:680–686 Eastmond PJ 2006 SUGAR-DEPENDENT1 encodes a patatin domain triacylglycerol lipase that initiates storage oil breakdown in germinating Arabidopsis seeds. Plant Cell 18:665– 675 Grönke S, Mildner A, Fellert S et al 2005 Brummer lipase is an evolutionary conserved fat storage regulator in Drosophila. Cell Metab 1:323–330 Han GS, Wu WI, Carman GM 2006 The Saccharomyces cerevisiae Lipin homolog is a Mg2+ dependent phosphatidate phosphatase enzyme. J Biol Chem 281:9210–9218 Jandrositz A, Petschnigg J, Zimmermann R et al 2005 The lipid droplet enzyme Tgl1p hydrolyzes both steryl esters and triglycerides in the yeast, Saccharomyces cerevisiae. Biochim Biophys Acta 1735:50–58 Jensen-Pergakes K, Guo Z, Giattina M, Sturley SL, Bard M 2001 Transcriptional regulation of the two sterol esterification genes in the yeast Saccharomyces cerevisiae. J Bacteriol 183:4950– 4957 Köffel R, Schneiter R 2006 Yeh1 constitutes the major steryl ester hydrolase under heme deficient conditions in Saccharomyces cerevisiae. Eukaryot Cell 5:1018–1025 Köffel R, Tiwari R, Falquet L, Schneiter R 2005 The Saccharomyces cerevisiae YLL012/YEH1, YLR020/YEH2, and TGL1 genes encode a novel family of membrane-anchored lipases that are required for steryl ester hydrolysis. Mol Cell Biol 25:1655–1668 Kostner GM, Laggner P 1989 Chemical and physical properties of lipoproteins. In: Fruchart JC, Sheperd J (eds) Human plasma lipoproteins (Clinical Biochemistry). Walter de Gruyter, Berlin-New York, p 23–54 Kurat CF, Natter K, Petschnigg J et al 2005 Obese yeast: triglyceride lipolysis is functionally conserved from mammals to yeast. J Biol Chem 281:491–500 Lardizabal KD, Mai JT, Wagner NW, Wyrick A, Voelker T, Hawkins DJ 2001 DGAT2 is a new diacylglycerol acyltransferase gene family. Purification, cloning, and expression in insect cells of two polypeptides from Mortierella ramanniana with diacylglycerol acyltransferase activity. J Biol Chem 276:38862–38869 Leber R, Landl K, Zinser E et al 1998 Dual localization of squalene epoxidase, Erg1p, in yeast reflects a relationship between the endoplasmic reticulum and lipid particles. Mol Biol Cell 9:375–386 Low CP, Liew LP, Pervaiz S, Yang H 2005 Apoptosis and lipoapoptosis in the fission yeast Schizosaccharomyces pombe. FEMS Yeast Res 5:1199–1206 Müllner H, Deutsch G, Leitner E, Ingolic E, Daum G 2005 YEH2/YLR020c encodes a novel steryl ester hydrolase of the yeast Saccharomyces cerevisiae. J Biol Chem 280:13321–13328
LIPID STORAGE IN YEAST
151
Neugnot V, Moulin G, Dubreucq E, Bigey F 2002 The lipase/acyltransferase from Candida parapsilosis: Molecular cloning and characterization of purified recombinant enzymes. Eur J Biochem 269:1734–1745 Oelkers P, Tinkelenberg A, Erdeniz N, Cromley D, Billheimer JT, Sturley SL 2000 A lecithin cholesterol acyltransferase-like gene mediates diacylglycerol esterification in yeast. J Biol Chem 275:15609–15612 Oelkers P, Cromley D, Padamsee M, Billheimer JT, Sturley SL 2002 The DGA1 gene determines a second triglyceride synthetic pathway in yeast. J Biol Chem 277:8877– 8881 Oshiro J, Han GS, Carman GM 2003 Diacylglycerol pyrophosphate phosphatase in Saccharomyces cerevisiae. Biochim Biophys Acta 1635:1–9 Sandager L, Gustavsson MH, Stahl U et al 2002 Storage lipid synthesis is non-essential in yeast. J Biol Chem 277:6478–6482 Sorger D, Daum G 2002 Synthesis of triacylglycerols by the acyl-Coenzyme A : diacylglycerol acyltransferase Dga1p in lipid particles of the yeast Saccharomyces cerevisiae. J Bacteriol 184:519–524 Sorger D, Daum G 2003 Triacylglycerol biosynthesis in yeast. Appl Microbiol Biotechnol 61:289–299 Sorger D, Athenstaedt K, Hrastnik C, Daum G 2004 A yeast strain lacking lipid particles bears a defect in ergosterol formation. J Biol Chem 279:31190–31196 Toke DA, Bennett WL, Oshiro J, Wu WI, Voelker DR, Carman GM 1998 Isolation and characterization of the Saccharomyces cerevisiae LPP1 gene encoding a Mg2+ -independent phosphatidate phosphatase. J Biol Chem 273:14331–14338 Yang H, Bard M, Bruner DA et al 1996 Sterol esterification in yeast: a two-gene process. Science 272:1353–1356 Yu C, Kennedy NJ, Chang CCY, Rothblatt JA 1996 Molecular cloning and characterization of two isoforms of Saccharomyces cerevisiae acyl-CoA : sterol acyltransferase. J Biol Chem 271:24157–24163 Zimmermann R, Strauss JG, Haemmerle G et al 2004 Fat mobilization in adipose tissue is promoted by adipose triglyceride lipase. Science 306:1383–1386 Zinser E, Paltauf F, Daum G 1993 Sterol composition of yeast organelle membranes and subcellular distribution of enzymes involved in sterol metabolism. J Bacteriol 175: 2853–2858 Zweytick D, Leitner E, Kohlwein SD, Yu C, Rothblatt J, Daum G 2000a Contribution of Are1p and Are2p to steryl ester synthesis in the yeast Saccharomyces cerevisiae. Eur J Biochem 267: 1075–1082 Zweytick D, Athenstaedt K, Daum G 2000b Intracellular lipid particles of eukaryotic cells. Biochim Biophys Acta 1469:101–120
DISCUSSION Tabas: You work in the same city as Rudi Zechner. Is the protein, diacylglyceride hydrolase, that you identified the homologue of what Rudi has found? Daum: Yes. More homologues in Drosophila and Arabidopsis have also been identified in other laboratories. This is a good justification for using yeast as a model system. The Tgl3p from yeast was the first enzyme of this kind. Tabas: Is there a homologue of hormone-sensitive lipase in yeast?
152
DAUM ET AL
Daum: We’ve detected three triacylglycerol hydrolases in the yeast, namely Tgl3p, Tgl4p and Tgl5p. None of these enzymes is a sequence homologue of the human hormone-sensitive lipase, but they are functionally related. Yang: I’d like to mention a recent finding of ours, which we feel is exciting. My lab works on the biogenesis of lipid droplets/lipid bodies. We think lipid droplets are made in the ER and then there is a fusion event giving rise to the mature lipid droplet. A recent paper suggested that the PAT perilipin family of proteins might mediate the budding of the lipid droplets from the ER. They divided the PAT proteins into two families. There are the constitutive PAT proteins which associate with the lipid bodies constitutively. These include adipophilin and perilipin. Then there are also so-called exchangeable PAT proteins. These include Tip47 and S12. It is thought that the exchangeable PAT proteins mediate the budding of lipid droplets, although this is just a hypothesis. We have also used yeast as a model system. We have a collection of all the yeast genes deleted. There are about 5000 such mutants. We used a lipophilic dye to look at these. There are few lipid droplets in some of these mutants, which we call slim mutants. There are others that have more lipid droplets than wild-type, which we call chubby mutants. After doing this for three years we have 30 mutants with significantly reduced lipid droplets, and 150 mutants with too many. Where do we go next? We found that no single mutant leads to the disappearance of the lipid particles, which was disappointing. Most of the ER stress mutants are chubby mutants; more ER stress results in more lipid droplets. We were also looking for mutants that affect the budding step. I initially thought that if a gene can affect budding, maybe we wouldn’t see any lipid droplets at all. But we overlooked one fact: even if these lipid droplets can’t bud from the ER, they still accumulate in the ER. They still show up in our screens. The hypothesis is that if the lipid droplets can’t bud off the ER, if you stress the cells and give them lots of fatty acids, the physical properties of the ER will change. We tested this in a lot of mutants. For example, we found three mutants from which we purified the ER in sucrose gradients. We found that buoyant density of the ER is changed in some of these mutants. We also did a sucrose density gradient and followed distribution of a marker for the ER in these mutant cells. There was a dramatic shift of ER markers to the light fraction in some of the mutants. This doesn’t suggest that these proteins function directly in the budding lipid particles, but it is showing that they are involved in budding in some way. Spiegelman: In these screens, did you find any regulatory mutants? And in particular, did you find any that might control transcriptional programmes? Yang: Yes, we found some transcription factors in the fat mutants. Spiegelman: Are these organized in yeast like they are in mammals, where they are PPARγ and SREBP targets? Yang: I’m not sure.
LIPID STORAGE IN YEAST
153
Spiegelman: Are they organized in a programmatic way, which you could understand? The lipogenics in mammals don’t all have identical regulation, but they will overlap. Is this true in yeast? Black: If you take a look at some of the control of OLE1, which is the orthologue of the delta 9 desaturase in yeast, it is controlled through a transcriptional regulatory network that is very SREBP like. There is a typical cleavage and then a migration to the nucleus. It is a fairly complex set of factors. If you want to compare it with an SREBP paradigm, this is kind of the mechanism it is forming. Daum: What is the evidence for growth of lipid particles by fusion? It could be either fusion or delivery. Yang: I agree with both models. Daum: In the lipid particle population we find two types: small ones and a few large ones. It is unclear whether the large ones have a specific function. We are working on this aspect with our colleagues from our biophysics department. Shi: Three years ago there was a Nature paper (Ashrafi et al 2003) on RNAi in Caenorhabditis elegans. They did a systematic knockout of almost every gene, 16 000 in all. Do the majority of your genes have homologues in C. elegans? Yang: In that screen you are looking at a multicellular tissue, so there is development involved. In our case it is much simpler: we are just looking within the cell. There may be a few overlaps, but not too many. Bernlohr: Is the budding process energy dependent, or is it a spontaneous process based on some size parameter? Daum: We know that the budding process is dependent on the synthesis of the lipids. This is indirectly an energy dependent process because it needs activation of fatty acids and then the reaction with glycerol or sterols. Ongoing synthesis of these components would be one driving force. Bernlohr: Small GTP-binding proteins could be involved in this maturation. You would find those in your screens. Shi: In the mutant where the ER disappears, does the cell divide normally? Where does the ER go? Yang: The cells are very sick. The ER just becomes lighter. The morphology has changed. Daum: On the other hand the mutants that lack the triacylglycerols and sterol esters look like wild-type. They grow a bit more slowly but have few other defects. They are a little more sensitive to exogenous fatty acids, but in the normal life in the lab it is almost like wild-type. Bernlohr: In the ATGL story there is the suggestion that the enzyme is evolutionarily related to the phospolipase A2, and this would generate a hydrolytic position at sn-2 and possibly the diacylglycerol that is made is 1,3-diacylglycerol. In your system have you had a chance to analyse the positional specificity? Daum: We have not yet checked this, but are planning to.
154
DAUM ET AL
Black: The hydrolase that doesn’t like to recycle the very long chain C26 fatty acids; one of the hallmarks of the fatty acid transport proteins is this very long chain acyl CoA activity, including the yeast orthologue. Have you made the FAT1 deletion in combination with those hydrolase mutants and seen what the phenotype is? Daum: No, but that’s a good idea. Shi: You didn’t mention the acylglycerophospholipids. You concentrated on neutral lipids. Where do they interconnect in yeast? Daum: One connection is Lro1p, which transfers one fatty acid from phosphatidylethanolamine or phosphatidylcholine to diacylglycerol thus forming triacylglycerol. It is thought that this might be a regulatory enzyme that gives us a balance between excess of phospholipids on one hand and storage of neutral lipids on the other. This may be an important link. There is no transfer of fatty acids from phospholipids to sterols in the yeast, as is known in the mammalian system. Reference Ashrafi K, Chang FY, Watts JL et al 2003 Genome-wide RNAi analysis of Caenorhabditis elegans fat regulatory genes. Nature 421:268–272
Fatty Acids and Lipotoxicity in Obesity and Diabetes: Novartis Foundation Symposium 286, Volume 286. Edited by Gregory Bock and Jamie Goode Copyright Novartis Foundation 2007. ISBN: 978-0-470-05764-3
Cide proteins and the development of obesity John Zhong Li and Peng Li*1 Department of Biolog y, Hong Kong University of Science and Technolog y, Clearwater Bay, Hong Kong and *Protein Science Laboratory of Ministry of Education, Department of Biological Sciences and Biotechnolog y, Tsinghua University, Beijing, China
Abstract. Obesity has become the most prevalent chronic disorder that affects large populations throughout the world. Obesity is due largely to the imbalance between energy intake and expenditure. Energy balance is regulated by interactions between various tissues such as adipose tissues, liver and skeletal muscle. Adipose tissues, which include brown adipose tissue (BAT) and white adipose tissue (WAT), play crucial roles in maintaining energy homeostasis. While WAT stores energy in the form of triglycerides; BAT increases energy expenditure through thermogenesis. Cide proteins including Cidea, Cideb and Fsp27, are expressed at high levels in BAT, liver and WAT, respectively. Cidea−/− mice exhibit increased lipolysis in BAT and are resistant to high fat diet-induced obesity and diabetes. Our recent data suggest that Cideb also plays an important role in the development of obesity by regulating various metabolic pathways in liver. The molecular mechanism of Cide protein regulation of metabolic networks and obesity is discussed. 2007 Fatty acids and lipotoxicity in obesity and diabetes. Wiley, Chichester (Novartis Foundation Symposium 286) p 155–161
The number of obese individuals has increased rapidly over the last few years not only in developed countries but also in rapidly developing countries such China. Obesity develops as a result of energy intake exceeding energy expenditure (Spiegelman & Flier 2001) and is a major risk factor for many metabolic diseases such as hypertension, stroke, liver steatosis, cancer and inflammatory diseases (Kopelman 2000). The development of obesity is regulated by multiple pathways and involves functional interactions among various tissues such as brain, adipose tissue, skeletal muscle, liver and intestine. Brain is the central organ for controlling energy intake such as food intake (Schwartz & Porte 2005). Adipose tissues can be divided into 1
This paper was presented at the symposium by Peng Li, to whom correspondence should be addressed. 155
156
LI & LI
brown adipose tissue (BAT) and white adipose tissue (WAT). Both BAT and WAT contain an abundant amount of lipid and can serve as energy storage organs. BAT plays a unique role in energy expenditure by uncoupling oxidative phosphorylation and dissipating energy as heat to maintain core body temperature in animals when exposed to cold (Himms-Hagan 1989). The thermogenic activity of BAT is mediated by uncoupling protein 1 (UCP1), a mature brown adipocyte-specific protein that is localized to the inner mitochondrial membrane (Zisman et al 2000). The primary role of WAT is to store energy in the form of triglycerides (TAG) in lipid droplets and immobilize the energy in time of needs such as starvation. Adipose tissue can also serve as an endocrine organ to secret crucial hormones such as leptin and adiponectin for the control of energy homeostasis (Ahima & Flier 2000). Liver plays a central role in energy homeostasis as it is the main organ for lipid de novo synthesis, lipid uptake and secretion, fatty acid oxidation, as well as the production of ketone bodies (Leonhardt & Langhans 2004). It is also an important organ for glucose synthesis (gluconeogenesis) and storage (glycogen synthesis). Cide proteins, including Cidea, Cideb and Fsp27 (Cidec in human), were originally identified by their sequence homology to the N-terminal region of DNA fragmentation factor DFF40/45 (Inohara et al 1998). When overexpressed in heterologous cells such as 293T, HeLa and CHO cells, both Cidea and Cideb were able to induce caspase independent cell death (Inohara et al 1998, Chen et al 2000). The mechanism by which high levels of Cide proteins induce cell death is unclear. Unlike DFF40/45 proteins that have high levels of expression in all tissues, we observed that Cidea is expressed at higher levels in BAT (Zhou et al 2003). Cideb mRNAs and proteins were detected in various tissues with highest levels of expression in the liver, moderate levels in the kidney and lower levels in the stomach and small intestine (Inohara et al 1998). Fsp27 was originally identified in a differentiated adiopocyte cell line (Danesch et al 1992), and its transcripts were detected at high levels in BAT, WAT and at lower levels in skeletal muscle (Zhou et al 2003). When over-expressed, Cideb protein can form a homodimer that is mediated by its C-terminal region (Chen et al 2000). Nuclear magnetic resonance (NMR) structural analysis indicated that the Cide-N domain, which is conserved among all Cide proteins and DFF, comprises a novel bipolar structure with two opposite charged regions on the surface that could serve as a potential protein–protein interaction interface (Lugovskoy et al 1999). The unique expression pattern of Cide proteins in metabolically active tissues, their structural properties and their cell death-inducing activity indicate that this class of proteins may play a vital role in maintaining cell survival and normal metabolic activity. To understand the physiological role of Cide proteins, we generated Cidea and Cideb null mice by homologous recombination. Interestingly, we observed that mice deficient in Cidea or Cideb exhibited higher energy expenditure and were resistant to high fat diet-induced obesity and diabetes (Zhou et al 2003, Li et al 2007). We compared the detailed
CIDE PROTEINS AND OBESITY TABLE 1
157
A comparison of the phenotype of Cidea and Cideb null mice
Gene name
Cidea
Cideb
Tissue distribution Gene modification Adiposity index Morphology of WAT Morphology of BAT Lipid content Liver morphology
BAT Knock out Decreased Smaller lipid droplet Smaller lipid droplet Decreased in BAT No change
Serum TAG Serum NEFA Glucose disposal Insulin sensitivity
Deceased Decreased Increased No change
Serum insulin Serum glucose Serum ketone body Serum adiponectin Serum resistin Serum leptin Food intake Body temperature Whole body metabolic rate Fatty acid oxidation Fatty acid synthesis Possible mechanism
No change Decreased Not determined Not determined Not determined Decreased No change Increased Increased Not determined Not determined Increased lipolysis and thermogensis in BAT, possible interaction with UCP1
Liver, Kidney, Small intestine Knock out Decreased Smaller lipid droplet No change Decreased in liver Smaller lipid droplet after HFD feeding Decreased Decreased Increased Increased hepatic insulin sensitivity Decreased Decreased Increased Increased No change Decreased Increased Not determined Increased Increased Decreased Reduced expression levels of SREBP-1c and its downstream target genes FAS, ACC and SCD in the liver
phenotype of Cidea and Cideb null mice (Table 1) and analysed the underlying mechanism of Cide protein regulation of energy homeostasis and the development of obesity and diabetes. Although we had convincing evidence that overexpression of Cide proteins in various cell lines induced cell death, no difference in cell number, cell differentiation or apoptotic activity was observed in brown adipocytes of Cidea null mice and in Cideb-deficient hepatocytes, suggesting that Cide proteins may not play a direct role in regulating cell survival during adipocyte and hepatocyte development and differentiation. Intriguingly, we observed that both Cidea and Cideb null mice exhibited a typical lean phenotype based on the following observations. First, mice deficient in Cidea or Cideb have a significantly lower adiposity index. Second, levels of plasma triglycerides (TAG) and free fatty acids (NEFA) in Cidea or Cideb null mice are significantly lower than that of wild-type mice. Third, the amount
158
LI & LI
