Current Topics in Developmental Biology, Volume 63 (Current Topics in Developmental Biology)

260 Contents of Previous Volumes Volume 63 1. Early Events in the DNA Damage Response Irene Ward and Junjie Chen 2. ...

Author: Gerald P. Schatten

21 downloads 890 Views 3MB Size Report

This content was uploaded by our users and we assume good faith they have the permission to share this book. If you own the copyright to this book and it is wrongfully on our website, we offer a simple DMCA procedure to remove your content from our site. Start by pressing the button below!
Report copyright / DMCA form

DOWNLOAD PDF

260

Contents of Previous Volumes

Volume 63 1. Early Events in the DNA Damage Response Irene Ward and Junjie Chen

2. Afrotherian Origins and Interrelationships: New Views and Future Prospects Terence J. Robinson and Erik R. Seiffert

3. The Role of Antisense Transcription in the Regulation of X-Inactivation Claire Rougeulle and Philip Avner

4. The Genetics of Hiding the Corpse: Engulfment and Degradation of Apoptotic Cells in C. elegans and D. melanogaster Zheng Zhou, Paolo M. Mangahas, and Xiaomeng Yu

5. Beginning and Ending an Actin Filament: Control at the Barbed End Sally H. Zigmond

6. Life Extension in the Dwarf Mouse Andrzej Bartke and Holly Brown-Borg

Volume 64 1. Stem/Progenitor Cells in Lung Morphogenesis, Repair, and Regeneration David Warburton, Mary Anne Berberich, and Barbara Driscoll

2. Lessons from a Canine Model of Compensatory Lung Growth Connie C. W. Hsia

3. Airway Glandular Development and Stem Cells Xiaoming Liu, Ryan R. Driskell, and John F. Engelhardt

4. Gene Expression Studies in Lung Development and Lung Stem Cell Biology Thomas J. Mariani and Naftali Kaminski

5. Mechanisms and Regulation of Lung Vascular Development Michelle Haynes Pauling and Thiennu H. Vu

6. The Engineering of Tissues Using Progenitor Cells Nancy L. Parenteau, Lawrence Rosenberg, and Janet Hardin-Young

Series Editor Gerald P. Schatten Director, PITTSBURGH DEVELOPMENTAL CENTER Deputy Director, Magee-Women’s Research Institute Professor and Vice-Chair of Ob-Gyn Reproductive Sci. & Cell Biol.-Physiology University of Pittsburgh School of Medicine Pittsburgh, Pennsylvania 15213

Editorial Board Peter Gru¨ss Max-Planck-Institute of Biophysical Chemistry Go¨ttingen, Germany

Phillip Ingham University of Sheffield, United Kingdom

Mary Lou King University of Miami, Florida

Story C. Landis National Institutes of Health National Institute of Neurological Disorders and Stroke Bethesda, Maryland

David R. McClay Duke University, Durham, North Carolina

Yoshitaka Nagahama National Institute for Basic Biology, Okazaki, Japan

Susan Strome Indiana University, Bloomington, Indiana

Virginia Walbot Stanford University, Palo Alto, California

Founding Editors A. A. Moscona Alberto Monroy

Contributors Numbers in parentheses indicate the pages on which the authors’ contributions begin.

Philip Avner (61), Unite´ de Geńe´tique Molećulaire Murine, Institut Pasteur, 75015 Paris, France Andrzej Bartke (189), Geriatrics Research, Department of Medicine, Southern Illinois University School of Medicine, Springfield, Illinois 62794 Holly Brown-Borg (189), Department of Pharmacology, Physiology and Therapeutics, University of North Dakota School of Medicine and Health Sciences, Grand Forks, North Dakota 58203 Junjie Chen (1), Division of Oncology Research, Mayo Clinic, Rochester, Minnesota 55905 Paolo M. Mangahas (91), Program in Developmental Biology, Baylor College of Medicine, Houston, Texas 77030 Terence J. Robinson (37), Evolutionary Genomics Group, Department of Zoology, University of Stellenbosch, Matieland 7602, South Africa Claire Rougeulle (61), Unite´ de Geńe´tique Molećulaire Murine, Institut Pasteur, 75015 Paris, France Erik R. SeiVert (37), Department of Earth Sciences, Oxford University, Oxford OX1 3PK, United Kingdom Irene Ward (1), Division of Oncology Research, Mayo Clinic, Rochester, Minnesota 55905 Xiaomeng Yu (91), Verna and Marrs McLean Department of Biochemistry and Molecular Biology, Baylor College of Medicine, Houston, Texas 77030 Zheng Zhou (91), Verna and Marrs McLean Department of Biochemistry and Molecular Biology and Program in Developmental Biology, Baylor College of Medicine, Houston, Texas 77030 Sally H. Zigmond (145), Biology Department, University of Pennsylvania, Philadelphia, Pennsylvania 19104