of total lipid in BAT of Cidea null mice and in liver of Cideb null mice dramatically reduced. Morphologically, the size of white adipocytes, especially the size of lipid droplets, in both Cidea- and Cideb-deficient mice was reduced. These data clearly suggest that both Cidea and Cideb play important roles in the development of obesity. Consistent with our observations, Cidea was shown to mediate human obesity by regulating human adipocyte lipolysis (Nordstrom et al 2005) and a V115F polymorphism in human was found to be associated with obesity in certain populations (Dahlman et al 2005). The anti-obesity effect of Cidea and Cideb deficiency is in part due to their increased energy expenditure as they all had a higher rate of whole body metabolism. However, the underlying mechanism of increased energy expenditure in Cidea and Cideb null mice appears to be different. Cidea null mice had a higher core body temperature and increased thermogenesis, while Cideb-deficient mice showed increased fatty acid oxidation. The increased thermogenesis in Cidea null mice may be partly contributed to by higher UCP1 activity in brown adipocytes (Zhou et al 2003). Increased fatty acid oxidation in Cideb-deficient hepatocytes is likely due to the decreased levels of ACC2, as it was shown to be an inhibitor of mitochondrial fatty acid oxidation (Abu-Elheiga et al 2003). The lean phenotype of Cideb-deficient mice could also be due to the reduced lipogenesis and decreased expression levels of SREBP1c and its downstream target genes ACC, FAS and SCD1. However, at this point, it is uncertain as to how these effects in BAT of Cidea null mice and the liver of Cideb null mice can lead to such a profound influence on adiposity. The fall in plasma TAG and NEFAs in both Cidea and Cideb-null mice and their resistance to diet-induced obesity raised the possibility that their insulin sensitivity might also be increased. Indeed, both Cidea and Cideb deficient mice showed an increased rate of glucose uptake. Although no difference in plasma insulin levels was observed in Cidea-deficient mice, these animals showed decreased plasma glucose levels when they were fed with a high fat diet. More interestingly, we observed drastic reduction in the levels of plasma insulin in Cideb null mice. Reduced insulin levels did not result from impaired β -cell function in Cideb mutant mice, as the stimulated plasma insulin levels in response to acute glucose loading were similar. Reduced insulin levels, improved glucose disposal and lower blood glucose levels in ITT experiments suggest that Cideb null mice have enhanced insulin sensitivity. The mechanism of increased glucose uptake in Cidea null mice is currently not clear. In Cideb null mice, we observed significantly increased levels of IRS1 tyrosine phosphorylation and AKT phosphorylation in the liver after insulin stimulation, suggestive of improved hepatic insulin sensitivity. Although the precise molecular nature of Cidea in regulating thermogenesis and Cideb in the regulation of gene expression of SREBP1c and its downstream targets remains to be resolved, our data reveal that Cide proteins play crucial roles in the develop-
CIDE PROTEINS AND OBESITY
159
ment of obesity and provide novel therapeutic targets for various metabolic disorders such as obesity, diabetes and fatty liver. References Abu-Elheiga L, Oh W, Kordari P, Wakil SJ 2003 Acetyl-CoA carboxylase 2 mutant mice are protected against obesity and diabetes induced by high-fat/high-carbohydrate diets. Proc Natl Acad Sci USA 100:10207–10212 Ahima RS, Flier JS 2000 Adipose tissue as an endocrine organ. Trends Endocrinol Metab 11:327–332 Chen Z, Guo K, Toh SY, Zhou Z, Li P 2000 Mitochondria localization and dimerization are required for CIDE-B to induce apoptosis. J Biol Chem 275:22619–22622 Dahlman I, Kaaman M, Jiao H, Kere J, Laakso M, Arner P 2005 The CIDEA gene V115F polymorphism is associated with obesity in Swedish subjects. Diabetes 54:3032–3034 Danesch U, Hoeck W, Ringold GM 1992 Cloning and transcriptional regulation of a novel adipocyte-specific gene, FSP27. CAAT-enhancer-binding protein (C/EBP) and C/EBP-like proteins interact with sequences required for differentiation-dependent expression. J Biol Chem 267:7185–7193 Himms-Hagen J 1989 Role of thermogenesis in the regulation of energy balance in relation to obesity. Can J Physiol Pharmacol 67:394–401 Inohara N, Koseki T, Chen S, Wu X, Nunez G 1998 CIDE, a novel family of cell death activators with homology to the 45 kDa subunit of the DNA fragmentation factor. EMBO J 17:2526–2533 Kopelman PG 2000 Obesity as a medical problem. Nature 404:635–643 Leonhardt M, Langhans W 2004 Fatty acid oxidation and control of food intake. Physiol Behav 83:645–651 Li JZ, Ye J, Xue B et al 2007 Cideb regulates diet-induced obesity, liver steatosis and insulin sensitivity by controlling lipogenesis and fatty acid oxidation. Diabetes in press Lugovskoy AA, Zhou P, Chou JJ, McCarty JS, Li P, Wagner G 1999 Solution structure of the CIDE-N domain of CIDE-B and a model for CIDE-N/CIDE-N interactions in the DNA fragmentation pathway of apoptosis. Cell 99:747–755 Nordstrom EA, Ryden M, Backlund EC et al 2005 A human-specific role of cell death-inducing DFFA (DNA Fragmentation Factor-alpha)-like effector A (CIDEA) in adipocyte lipolysis and obesity. Diabetes 54:1726–1734 Schwartz MW, Porte D Jr 2005 Diabetes, obesity, and the brain. Science 307:375–379 Spiegelman BM, Flier JS 2001 Obesity and the regulation of energy balance. Cell 104:531–543 Zhou Z, Yon Toh S, Chen Z et al 2003 Cidea-deficient mice have lean phenotype and are resistant to obesity. Nat Genet 35:49–56 Zisman A, Peroni OD, Abel ED et al 2000 Targeted disruption of the glucose transporter 4 selectively in muscle causes insulin resistance and glucose intolerance. Nat Med 6:924– 928
DISCUSSION Bernlohr: Is Cidea mitochondrial in these experiments. P Li: When we overexpress it, Cidea is colocalized with mitochondrial markers in the late stage of the expression.
160
LI & LI
Bernlohr: Since it is a small protein, does Cidea have any potential for electron transport on its own? P Li: I have never tested that. For the mitochondrial localization, we have a little more insight with the antibodies for Cidea. The problem with Cidea is that we don’t have good antibodies. When we overexpress it in the cells it tends to overlap with mitochondria. These cells die very quickly in the late stage of transfection. When we saw the mitochondrial localization, we went back to see whether this was original localization or because of the effect on cell death and the morphological changes. We are still working on this. Black: For the Cidea knockout mice you talked about adaptive thermogenesis. What happens if you put the mice in the cold? P Li: We put Cidea knockouts in the cold for 24 h. They had 0.8–1 °C body temperature increases. If we compare this with UCP1, which drops dramatically, the body temperature actually increases slightly in the Cidea knockouts. Black: I was curious about the Cideb knockout. There is an increase in β oxidation, and I noticed that the acyl CoA synthetase 4 is reduced approximately twofold. Did you see any of the acyl CoA synthetases that were commensurately up-regulated with the genes typical of β oxidation? P Li: We checked extensively and didn’t fi nd any genes up-regulated from the β oxidation pathway. We see 20–30% up-regulation in the fatty acid oxidation in isolated hepatocytes from these knockouts. We thought this might be due to the decreased level of ACC2. People are still uncertain of the role of ACC2 in hepatocytes. Shi: The difficulty of working with Cide proteins is a lack of information on their cellular functions. Have you done experiments to see whether the protein itself is phosphorylated when stimulated by insulin or other hormones? P Li: We treated hepatocytes with β agonists and insulin and found no changes in phosphorylation. Shi: How did you pick up on Fsp27 originally? What information led you to look at it? P Li: It was the Fsp27 expression in white adipose tissue, its homology to Cidea and our original phenotype for the Cidea knock out mice. Shi: Is Fsp27 secreted? P Li: No. Not only isn’t it excreted, it is also associated with some kinds of membrane particles. Glass: What about transcription factors involved in mitochondrial biogenesis, such as the nuclear respiratory factors? P Li: I didn’t check NRF1. Spiegelman: I believe Cidea is a PGC1α target, but we haven’t looked at what transcription factors are involved in that.
CIDE PROTEINS AND OBESITY
161
P Li: We don’t know what this gene does to regulate the transcriptional programme. We were surprised to fi nd it does, because we never noted a nuclear localization. Spiegelman: That’s not so surprising. It is a two-way conversation from the nucleus to the mitochondrion and from the mitochondrion back to the nucleus.
Fatty Acids and Lipotoxicity in Obesity and Diabetes: Novartis Foundation Symposium 286, Volume 286. Edited by Gregory Bock and Jamie Goode Copyright Novartis Foundation 2007. ISBN: 978-0-470-05764-3
General discussion I Visualizing brown adipose tissue with FDG-PET Saito: I’d like to make a short presentation on human brown adipose tissue (BAT) visualized by FDG-PET ([18F]-2-fluoro-deoxy-D-glucose/positron emission tomography). It is well accepted that BAT/UCP1 contributes to whole body fatty acid oxidation and energy expenditure, and thereby regulation of body fat content. This activity is under the control of the sympathetic nerve– β adrenoceptor system. This system also activates the glucose utilization in this specific tissue. We recently reconfirmed that the β -adrenergically activated glucose utilization in BAT completely disappears in the UCP knockout mouse, showing that the glucose utilization in BAT is UCP1-dependent. Thus, BAT-UCP1 is a target for body fat control and the regulation of glucose metabolism. But this came from mouse or rat studies. Is this also true in humans? I’m afraid that almost all of you here may be doubtful about BAT in humans. In fact, it is quite difficult to detect BAT in human subjects by conventional histological/microscopical methods. We have tried a new method for detecting human BAT: that is FDG-PET. As you know, by this method we can monitor tissue glucose utilization by detecting [18F]FDG. It is used for the early detection of tumours, because tumour tissues accumulate more FDG than normal tissues. Sometimes, significant accumulation of FDG is observed in the shoulder and thoracic spine regions where there are no tumours. By simultaneous CT examinations we identified that this accumulation is in adipose tissue. Then, is it white or brown fat? If it is brown, we’d expect that cold exposure or some comparable sympathetic stimulation increases this FDG accumulation. To test this idea, using healthy volunteers, we examined the effects of acute exposure on the FDG accumulation. When the subjects were kept at 28 °C there was no accumulation. But when the same subjects were kept in a cold environment at 19 °C for 2 h, a clear FDG accumulation was detected. This must be brown fat activated by cold exposure. We expect that FDG-PET will be a powerful tool for evaluating human brown fat. Spiegelman: So why should people accept that it is brown adipose tissue? It clearly is enhanced glucose uptake dependent on the cold. There are molecular markers of fat. There is oxygen consumption and uncoupled respiration that is characteristic of fat. I would like to believe what you have shown is brown fat, but why should I? I am playing the devil’s advocate: suppose someone was going to say that white fat is differentially innervated with sympathetic nerves. There is in fact sympathetic innervation in white fat. What you have could be differential stimulation of the white fat cells, which stimulates glucose uptake. 162
GENERAL DISCUSSION I
163
Saito: We have only mouse data following cold acclimation or chronic β3 adrenergic stimulation, where there is a conversion of white fat to brown fat. Glass: In the setting where human adults do accumulate brown fat, such as pheochromocytoma, this occurs in stereoptypical locations. Is this pattern what you’d expect to see in a human with pheochromocytoma? Saito: There have been papers describing that in pheochromocytoma there is a large amount of brown fat, and the pattern of deposition is similar to what we saw. O’Rahilly: This should be quite common, because FDG-PET is used frequently in pheochromocytoma localization these days. Sitting in the databases there should be a whole bunch of information. Spiegelman: I have done a scan of the literature. There have not been papers on this subject. O’Rahilly: I bet radiologists have the information. Saito: Radiologists have focused on this problem because it has led to misdiagnosis of tumours. O’Rahilly: If you think this is subcutaneous can you do a needle biopsy and measure UCP1, to prove it. Spiegelman: People are not going to accept just increased PET signal as indicative of brown fat. Muoio: What about β3 agonists? Spiegelman: There is no question that chronic stimulation of β3 agonists can stimulate brown adipose tissue. Usually something is proved to correlate with a diagnostic and then the diagnostic tool is used. Shi: What happens if you put babies of different ages in there? O’Rahilly: You can’t use PET in children. It is highly radioactive. There are no β3 agonists licensed for use in humans.
Fatty Acids and Lipotoxicity in Obesity and Diabetes: Novartis Foundation Symposium 286, Volume 286. Edited by Gregory Bock and Jamie Goode Copyright Novartis Foundation 2007. ISBN: 978-0-470-05764-3
Adiponectin and adiponectin receptors in obesity-linked insulin resistance Takashi Kadowaki*†, Toshimasa Yamauchi*‡, Naoto Kubota*†‡, Kazuo Hara*† and Kohjiro Ueki* * Department of Metabolic Diseases, Graduate School of Medicine, University of Tokyo, Tokyo, Japan, † Division of Applied Nutrition, National Institute of Health and Nutrition, Tokyo, Japan and ‡ Department of Integrated Molecular Science on Metabolic Diseases, Graduate School of Medicine, University of Tokyo, Tokyo, Japan
Abstract. Adiponectin is an abundantly expressed adipokine in adipose tissue and has direct insulin sensitizing activity. We have proposed the following adiponectin hypothesis. Interactions of genetic factors such as single nucleotide polymorphisms (SNPs) in the Adiponectin gene and environmental factors causing obesity result in hypoadiponectinaemia, which appears to play an important causal role in obesity-linked insulin resistance, type 2 diabetes and the metabolic syndrome. We have cloned the adiponectin receptors, AdipoR1 and AdipoR2, which mediate the antidiabetic metabolic actions of adiponectin. AdipoR1 and AdipoR2 are down-regulated in obesity-linked insulin resistance. Up-regulation of adiponectin or adiponectin receptors may represent potential versatile therapeutic targets to combat obesity-linked diseases characterized by insulin resistance. 2007 Fatty acids and lipotoxicity in obesity and diabetes. Wiley, Chichester (Novartis Foundation Symposium 286) p 164–182
Adiponectin hypothesis in the pathogenesis of obesity-linked insulin resistance and metabolic syndrome Obesity-linked insulin resistance is a key feature of type 2 diabetes, metabolic syndrome and cardiovascular diseases (Reaven 1998, National Cholesterol Education Program 2001, International Diabetes Federation 2005). White adipose tissue (WAT) has been recognized as an important endocrine organ that secretes a number of biologically active ‘adipokines’ (Hotamisligil et al 1993, Zhang et al 1994, Steppan et al 2001, Lazar 2006). Dysregulation of these adipokines have been shown to affect insulin sensitivity through modulation of insulin signalling pathway such as phosphorylation of the insulin receptor substrate (IRS) proteins (e.g. IRS1 and IRS2) (Hotamisligil 1993) and the molecules involved in glucose and lipid metabolism (Shulman 2000). Of these adipokines, adiponectin has 164
ADIPONECTIN
165 Environmental factors
Genetic factors
Variations of adiponectin gene (SNP 276 G allele)
Environmental factors causing obesity (HFdiet,etc)
(Diabetes 51:536-540, 2002)
(Nat Med 7:941-964, 2001) (Nat Genet 30:221-226, 2002)
Hypoadiponectinemia muscle (J Biol Chem 277:25863-25866, 2002) (Nature 423:762-764, 2003)
Insulin Resistance Metabolic Syndrome (Nat Med 7:941-946, 2001; Nat Med 8:1288-1295, 2002)
Atherosclerosis (J Biol Chem 278:2461-2468, 2003)
FIG. 1. Adiponectin hypothesis for insulin resistance, metabolic syndrome and atherosclerosis (Kadowaki & Yamauchi 2005).
recently attracted much attention because of its antidiabetic and antiatherogenic effects (Yamauchi et al 2001, Berg et al 2001, 2005, Fruebis et al 2001, Kadowaki & Yamauchi 2005, Okamoto et al 2006, Kadowaki et al 2006). Indeed, a decrease in the circulating levels of adiponectin by genetic and environmental factors has been shown to contribute to the development of diabetes and the metabolic syndrome. Based on the significant body of evidence, we have proposed the ‘adiponectin hypothesis,’ in which reduced plasma adiponectin levels caused by interactions between genetic factors and environmental factors causing obesity may play a crucial role in the development of insulin resistance, type 2 diabetes, the metabolic syndrome and atherosclerosis (Kadowaki & Yamauchi 2005, Kadowaki et al 2006) (Fig. 1). Up-regulation of adiponectin as a therapeutic target According to the adiponectin hypothesis (Kadowaki & Yamauchi 2005, Kadowaki et al 2006), a therapeutic strategy for the treatment of insulin resistance, type 2 diabetes, the metabolic syndrome and cardiovascular disease may include the upregulation of plasma adiponectin levels. Thiazolidinediones (TZDs) are known to improve systemic insulin sensitivity in obesity-linked insulin resistance and diabetes. TZDs have been widely used as therapeutic agents for the treatment of type 2 diabetes (Nolan et al 1994, Yki-Jarvinen 2004). Plasma adiponectin levels have been shown to be up-regulated by TZDs (Yamauchi et al 2001, Hirose et al
166
KADOWAKI ET AL
2002, Maeda et al 2001), and high molecular weight (HMW) adiponectin is a predominant form of adiponectin upregulated by TZDs (Pajvani et al 2004). Since adiponectin is an insulin-sensitizing adipokine, it is reasonable to speculate that the action whereby TZDs increase insulin sensitivity is mediated, at least in part, by increased plasma adiponectin levels. Adiponectin-deficient (Adipo −/−) ob/ob mice with a C57BL/6 background (Kubota et al 2002) were used to investigate whether the PPARγ agonist pioglitazone is capable of ameliorating insulin resistance in the absence of adiponectin (Kubota et al 2006). Low-dose (10 mg/kg) pioglitazone treatment significantly upregulated serum adiponectin levels in ob/ob mice and significantly ameliorated the insulin resistance and diabetes. In contrast, this dose of pioglitazone failed to ameliorate the insulin resistance and diabetes in Adipo −/−ob/ob mice (Kubota et al 2006). With high-dose (30 mg/kg) pioglitazone treatment, the insulin resistance and diabetes of ob/ob mice were again significantly ameliorated. Interestingly, Adipo −/−ob/ob mice also displayed significant amelioration of insulin resistance and diabetes. The serum FFA and triglyceride levels as well as adipocyte sizes in ob/ob and Adipo −/−ob/ob mice were unchanged after low-dose pioglitazone treatment but were significantly reduced to a similar degree after high-dose pioglitazone treatment. Moreover, the expression of TNF- α and resistin in adipose tissues of ob/ob and Adipo −/−ob/ob mice were unchanged after low-dose pioglitazone but were decreased after high-dose pioglitazone. Thus, a low dose of pioglitazone increases adiponectin levels in part via activation of Adiponectin gene transcription (Yamauchi et al 2001, Iwaki et al 2003) without stimulating adipocyte differentiation, thereby increasing AMPK activation, decreasing gluconeogenesis in the liver, and ameliorating insulin resistance and type 2 diabetes (Fig. 2). On the other hand, a high dose of pioglitazone induces Adipose tissue
Liver adiponectin ↑
Ob/Ob Low dose TZD
Insulin ↓ resistance
Gluconeogenesis ↓ AMPK ↑
Transcription ↑
PPAR-g High dose TZD Ob/Ob AdipoKOOb/Ob
Adipocyte differentiation ↑ adipocyte size ↓ FFA ↓ TNF ↓ Resistin ↓
Skeletal muscle Insulin ↓ resistance
FIG. 2. TZDs ameliorate insulin resistance and diabetes by both adiponectin dependent and independent pathways (Kubota et al 2006).