ix

Preface This volume of Current Topics in Developmental Biology showcases an exciting array of topics in our field, from the protein response to DNA damage, to the controversy of mammalian taxonomy, to the role of antisense transcription in X inactivation, to the disposal of apoptotic cells, to actin filament formation, to longevity in dwarf mouse strains. For the developmental biology student seeking an exciting niche to study, this volume highlights a wealth of opportunity. Early Events in the DNA Damage Response by Irene Ward and Junjie Chen of the Mayo Clinic explores the proteins that respond to doublestrand breaks and replication arrest in DNA, and reveals what is becoming an increasingly complex picture of how these molecules identify and mediate DNA damage, which may lead to insights to and interventions for cancer. In Afrotherian Origins and Interrelationships: New Views and Future Prospects by Terence Robinson of the University of Stellenbosch and Erik SeiVert of Duke University, the authors consider the controversial mammalian clade Afrotheria, a diverse collection including the mighty elephant, the sea-going manatee, and the aardvark. Do these mammals truly have the same ancestral phyla in common? While genetically it would appear so, the morphological data is confusing. The authors encourage both more sophisticated molecular testing and continued study of the fossil record to resolve this question. The Role of Antisense Transcription in the Regulation of X-Inactivation by Claire Rougeulle and Philip Avner of Institut Pasteur is a sweeping review of our present understanding of how the group of Tsix antisense transcripts contribute to imprinted and random inactivation. As our knowledge of the function of non-coding RNAs increases, the authors counsel that we reconsider labeling such portions of the genome as ‘‘junk DNA.’’ The Genetics of Hiding the Corpse: Engulfment and Degradation of Apoptotic Cells in C. elegans and D. melanogaster by Zheng Zhou, Paolo Manghas and Xiaomeng Yu of Baylor examines the proteins and receptors that make dying cells recognizable, and those responsible for initiating disposal by neighbor cells, with important implications regarding these processes in mammals, since phagocytosis impacts such mechanisms as inflammation and immune response. In Beginning and Ending an Actin Filament: Control at the Barbed End by Sally Zigmond of the University of Pennsylvania describes the mechanisms whereby new filaments are formed and how they are elongated, and how filaments are capped. A suite of proteins acting as a complex are xi

xii

Preface

responsible for this interplay, similar to the protein interplay inherent in cell migration and, probably, in other cellular dynamics. Finally, in Life Extension in the Dwarf Mouse by Andrzej Bartke of Southern Illinois University and Holly Brown-Borg of the University of North Dakota, the authors consider the common factors contributing to longevity in several lines of dwarf mice. In many, the reduced synthesis of insulin-like growth factor seems to result in reduced cellular aging via oxidative stress, probably from reduced metabolic function. Intriguingly, animals subject to caloric restriction display a similar heightened response to oxidative stress, including a lower incidence of cancer. This volume has benefited from the ongoing cooperation of a team of participants who are jointly responsible for the content and quality of its material. The authors deserve the full credit for their success in covering their subjects in depth yet with clarity, and for challenging the reader to think about these topics in new ways. The members of the Editorial Board are thanked for their suggestions of topics and authors. I also thank Leah KauVman for her fabulous editorial insight and Anna Vacca for her exemplary administrative support. Finally, we are grateful to everyone at the Pittsburgh Development Center of Magee-Womens Research Institute here at the University of Pittsburgh School of Medicine for providing intellectual and infrastructural support for Current Topics in Developmental Biology. Jerry Schatten Pittsburgh Development Center, Pennsylvania

1

____________________________________________________________________________

Early Events in the DNA Damage Response Irene Ward and Junjie Chen Division of Oncology Research Mayo Clinic Rochester, Minnesota 55905

I. Introduction II. Formation of Multiprotein Complexes A. ATM and ATR B. MRN Complex C. MDC1/NFBD1 D. 53BP1 E. H2AX F. DNA-PK G. Rad51/BRCA1/BRCA2 H. Rad17, the 9-1-1 Complex, and Claspin III. Concluding Remarks Acknowledgments References

The ability to sense DNA damage and activate response pathways that coordinate cell cycle progression and DNA repair is essential for the maintenance of genomic integrity and the viability of organisms. During the last couple of years, several proteins have been identified that participate very early in the DNA damage response. Here we review the current understanding of the mechanisms by which mammalian cells detect DNA lesions, especially double-strand breaks, and mediate the signal to downstream transducers. C 2004, Elsevier Inc.

I. Introduction DNA constantly encounters potentially deleterious assaults from both environmental and endogenous sources. To protect the integrity of their DNA, cells have evolved a variety of response pathways that initiate repair and carefully coordinate it with DNA transcription, replication, and cell-cycle progression. The main repair strategies are direct reversal of lesions, excision of damaged DNA, and rejoining of DNA breaks (Table I). Direct repair of certain alkylation adducts or UV-induced photolesions by specialized single enzymes is the simplest and perhaps oldest repair Current Topics in Developmental Biology, Vol. 63 Copyright 2004, Elsevier Inc. All rights reserved. 0070-2153/04 $35.00

1

2

Ward and Chen

Table I Overview of the Major DNA Repair Mechanisms Main inducer

Type of damage

Ultraviolet light

CPDs, 6-4PPs

Ultraviolet light certain chemotherapeutic drugs or environmental toxins (e.g., cisplatin or PAHs) Oxygen radicals and other products from cellular metabolism (oxidation, hydrolysis, methylation) Errant replication

CPDs, 6-4PPs intrastrand adducts or other bulky adducts

Bistranded BER-induced SSBs, recombination, replication fork collapse, ionizing radiation Cisplatin

Repair pathway Direct repair (photoreactivation) Nucleotide excision repair (NER)

Non-bulky base modifications

Base excision repair (BER)

Mispaired bases, insertions, deletions Double-strand breaks (DSBs)

Mismatch repair (MMR) Non-homologous end-joining (NHEJ) and/or homologous recombination (HR)

Interstrand crosslinks

Abbreviations: CPDs, cyclobutane pyrimidine dimers; 6–4PPs, 6–4 photoproducts; PAHs, polycyclic aromatic hydrocarbons; SSBs, single-strand breaks.

mechanism. It is conserved from bacteria to vertebrates, although humans seem to lack photolyases, the enzymes that reverse UV damage. UV lesions in humans are solely targeted by the nucleotide excision repair (NER) pathway. This versatile pathway also repairs various other bulky, helix-distorting lesions that arise, for instance, from exposure to genotoxic compounds such as polycyclic aromatic hydrocarbons (PAH). NER is a multistep process that comprises recognition of disrupted base pairing followed by unwinding of the DNA helix around the lesion and dual incision. The damaged oligonucleotide patch is subsequently excised, and the remaining gap is filled by regular DNA replication using the intact complementary strand as a template. A subpathway of NER, termed transcription-coupled repair (TCR) (versus global genomic repair [GGR]), targets damage that blocks DNA transcription and involves displacement of the stalled RNA polymerase (reviewed in Cleaver et al., 2001). In addition, cells can use special polymerases to read through a lesion that blocks the normal replication machinery, although this aberrant translesion synthesis often comes at the expense of inserting point mutations (reviewed in Goodman and Tippin, 2000). Another excision repair pathway, mismatch repair (MMR), targets mispaired bases and nucleotide insertion/deletion loops that arise during errant