ADIPONECTIN
167
adipocyte differentiation, leading to an increase in the number of small adipocytes, which is associated with decreased serum free fatty acid (FFA) levels and decreased tumour necrosis factor (TNF) α and resistin expression, which may contribute to amelioration of insulin resistance in skeletal muscle (Fig. 2) (Kubota et al 2006). We propose that there are two different pathways in the amelioration of insulin resistance induced by TZDs such as pioglitazone. One involves an adiponectindependent pathway and the other an adiponectin-independent pathway (Fig. 2). Scherer’s group also demonstrated that rosiglitazone ameliorated glucose intolerance via both adiponectin-dependent and -independent pathways (Nawrocki et al 2006). Other pharmacological agents as well as lifestyle changes have also been reported to be associated with the up-regulation of plasma adiponectin levels (reviewed in Table 1 of Kadowaki et al 2006). Cloning, function, and regulation of adiponectin receptors In order to further understand the molecular mechanism of adiponectin action, we isolated cDNA for human adiponectin receptors (Yamauchi et al 2003). The cDNA encoded a protein designated adiponectin receptor 1 (AdipoR1) (Yamauchi et al 2003). This protein is structurally conserved from yeast to humans (especially in the 7 transmembrane domains). Interestingly, the yeast homologue (YOL002c) plays a key role in metabolic pathways that regulate lipid metabolism, such as fatty acid oxidation (Karpichev et al 2002). Moreover, we found a gene that was significantly homologous (67% amino acid identity) with AdipoR1, which was termed AdipoR2 (Yamauchi et al 2003). The existence of two distinct receptors for adiponectin was consistent with our previous assumption based upon binding studies (Yamauchi et al 2002). AdipoR1 is ubiquitously expressed, including in skeletal muscle and liver, whereas AdipoR2 is most abundantly expressed in the liver. AdipoR1 and AdipoR2 appear to be integral membrane proteins; the Nterminus is internal and the C-terminus is external—opposite to the topology of all other reported G protein-coupled receptors (GPCRs) (Yamauchi et al 2003) (Fig. 3). Expression of AdipoR1 and AdipoR2 or suppression of AdipoR1 and AdipoR2 expression supports our conclusion that AdipoR1 and AdipoR2 serve as receptors for adiponectin and mediate increased AMP kinase, PPARα and p38 MAP kinase activities as well as fatty-acid oxidation and glucose uptake by adiponectin. Recently, a two-hybrid study revealed that the C-terminal extracellular domain of AdipoR1 interacted with adiponectin, whereas the N-terminal cytoplasmic domain of AdipoR1 interacted with APPL (adaptor protein containing pleckstrin homology domain, phosphotyrosine-binding domain, and leucine zipper motif) (Mao et al 2006). Moreover, interaction of APPL with AdipoR1 in mammalian cells was stimulated by adiponectin binding, and this interaction appeared to play an important role in adiponectin-mediated AMPK activation and downstream
168
KADOWAKI ET AL
AdipoR1
AdipoR2
(66.7%homology)
Extracellular
Intracellular Hydrophobic Charged
Polar Glycine
kb 9.5 7.5 4.4
kb 7.5
2.4
AdipoR1
1.35
4.4
AdipoR2
Brain Heart Kidney Liver Lung Muscle Spleen Testis
0.24
1.35
Brain Heart Kidney Liver Lung Muscle Spleen Testis
2.4
FIG. 3. Cloning of adiponectin receptors (Yamauchi et al 2003).
effects (Mao et al 2006).The expression levels of both AdipoR1 and AdipoR2 were significantly decreased in insulin-resistant ob/ob mice, probably in part because of obesity-linked hyperinsulinaemia (Tsuchida et al 2004) (Fig. 4). Moreover, adiponectin-induced activation of AMPK was significantly decreased, for example, in the skeletal muscle of ob/ob mice, suggesting that adiponectin resistance is present in ob/ob mice (Tsuchida et al 2004). Thus, obesity decreases not only plasma adiponectin levels but also AdipoR1/R2 expression, thereby causing adiponectin resistance and leading to insulin resistance, which in turn aggravates hyperinsulinaemia, forming a ‘vicious cycle’ (Tsuchida et al 2004) (Fig. 4). Expression of adiponectin receptors in the liver ameliorated diabetes Consistent with the data in ob/ob mice, expression levels of AdipoR1 or AdipoR2 were decreased to approximately 65% or 55%, respectively, in the liver of db/db mice as compared with wild-type control B6 mice. To determine the role of decreased expression levels of AdipoRs in the development of the insulin resistance and diabetes observed in obese mice, we studied the effects of adenovirusmediated restoration of AdipoR1 expression in db/db mice. In fact, adenovirusmediated overexpression of AdipoR1 or AdipoR2 in the liver of db/db mice significantly improved insulin resistance and diabetes in db/db mice (Yamauchi et al 2007) (Fig. 5).
ADIPONECTIN
169
(Hypothesis)
Obesity Decreased adiponectin levels
Hyperinsulinemia Oxidative stress Inflammation
Decreased adiponectin receptor expression
“Adiponectin resistance”
Decreased adiponectin effects Insulin resistance FIG. 4.
“Vicious Cycle”
Obesity, adiponectin resistance and insulin resistance (Tsuchida et al 2004).
db/db mice AdipoR1 in liver
800 mg/dL
Ratio
3
Plasma glucose during OGTT
**
2 1
AdipoR2 in liver
*
**
**
0
Adenovirus:
**
600
**
4 2
0 Adenovirus: LacZ AdipoR2
50
0
100
LacZ AdipoR1
150 min
Plasma glucose during OGTT mg/dL
Ratio
400 200
0 Adenovirus: LacZ AdipoR1
6
*
600
*
400
** *
**
200 0
0
Adenovirus:
**
50
100
LacZ AdipoR2
150 min
FIG. 5. Expression of AdipoR1 or AdipoR2 in the liver of db/db mice significantly ameliorated diabetes (Yamauchi et al 2007).
AdipoR1 increases AMPK activation by adiponectin in liver Expression of AdipoR1 resulted in significantly increased activation of AMPK in the liver by adiponectin, whereas expression of AdipoR2 did not. Activation of AMPK in the liver has been reported to reduce the expression of genes encoding hepatic gluconeogenic enzymes such as glucose-6-phosphatase (G6pc) and
170
KADOWAKI ET AL
WAT Secretion Adiponectin
AdipoR1
Liver
AMPK ↑ PEPCK ↓ G6Pase ↓
AdipoR2
PPARα ↑ ACO ↑ , UCP2 ↑
fatty-acid SREBP1c ↓ oxidation ↑
gluconeogenesis ↓
fasting glucose ↓
energy expenditure ↑
TG content ↓
Increased insulin sensitivity FIG. 6. AdipoR1 appears to mediate its biological effects via AMPK, whereas AdipoR2 appears to mediate them via PPARα (Yamauchi et al 2007).
phosphoenolpyruvate carboxykinase 1 (Pck1) (Lochhead et al 2000) as well as genes encoding molecules involved in lipogenesis such as sterol regulatory elementbinding protein 1c (Srebf1) (Woods et al 2000). In fact, expression of AdipoR1 significantly decreased the expression of G6pc, Pck1 and Srebf1 in the liver of db/db mice, which may be a mechanism by which restoration of AdipoR1 in the liver reduced endogenous glucose production (EGP), apparently increased glucose infusion rate (GIR) and improved diabetes (Yamauchi et al 2007) (Fig. 6). In contrast, expression of AdipoR2 had little effect on the expression levels of G6pc, Pck1 or Srebf1. These results suggested that AdipoR1 may be more involved in the activation of AMPK by adiponectin than AdipoR2 in liver in vivo (Yamauchi et al 2007) (Fig. 6). AdipoR2 increases PPARa target genes in liver Expression of AdipoR2 significantly increased the expression of genes encoding molecules involved in glucose uptake such as glucokinase (Gck) (Matshinsky et al 2006), unlike the molecules involved in gluconeogenesis, which appeared to be one possible mechanism by which AdipoR2 expression in the liver apparently increased GIR and improved diabetes. On the other hand, expression of AdipoR1
ADIPONECTIN
171
had little effect on the expression levels of Gck. Expression of AdipoR2 in the liver of db/db mice increased PPARα (Ppara) itself (Yamauchi et al 2007) (Fig. 6) and its target genes (Kersten et al 2000) such as Acox1 (acyl-CoA oxidase) and Ucp2 (uncoupling protein 2), whereas expression of AdipoR1 in the liver of db/db mice had little effect on PPARα itself or its target genes such as Acox1 and Ucp2. These observations suggested that AdipoR2 may be more involved in activation of the PPARα pathways by adiponectin than AdipoR1. Adenovirus-mediated expression of AdipoR1 or AdipoR2 in the liver of db/db mice significantly increased fatty-acid oxidation (Fig. 6), and tended to decrease hepatic triglyceride content (Fig. 6), which may be one mechanism by which expression of AdipoR1 or AdipoR2 in the liver improved insulin resistance and diabetes (Yamauchi et al 2007) (Fig. 6). The present data suggest that down-regulation of AdipoR1 and AdipoR2 in obesity plays causal roles, at least in part, in the development of insulin resistance and diabetes. AdipoR1 and AdipoR2 serve as the major adiponectin receptors in vivo We generated AdipoR1 knockout mice, AdipoR2 knockout mice and AdipoR1/R2 double knockout mice (Yamauchi et al 2007). AdipoR1 knockout mice showed significantly impaired glucose tolerance and insulin resistance (Yamauchi et al 2007) (Fig. 7). EGP was significantly increased and GIR was significantly decreased AdipoR2 KO
Double KO
400 Plasma glucose
400 Plasma glucose
400 Plasma glucose
300
300
300
01530 60
∗∗
200 100
120min
Plasma insulin
3
2
∗
2
1 ∗∗ 0
15
30min
120min
∗
0
15
30min
∗
∗
∗
0 1530 60
3
Plasma insulin
1 0
0
Wild-type Double KO
∗∗
200 100
01530 60
3
0
mg/dL
∗
Wild-type AdipoR2 KO
ng/mL
ng/mL
0
∗ ∗
mg/dL
200 100
Wild-type AdipoR1 KO
∗∗
ng/mL
mg/dL
AdipoR1 KO
∗
120min
Plasma insulin
2 ∗∗
1 0
∗
∗
0
15
30min
AdipoR1 and AdipoR2 play physiological roles in the regulation of insulin sensitivity and glucose metabolism FIG. 7. AdipoR1 knockout (KO) and double KO mice showed insulin resistance and impaired glucose tolerance, whereas AdipoR2 KO mice showed insulin resistance without glucose intolerance (Yamauchi et al 2007).
172
KADOWAKI ET AL
in AdipoR1 knockout mice as compared with the wild-type mice. These observations indicate increased hepatic glucose production and insulin resistance in liver of AdipoR1 knockout mice. Although glucose intolerance was not observed in AdipoR2 knockout mice, plasma insulin levels were found to be significantly higher in the AdipoR2 knockout mice than in the wild-type mice (Fig. 7), suggesting the presence of insulin resistance in the AdipoR2 knockout mice. In contrast to AdipoR1 knockout mice, EGP was not significantly higher in AdipoR2 knockout mice. However, GIR was significantly decreased, and Rd tended to be decreased in AdipoR2 knockout mice. AdipoR1/R2 double knockout mice exhibited significantly impaired glucose tolerance and insulin resistance (Yamauchi et al 2007) (Fig. 7). There is a significant elevation of the insulin resistance index in the AdipoR1/R2 double knockout mice compared to the AdipoR1 knockout mice (Fig. 7), and this can be attributed to the contribution of the AdipoR2 deficiency (Fig. 7). These findings provided the first direct evidence that AdipoR1 and AdipoR2 do indeed play important physiological roles in the regulation of insulin sensitivity in vivo (Yamauchi et al 2007). Liver is a major target of adiponectin action (Berg et al 2001). We detected no appreciable adiponectin specific binding activity in the hepatocytes from AdipoR1/R2 double knockout mice (Yamauchi et al 2007) (Fig. 8, left), indicating undetectable levels of functional adiponectin receptors in hepatocytes from AdipoR1/R2 double knockout mice. Consistent with this, the glucose lowering effect of adiponectin was completely abrogated in AdipoR1/R2 double knockout mice (Yamauchi et al 2007) (Fig. 8, right). Thus, AdipoR1 and AdipoR2
Adiponectin binding activities in hepatocytes 120
100
Wild-type Double KO
110
75
100 %
%
Glucose lowering effect of adiponectin
50
Wild-type Double KO
90 80
25 0
** 2.5
**
**
8.8 15
**
**
27.5 52.5
∗
70 60
0
1
2
∗∗ 3
4
∗∗ 5
6
7
8 hr
AdipoR1 and AdipoR2 serve as the major adiponectin receptors in vivo FIG. 8. Adiponectin binding and glucose lowering effect of adiponectin were completely abrogated in AdipoR1/AdipoR2 double knockout mice (Yamauchi et al 2007).
ADIPONECTIN
173
are the major adiponectin receptors in vivo, which mediate the major, if not the entire, part of adiponectin binding and adiponectin actions. Adiponectin receptors as therapeutic targets Our data indicated that not only strategies to increase AdipoR1/R2 but also agonism of AdipoR1/R2 may be logical approaches with which to provide a novel treatment modality for insulin resistance and type 2 diabetes linked to obesity (Yamauchi et al 2007). Up-regulation of adiponectin receptors Previously, Staels’s group reported that adiponectin receptor expression in macrophages may be regulated by agonists of the nuclear receptors PPARα , PPARγ, and liver X receptor (Chinetti et al 2004). We have shown that, in KKAy mice, a PPARα agonist reversed decreases in AdipoR1 and AdipoR2 expression, which was lower in WAT of KKAy mice than in that of wild-type control KK mice (Tsuchida et al 2005). These data suggested that dual activation of PPARγ and PPARα enhanced the action of adiponectin by increasing both total and HMW adiponectin level and adiponectin receptor number, which can ameliorate obesitylinked insulin resistance. Development of adiponectin receptor agonists Osmotin, which is ubiquitous in fruits and vegetables, is a member of the pathogenesis related-5 (PR-5) family of plant defence proteins that induce apoptosis in yeast. PR-5 family proteins are extremely stable and may remain active even when in contact with the human digestive or respiratory system. Bressan’s group identified that PHO36/YOL002c, the yeast homologue of AdipoR1, is a receptor for osmotin (Narasimhan et al 2005). Moreover, X-ray crystallographic studies revealed that osmotin showed similarity to globular adiponectin in 3D structure (Narasimhan et al 2005). Interestingly, osmotin activates AMPK via adiponectin receptors in mammalian C2C12 myocytes (Narasimhan et al 2005). These data raise the possibility that further research examining similarities in adiponectin and osmotin may facilitate the development of potential adiponectin receptor agonists (Narasimhan et al 2005). Summary Adiponectin is an adipokine that exerts a potent insulin-sensitizing effect by binding to its receptors such as AdipoR1 and AdipoR2, leading to activation of
174
KADOWAKI ET AL
AMPK, PPARα , and presumably some other unknown signalling pathways. In obesity-linked insulin resistance, both adiponectin and adiponectin receptors are down-regulated. Up-regulation of adiponectin as well as up-regulation of adiponectin receptors or enhancing adiponectin receptor functions may become a versatile therapeutic strategy for obesity-linked insulin resistance, such as type 2 diabetes, the metabolic syndrome, and cardiovascular diseases.
Acknowledgments This work was supported by the Program for Promotion of Fundamental Studies in Health Sciences of the Organization for Pharmaceutical Safety and Research of Japan, a grant from the Human Science Foundation (to T. Kadowaki) and a Grant-in Aid for Creative Scientific Research (10NP0201) from the Japan Society for the Promotion of Science (to T. Kadowaki).
References Berg AH, Scherer PE 2005 Adipose tissue, inflammation, and cardiovascular disease. Circ Res 96:939–949 Berg AH, Combs TP, Du X, Brownlee M, Scherer PE 2001 The adipocyte-secreted protein Acrp30 enhances hepatic insulin action. Nat Med 7:947–953 Chinetti G, Zawadski C, Fruchart JC, Staels B 2004 Expression of adiponectin receptors in human macrophages and regulation by agonists of the nuclear receptors PPARalpha, PPARgamma, and LXR. Biochem Biophys Res Commun 314:151–158 Fruebis J, Tsao TS, Javorschi S et al 2001 Proteolytic cleavage product of 30-kDa adipocyte complement-related protein increases fatty acid oxidation in muscle and causes weight loss in mice. Proc Natl Acad Sci USA 98:2005–2010 Hirose H, Kawai T, Yamamoto Y et al 2002 Effects of pioglitazone on metabolic parameters, body fat distribution, and serum adiponectin levels in Japanese male patients with type 2 diabetes. Metabolism 51:314–317 Hotamisligil GS, Shargill NS, Spiegelman BM 1993 Adipose expression of tumor necrosis factor-alpha: direct role in obesity-linked insulin resistance. Science 259:87–91 International Diabetes Federation 2005 A new worldwide defi nition of the metabolic syndrome. http://www.idf.org/home/index.cfm?unode=32EF2063-B966-468F-928CA5682A4E3910 Iwaki M, Matsuda M, Maeda N et al 2003 Induction of adiponectin, a fat-derived antidiabetic and antiatherogenic factor, by nuclear receptors. Diabetes 52:1655–1663 Kadowaki T, Yamauchi T 2005 Adiponectin and adiponectin receptors. Endocr Rev 26:439–451 Kadowaki T, Yamauchi T, Kubota N, Hara K, Ueki K, Tobe K 2006 Adiponectin and adiponectin receptors in insulin resistance, diabetes, and the metabolic syndrome. J Clin Invest 116:1784–92 Karpichev IV, Cornivelli L, Small GM 2002 Multiple regulatory roles of a novel Saccharomyces cerevisiae protein, encoded by YOL002c, in lipid and phosphate metabolism. J Biol Chem 277:19609–19617 Kersten S, Desvergne B, Wahli W 2000 Roles of PPARs in health and disease. Nature 405:421–424
ADIPONECTIN
175
Kubota N, Terauchi Y, Yamauchi T et al 2002 Disruption of adiponectin causes insulin resistance and neointimal formation. J Biol Chem 277:25863–25866 Kubota N, Terauchi Y, Kubota T et al 2006 Pioglitazone ameliorates insulin resistance and diabetes by both adiponectin-dependent and -independent pathways. J Biol Chem 281:8748–8755. Lazar MA 2006 The humoral side of insulin resistance. Nat Med 12:43–44 Lochhead PA, Salt IP, Walker KS, Hardie DG, Sutherland C 2000 5-aminoimidazole-4carboxamide riboside mimics the effects of insulin on the expression of the 2 key gluconeogenic genes PEPCK and glucose-6-phosphatase. Diabetes 49:896–903 Maeda N, Takahashi M, Funahashi T et al 2001 PPARγ ligands increase expression and plasma concentrations of adiponectin, an adipose-derived protein. Diabetes 50:2094–2099 Mao X, Kikani CK, Riojas RA et al 2006 APPL1 binds to adiponectin receptors and mediates adiponectin signalling and function. Nat Cell Biol 8:516–523 Matschinsky FM, Magnuson MA, Zelent D et al 2006 The network of glucokinase-expressing cells in glucose homeostasis and the potential of glucokinase activators for diabetes therapy. Diabetes 55:1–12 Narasimhan ML, Coca MA, Jin J et al 2005 Osmotin is a homolog of mammalian adiponectin and controls apoptosis in yeast through a homolog of mammalian adiponectin receptor. Mol Cell 17:171–180 National Cholesterol Education Program 2001 Executive Summary of the Third Report of the National Cholesterol Education Program (NCEP) Expert Panel on Detection, Evaluation, and Treatment of High Blood Cholesterol in Adults (Adult Treatment Panel III). JAMA 285:2486–2497 Nawrocki AR, Rajala MW, Tomas E et al 2006 Mice lacking adiponectin show decreased hepatic insulin sensitivity and reduced responsiveness to peroxisome proliferator-activated receptor gamma agonists. J Biol Chem 281:2654–2660 Nolan JJ, Ludvik B, Beerdsen P, Joyce M, Olefsky J 1994 Improvement in glucose tolerance and insulin resistance in obese subjects treated with troglitazone. N Engl J Med 331:1188–1193. Okamoto Y, Kihara S, Funahashi T, Matsuzawa Y, Libby P 2006 Adiponectin: a key adipocytokine in metabolic syndrome. Clin Sci (Lond) 110:267–78 Pajvani UB, Hawkins M, Combs TP et al 2004 Complex distribution, not absolute amount of adiponectin, correlates with thiazolidinedione-mediated improvement in insulin sensitivity. J Biol Chem 279:12152–12162 Reaven GM 1998 Role of insulin resistance in human disease. Diabetes 37:1595–1607 Shulman GI 2000 Cellular mechanisms of insulin resistance. J Clin Invest 106:171–176 Steppan CM, Bailey ST, Bhat S et al 2001 The hormone resistin links obesity to diabetes. Nature 409:307–312 Tsuchida A, Yamauchi T, Ito Y et al 2004 Insulin/Foxo1 pathway regulates expression levels of adiponectin receptors and adiponectin sensitivity. J Biol Chem 279:30817–30822 Tsuchida A, Yamauchi T, Takekawa S et al 2005 Peroxisome proliferator-activated receptor (PPAR) alpha activation increases adiponectin receptors and reduces obesity-related inflammation in adipose tissue: comparison of activation of PPARalpha, PPARgamma, and their combination. Diabetes 54:3358–3370 Woods A, Azzout-Marniche D, Foretz M et al 2000 Characterization of the role of AMPactivated protein kinase in the regulation of glucose-activated gene expression using constitutively active and dominant negative forms of the kinase. Mol Cell Biol 20:6704–6711 Yamauchi T, Kamon J, Waki H et al 2001 The fat-derived hormone adiponectin reverses insulin resistance associated with both lipoatrophy and obesity. Nat Med 7:941–946 Yamauchi T, Kamon J, Minokoshi Y et al 2002 Adiponectin stimulates glucose utilization and fatty-acid oxidation by activating AMP-activated protein kinase. Nat Med 8:1288–1295
176
KADOWAKI ET AL
Yamauchi T, Kamon J, Ito Y et al 2003 Cloning of adiponectin receptors that mediate antidiabetic metabolic effects. Nature 423:762–769 Yamauchi T, Nio Y, Maki T et al 2007 Targeted disruption of AdipoR1 and AdipoR2 causes abrogation of adiponectin binding and metabolic actions. Nat Med 13:332–339 Yki-Jarvinen H 2004 Thiazolidinediones. N Engl J Med 351:1106–1118 Zhang Y, Proenca R, Maffei M, Barone M, Leopold L, Friedman JM 1994 Positional cloning of the mouse obese gene and its human homologue. Nature 372:425–432
DISCUSSION O’Rahilly: I’d be interested to know how much adiponectin deficiency might be responsible for the insulin resistance seen in lipodystrophy. Have you had a chance to study lipodystrophic models of mice, where if you restore leptin it partly restores metabolism? Would giving back adiponectin have this effect? Kadowaki: We addressed this issue in our previous paper (Yamauchi et al 2001). Insulin resistance in lipodystrophic mice was partially ameliorated by leptin replenishment, and partially ameliorated by adiponectin replenishment. It was completely ameliorated by a combination of the two. We think leptin and adiponectin are the two major insulin sensitizing factors in adipocytes. Tabas: Particularly with the PPARα effect, is there any evidence that adiponectin could be acting by delivering another molecule to the cell, rather than by directly binding to the receptor and causing a direct signalling action? Is there any evidence to suggest that adiponectin, particularly with its lipid binding pocket, might be delivering something to the cell that is responsible for at least some of its effects? Kadowaki: It is clear that some of the adiponectin effects are mediated by PPARα . However, as you suggest, we can’t exclude the possibility that PPARα activation by adiponectin is mediated via an indirect mechanism, particularly when the time-course of PPARα activation is slower than that of AMPK activation. Spiegelman: I was interested in the two-dose experiment you did with pioglitazone. With regard to these two mechanisms, this leads to a clear-cut prediction. One is a gene expression change and one is a differentiation change, and these would predict that there should be a distinct time difference in those responses. Adiponectin is presumably modulated as a gene expression change, but changes in adipocyte number and size sufficient to have an important effect presumably take days. With 10 mg/kg and 30 mg/kg, can you show a distinct time difference? Kadowaki: That is a good point. We carried out those studies as a two-week experiment. If we were to do a one week experiment, then the data might be different. Spiegelman: Your hypothesis is attractive. It would be worth fi lling it out with a consistent time-course. It makes sense.
ADIPONECTIN
177
Glass: In our analysis of the dose–response of activation of PPAR with the TZDs, we are looking at the positive regulation of a gene such as CD36 as opposed to the ability of a TZD to exert an anti-inflammatory effect. For reasons that we don’t understand there is about a threefold difference in the dose required for 50% maximal effect. Your experiment would perfectly correlate with a specific effect on activation at low doses versus adding on an effect on anti-inflammation at the higher doses. Kadowaki: It is very interesting that the anti-inflammatory action required a threefold higher dose of TZD. Spiegelman: Are you are interpreting this in terms of agonist action instead of repressive actions? Glass: Yes. I don’t understand why the dose for 50% potency should be different. Spiegelman: You are proposing a different molecular mechanism for those two pathways, so it is plausible. Glass: I would predict that you would not see much anti-inflammatory effect at the low dose and a more potent effect at the high dose. Tabas: Could the high dose effect be independent of PPARγ ? Glass: This is hard to answer because there is some effect of high-dose TZDs on PPARδ in the mouse. It also depends on the ligand. In PPARγ knockout macrophages these drugs are much less effective, even at higher doses. Most of the activity if not all of it is PPARγ dependent. Kadowaki: This point is well taken. We should include your idea in our interpretations. When we used low doses of pioglitazone, adiponectin was up-regulated but TNFα and MCP1 were not affected. A higher dose of pioglitazone not only further up-regulated adiponectin but also down-regulated insulin resistance causing adipokines such as TNFα and MCP1. Because adipocyte differentiation occurs simultaneously under these conditions, we interpreted these results as being a consequence of adipocyte differentiation. Another interpretation might be the differences in the TZD dose–response curves between the upregulation of adiponectin and down-regulation of insulin resistance causing adipokines. Glass: I think that is quite likely. Tabas: The TZDs show pretty strong sexual dimorphism in mice. Are the females more responsive? Glass: For atherosclerosis, the males are much more responsive. Tabas: In the TZD effects that you are seeing, both at low and high doses, did you notice a sexual dimorphism? Kadowaki: We did these experiments only in male mice. Carling: I wanted to ask about the mechanism underlying the activation of AMPK by adiponectin. Previously, you have shown that adiponectin activates
178
KADOWAKI ET AL
AMPK in the muscle (Yamauchi et al 2002). Have you been able to look in the adiponectin R1 and R2 knockouts at whether AMPK is activated by adiponectin? Kadowaki: We are doing that right now. The major site of insulin resistance was not in the skeletal muscle but in the liver in the AdipoR1 knockout mice. We have not yet looked at skeletal muscle. In the adiponectin knockout mice we did those experiments. They show liver insulin resistance, and decreased AMPK in the liver. But in the skeletal muscle AMPK activity was up-regulated in contrast to our original expectations. It turns out that adiponectin antagonizes the leptin effect centrally, which results in activation of leptin-induced activation of AMPK in skeletal muscle in adiponectin knockout mice. Carling: So, from these studies it appears that AdipoR1 activates AMPK. Can you show that in the liver? Is it just that AdipoR2 is very low in the liver? Kadowaki: AdipoR1 and AdipoR2 are both expressed. We have data comparing adiponectin binding and adiponectin-induced AMPK activation. We think that the inability of adiponectin to activate AMPK in AdipoR1 knockout mice is the rather specific consequence of lack of AdipoR1, since adiponectin binding was partially retained because of the presence of AdipoR2. X Li: Regarding the AMPK activation and the PPARα activation, there are some studies showing that adiponectin could increase cAMP levels in endothelial cells. We have also observed that adiponectin can increase cAMP levels in GT1-7 cells, a cell line derived from the hypothalamus that can secrete GnRH. Is cAMP possibly involved in participating in PPARα or AMPK activation pathways? Kadowaki: I think that increased cAMP in response to adiponectin has only been demonstrated in particular cell types. I don’t think this has been reported in liver or skeletal muscle. We haven’t done careful studies to see whether the adiponectin can stimulate cAMP or the downstream pathway in our experiments. Spiegelman: The cytokines and cytokine-like molecules function as an inter-regulated network in most syndromes, such as rheumatoid arthritis. I am interested that you put MCP1 in a unidirectional cascade with the other cytokines. How much do you think it is a distinct, unidirectional inflammatory pathway through MCP? Kadowaki: Your point is well taken. There are reports which argue that TNFα may up-regulate MCP1. In that sense, there is an argument that TNFα could be an initiator, but there is no evidence that overexpression of TNFα causes macrophage infi ltration in vivo. As a tentative hypothesis we put MCP1 upstream of TNFα , but the other way may also be the case. Although I didn’t show you the data, MCP1 overexpression does not seem to grossly interfere with the adiponectin pathway. Spiegelman: So it is clearly downstream of adiponectin. Chris Glass, what do you think the hierarchy is there?
ADIPONECTIN
179
Glass: There is a chicken and egg problem. It is hard to figure out what gets the ball rolling, but once the ball is rolling it tends to accelerate. Once you make a cytokine such as MCP1, it will recruit more macrophages which will then make more cytokines. It is a positive feedback loop. From a therapeutic standpoint this is interesting, because in a positive feedback loop with many links, you only have to cut one link to have a therapeutic effect. Spiegelman: In rheumatoid arthritis in human patients this hasn’t been the case. The benefit hasn’t quite been what was hoped for. Glass: You then have to wonder how many of the links are parallel links. There could be redundant factors that are involved in sequential steps. MCP1 will probably turn out to be an important chemoattractant molecule in adipose tissue, but not the only one. Interestingly, the receptor for MCP1, CCR2, is strongly negatively regulated by TZDs. This would be an interesting point to intervene. O’Rahilly: Takashi Kadowaki just told us about lipodystrophy, and it seems you don’t need to invoke any adipose tissue-based inflammatory response to adiponectin to get a beneficial response. Therefore that whole set of players is gone, yet you still get a beneficial effect of adiponectin replenishment in states of adiponectin insufficiency. Perhaps one doesn’t need to invoke additional players in linking adipose tissue dysfunction with insulin resistance. Glass: I don’t think the inflammatory mediators are necessarily initiators, but they could be amplifiers. Spiegelman: It is hard to ignore a large body of data suggesting that they are at least amplifiers. Kadowaki: Do lipotrophic mice or humans show increased serum cytokine levels? Speigelman: I don’t think anyone has looked at this. Bernlohr: R1 mediated signalling is via AMPK. Is R2 signalling also mediated by AMPK? Kadowaki: No, it is mediated by PPARα . Bernlohr: Is AMPK an intermediate? Kadowaki: We don’t think so. Muoio: Is the PGC1-driven antioxidant programme ubiquitous? Spiegelman: Yes, it has been every place we have looked. Muoio: There was a paper in Cell Metabolism some months ago by a group at Pennington that established an interaction between adiponectin and PGC1α (Civitarese et al 2006). In any of your models do you find evidence that adiponectin regulates PGC1α ? If not, perhaps you could speculate on what is driving the antioxidant programme that you described? Kadowaki: We have been studying PGC1 expression with respect to adiponectin action. So far, PGC1 induction seems to be downstream of adiponectin action. At least some of the effects mediated by adiponectin may require PGC1, and PGC1 action may mediate some of the adiponectin actions.