1. Early Events in the DNA Damage Response

3

DNA replication. Strand discrimination in eukaryotic cells is not yet fully understood but is thought to occur by contact of MMR proteins with the replication machinery (reviewed in Schofield and Hsieh, 2003). Non-bulky base modifications, which are primarily caused by the normal cellular metabolism processes such as oxidation, hydrolysis, and nonenzymatic methylation as well as by the intrinsic molecular instability of the DNA itself, are mainly removed by the base excision repair pathway (BER). In BER, specific DNA glycosylases recognize and excise the modified base. The resulting abasic sugar is cleaved by an endonuclease. DNA pol subsequently removes the 50 -terminal deoxyribose-phosphate residue and fills the single-nucleotide gap. The remaining nick is then sealed by a DNA ligase (reviewed in Memisoglu and Samson, 2000). If single base lesions occur closely spaced on opposite strands, processing by BER can give raise to double-strand breaks (DSBs). Such bistranded damage clusters can form as a consequence of endogenous base damage or result from free radicals generated during radiolysis of water upon exposure of cells to ionizing radiation (IR) (Sutherland et al., 2003; Wallace, 1998). IR can also introduce DSBs directly by depositing energy within the DNA and causing multiple breaks. Other important sources of DSB include HO endonuclease-induced DSBs that start mating type switch in yeast (Haber, 1992) and Spo11 transesterase-induced DSBs that initiate meiotic recombination in yeast and mammals (Mahadevaiah et al., 2001; Sun et al., 1989). DSBs are also introduced during the process of V(D)J recombination and class switch recombination (CSR), which is part of the normal development of the immune repertoire in B and T lymphocytes (Gellert et al., 1992; Honjo et al., 2002). Moreover, DSBs arise frequently during DNA replication when replication forks encounter single-strand breaks and collapse (Thompson and Schild, 2002). DSBs are more challenging to repair than other DNA lesions and are considered the most toxic type of DNA damage. If left unrepaired or repaired improperly, they cause chromosomal aberrations such as translocations, amplifications, or deletions, which may be lethal or result in oncogenic transformation (Difilippantonio et al., 2002; Zhu et al., 2002). The two major pathways of DNA DSB repair are homologous recombination (HR), a highly accurate process that requires large regions of homologous sequence as a template, and nonhomologous DNA endjoining (NHEJ), which simply joins broken ends together, thereby often generating deletions, insertions, or base pair substitutions. If substantial regions of homology flank a DSB, cells can use a third repair pathway termed single-strand annealing (SSA), which involves the interaction of the two repeats and results in the loss of one flanking region plus the intervening DNA (Lin et al., 1984). Similarly, very small, so-called microhomology regions can be used by a subpathway of NHEJ, which has also been designated

4

Ward and Chen

microhomology-driven SSA (Gottlich et al., 1998), direct-repeat end-joining (Thacker et al., 1999), and error-prone NHEJ (PfeiVer et al., 2000). In diploid yeast, DNA DSBs seem to be repaired almost exclusively through high-fidelity HR. Mammalian cells use recombinational repair as well (Liang et al., 1998), although NHEJ makes an important contribution to DSB repair, especially during the G0 and G1 phases of the cell cycle when no sister chromatid is available (Lee et al., 1997; Takata et al., 1998). In addition, the relative contribution of HR and NHEJ appears to change with the developmental stage of a cell, with HR being the major repair pathway in embryonic cells, while NHEJ dominates in diVerentiated somatic cells (Essers et al., 2000). Repair of a DSB by HR involves 50 ! 30 resection of the broken DNA ends followed by identification and invasion of the homologous sequence at the sister chromatid or homologous chromosome. The 30 overhangs of the invading strands then serve as primers for DNA synthesis, using the intact strand as a template. In contrast, NHEJ comprises simply the alignment of DSBs, which may have to be modified by nucleases and/or polymerases to obtain compatible ends that can then be ligated. To allow time for repair of the various types of DNA lesions and to prevent damage from being passed onto daughter cells, cells activate socalled checkpoint signaling pathways that sense DNA lesions, amplify the signal, and transiently arrest or slow cell cycle progression. In addition, checkpoint pathways induce transcriptional programs and enhance DNA repair pathways. Although over the past decades much progress has been made in dissecting the diVerent DNA damage response pathways, less is known about the initial events that trigger cell cycle checkpoints and stimulate DNA repair. In this chapter, we focus on the proteins that participate early in the response to DNA DSBs (e.g., ATM, DNA-PK, MRN complex, H2AX, MDC1, 53BP1, Chk2) and/or replication arrest (e.g., ATR, Rad17, 9-1-1 complex, Chk1) and discuss their role in safeguarding genome integrity.

II. Formation of Multiprotein Complexes The dynamic formation of large multiprotein complexes at the region surrounding DNA lesions provided important insight into the early events in response to DNA damage. Among the first proteins that relocalize to these nuclear foci are MDC1/NFBD1, 53BP1, and the Mre11-Rad50-NBS1 (MRN) complex (Fig. 1). Induction of DNA DSBs in defined subnuclear volumes using ultrasoft X rays demonstrated that these foci indeed form at sites of DNA strand breaks (Nelms et al., 1998). Moreover, immunofluorescence analyses showed that the proteins colocalize extensively with foci


5

Figure 1 Formation of multiprotein complexes at the sites of DNA double-strand breaks. Exposure of cells to ionizing radiation results in the rapid recruitment of numerous proteins to the sites of DNA lesions. The ATM (ataxia telangiectasia mutated) kinase, which is central to this response, initiates a cascade of phosphorylation events (P) that activate cell cycle checkpoint pathways and, if necessary, apoptosis. How ATM participates in DNA repair is not well defined. In contrast, the related DNA-PK kinase, consisting of the DNA-PKcs and KU70/ Ku80 subunits, attaches to DNA ends and is essential for nonhomologous DNA end-joining.

formed by phsophorylated H2AX ( -H2AX), a variant of histone H2A that is randomly incorporated in approximately 20–30% of nucleosomes (Rogakou et al., 1998). H2AX phosphorylation is damage dependent, and experiments using a pulsed microbeam laser to introduce DNA doublestrand breaks into specific partial nuclear volumes of cells revealed that H2AX phosphorylation is confined to megabase areas surrounding strand breaks (Rogakou et al., 1999). Phosphorylation of H2AX in response to IR is mediated by ATM (ataxia telangiectasia mutated) (Burma et al., 2001), while the related ATR (ATM and Rad3-related) kinase phosphorylates H2AX in response to replication arrest (Ward and Chen, 2001). ATM and ATR have also been shown to phosphorylate numerous other proteins recruited to sites of DNA damage, including MDC1/NFBD1, 53BP1, NBS1 and members of the Rad9-Rad1-Hus1 (9-1-1) complex, and are thought to be key regulators in the DNA damage response.