180
KADOWAKI ET AL
Spiegelman: I should interject here that AMPK is a strong regulator of PGC1α . Much like the action of AMPK it drives PGC1α in different directions in different tissues. Muoio: Is there strong evidence that AMPK is a potent activator/regulator? Spiegelman: Yes, we have some unpublished data, but what is in the literature now is that it induces PGC1α in muscle. Muoio: The data that I have seen are not so convincing; AMPK-mediated induction of PGC1α mRNA appears to be relatively modest. Spiegelman: AMPK modifies PGC1α . It does turn on its expression, but it also modifies the protein. It is a two to threefold effect in these papers, but I can tell you that there is a lot more going on than just the induction of the gene. Shi: I want to raise the issue of adiponectin resistance. It is a very different concept from traditional resistance, such as insulin resistance that is always associated with hyperinsulinaemia. In adiponectin resistance you only see the downregulation of the receptor expression. Is there down-regulation of the signal transduction pathways? Those mice are still very responsive to adiponectin. You talk about a therapeutic target. Let’s say that adiponectin indeed improves insulin sensitivity. It is a large and abundant protein. How can it be trimmed into smaller active fragments? Kadowaki: Ob/Ob mice show down-regulation of their adiponectin receptors. This is significant in skeletal muscle and white adipose tissue, while it is relatively modest in the liver (40–50%). When we injected adiponectin, it stimulated AMPK activity and the activation was blunted in ob/ob mice. We don’t know whether this is because the receptor is down-regulated, or whether there is some specific post receptor defect. We haven’t been able to do a careful dose–response curve experiment to determine the half-maximal stimulation dose in glucose lowering in wildtype and adiponectin resistant mice. Shi: When I asked about the lack of adiponectinaemia, I was referring to hyperinsulinaemia associated with insulin resistance. Kadowaki: It is different from the usual insulin-resistance state, which is characterized by hyperinsulinaemia. Obesity is accompanied by hypoadiponectinaemia and partial, but not complete, adiponectin resistance. Adiponectin administration should have some effect. Spiegelman: It is also so abundant to begin with: adiponectin is circulating at ridiculously high levels. Bernhlohr: If it is so abundant, in your adiponectin null animals is there reduced ER stress or ER burden? Kadowaki: Plasma adiponectin levels are not up-regulated in the adiponectin receptor knockout mice. It is possible that AdipoR1 and R2, even though they bind adiponectin and mediate its biological functions, are not receptors for mediating its degradation or internalization. Until I listened to Ira Tabas’ paper, I didn’t know
ADIPONECTIN
181
the connection between ER stress and adiponectin. I’m interested in this now. Also, the most active form of adiponectin is the high molecular weight form. Usually, this molecule exists as a 12-mer or 18-mer, but according to our analysis the 18-mer is the predominant form. Even though adiponectin concentrations are high when it is measured as a monomer, the effective molar concentration will be something like one twentieth of this. Spiegelman: One of the most important issues relates to cancer and carcinogenesis, although this is not something we are concentrating on here. I work at a cancer institute and I have spent years trying to interest colleagues in metabolic disease. On the other hand, for this group here, the connection between metabolic disease and cancer is a time bomb that we hardly ever discuss. Within the next decade it will become the most important preventable form of human cancer. Tabas: Can you be more specific? What connection are you talking about? Spiegelman: People with obesity and obesity-linked metabolic disease have a greatly increased frequency of breast cancer, colon cancer and prostate cancer, such that the American Cancer Society have declared that obesity is a primary interest of theirs. As cigarette smoking has reduced in the USA, obesity is exceeding cigarette smoking as a cause of human cancer. Glass: It is interesting that two of the three cancers you mentioned are characterized by inflammatory infi ltrates. At least in the case of colon cancer there is evidence that knocking out the IKK pathway reduces experimental models of colon cancer. In prostate cancer the inflammatory infi ltrate correlates tightly with when prostate cancers become resistant to anti-androgen therapy. The connection may have something to do with inflammation. Spiegelman: I would urge you to take an interest in the connection to carcinogenesis in some of your animal models that are lacking adiponectin. O’Rahilly: Homo sapiens is a neglected experimental model. Pretty much all our patients with severe insulin resistance who have had colonoscopies have got polyposis, irrespective of the cause of the resistance. This is probably the result of the trophic effect of massive hyperinsulinaemia, either through the IGF receptor or a hybrid receptor. Spiegelman: Or through the inflammatory route. O’Rahilly: These people have mutations in the insulin receptor and we don’t see much inflammation. The inflammatory hypothesis has become so dominant in our thinking it may obfuscate some of the more traditional models. Insulin is a good survival factor for these cells. In some of these patients we know the initiating factor in their hyperinsulinaemia and it doesn’t involve inflammation. Kadowaki: Bruce, you raise a very important question. Our working hypothesis is that increased adiponectin linked to obesity may play a role in obesity-linked carcinomas. We have been doing some experiments in collaboration with surgeons at Tokyo University Medical School, observing decreased proliferation and
182
KADOWAKI ET AL
increased apoptosis in certain cancer cell lines. We then went on to put these carcinomas in nude mice and have been able to show the growth-inhibitory effect of adiponectin treatment. There is a significant reduction of tumour size in nude mice by adiponectin. Spiegelman: There was a controversy about PPARγ and cancer. I think we eventually showed in genetic models that PPARγ is an anticancer agent. A model you should look at is the azoxymethane-induced colon cancer, which can easily applied to any genetic model you have for adiponectin. Kadowaki: We previously published a paper using this method with PPARγ heterozygote knockout mice, which showed increased carcinogenesis (Lu et al 2005). Tabas: As a quick segue, that raises an unanswered question: someone asked about therapeutic approaches. Shi: Adiponectin is a large protein and would cost a fortune to make. Spiegelman: If PPARγ regulates it, instead of a new therapeutic modality, if this is all true and plays out we have molecular mechanistic understanding of a therapy that was discovered in Japan in the 1970s. PPARγ ligands regulate this pretty well already. Shi: If you could make a smaller protein analogue that did the same thing, this would make adiponectin a possible therapeutic protein. Kadowaki: For the exact reasons that you pointed out, we are not working on a direct injection strategy for adiponectin treatment. Instead we are screening for adiponectin receptor agonists. Also, down-regulation of adiponectin is a major mechanism. We now have some evidence that increased inflammation and oxidative stress may play a major role in the down-regulation of plasma adiponectin, in particular the high molecular weight form. If we can reduce inflammation or oxidative stress by some means, this will up-regulate plasma adiponectin and in particular the high molecular weight form. O’Rahilly: I know a way of increasing adiponectin 10-fold in humans: giving people anti-insulin receptor antibodies. This has problems, but it is telling us something important about what is regulating adiponectin. Whatever the schema we have, it has to take into account that blockading insulin action in humans is the most potent way of increasing adiponectin. References Civitarese AE, Ukropcova B, Carling S 2006 Role of adiponectin in human skeletal muscle bioenergetics. Cell Metab 4:75–87 Lu J, Imamura K, Nomura S et al 2005 Chemopreventive effect of peroxisome proliferatoractivated receptor gamma on gastric carcinogenesis in mice. Cancer Res 65:4769–4774 Yamauchi T, Kamon J, Waki H et al 2001 The fat-derived hormone adiponectin reverses insulin resistance associated with both lipoatrophy and obesity. Nat Med 7:941–946 Yamauchi T, Kamon J, Minokoshi Y et al 2002 Adiponectin stimulates glucose utilization and fatty-acid oxidation by activating AMP-activated protein kinase. Nat Med 8:1288–1295
Fatty Acids and Lipotoxicity in Obesity and Diabetes: Novartis Foundation Symposium 286, Volume 286. Edited by Gregory Bock and Jamie Goode Copyright Novartis Foundation 2007. ISBN: 978-0-470-05764-3
Anti-inflammatory and antidiabetic roles of PPARγ Gabriel Pascual*§, Amy L. Sullivan*§, Sumito Ogawa*, Amir Gamliel†, Valentina Perissi†, Michael G. Rosenfeld†‡ and Christopher K. Glass*‡1 * Department of Cellular and Molecular Medicine, †Howard Hughes Medical Institute, ‡ Department of Medicine, § Biomedical Sciences Graduate Program, University of California, San Diego, 9500 Gilman Dr., La Jolla, CA 92093, USA
Abstract. The peroxisome proliferator-activated receptor γ (PPARγ ) regulates adipocyte differentiation and glucose homeostasis and is the molecular target of thiazolidinediones (TZDs) that act as insulin-sensitizers in patients with type 2 diabetes. PPARγ is also expressed in macrophages and negatively regulates the programme of macrophage activation by repressing a subset of AP1 and NF-κ B-dependent genes. Recent genetic, molecular and biochemical studies support the idea that PPARγ inhibits inflammatory gene expression in activated macrophages by a NCoR/sumoylation-dependent pathway. Sumoylation of PPARγ targets it to NCoR corepressor complexes that are bound to inflammatory response gene promoters and prevents their signal-dependent clearance that is normally a prerequisite for transcriptional activation. As a consequence, genes remain in a repressed state. Because the ligand-induced allosteric changes that promote entry of PPARγ into this transrepression pathway are distinct from those that mediate interactions with conventional coactivators, these fi ndings may facilitate the development of novel PPARγ ligands that retain antidiabetic activities but have reduced side effects. 2007 Fatty acids and lipotoxicity in obesity and diabetes. Wiley, Chichester (Novartis Foundation Symposium 286) p 183–199
PPARγ, macrophages and insulin resistance Inflammation is increasingly recognized as a central component of insulin resistance (Lazar 2005, Pickup & Mattock 2003, Pickup et al 1997, Shoelson et al 2003, Wellen & Hotamisligil 2003, 2005). Pharmacological or genetic inhibition of pathways that underlie inflammatory responses protect experimental animals from diet-induced insulin resistance, suggesting a direct pathogenic role of inflamma1
This paper was presented at the symposium by Christopher K. Glass, to whom correspondence should be addressed 183
184
PASCUAL ET AL
tion in promoting an insulin-resistant state (Arkan et al 2005, Hirosumi et al 2002, Hundal et al 2002, Pickup & Mattock 2003, Yuan et al 2001). While adipocytes themselves are sources of proinflammatory cytokines, recent studies indicate that macrophages are present in much higher numbers in adipose tissue of obese, insulin resistant animals than in adipose tissue of lean animals, and may be the major source of some of the inflammatory mediators that are linked to insulin resistance (Weisberg et al 2003, Xu et al 2003). Recent studies demonstrated that myeloid-specific disruption of the IKK β gene in macrophages, the major I-κ B kinase involved in activation of NF-κ B, resulted in protection from diet-induced insulin resistance (Arkan et al 2005). These findings are consistent with the hypothesis that the macrophage response to inflammatory stimuli through activation of NF-κ B is a determinant of insulin resistance. The peroxisome proliferator-activated receptor γ (PPARγ ) is the molecular target of thiazolidinediones (TZDs) (Lehmann et al 1995), which are insulinsensitizing drugs (Nolan et al 1994), but the mechanisms responsible for their anti-diabetic actions remain poorly understood. TZDs are likely to exert insulinsensitizing effects by both positively and negatively regulating gene expression in several tissues (Picard & Auwerx 2002, Spiegelman 1998, Willson et al 2001), with adipose tissue being a major anatomical site of PPARγ expression. Positive regulation of gene expression in adipocytes drives both adipogenic and insulin-sensitizing programmes of gene expression, but these gene networks remain poorly described and the ability to discriminate subsets of PPARγ -sensitive genes with selective PPARγ modulators is largely unexplored. PPARγ agonists inhibit the production of numerous inflammatory mediators in macrophages (Welch et al 2003) and antagonize the effects of tumour necrosis factor α (TNFα ) in adipocytes (Iwata et al 2001, Ruan et al 2003, Souza et al 1998), suggesting that these actions may contribute to their insulin-sensitizing activities. Mediators of insulin resistance Adipose tissue secretes a large number of proteins that influence insulin action. Pro-inflammatory mediators that have been linked to insulin resistance include TNFα , resistin, interleukin (IL)6, MCP1, PAI1 and angiotensin (reviewed in Lazar 2005, Wellen & Hotamisligil 2005). The relative contributions of adipocytes and macrophages to the production of cytokines that are responsible for insulin resistance has not been clearly resolved, but macrophages appear to be responsible for a significant fraction of the TNFα produced in adipose tissue in the mouse. Many of these factors regulate programmes of gene expression in adipocytes, macrophages and other cell types through signalling pathways that lead to activation of JNK and IKK. Each of these kinases can act directly on insulin signalling pathways by phosphorylating IRS1 and inhibiting its ability to transduce insulin recep-
PPARγ
185
tor signalling (Gao et al 2002, 2003, Hotamisligil et al 1994, 1996). JNK and IKK indirectly contribute to insulin resistance by regulating AP1 and NF-κ Bdependent programmes of gene expression, respectively, that amplify inflammatory responses. In addition to cytokines, free fatty acids have been shown to exert proinflammatory effects, in part by activating JNK1 and IKKβ in adipocytes and endothelial cells (Gao et al 2002, 2004, Sinha et al 2004). The impact of free fatty acids is potentially complex, however, because their metabolites may also serve as activating ligands for PPARs that exert anti-inflammatory roles (Forman et al 1997, 1995, Ziouzenkova et al 2003). Other potential contributions to insulin resistance include production of reactive oxygen species (Houstis et al 2006) and endoplasmic reticulum stress, each of which can lead to IKK and JNK activation (Wellen & Hotamisligil 2005). Roles of NCoR in the control of inflammation Signal-dependent induction of inflammatory response genes requires the activation of transcription factors such as NF-κ B, their binding to specific regulatory elements in target genes, and the recruitment of coactivator proteins that are required for transcriptional activation. Recent studies of the nuclear receptor corepressor NCoR and the related factor SMRT indicate that for many inflammatory response genes, there is an additional derepression step that is a prerequisite to transcriptional activation (Hoberg et al 2004, Ogawa et al 2004, Perissi et al 2004). NCoR and SMRT were initially identified as proteins that interact with unliganded retinoic acid and thyroid hormone receptors and mediate their active repression functions (Chen & Evans 1995, Horlein et al 1995). Addition of agonists results in dissociation of NCoR and SMRT from retinoic acid or thyroid hormone receptors in exchange for coactivator complexes and a consequent switch in transcriptional function from repression to activation (Glass & Rosenfeld 2000). NCoR and SMRT have been found to be present in cells in several complexes, including a major complex containing HDAC3, Tbl1, TblR1 and GPS2 (Guenther et al 2000, Li et al 2000, Yoon et al 2003, Zhang et al 2002). The Tbl1 and TblR1 components of this complex appear to contribute to stability of binding to chromatin through histone interactions, with the HDAC3 component contributing to active repression through histone deacetylase activity (Yoon et al 2003). TblR1 has also recently been suggested to play a critical role in the clearance of this complex from nuclear receptor target genes by directing its proteosome-dependent proteolysis involving the ubiquitin-conjugating enzyme UbcH5 (Perissi et al 2004). To identify endogenous transcriptional programmes regulated by NCoR in macrophages, we used gene expression profi ling to compare wild type and NCoR−/− macrophages (Ogawa et al 2004). Unexpectedly, these studies revealed that NCoR acts as a transcriptional checkpoint for NF-κ B and AP1-dependent gene
186
PASCUAL ET AL
networks that regulate diverse biological processes including inflammation, cell migration and collagen catabolism, with loss of NCoR resulting in derepression of AP1 and NF-κ B target genes. Derepressed genes included chemokines, chemokine receptors and matrix metalloproteinases that are linked to the development of insulinresistance and atherosclerosis. Consistent with this pattern of altered gene expression, NCoR−/− macrophages exhibited a partially activated phenotype, exemplified by enhanced ability to migrate through a gelatin barrier (Ogawa et al 2004). These findings raise the possibility that NCoR complexes play key roles in controlling inflammatory transcriptional programmes that contribute to the development of insulin resistance and its clinical consequences. Genome-wide analysis of PPARg target genes in macrophages We utilized gene expression profi ling to identify transcriptional targets of PPARγ in primary mouse peritoneal macrophages. In contrast to adipocytes, in which PPARγ is a potent activator of gene expression, relatively few genes were induced by TZDs in macrophages (Fig. 1a). A significant fraction of the positively regulated
a
Rosi-inducible >2x control
b
7/8500 genes
FIG. 1. Positive and negative regulation of macrophage gene expression by the PPARγ agonist rosiglitazone. (a) Mouse peritoneal macrophages were treated with 1 µ M rosiglitazone or control vehicle for 16 hours. RNA was isolated and subjected to microarray analysis using Codelink Mouse Uniset I microarrays. Gene expression values are plotted for the two treatment conditions. (b) Mouse peritoneal macrophages were treated with LPS (0.1 mg/ml) for six hours in the presence or absence of rosiglitazone (1 µ M). RNA was isolated and analysed using Codelink Mouse Uniset I microarrays as in (a). Expression profi les for the 311 genes induced more than fivefold by LPS are plotted on the left. Expression profi les for 108 of these genes exhibiting significant repression by rosiglitazone are plotted on the right. Panel (a) is adapted from Welch et al (2003).
PPARγ
187
genes are involved in lipid transport and metabolism, including CD36 and ABCG1. This functionally related set of genes might play important roles in supporting macrophage energy requirements and maintaining overall lipid homeostasis. To examine the role of PPARγ in negative regulation of gene expression, we investigated the ability of PPARγ agonists to inhibit transcriptional responses to lipopolysaccharide (LPS), which initiates inflammatory responses through activation of toll-like receptor 4 (TLR4) (Akira & Takeda 2004) and poly I : C, which initiates inflammatory responses through activation of TLR3 (Takeda & Akira 2004). Systemic administration of LPS to humans and experimental animals induces an insulin-resistant state (Agwunobi et al 2000, Virkamaki & Yki-Jarvinen 1994) and recent studies suggest roles of TLR4 in mediating inflammatory responses to free fatty acids (FFAs). An example of an experiment evaluating LPS responses of primary mouse peritoneal macrophages to the TLR4 ligand LPS in the presence and absence of the PPARγ agonist rosiglitazone is illustrated in Fig. 1b, c. In this experiment, approximately a third of the more than 300 strongly induced (>5-fold) LPS target genes were significantly repressed by rosiglitazone. The finding that many of the LPS responsive genes were sensitive to PPARγ mediated repression, while others were PPARγ -resistant, raised a number of biological and mechanistic questions. At a mechanist level, we wish to understand how PPARγ discriminates between sensitive and resistant genes. To explore whether the mechanisms are conserved with other nuclear receptors, we performed a comparative analysis of the impact of six different nuclear receptor ligands on transcriptional responses of macrophages to LPS and the TLR3specific ligand, poly I : C. These studies demonstrated that GR, LXR and PPARγ repress overlapping but distinct sets of TLR-inducible genes (Ogawa et al 2005). These results imply the utilization of distinct mechanisms for transrepression by each receptor. Of particular interest, the pattern of repression was different when macrophages were stimulated with the TLR3 ligand, poly I : C, than when stimulated by LPS. Together, these experiments indicated that nuclear receptors inhibit the TLR-dependent programme of gene expression in a receptor, promoter, and signal-dependent manner (Ogawa et al 2005). The NCoR/HDAC3/TBL complex is a target of PPARg -mediated transrepression A major goal of our group has been to define the molecular mechanisms responsible for PPARγ -dependent inhibition of inflammatory responses in macrophages. We have studied the inducible nitric oxide synthase (iNOS) gene as a model because it is one of several inflammatory response genes expressed by macrophages that are linked to the pathogenesis of insulin resistance (Kuhlencordt et al 2001, Perreault & Marette 2001). The iNOS gene is strongly induced in macrophages by
188
PASCUAL ET AL
lipopolysaccharide (LPS) (Lowenstein et al 1993) and is negatively regulated by PPARγ agonists in a PPARγ -dependent manner (Ricote et al 1998, Welch et al 2003). The observation that NCoR complexes are bound to a subset of inflammatory response genes raised the possibility that they might be involved in repression by PPARγ. Consistent with this possibility, inhibition of NCoR expression using an NCoR-specific siRNA resulted in a complete reversal of iNOS transrepression by rosiglitazone but not by a control siRNA (Fig. 2). Intriguingly, similar results were obtained utilizing a selective PPARγ modulator (GW0072) that is effective at dissociating NCoR from PPARγ, but only weakly stimulates interactions with nuclear receptor coactivators (Oberfield et al 1999) (Fig. 2). Chromatin immunoprecipitation (ChIP) experiments confirmed that NCoR, HDAC3, TBL1 and TBLR1 were present on the iNOS promoter under basal conditions and that the NCoR and HDAC3 components of this complex were cleared following LPS stimulation (Fig. 3). However in cells treated with rosiglitazone or GW0072, both NCoR and HDAC3 remained on the iNOS promoter after LPS stimulation (Fig. 3). Time course experiments demonstrated that the E2 ligase Ubc5 was rapidly recruited to the iNOS promoter following LPS stimulation in the absence of rosiglitazone, coincident with loss of NCoR. In contrast, Ubc5 was not recruited to the promoter in the presence of Ro. These results suggest that PPARγ acts to repress LPS induction of the iNOS gene by preventing recruitment of the Ubc5/19S proteosome machinery required for the clearance of NCoR and HDAC3. ChIP experiments performed in primary macrophages evaluating both the iNOS promoter and the CD36 promoter, which is positively regulated by PPARγ (Tontonoz et al 1998), further revealed that PPARγ was recruited to both promoters in a ligand-dependent manner. As expected, the p65 component of NF-κ B was recruited exclusively to the iNOS promoter in response to LPS, and
LPS Ro GW0072 Lamin siRNA NCoR siRNA
-
+ -
+ + -
+ + + -
+ + +
+ + -
+ + + -
+ + +
+
+ +
iNOS GAPDH
FIG. 2. NCoR is required for PPARγ -dependent repression of iNOS activation. Primary mouse peritoneal macrophages were treated with LPS in the presence or absence of rosiglitazone (Ro) or the selective PPARγ modulator GW0072. RNA was isolated after 6 hours and analyzed for iNOS and GAPDH expression by northern blotting. From Pascual et al (2005).