6

Ward and Chen

A. ATM and ATR ATM and ATR are conserved serine-threonine kinases characterized by a Cterminal catalytic motif containing a phosphatidylinositol 3-kinase domain. The gene that encodes ATM is mutated in the severe auotosomal recessive disorder ataxia telangiectasia (A-T). A-T patients suVer from progressive cerebellar degeneration, immunodeficiency, growth retardation, hypogonadism, chromosomal instability, and cancer predisposition (Gatti et al., 2001). At the cellular level, A-T cells show hypersensitivity to IR, radio-resistant DNA synthesis (RDS), and a high frequency of chromosome aberrations (Abraham, 2001; Shiloh, 2003). ATR deficiency is even more severe, resulting in early embryonic lethality in mice (Brown and Baltimore, 2000). Partial loss of ATR activity has been associated with Seckel syndrome, a rare inherited disorder characterized by intrauterine growth retardation and microcephaly (O’Driscoll et al., 2003). ATM is primarily activated in response to DSBs, while ATR reacts to a wider range of lesions, including stalled replication forks. Both proteins are implicated in the sensing of DNA damage and/or the transducing of the damage signal and have been shown to associate with DNA in vitro (Smith et al., 1999; Suzuki et al., 1999; Unsal-Kacmaz et al., 2002). Moreover, ATR undergoes dramatic relocation to sites of stalled replication forks in response to replication stress (Tibbetts et al., 2000). Similarly, detergent extraction revealed rapid changes in the subcellular localization of ATM in response to radiomimetic agents, suggesting that a fraction of the ATM pool associates with sites of DNA DSBs (Andegeko et al., 2001). It was therefore thought that both ATM and ATR might be activated through interaction with DNA or DNA-associated sensing units. However, a study by Bakkenist and Kastan (2003) showed that ATM exists as an inactive dimer or multimer in undamaged cells with the kinase domain of each molecule bound to the FAT (FRAP/ATM/TRRAP) domain of another ATM molecule. DSBspecific alterations in the higher order chromatin structure or exposure of cells to hypotonic stress or chromatin-modifying agents result in the dissociation of the ATM molecules. Dimer dissociation is induced independent of direct DNA binding by intermolecular autophosphorylation of ATM on Ser 1981 and results in monomers with accessible kinase domains that are free to migrate and phosphorylate substrates. Their finding is supported by an in vitro study showing that ATM can be activated by ATP in the absence of DNA by a mechanism involving autophosphorylation (Kozlov et al., 2003). It remains to be seen whether ATR becomes activated by a similar mechanism. The in vitro kinase activity of ATR seems not to increase after exposure of cells to various genotoxic agents (Abraham, 2001), although kinase-dead ATR failed to relocalize in response to DNA damage (Barr et al., 2003). In vitro studies suggest that ATR-interacting protein (ATRIP),


7

which forms a stable complex with ATR and regulates ATR expression (Cortez et al., 2001), targets ATR to sites of DNA damage. ATRIP has a high aYnity to replication protein A (RPA)-coated single-stranded DNA (ssDNA), suggesting that this is the critical structure that recruits the ATR– ATRIP complex (Zou and Elledge, 2003). Studies using a cell-free system derived from Xenopus laevis eggs (Lupardus et al., 2002) as well as findings in our own laboratory (Ward et al., 2004a) indicate that ATR activation is primarily linked to damage-induced replication arrest, suggesting that DNA nicks per se are insuYcient to recruit ATR. Once activated, ATR and ATM amplify the damage signal by phosphorylating various substrates. Both kinases share the same minimal essential phosphorylation consensus sequence (Kim et al., 1999); substrate selection might be based on spatiotemporal interactions. The first ATM phosphorylation site identified was serine residue 15 (Ser 15) of p53 (Banin et al., 1998; Canman et al., 1998; Khanna et al., 1998), which is involved in enhancing the transcriptional transactivation activity of p53 (Dumaz et al., 1999). ATM had been previously implicated in p53 regulation, because p53 accumulation is diminished or severely delayed in irradiated A-T cells (Kastan et al., 1992). However, IR-induced Ser 15 phosphorylation is not completely suppressed in A-T cells (Canman et al., 1998), and subsequent in vivo studies showed that ATR participates in the late IR-induced phosphorylation of Ser 15 (Cliby et al., 1998) and phosphorylates this site in response to UV irradiation (Tibbetts et al., 1999). ATM also regulates p53 activity via phosphorylation of MDM2 on Ser 395. MDM2 is an E3 ubiquitin ligase that targets p53 for nuclear export and proteosome-mediated degradation. Phosphorylation of MDM2 at Ser 395 allows the interaction with p53 but inhibits p53 nuclear export and subsequent degradation (Maya et al., 2001). Another control measure involves ATM phosphorylation of the eVector kinase Chk2 on Thr 68 (Ahn et al., 2000; Matsuoka et al., 2000; Melchionna et al., 2000), which in turn phosphorylates p53 on Ser 20, thereby blocking the interaction of p53 with the negative regulator MDM2 (Chehab et al., 2000; Hirao et al., 2000; Shieh et al., 2000). However, a physiological role of Chk2 in regulating p53 has been questioned (Ahn et al., 2003), although lossof-function mutations in Chk2 have been found in a subset of patients with Li-Fraumeni syndrome, a highly penetrant cancer phenotype usually associated with inherited mutations in p53 (Bell et al., 1999). Together, activation of p53 in response to DNA damage results in the upregulation of a number of p53 target genes, including the cyclin-dependent kinase inhibitor p21waf1/CIP1, which suppresses cyclin E/Cdk2 kinase activity and prevents G1-to-S-phase progression (G1 checkpoint) (Deng et al., 1995). However, cyclin E/Cdk2 kinase activity can also be suppressed in cells lacking p53 or p21. This fact led to the discovery of a second ATMor ATR-dependent pathway that is involved in the rapid initiation of the G1

8

Ward and Chen

checkpoint (Bartek et al., 2001). This pathway targets Cdc25A, a phosphatase that activates cyclin E/Cdk2 by dephosphorylating Thr14 and Tyr 15 of Cdk2. Phosphorylation of Chk2 or Chk1 by ATM and ATR, respectively, results in the phosphorylation and subsequent ubiquination of Cdc25A and prevents the activation of cyclin E/Cdk2 (Bartek et al., 2001). Entry into S-phase is blocked by preventing Cdk2-mediated loading of Cdc45 onto replication origins, a prerequisite for the recruitment of DNA polymerase (Costanzo et al., 2000; Falck et al., 2001). The Cdc25A degradation pathway and prevention of cyclin E/cdk2 and cyclin A/cdk2 activation are also critical for the intra-S-phase checkpoint. S-phase cells transiently decrease the rate of DNA synthesis in response to radiation by blocking the firing of late replication origins and decreasing the rate of strand elongation (Larner et al., 1999; Painter et al., 1980). In contrast, cells from A-T patients fail to slow down DNA synthesis after exposure to IR and show a typical RDS phenotype (Painter et al., 1980). However, full execution of the intra-S-phase checkpoint requires the activation of a second ATM-dependent pathway (Falck et al., 2002). This parallel pathway includes phosphorylation of NBS1, the product of the gene aVected in Nijmegen breakage syndrome (NBS), a chromosomal instability syndrome. ATM phosphorylates NBS1 at multiple sites, including Ser 278, Ser 343, Ser 397, and Ser 615 (Gatei et al., 2000; Lim et al., 2000; Wu et al., 2000; Zhao et al., 2000). In addition, ATM phosphorylates BRCA1, the product of the breast cancer susceptibility gene 1, at multiple sites, including Ser 1189, Ser 1524, and Ser 1542 (Cortez et al., 1999) as well as SMC1, structural maintenance of chromosome protein 1, at Ser 957 and Ser 966 (Kim et al., 2002; Yazdi et al., 2002). Phosphorylation of all three proteins is required for full activation of the intra-S-phase checkpoint (Falck et al., 2002; Kim et al., 2002; Yazdi et al., 2002). Moreover, NBS1 and BRCA1 are required for optimal phosphorylation of SMC1 by ATM (Kim et al., 2002; Yazdi et al., 2002). SMC1 is a component of the cohesin complex, essential for sister chromatid cohesion during mitosis (reviewed in Hirano et al., 2002). The establishment of sister chromatid cohesion has been linked to proliferating cell nuclear antigen (PCNA)-dependent DNA replication (Skibbens et al., 1999), but the precise details of how the NBS1–SMC1 branch of the intra-S-phase checkpoint inhibits DNA synthesis remain to be elucidated. There is evidence that both S-phase checkpoint pathways can also be triggered by ATR. Inhibition of replicon initiation upon UV irradiation is severly attenuated by overexpression of kinase-inactive ATR. Similarly, pretreatment of cells with caVeine or UCN-01, inhibitors of ATR and Chk1, respectively, abolished the S-phase checkpoint response to UV treatment (HeVernan et al., 2002). Moreover, ATM-independent phosphorylation of SMC1 on Ser 957 and Ser 966 has been observed in response to UV- or hydroxyurea-induced replication arrest (Kim et al., 2002).