PPARγ
189 -159
-15 +1 iNOS gene κB element
LPS Ro GW0072
-
ChIP - iNOS + + + + + -
10% Input IgG αNCoR αHDAC3 αTBL1 αTBLR1
FIG. 3. Rosiglitazone (Ro) and GW0073 inhibit LPS-induced release of NCoR from the iNOS promoter. Primary mouse peritoneal macrophages were treated with LPS in the presence or absence of Ro or the selective PPARγ modulator GW0072. Cells were fi xed with formaldehyde after one hour and occupancy of the iNOS promoter by NCoR, HDAC3, Tbl1 and TblR1 was evaluated by ChIP assay. Data are from Pascual et al (2005).
this recruitment was not affected by rosiglitazone treatment. Similar results were observed for four other LPS-inducible genes that also exhibited increased levels of expression under unstimulated conditions in NCoR-deficient macrophages, including Ccl7, Cxcl10 and Tgtp. Although not all PPARγ -sensitive promoters exhibited this pattern, these findings suggest that prevention of NCoR clearance is a broadly used mechanism mediating repressive actions of PPARγ. Ligand-dependent sumoylation of PPARg is required for transrepression The observation that ligand-dependent recruitment of PPARγ to LPS-responsive promoters required NCoR raised a paradox, because the binding of rosiglitazone or GW0072 normally disrupts direct interactions between NCoR and PPARγ (Oberfield et al 1999). To identify PPARγ -interacting proteins that might potentially resolve this paradox, a yeast two-hybrid screen was performed using a library constructed from mRNA derived from primary macrophages. One of the clones isolated in this screen encoded the initial 208 amino acids of PIAS1 (protein inhibitor of activated STAT1), initially identified as a suppressor of interferon-dependent transcription (Liu et al 1998). PIAS1 belongs to a family of sumo E3 ligases that contain a central RING domain shown to be important for sumo ligase activity
190
PASCUAL ET AL
(Jackson 2001). Co-immunoprecipitation experiments using antibodies directed against endogenous PPARγ and PIAS1 demonstrated a basal interaction in primary macrophages that was enhanced by treatment with rosiglitazone. Sumoylation of transcription factors has previously been correlated with impaired transcriptional activation and/or transcriptional repression (Kotaja et al 2002, Ling et al 2004, Nishida & Yasuda 2002). To investigate whether PIAS1 was involved in PPARγ -mediated transrepression, PIAS1-specific siRNAs were developed that achieved greater than 80% reduction of PIAS1 mRNA and protein in both primary macrophages and RAW264.7 cells. QT-PCR analysis revealed that siRNA-mediated knockdown of PIAS1 in primary macrophages abolished PPARγ transrepression of the endogenous iNOS gene (Fig. 4a). Knockdown of PIAS1 expression also prevented the recruitment of PPARγ to the iNOS promoter following rosiglitazone treatment, suggesting that a sumoylation-dependent step is required for PPARγ to bind to the NCoR complex and prevent corepressor a Normalized mRNA
90 60 30 0
LPS Rosi si cont siNCoR siPIAS1
-
+ -
+ + + -
+ + + -
+ + +
-
+ -
+ + + -
+ + + -
+ + +
120
Normalized mRNA
b
iNOS
120
90 60 30 0
LPS Rosi si cont siNCoR siUbc9
FIG. 4. PIAS1 and Ubc9 are required for ligand-dependent transrepression of the iNOS gene. Primary mouse peritoneal macrophages were transfected with control siRNAs or siRNAs directed against PIAS1 (panel a) or Ubc9 (panel b). Forty eight hours later, cells were treated with LPS in the presence or absence of rosiglitazone (Ro). RNA was isolated 6 hours later and expression of iNOS mRNA was determined by quantitative PCR.
PPARγ
191
clearance. The sumoylation pathway involves a rate-limiting E2 ligase, Ubc9, which catalyses the transfer of SUMO groups to target proteins, a reaction that is thought to be stabilized and assisted by E3 ligase family members such as PIAS1 (Muller et al 2001). Ultimately, the SUMO group is conjugated to a molecular target, resulting in a specific functional outcome for that substrate (Jackson 2001). To examine whether these enzymatic activities were required for PPARγ dependent transrepression, primary macrophages were transfected with a siRNA pool directed against murine Ubc9. Knockdown of Ubc9 expression resulted in loss of PPARγ -mediated transrepression of the endogenous iNOS gene as determined by QT-PCR (Fig. 4b). Recent reports have indicated that PPARγ can be sumoylated within the Nterminal AF1 domain of PPARγ 2 at K107 (equivalent to K77 of PPARγ1) and that this modification inhibits PPARγ activity on positively-regulated target genes (Ohshima et al 2004). Examination of the primary amino acid sequence of mouse and human PPARγ revealed an additional sumoylation consensus sequence ψKXE/ D (Muller et al 2001) at K365 (corresponding to K367 in the human PPARγ sequence). Intriguingly, crystal structures of the apo and rosiglitazone-bound forms of PPARγ indicated that the primary amine group of K365 was orientated towards the interior of the LBD in the apo form, but solvent exposed in the rosiglitazone-bound form. A similar shift occurs in the structure of PPARγ bound to GW0072 (Oberfield et al 1999). Because this amino group is the point of covalent attachment of sumo, the PPARγ crystal structures suggested that K365 would be sumoylated in a ligand-dependent manner. To test this hypothesis, K365 of PPARγ was mutated to arginine and the wild-type and mutant proteins were tested for sumoylation in vivo and in vitro. Wild-type PPARγ was found to exhibit a basal level of sumoylation by Myc-tagged Sumo1 that was significantly enhanced following treatment with rosiglitazone. In contrast, the basal level of sumoylation of PPARγK365R was not enhanced by rosiglitazone treatment. Most of the ligandindependent sumoylation was abolished by mutation of PPARγ K77 to arginine (data not shown), consistent with previous findings (Ohshima et al 2004). Functional studies in RAW264.7 cells demonstrated that PPARγK365R was defective for inhibition of the iNOS promoter, while PPARγK77R retained full transrepression activity. PPARγK77R exhibited enhanced transactivation function consistent with previous findings (Ohshima et al 2004), while PPARγK365R exhibited approximately the same activity as wild-type PPARγ on the positively regulated Aox-TK luciferase promoter. ChIP assays demonstrated that wild type PPARγ and PPARγK77R were efficiently recruited to the iNOS promoter in response to rosiglitazone, while PPARγK365R was not. In contrast, wild-type PPARγ and each of the PPARγ mutants were recruited to the positively regulated CD36 promoter. In concert, these studies define a novel pathway mediating ligand-dependent transrepression of inflammatory response genes by PPARγ in macrophages (Fig.
192
PASCUAL ET AL
5). Genes subject to transrepression by this pathway are marked in the basal state by the presence of NCoR/HDAC3/TBL corepressor complexes. In the absence of a PPARγ agonist, LPS signalling results in the clearance of the NCoR and HDAC3 components of this complex in a TBL1-, TBLR1- and Ubc5-dependent manner, consistent with the clearance mechanism recently described for nuclear receptor target genes (Perissi et al 2004). Activation is achieved by exchange of the NCoR corepressor complex for coactivator complexes that mediate transcriptional activation. The PPARγ -regulated transrepression pathway is initiated by ligand-induced site-specific sumoylation of the ligand-binding domain. This specific sumoylation event targets PPARγ to NCoR complexes associated with LPS target genes and blocks the ability of Ubc5 to be recruited in response to LPS
De-repression
Activation 19S Proteasome
HDAC3
Ubc5
NCoR
TBL1 TBLR1
LPS
HDAC3
Coactivator Complexes
p65 p50
NCoR TBL1 TBLR1
iNOS
TBL1 TBLR1
Repression checkpoint
Activation
LPS + Ro 19S Proteasome
Ubc9 PIAS1
Ro
Ubc5
PIAS1 K
Su
Su
Su HDAC3
Transrepression
p65 p50
NCoR
PPARγ
TBL1 TBLR1
Trans-repression
FIG. 5. Model for LPS activation and PPARγ -dependent repression of the iNOS gene. See text for explanation. Adapted from Pascual et al (2005).
PPARγ
193
signals. As a consequence, NCoR complexes are not cleared from the promoter, resulting in a failure to relieve the repression checkpoint and maintenance of these genes in a repressed state (Pascual et al 2005). These fi ndings are of potential importance with respect to developing new drugs that regulate PPARγ activity because the ligand-induced allosteric change in receptor conformation that results in sumoylation is distinct from previously described allosteric changes that are involved in regulating coactivator or corepressor interactions. This is illustrated by the ability of both rosiglitazone, a conventional agonist, and GW0072, a partial agonist/antagonist that only weakly recruits coactivators, to fully activate this sumoylation pathway. It should therefore be possible to fully separate transactivation and transrepression activities of ligands to allow investigation of the relative importance of these two activities in various disease states, including insulin resistance and diabetes. Conclusions It is now clear that inflammation is a central component of insulin resistance in the setting of obesity, and that inhibition of inflammation is of therapeutic benefit in animals and humans. It is likely that a combination of cytokines, lipids and other mediators contribute to the insulin-resistant state. As a consequence, efforts to restore insulin sensitivity are more likely to be effective if they target multiple components of the inflammatory response, rather than specific mediators such as TNFα , MCP1 or resistin. At the same time, the therapeutic goal of inhibiting inflammation to promote insulin sensitivity cannot come at the cost of impaired immunity. Based on the growing evidence that TZDs act as insulin sensitizers, in part by counter-regulating obesity-induced inflammatory responses in macrophages, it will be important to define the molecular mechanisms that mediate transrepression of inflammatory response genes by PPARγ. The identification of these mechanisms will not only be of broad interest with respect to understanding underlying mechanisms that influence the development of insulin resistance, but will be directly relevant to the development of improved drugs for its prevention and treatment.
References Agwunobi AO, Reid C, Maycock P, Little RA, Carlson GL 2000 Insulin resistance and substrate utilization in human endotoxemia. J Clin Endocrinol Metab 85:3770–3778 Akira S, Takeda K 2004 Toll-like receptor signalling. Nat Rev Immunol 4:499–511 Arkan MC, Hevener AL, Greten FR et al 2005 IKK-beta links inflammation to obesity-induced insulin resistance. Nat Med 11:191–198 Chen JD, Evans RM 1995 A transcriptional co-repressor that interacts with nuclear hormone receptors. Nature 377:454–457
194
PASCUAL ET AL
Forman BM, Tontonoz P, Chen J, Brun RP, Spiegelman BM, Evans RM 1995 15-Deoxy- δ12, 14–prostaglandin J2 is a ligand for the adipocyte determination factor PPARγ. Cell 83:803–812 Forman BM, Chen J, Evans RM 1997 Hypolipidemic drugs, polyunsaturated fatty acids, and eicosanoids are ligands for peroxisome proliferator-activated receptors α and δ. Proc Natl Acad Sci USA 94:4312–4317 Gao Z, Hwang D, Bataille F et al 2002 Serine phosphorylation of insulin receptor substrate 1 by inhibitor kappa B kinase complex. J Biol Chem 277:48115–48121 Gao Z, Zuberi A, Quon MJ, Dong Z, Ye J 2003 Aspirin inhibits serine phosphorylation of insulin receptor substrate 1 in tumor necrosis factor-treated cells through targeting multiple serine kinases. J Biol Chem 278:24944–24950 Gao Z, Zhang X, Zuberi A et al 2004 Inhibition of insulin sensitivity by free fatty acids requires activation of multiple serine kinases in 3T3-L1 adipocytes. Mol Endocrinol 18:2024–2034 Glass CK, Rosenfeld MG 2000 The coregulator exchange in transcriptional functions of nuclear receptors. Genes Dev 14:121–141 Guenther M, Lane W, Fischle W, Verdin E, Lazar M 2000 A core SMRT corepressor complex containing HDAC3 and TBL1, a WD40-repeat protein linked to deafness. Genes Dev 14:1048–1057 Hirosumi J, Tuncman G, Chang L et al 2002 A central role for JNK in obesity and insulin resistance. Nature 420:333–336 Hoberg JE, Yeung F, Mayo MW 2004 SMRT derepression by the Iκ B kinase α : a prerequisite to NF-κ B transcription and survival. Mol Cell 16:245–255 Horlein AJ, Naar AM, Heinzel T et al 1995 Ligand-independent repression by the thyroid hormone receptor mediated by a nuclear receptor co-repressor. Nature 377:397–404 Hotamisligil GS, Murray DL, Choy LN, Spiegelman BM 1994 Tumor necrosis factor alpha inhibits signaling from the insulin receptor. Proc Natl Acad Sci USA 91:4854–4858 Hotamisligil GS, Peraldi P, Budavari A, Ellis R, White MF, Spiegelman BM 1996 IRS-1-mediated inhibition of insulin receptor tyrosine kinase activity in TNF-alpha- and obesity-induced insulin resistance. Science 271:665–668 Houstis N, Rosen ED, Lander ES 2006 Reactive oxygen species have a causal role in multiple forms of insulin resistance. Nature 440:944–948 Hundal RS, Petersen KF, Mayerson AB et al 2002 Mechanism by which high-dose aspirin improves glucose metabolism in type 2 diabetes. J Clin Invest 109:1321–1326 Iwata M, Haruta T, Usui I et al 2001 Pioglitazone ameliorates tumor necrosis factor-alphainduced insulin resistance by a mechanism independent of adipogenic activity of peroxisome proliferator activated receptor-gamma. Diabetes 50:1083–1092 Jackson PK 2001 A new RING for SUMO: wrestling transcriptional responses into nuclear bodies with PIAS family E3 SUMO ligases. Genes Dev 15:3053–3058 Kotaja N, Karvonen U, Janne OA, Palvimo JJ 2002 PIAS proteins modulate transcription factors by functioning as SUMO-1 ligases. Mol Cell Biol 22:5222–5234 Kuhlencordt PJ, Chen J, Han F, Astern J, Huang PL 2001 Genetic deficiency of inducible nitric oxide synthase reduces atherosclerosis and lowers plasma lipid peroxides in apolipoprotein E-knockout mice. Circulation 103:3099–3104 Lazar MA 2005 How obesity causes diabetes: not a tall tale. Science 307:373–375 Lehmann JM, Moore LB, Smith-Oliver TA, Wilkison WO, Willson TM, Kliewer SA 1995 An antidiabetic thiazolidinedione is a high affinity ligand for peroxisome proliferator activated receptor γ (PPARγ ). J Biol Chem 270:12953–12956 Li J, Wang J, Nawaz Z, Liu J, Qin J, Wong J 2000 Both corepressor proteins SMRT and N-CoR exist in large protein complexes containing HDAC3. EMBO J 19:4342–4350 Ling Y, Sankpal UT, Robertson AK, McNally JG, Karpova T, Robertson KD 2004 Modification of de novo DNA methyltransferase 3a (Dnmt3a) by SUMO-1 modulates its interaction
PPARγ
195
with histone deacetylases (HDACs) and its capacity to repress transcription. Nucleic Acids Res 32:598–610 Liu B, Liao J, Rao X et al 1998 Inhibition of Stat1-mediated gene activation by PIAS1. Proc Natl Acad Sci USA 95:10626–10631 Lowenstein CJ, Alley EW, Raval P et al 1993 Macrophage nitric oxide synthase gene: two upstream regions mediate induction by interferon γ and lipopolysaccharide. Proc Natl Acad Sci USA 90:9730–9734 Muller S, Hoege C, Pyrowolakis G, Jentsch S 2001 SUMO, ubiquitin’s mysterious cousin. Nat Rev Mol Cell Biol 2:202–210 Nishida T, Yasuda H 2002 PIAS1 and PIASxα function as SUMO-E3 ligases toward androgen receptor and repress androgen receptor-dependent transcription. J Biol Chem 277:41311–41317 Nolan JJ, Ludvik B, Beerdsen P, Joyce M, Olefsky J 1994 Improvement in glucose tolerance and insulin resistance in obese subjects treated with troglitazone. N Engl J Med 331: 1188–1193 Oberfield JL, Collins JL, Holmes CP et al 1999 A peroxisome proliferator activated receptor γ ligand inhibits adipocyte differentiation. Proc Natl Acad Sci USA 96:6102– 6106 Ogawa S, Lozach J, Jepsen K et al 2004 A nuclear receptor corepressor transcriptional checkpoint controlling activator protein 1-dependent gene networks required for macrophage activation. Proc Natl Acad Sci USA 101:14461–14466 Ogawa S, Lozach J, Benner C et al 2005 Molecular determinants of crosstalk between nuclear receptors and Toll-like receptors mediating counter-regulation of inflammatory responses. Cell 122:707–721 Ohshima T, Koga H, Shimotohno K 2004 Transcriptional activity of peroxisome proliferatoractivated receptor γ is modulated by SUMO-1 modification. J Biol Chem 279:29551– 29557 Pascual G, Fong AL, Ogawa S et al 2005 A SUMOylation-dependent pathway mediates transrepression of inflammatory response genes by PPAR-gamma. Nature 437:759–763 Perissi V, Aggarwal A, Glass CK, Rose DW, Rosenfeld MG 2004 A corepressor/coactivator exchange complex required for transcriptional activation by nuclear receptors and other regulated transcription factors. Cell 116:511–526 Perreault M, Marette A 2001 Targeted disruption of inducible nitric oxide synthase protects against obesity-linked insulin resistance in muscle. Nat Med 7:1138–1143 Picard F, Auwerx J 2002 PPARγ and glucose homeostasis. Annu Rev Nutr 22:167–197 Pickup JC, Mattock MB 2003 Activation of the innate immune system as a predictor of cardiovascular mortality in type 2 diabetes mellitus. Diabet Med 20:723–726 Pickup JC, Mattock MB, Chusney GD, Burt D 1997 NIDDM as a disease of the innate immune system: association of acute-phase reactants and interleukin-6 with metabolic syndrome X. Diabetologia 40:1286–1292 Ricote M, Li AC, Willson TM, Kelly CJ, Glass CK 1998 The peroxisome proliferator-activated receptor- γ is a negative regulator of macrophage activation. Nature 391:79–82 Ruan H, Pownall HJ, Lodish HF 2003 Troglitazone antagonizes tumor necrosis factor-alphainduced reprogramming of adipocyte gene expression by inhibiting the transcriptional regulatory functions of NF-kappaB. J Biol Chem 278:28181–28192 Shoelson SE, Lee J, Yuan M 2003 Inflammation and the IKK beta/I kappa B/NF-kappa B axis in obesity- and diet-induced insulin resistance. Int J Obes Relat Metab Disord 27 Suppl 3:S49–52 Sinha S, Perdomo G, Brown NF, O’Doherty RM 2004 Fatty acid-induced insulin resistance in L6 myotubes is prevented by inhibition of activation and nuclear localization of nuclear factor kappa B. J Biol Chem 279:41294–41301
196
PASCUAL ET AL
Souza SC, Yamamoto MT, Franciosa MD, Lien P, Greenberg AS 1998 BRL 49653 blocks the lipolytic actions of tumor necrosis factor-alpha: a potential new insulin-sensitizing mechanism for thiazolidinediones. Diabetes 47:691–695 Spiegelman BM 1998 PPARγ : adipogenic regulator and thiazolidinedione receptor. Diabetes 47:507–514 Takeda K, Akira S 2004 TLR signaling pathways. Semin Immunol 16:3–9 Tontonoz P, Nagy L, Alvarez JGA, Thomazy VA, Evans RM 1998 PPARγ promotes monocyte/macrophage differentiation and uptake of oxidized LDL. Cell 93:241–252 Virkamaki A, Yki-Jarvinen H 1994 Mechanisms of insulin resistance during acute endotoxemia. Endocrinology 134:2072–2078 Weisberg SP, McCann D, Desai M, Rosenbaum M, Leibel RL, Ferrante AW Jr 2003 Obesity is associated with macrophage accumulation in adipose tissue. J Clin Invest 112:1796–1808 Welch JS, Ricote M, Akiyama TE, Gonzalez FJ, Glass CK 2003 PPARγ and PPARδ negatively regulate specific subsets of lipopolysaccharide and IFNγ target genes in macrophages. Proc Natl Acad Sci USA 100:6712–6717 Wellen KE, Hotamisligil GS 2003 Obesity-induced inflammatory changes in adipose tissue. J Clin Invest 112:1785–1788 Wellen KE, Hotamisligil GS 2005 Inflammation, stress, and diabetes. J Clin Invest 115–1111–1119 Willson TM, Lambert MH, Kliewer SA 2001 Peroxisome proliferator-activated receptor gamma and metabolic disease. Annu Rev Biochem 70:341–367 Xu H, Barnes GT, Yang Q et al 2003 Chronic inflammation in fat plays a crucial role in the development of obesity-related insulin resistance. J Clin Invest 112:1821–1830 Yoon HG, Chan DW, Huang ZQ et al 2003 Purification and functional characterization of the human N-CoR complex: the roles of HDAC3, TBL1 and TBLR1. EMBO J 22:1336–1346 Yuan M, Konstantopoulos N, Lee J et al 2001 Reversal of obesity- and diet-induced insulin resistance with salicylates or targeted disruption of Ikkbeta. Science 293:1673–1677 Zhang J, Kalkum M, Chait BT, Roeder RG 2002 The N-CoR-HDAC3 nuclear receptor corepressor complex inhibits the JNK pathway through the integral subunit GPS2. Mol Cell 9:611–623 Ziouzenkova O, Perrey S, Asatryan L et al 2003 Lipolysis of triglyceride-rich lipoproteins generates PPAR ligands: evidence for an antiinflammatory role for lipoprotein lipase. Proc Natl Acad Sci USA 100:2730–2735
DISCUSSION Attie: You might have provided an explanation for data that I showed earlier, where we saw islets in the SCD1 knockout mice that had super-induction of CD36 and LPL. Are those targets of the trans-repression pathway that you are describing? Glass: In the absence of a PPARγ ligand, CD36 would probably have NCoR complexes on it. It is possible that inflammatory signals could remove this complex independent of ligands, resulting in partial activation. Spiegelman: That is a direct positive regulation. Attie: I want to get the directionality right here. You were starting with an unliganded receptor. Glass: There are two NCoR-dependent processes. One involves the unliganded receptor on direct target genes. The control here is that the ligand causes the NCoR
PPARγ
197
complex to come off. There is another NCoR-dependent process on inflammatory response genes. This involves the receptor being tethered to the gene through SUMO. In this case it is preventing the signal-dependent dissociation of the NCoR complex from the promoters of inflammatory response genes. Attie: You said this was an oleate-dependent process. Glass: That’s correct. If you knock out the mechanism to make oleate, it is possible that you could affect this process. We haven’t looked in pancreatic β cells or any of the other ‘sexy’ cells at this. Attie: I wonder whether that mechanism could be mediating some of the unexpected effects of the SCD1 knockout. O’Rahilly: Other than the metabolic effects of knocking out PPARγ in the macrophage, from the range of gene transcription effects that are altered, could you predict what inflammatory pathologies you might expect? Have you tested them in your mice? As our patient population grows, it would be interesting to be able to look and see what range of autoimmune and inflammatory pathologies we might expect to see enriched. Glass: We haven’t done much on this. There is a paper on an adjuvant arthritis model (Setoguchi et al 2001) in which haploinsufficiency for PPARγ seems to lead to a marked increase in severity of disease. I think this is the best genetic example. The problem with PPARγ is that the total knockout is lethal. Tabas: What concentrations of LPS are you using in your experiments? Glass: The LIPID MAPS consortium went to the trouble of making Kdo2-lipid A, which is a highly specific TLR4 agonist. We typically use 100 ng/ml. Tabas: Are the results you get mimicked by using pure lipid A? Glass: Yes. That is as pure a ligand as we can get. Everything that we are looking at is going through TLR4. Tabas: In the Myd88 or Toll receptor knockout you would suppress the inflammatory effect, then. You showed the oleic acid data and volunteered the idea that you don’t think this is it per se because of the chronicity and amount needed. If I can read your mind, are you thinking that the oleic acid is a precursor? Glass: Yes, although if you are estimating that it is only 60–70 nM, that is pretty low. Yet, if we were to look at the ability of oleic acid to bind 60–70 nM, it wouldn’t happen. Spiegelman: It wouldn’t even come near that. Glass: This is where people tend to get stuck. Tabas: There is another system I am aware of where a similar dilemma occurred. It turned out that the answer was cannabinoids. Have you ever explored the idea that the fatty acids are getting metabolized to a cannabinoid-like agonist? Glass: We haven’t looked at that. Spiegelman: One of the problems we all face in this field is not that it is hard to find PPARγ agonists (it is easy to do this: many small hydrophobic compounds are
198
PASCUAL ET AL
weak agonists at high concentrations), the problem is trying to understand the biological significance of this. It is easy to find agonists; it is hard to make a convincing case that your agonist is regulating the biology of lipogenesis (in our case) and inflammation (in Chris Glass’ case). Cannabinoids are weak agonists of PPARγ. Glass: If you plate a macrophage under standard culture conditions, there isn’t much substrate for it to work on. If you were to try to find a ligand for PPARγ by just looking at the standard cultured macrophage you would have a hard time finding it. This is something we did learn that was useful from our LIPID MAPS experience. Our colleagues studying arachidonic acid metabolism threw their hands up in disgust because they could find very little synthesis of arachidonic acid metabolites under standard culture conditions unless additional substrate was provided. Our current thinking is that we have to provide exogenous substrate to more closely model the in vivo situation. Spiegelman: You could also hit them with a receptor agonist that will stimulate those pathways, so it doesn’t have to be subject only to the lipids in the medium. Kim: Are there any differences in fat cells versus macrophages in terms of the sumoylation of PPARγ, with the same dose of TZDs? Glass: That’s a good question. We think this pathway also operates in fat cells. But in fat cells it is not NCoR that appears to be the key co-repressor. Rather it appears to be SMRT, which is the dominant member of that co-repressor family. To look at the ability of PPARγ to suppress inflammatory mediators in a 3T3L1 adipocyte model system you need to knock out SMRT, not NCoR. Bernlohr: In all the models of adipose physiology, fatty acids produced by the adipocytes activate the macrophages and then there is the chemokine production. In a co-culture system do activated macrophages make lipid species that play a biological role? That is, if you put your LIPID MAPS hat on, in that co-culture system are there families of hydrophobic we should be thinking about that are derived from the macrophages that are functioning? Glass: That is what we are thinking. For us, this is the big advantage being in the LIPID MAPS group. Others are helping us think about how to address this question. It is hard to recapitulate in vitro the in vivo system, but we have looked at some of the metabolism of fatty acids and our thinking is that macrophages are responding to what they are seeing. It is probably the milieu of fatty acids that are brought into the cell that then are acted on by whatever enzymatic machinery is in the macrophage that can generate ligands for the receptors. Bernlohr: I’m thinking of things that the macrophages spit out. For example, adipocytes generate large amounts of PGE2. Are the macrophages doing similar things? Glass: In the TLR-induced system that we are using, they do turn on COX2 tremendously, and if they have substrate they make huge amounts of this.
PPARγ
199
Tabas: When you showed your diagram of how you did your conditioned medium transfer experiments, there is a cocktail of FFAs. Then in a later experiment you showed specific ones. What was in the cocktail? Glass: This was a cocktail of the five most prevalent fatty acids in serum. They were used at equal molar concentrations at a level supposed to mimic physiology. Now we are measuring free fatty acid concentrations of serum in adipose tissue, and measuring the fatty acid composition of the triglycerides. It is probably the fatty acid composition of the triglycerides that is released when adipocytes are stimulated in a fasting condition. We have put mice on different diets and will measure the fatty acids. We will use that information to build our cocktails. We think the receptor is sensing a combination of fatty acids. We are trying to recapitulate that to see whether the fatty acids that are in the obese animals have a different impact on this pathway than the fatty acids in a lean animal. Muoio: How do the macrophages process fatty acids? Does PPARγ alter fatty acid metabolism as well as the inflammatory responses? Is β -oxidation an important pathway in this cell type? Do changes in lipid metabolism affect respiratory bursts and production of reactive oxygen species? Glass: There is a lot of information on this, but we have a good chance to expand our understanding of much of that. Many of the studies of fatty acid metabolism in macrophages was initially driven by foam cell formation and what macrophages do with cholesterol to build cholesterol esters, the primary lipid in the foam cell. They definitely are able to oxidise fatty acids. There are two types of macrophages, an M1 and M2 type. A recent paper addresses this (Vats et al 2006). Spiegelman: I believe the respiratory burst is a classic example of non-mitochondrial generation of ROS. Glass: When macrophages deal with parasites they become M2 cells. This is a long haul. They tend to use fatty acids for energy when they differentiate down that pathway. Muoio: Does PPARγ turn on the triglyceride biosynthesis programme in the macrophages? Glass: It does part of it, certainly. Spiegelman: You get fatty looking monocytes when you activate PPARγ. References Setoguchi K, Misaki Y, Terauchi Y et al 2001 Peroxisome proliferator-activated receptorgamma haploinsufficiency enhances B cell proliferative responses and exacerbates experimentally induced arthritis. J Clin Invest 108:1667–1675 Vats D, Mukundan L, Odegaard JI et al 2006 Oxidative metabolism and PGC-1beta attenuate macrophage-mediated inflammation. Cell Metab 4:13–24 (Erratum: 2006 Cell Metab 4:255)
Fatty Acids and Lipotoxicity in Obesity and Diabetes: Novartis Foundation Symposium 286, Volume 286. Edited by Gregory Bock and Jamie Goode Copyright Novartis Foundation 2007. ISBN: 978-0-470-05764-3
Final Discussion Nutrition, ageing and lipotoxicity Tabas: In the discussion of a paper we wrote a few years ago we were interested in fatty acid metabolism in macrophages with regard to eicosanoid metabolism. We showed that the arachidonic acid content of LDL, one of the lipoproteins in a modified form taken up by macrophages, makes a profound difference to the eicosanoid production in macrophages (Leventhal et al 2004). Lipoprotein-derived arachidonic acid is a very important variable. We could make LDL of various compositions with respect to arachidonic acid. We speculated in the discussion of the paper: could these variations occur naturally depending upon the diet? Black: As our diet has shifted, the amount of omega 6 fatty acids we have taken in far exceeds what would have occurred in history. We have moved from a 3 : 1 ratio of omega 6 to omega 3 fatty acids to perhaps a 10 : 1 ratio. This change in diet may be exacerbating a lot of these conditions we are seeing in a pathophysiological context. Spiegelman: The problem is that these are moving in parallel universes scientifically. All of us on the molecular and cellular side have an obligation to take an interest in the nutritional aspects, but right now these parallel universes are sliding past one another a bit. O’Rahilly: In 2001 we published a paper in which we looked at 2000 people in Ely, Cambridgeshire, using dietary histories that were confirmed by plasma-based assays (Halsall et al 2001). We showed that the relationship between the common PPARγ Pro12Ala polymorphism and protection against obesity and hyperinsulinaemia was dependent on dietary fat intake. There was clear evidence of interaction between nutrition and genotype in determining BMI. Black: Did you look at the fatty acid profi les in the diet? O’Rahilly: Not extensively enough. This was mostly based on well validated dietary history, with some validation by biochemical analysis. Bernlohr: What influence does age play in all of this? We make measurements in young mice, but age does affect expression of the genes and the proteins that are available, and thus which pathways tend to dominate. The nuclear receptors are a complex system of multiple equilibria. If you tip the balance even subtly, this can result in dramatic effects. Spiegelman: Mitochondrial biology is slowly going downhill with age. Bernlohr: The difference between 3 and 5% is trivial if you are talking about the return on your dollar investment in one week, but over a 40 year investment it is 200
FINAL DISCUSSION
201
significant. We aren’t dealing with the small changes that might be underlying metabolic disease. Tabas: The Framingham table for heart disease shows that you could be living a perfect life, but in the time it takes to go from 60 to 70 years old you are converted from low risk to a high risk just by ageing. There have been a number of studies where people have tried to look at the role of cellular senescence, but I’m not sure how useful this is. Spiegelman: Whereas humans typically die of cardiovascular disease, mice die of cancer and they are not particularly prone to age-related cardiovascular events. Rodents are not a good model in this respect. Black: Along those lines, if we begin to look at the epidemiology of cancer and the role of specific fatty acid classes, in many cases the omega 3 fatty acids seem to be protective against many cancers. The relationship between obesity and cancer may have its foundations in some relationship with diet. Tabas: Willa Hseuh has looked at the effect of ageing on insulin resistance. She works with mice that are either the typical 8–12 weeks old, or mice that are 1–2 years old. She looks at the ability of various insults to lead to insulin resistance. She has shown that the older mice are substantially more sensitive to insulin resistance. This might be an experimental model, and it would be possible to add additional insults to this. Glass: I have a comment on lipotoxicity. PPARγ is one example where we don’t know what the key lipids are that are involved. Do we know the lipids that are directly involved in lipotoxicity? Are they the major classes of lipids that are also key metabolic components, or are there specific subclasses of lipids? An inositol phospholipid that is 2% of the total pool turns out to be important in signalling. Spiegelman: Jerry Shulman and I have vigorous arguments about this. He implies that those lipids are causally linked to insulin resistance, and I am not convinced. He has shown that the acyl-CoA derivatives that accumulate in muscle and liver are highly correlated with insulin resistance, but I don’t see the data that they are causal. O’Rahilly: If we are going to talk about a topic, such as lipotoxicity, it is better that it clearly defined and has an experimental paradigm. Insulin resistance is easy because we can take insulin and measure a biological response. So can we agree on lipotoxicity when we see it? We make a general assumption that when we fatten mice and they get insulin resistance, there must be lipotoxicity occurring. We use this term in a vague, generalized sense. I would suggest as a model some metabolic situation where we would actually administer lipid and witness a reliable and repeatable response. Spiegelman: Shulman does this with clamps. He infuses in his clamp and in his model he can show the induction of insulin resistance. So he does have an
202
FINAL DISCUSSION
experimental paradigm that is convincing. If lipid is elevated in the context of a mouse clamp, a state of insulin resistance can be induced. Is this lipotoxicity? I would accept that as lipid-induced insulin resistance. I am not sure about the lipotoxicity part. Bernlohr: We have spent most of our time in this meeting focusing on fat, muscle, liver and macrophages. But lipotoxicity could have its greatest effect on endothelial cells. Lipid spill over that can’t be stored by adipose is taken up by endothelial cells and affects biology. Those cells normally don’t see a large lipid challenge, but in an obese individual they will challenged by a new set of conditions. Spiegelman: What about the brain? The ApoE ε4 allele polymorphism is probably the most convincing genetic variant in Alzheimer’s. I don’t think I’ve yet heard a convincing story about its mechanism, but the clinical data are good. Tabas: There are a number of interesting hypotheses to do with the repair of neuronal damage and the effect that lipid delivery may play in this. Clearly, the answer is not out yet. Muoio: Patrick Sullivan at Duke University works with a mouse model of Alzheimer’s disease that is precipitated by a specific genetic variant of apoprotein E. He finds that the development of Alzheimer’s as a result of the ApoE genotype depends on amount and type of dietary fatty acids. Black: As was mentioned earlier, we run in parallel universes. Even when you go to the bigger meetings, there is not a lot of cross-talk between the Nutrition Society and the American Society of Biochemistry and Molecular Biology. Spiegelman: It is hard to bring people from different fields together. We have been working hard on the PPARγ story for at least a decade, but it is hard to take the next step unless we can talk about specific molecules. Black: The Lipid Maps consortium is developing an idea of what these lipids are, but they are looking at what in essence are steady state lipids. Bioactive lipids will be short-lived, and if it is a bioactive lipid that is the effector here, then identifying this specific mediator still remains a difficult task. Spiegelman: This could be the explanation for why it has been so difficult to come to an answer. Black: The technologies are now coming online that might allow us to address these questions. But it will require having a group of students who want to learn mass spectroscopy, for example, as their primary expertise. We are lacking along these lines. Attie: If we look at hepatic steatosis, for example, the traditional view is that the triglyceride is causal for some of the consequences. The better imaging techniques have shown that many more people have hepatic steatosis than anyone had imagined, and many of them are not sick. Is the triglyceride a red herring? Is this really a marker of lipotoxicity? The same could be said of muscle.
FINAL DISCUSSION
203
Spiegelman: Physical activity in athletes induces muscle steatosis. Of course, it may not be the same steatosis. There are two places you see fat droplets in muscle: one is in fat people with diabetes; the other is trained atheletes. References Halsall DJ, Luan J, Saker P et al 2001 Uncoupling protein 3 genetic variants in human obesity: the c-55t promoter polymorphism is negatively correlated with body mass index in a UK Caucasian population. Int J Obes Relat Metab Disord 25:472–477 Leventhal AR, Leslie CC, Tabas I 2004 Suppression of macrophage eicosanoid synthesis by atherogenic lipoproteins is profoundly affected by cholesterol-fatty acyl esterification and the Niemann-Pick C pathway of lipid trafficking. J Biol Chem 279:8084–8092
Fatty Acids and Lipotoxicity in Obesity and Diabetes: Novartis Foundation Symposium 286, Volume 286. Edited by Gregory Bock and Jamie Goode Copyright Novartis Foundation 2007. ISBN: 978-0-470-05764-3
Contributor Index Non-participating co-authors are indicated by asterisks. Entries in bold indicate papers; other entries refer to discussion contributions
A
H
*Accili, D. 99 *Arellano, J. 99 *Athenstaedt, K. 142 Attie, A. D. 6, 10, 45, 46, 47, 54, 55, 56, 57, 84, 94, 122, 196, 197, 202
*Han, S. 99 Han, X. 98 *Hara, K. 164 Hotamisligil, G. S. 9, 10, 11, 21, 22, 39, 41, 42, 45, 46, 54, 56, 70, 82, 83, 86, 94, 95, 96, 97, 98, 110, 111, 112, 122, 123, 124, 125
B Bernlohr, D. A. 7, 42, 43, 83, 95, 96, 113, 121, 122, 123, 124, 125, 138, 139, 140, 153, 159, 160, 179, 180, 198, 200, 202 Black, P. N. 39, 121, 127, 138, 139, 140, 141, 153, 154, 160, 200, 201, 202 C Carling, D. 72, 81, 82, 83, 84, 177, 178 *Cui, D. 99 *Czabany, T. 142 D Daum, G. 68, 125, 142, 151, 152, 153, 154 *DiRusso, C. C. 127 F *Flowers, J. B. 47 *Flowers, M. T. 47 *Forcheron, F. 99 G *Gamliel, A. 183 Glass, C. K. 8, 20, 21, 42, 125, 160, 163, 177, 179, 181, 183, 196, 197, 198, 199, 201 *Grillitsch, K. 142 *Groen, A. K. 47
K Kadowaki, T. 11, 21, 22, 41, 44, 164, 176, 177, 178, 179, 180, 181, 182 Kim, J. B. 8, 68, 70, 84, 198 *Koves, T. R. 24 *Kubota, N. 164 *Kuipers, F. 47 L *Li, J. Z. 155 Li, P. 10, 81, 155, 159, 160, 161 Li, X. 178 *Li, Y. 99 *Liang, C.-P. 99 *Lobo, S. 113 M Muoio, D. M. 10, 11, 24, 38, 39, 40, 41, 42, 43, 44, 45, 69, 95, 98, 138, 141, 163, 179, 180, 199, 202 N *Ntambi, J. M. 47 204
CONTRIBUTOR INDEX O *Ogawa, S. 183 O’Rahilly, S. 9, 13, 20, 21, 22, 23, 54, 55, 56, 94, 97, 109, 110, 123, 138, 163, 176, 179, 181, 182, 197, 200, 201
205 139, 140, 152, 153, 160, 161, 162, 163, 176, 177, 178, 179, 180, 181, 182, 196, 197, 198, 199, 200, 201, 202, 203 *Sullivan, A. L. 183
P
T
*Pascual, G. 183 *Perissi, V. 183
Tabas, I. 69, 95, 96, 99, 109, 110, 111, 112, 122, 151, 176, 177, 181, 182, 197, 199, 200, 201, 202 *Tall, A. 99
R Reue, K. 58, 68, 69, 70, 71 *Rosenfeld, M. G. 183 S Sabin, M. 23, 124 Saito, M. 124, 162, 163 *Seimon, T. 99 Shi, Y. 7, 9, 10, 23, 43, 45, 57, 70, 97, 121, 122, 125, 139, 141, 153, 154, 160, 163, 180, 182 Spiegelman, B. M. 1, 3, 6, 7, 8, 9, 10, 11, 20, 22, 38, 39, 40, 41, 42, 43, 44, 45, 53, 55, 56, 68, 69, 70, 71, 84, 94, 96, 97, 98, 110, 111, 112, 122, 123, 124, 125,
U *Ueki, K. 164 W *Wagner, A. 142 Y *Yamauchi, T. 164 Yang, R. 55, 69, 139, 152, 153 Z Zhang, C.-Y. 11, 40, 57
Fatty Acids and Lipotoxicity in Obesity and Diabetes: Novartis Foundation Symposium 286, Volume 286. Edited by Gregory Bock and Jamie Goode Copyright Novartis Foundation 2007. ISBN: 978-0-470-05764-3
Subject Index
A ABCG1 187 acanthosis nigricans 16, 110 acetyl CoA carboxylase (ACC) 61, 83 ACC2 158, 160 acid soluble metabolites (ASM) 30 activator protein 1 (AP1) complexes 87 gene networks 185, 186 acyl carnitine 43 muscle profi ling 29 acyl-CoA 36, 138 acyl-CoA oxidase 171 acyl-CoA synthetases (ACSLs; FadD) 115– 117, 128, 129, 135, 139–141 features 132–134 ACSL1 116–117, 121, 128, 133 ACSL3-ACSL6 133 ACSL4 116, 160 acyl-CoA:cholesterol acyltransferase (ACAT) 100, 144 acylglycerophospholipids 154 adipocytes, model for fatty acid uptake in 117 adipokines 164 adiponectin 55, 69–70, 72, 82, 111–112, 123, 156, 176, 177–178, 180–181, 182 syndromes of severe insulin resistance 17–18, 21–22 up-regulation 165–167 adiponectin hypothesis 164–165 adiponectin receptors cloning, function and regulation 167–168 as therapeutic targets 173 up-regulation 173 development of adiponectin receptor agonists 173 expression in liver ameliorated diabetes 168–171 see also AdipoR1; AdipoR2 adiponectinaemia 180
AdipoR1 167, 168, 173 increase in AMPK activation by adiponectin in liver 169–170 as major adiponectin receptor in vivo 171–173 AdipoR2 167, 168, 173 deficiency 172 as major adiponectin receptor in vivo 171–173 PPARα target genes in liver 170–171 adipose tissue expandability 16 adipose-type FABP 124 AGPAT2 61 AGPAT2 deficiency 56 AKT2 16, 18 Alzheimer’s disease 202 5-amino-4-imidazole carboxamide (AICAR) 76 AMP 74 AMP-activated protein kinase 72–79, 81–84 regulation 73–76 downstream targets 76–77 AMP kinase (AMPK) 7, 78–79, 81–84, 112, 167–168, 173, 174, 176, 178, 179–180 angiotensin 184 α1 antitrypsin deficiency 97 apoptosis 101 UPR-SRA model 102–105 APPL1 83 Arabidopsis thaliana 146 arachidonic acid 200 ARE1 144 Are1p 144, 145 ARE2 144 Are2p 144, 145 arrogenes 5 atherosclerosis, macrophage death pathways in 99–107, 109–112 ATP production, PGC1 coactivators 10 atrogin 5 Ayr1p 143 206
SUBJECT INDEX 9-p-azidophenoxy nonanoic acid 133 azoxymethane-induced colonic cancer 182 B beta-oxidation, mitochondrial 34–36 brain-derived neurotrophic factor (BDNF) 15 breast cancer 181 brown adipose tissue (BAT) 156 visualizing with FDG-PET 162–63 β -OH-butyrylcarnitine (C4-OH) 31 C C/EBPα 59, 60 C75 76 Caenorhabditis elegans 130, 153 calcium/calmodulin dependent protein kinase kinase β (CaMKK β ) 73, 74, 75, 76, 79, 82 cAMP-response element binding (CREB) protein 5, 78 CREB-H 91 Candida parapsilosis 146 cannabinoids 197–198 canola oil 49 carbohydrate responsive element binding protein (ChREBP) 48 carnitine palmitoyltransferase 1 (CPT1) 26, 29, 34 catalase 5, 8 caveolin 115 CD36 see fatty acid translocase ceramide 25, 36, 40, 52 cholecystokinin 57 cholestasis 47–52 CHOP 103, 107 Cide proteins 155–159, 159–160 Cidea 156, 157–158, 159–161 Cideb 156, 157–158, Cidec 156 coconut oil 49 colon cancer 181 azoxymethane-induced 182 COX2 198 CPT1 43–44 Cxcl10 189 cyclodextrin 115 cycloheximide 96 cystathionine- β -synthase (CBS) 73 cystic fibrosis 98
207 D DFF 156 Dga1p 144 diacylglyceride hydrolase 151 diacylglycerol (DAG) 25, 40, 118, 143, 154 1,3-diacylglycerol 153 diacylglycerol acyltransferase (DAGT) 61, 144 DGAT1 69 diacylglycerol kinase 143 diacylglycerol transferase 60 dihydroxyacetone phosphate (DHAP) 143 dihydroxyacetone phosphate acyltransferases 143 Dpp1p 143 Drosophila melanogaster 130, 146 dyslipidaemia 25, 57 E E-3 ubiquitin ligases 5 efferocytosis 101, 102 EGP 171, 172 electron transport system 26, 28 endoplasmic reticulum stress 86–91, 94–98 Erg1p 145–146 ergosterol 144–145 ERRα 4 Escherichia coli 128, 129, 133 etomoxir 34, 41 exercise training, lipid tolerance and 32–34 F Faa1p 128, 134 Faa4p 128, 134 FABP 122, 123, 124, 139 FABP4 124 FABP5 124, 125 FABPpm 128 FadD 133 FATPs 117, 128, 135 features 130–132 FATP1 116, 117, 118, 121, 128, 113, 135, 139, 140 FATP2 130, 132, 138, 139, 141 FATP4 130, 135, 141 FATP5 135 FATP/ACSVL motif 130, 131 FATP/VLACS motif 130, 132
208 fatty acid synthase 60, 61 fatty acid translocase (FAT) (CD36) 114, 117, 118, 128, 141, 187, 196 fatty acid transport proteins see FATPs forskolin 122 FOXO3 5 Fsp27 156, 160 G G protein-coupled receptors (GPCRs) 167 G-3-P acyltransferases (GAT) 143 GADD153 103 GC-MS 139 ghrelin 76 GIR 171, 172 glucokinase gene 170–171 glucose-6-phosphatase gene 169–170 GLUT4 translocation 25 glutamate-oxaloacetate transaminase 42–43 glycerol-3-phosphate 143 glycerolipids 36 glycerophospholipids 144 GPS2 185 GPX 8 GPX1 5 growth retardation 15 GW0072 188, 189, 191, 193 H HDAC3 185, 188 heart-type FABP 124 hepatocarcinoma 44 hepatomegaly 56 hepatosteatosis 56 hexokinase 140 high-density lipoprotein (HDL) 105 histone acetyltransferases 69 HIV infection 61 HMG CoA synthase 32 β -hydroxyacyl-CoA dehydrogenase 44 β -hydroxybutyrate 10, 44 hypercholesterolaemia 47–52 hyperglycaemia 15, 105 hyperinsulinaemia 17, 54, 105, 110, 168, 180 hyperlipidaemia 25, 135 hypertension 23 hypertriglyceridaemia 17 hypoadiponectaemia 180
SUBJECT INDEX I incomplete fat oxidation 28 inducible nitric oxide synthase (iNOS) gene 187–188 inhibitor of NF-κ B kinase (IKK) 87–89, 91, 96, 118, 184, 185 insulin 87 insulin-like growth factor (IGF) receptor 110 insulin receptor (IR) 25 insulin receptor substrate (IRS) proteins 164 IRS1 8, 25, 87, 88, 118, 164, 184 IRS2 105, 164 insulin resistance, mediators of 184–185 insulin sensitivity, effect of lipin 1 on 66–67 insulin signalling, regulation by lipid metabolites 117–119 interleukins IL1 8, 114, 116 IL6 8, 10–11, 61, 118, 184 IL8 61 IL18 61 intramuscular triacylglycerol (mTAG) 26 intramyocellular lipid (IMCL) 62–64 isoproterenol 134 J c-Jun N-terminal kinase ( JNK) 25, 41–42, 87, 88, 89, 91, 96, 98, 105, 107, 122, 184, 185 IRE-1 dependent activation 90 JNK1 88, 119, 185 JNK2 88 JNK3 88 JNK binding protein ( JIP) 88 K kainic acid 5 Kdo2-lipid A 197 L lanosterol 144–145 LC-MS/MS 139 lecithin:cholesterol acyltransferase deficiency 50 leptin 72, 79, 123, 156, 176 leptin deficiency 14, 15
SUBJECT INDEX leptin receptor 14–15, 20 lipid cores 101 lipid particles (LP) 142 lipid rafts 114, 115 lipid supply, transcriptional adaptation to increased 27–28 lipin 1 action 65–66 as determinant of insulin sensitivity 61–64 lipid metabolism in muscle and liver 64–65 requirement for fat storage in adipose tissue 59–61 role in adipogenesis and lipid metabolism 58–67, 68–71 lipin 1A 70 lipin 1α 59, 60, 62 lipin 1B 70 lipin 1β 59, 60, 61, 62 lipin 2 59 lipin 3 59 lipodystrophy 23, 55–56, 59, 61, 176 lipopolysaccharide 39 lipoprotein lipase 49, 113 lipoprotein-X 49, 50 lipotoxicity 25, 26 liver-X-receptor (LXR) 48, 187 LKB1 73, 74, 75, 76, 81–82 glucose homeostasis 77–78 LMNA 61 low density lipoproteins (LDL) 100 LPIN1 61, 62 Lpp1p 143 Lrolp 143–144 M malate dehydrogenase 42 MCP1 178, 179, 184, 193 melanocortin 4 receptor deficiency 15 metabolic syndrome 23, 105, 118 metformin 78, 79, 84 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MTPT) 5 mitochondrial lipid overload 28–32 monoacyl glyceroltransferase 56 monounsaturated fatty acids (MUFA) 47, 49 Mortierella ramanniana 144 MURF 5
209 Mycobacterium tuberculosis 130 MyD88 pathways 42 N NADH shuttle systems 34 NADH/NAD ratio 42, 43 NCoR 188, 189, 197 role in control of inflammation 185–186 NCoR/HDAC3/TBL complex 187–189 necrosis 102 necrotic cores 101 NF-κ B 184, 185, 186 N-myristolyation 133 NRF1 4 NRF2 4 nutrition, ageing and lipotoxicity 200–203 O OLE1 13 oleate 39, 40, 52, 84, 197 oleoyl-CoA 47 osmotin 173 oxidative phosphorylation (OXPHOS) 4 8-OXO-guanine 5 oxygen regulated protein 150 (ORP150) 90 P p38 MAP kinase 167 Pah1p 143 PAI1 184 palmitate 39, 40, 52, 84 palmitoleate 52 palmitoleoyl-CoA 47 palmitoylation 133 palmitoyl-CoA 40, 47 PAT proteins 152 pathogenesis-related 5 (Pr-5) family proteins 173 PDK1 81 perilipin 122 peroxisome proliferator-activated receptors see PPARs peroxisome proliferator-activated receptor γ (PPARγ ) coactivator 1 see PGC1 coactivators PGC1 coactivators 3–5, 6–12 PGC1α 4–5, 28, 32–34, 40–41, 42, 69, 76, 78, 179–180 PGC1β 4, 5
210 Phe57 123 4-phenyl butyric acid (PBA) 90, 94 pheochromocytoma 163 phosphatidate phosphatase 143 phosphatidic acid (PA) 143 phosphatidic acid phosphatase-1 (PAP1) 65, 69, 70 phosphatidylcholine 144, 154 phosphatidylethanolamine 144, 154 phosphatidylinositol-3-kinase activation 118 phosphoenolpyruvate carboxykinase 1 gene 170 phospholipase A2 153 phospholipase C 143 phospholipase D 143 PIASI (protein inhibitor of activated STAT1) 189–191 pioglitazone 64, 166, 167, 177 PKA 83–84 PKC 41 Plc1p 143 Pld1p 143 polycystic ovary syndrome 16 PPAR agonists 44 PPARs 27, 177 PPARα 27, 34, 41, 44, 125, 167, 171, 173, 174, 176, 178 PPARα agonists 45 PPARδ 27, 125, 177 PPARγ 16, 18, 21, 22, 23, 27, 59, 60, 70, 114–115, 125, 152, 173, 177, 182, 201 anti-inflammatory and antidiabetic roles 183–193, 196–199 genome-wide analysis 186–187 ligand-dependent sumoylation 189–193 macrophages and insulin resistance 183–184 PPP1R3 23 PPP1R3A gene 17 proprotein 1 convertase deficiency 15 prostate cancer 181 protein kinase C (PKC) 25, 87 protein phosphatase 2C α (PP2C α ) 74 pyruvate phosphotransferase system 140 R reactive oxygen species (ROS) 38–39, 42, 43, 44, 95, 96, 199 PGC1 coactivators 5, 7, 8, 9–10 remnant lipoproteins 100 resistin 184, 193
SUBJECT INDEX rheumatoid arthritis 178, 179 rosiglitazone 70, 188, 189, 190, 191, 193 Rox3p 148 S Saccharomyces cerevisiae 130, 133, 142, 143, 146–147, 148, 149 PAP1 65 salt-inducible kinase (SIK) 1 and 2 78 scavenger receptor (SRA) 104 Scd1 deficiency 48, 49–50, 50–2, 53, 54 SERCA 103 serine palmitoyl transferase 55 skeletal muscle, lipid-induced metabolic dysfunction in 24–36, 38–46 Slc1p 143 Smp2 70 SMRT 185, 198 SOD1 5, 8 SOD2 5, 8, 96 sphingolipids 52 squalene epoxidase 145 STE hydrolases 148 STE synthases 144 stearate 52 stearoyl CoA desaturase 60 stearoyl CoA desaturase deficiency 47–52 sterol responsive element binding protein 152, 153 SREBP-1c 48, 158 steryl esters (STE) 142 biosynthesis 144–145 mobilization 148 SUMO 197 suppressor of cytokine signalling (SOCS) 89 T TAK1 73 taurine-conjugated ursodecoxycholic acid (TUDCA) 90 Tbl1 185, 188 TblR1 185, 188 Tgl1p 148 Tgl3p 146–147, 151, 152 Tgl4p 146–147, 152 Tgl5p 146–147 Tgtp 189 thapsigargin 96 thiazoladinediones (TZDs) 45, 64, 65, 67, 165–166, 177, 184, 198
SUBJECT INDEX Tip47 12 Toll receptors 42 toll-like receptors TLR4 187 TLR3 187 transducer of regulated CREB activity 2 (TORC2) 78 triacylglycerol lipases 146–147 triacylglycerol (TAG) 25, 142, 143, 145, 154 biosynthesis 40, 143–144 hydrolysis 146–147 tricarboxylic acid (TCA) cycle 26, 28, 30– 32, 34, 35–36, 39, 41 triolein 49 TrkB receptor 15 TUDCA 94, 98 tumour necrosis factor 8, 114 tumour necrosis factor α (TNFα ) 116, 118, 123, 178, 184, 193 tumour necrosis factor β (TNFβ ) 87, 88 tunicamycin 96 Type B insulin resistance 16, 18 U Ubc5 188 Ubc9 190 UbcH5 185 UDCA 98
211 uncoupling protein UCP1 4, 5, 156 UCP2/3 story 43 UCP3 5 unfolded protein response (UPR) 103 UPAR 95 UPR–CHOP pathways 104 UPR–SRA 102–105, 105, 106 V vectorial acylation 127–136, 138–141 FATP and ACSL 134 very long-chain acyl CoA synthetases (ACSVL) 131 VLDL 100 W white adipose tissue (WAT) 156, 164 Wolcoott-Rallison Syndrome 97 X XBP1 90 XBP1 haploinsufficiency 97 XBP splicing 95 XPP splicing 95 Y Yeh1p and Yeh2p 148