9

The activation of ATR by replicational stress made it a prime suspect in the control of the so-called replication checkpoint. This checkpoint ensures that cells do not enter mitosis prior to the completion of S-phase. However, recent work on conditional ATR knockout cells demonstrated that ATR (and ATM) is dispensable for the replication checkpoint. ATR-deficient cells prevent M-phase entry in response to aphidicolin-induced replication arrest and remain capable of entering mitosis once the replication inhibitor has been removed. However, ATR knockout cells that proceed to mitosis after aphidicolin treatment show a high rate of chromosome breaks, suggesting that ATR plays an essential role in stabilizing stalled replication forks and preventing the generation of DSBs (Brown and Baltimore, 2003). Entry into mitosis in response to IR is prevented by the G2 cell cycle checkpoint. Studies on A-T cells led to the discrimination of two molecularly distinct G2 checkpoints (Xu et al., 2002). One occurs early after IR and blocks G2-phase cells from progress into mitosis. This transient checkpoint depends on ATM and, to a lesser degree, on ATR. The second G2 checkpoint targets cells that had been in earlier phases of the cell cycle at the time of irradiation (most likely S-phase). This late G2 checkpoint response requires ATR and is ATM independent (Brown and Baltimore, 2003; Xu et al., 2002). In any case, IR leads to ATM-mediated phosphorylation of Chk2 on Thr 68 and/or ATR-mediated phopshorylation of Chk1 on Ser317 and Ser 345. Notably, phosphorylation of Chk1 also requires BRCA1 (Yarden et al., 2002). Activated Chk2 or Chk1 in turn phosphorylate the mitosis-promoting phosphatase Cdc25C on Ser 216. Phosphorylation of this residue creates a binding site for 14-3-3 proteins. The 14-3-3 bound form of Cdc25C is then sequestered in the cytoplasm, thereby prohibiting Cdc25C from dephosphorylating Cdc2 on Thr14 and Tyr 15. Without activation of the Cdc2/ cyclin B1 complex, cells remain in the G2-phase (Peng et al., 1997; Sanchez et al., 1997).

B. MRN Complex Given the central role of the ATM and ATR kinases in the DNA damage response, it comes as no surprise that their activity is tightly regulated. Two studies showed that the MRN complex is required for ATM activation (Carson et al., 2003; Uziel et al., 2003). This highly conserved complex is composed of the Mre11, Rad50, and Nbs1 (Xrs2 in Saccharomyces cerevisiae) proteins and has been linked to the detection and repair of DSBs as well as checkpoint signaling and DNA replication. All three proteins are essential and cause early embryonic lethality in mice (Luo et al., 1999; Xiao et al., 1997; Zhu et al., 2001). Hypomorphic mutations of Mre11 and NBS1 cause ataxia telangiectasia-like disease (A-TLD) (Stewart et al., 1999) and

10

Ward and Chen

Nijmegen breakage syndrom (NBS) (Carney et al., 1998), respectively. A-TLD is phenotypically very similar to A-T, although the clinical features develop slower. NBS does not lead to the cerebellar degeneration that leads to ataxia in A-T and A-TLD, but patients present with mental deficiency, microcephaly, immunodeficiency, and increased predisposition to lymphoid malignancies. In addition, the cellular phenotype of all three disorders is characterized by radiosensitivity, radioresistant DNA synthesis, and chromosome fragility (Stewart et al., 1999). These phenotypic similarities link the MRN complex to ATM function. Indeed, NBS1 is a substrate of ATM, and NBS1 phosphorylation is required not only for the intra-S-phase checkpoint but also for a full response of the G1 and early G2 checkpoints (Antoccia et al., 1999; Buscemi et al., 2001; Jongmans et al., 1997; Matsuura et al., 1998; Yamazaki et al., 1998). ATM-mediated phosphorylation of the MRN complex may occur at the sites of DNA damage because ATM is not required for the recruitment of NBS1 to the sites of DSBs (Lukas et al., 2003). MRN foci can be microscopically detected as early as 10 min after exposure of cells to IR (Mirzoeva et al., 2001). Mre11, with its two DNA-binding motifs, forms the core of the MRN complex that concentrates in these foci. Recent models suggest that Mre11 forms a homodimer (or multimer), which then interacts with Rad50 and NBS1. Rad50 contains an amino-terminal Walker A and a carboxy-terminal Walker B ATP-binding motif that are brought together by a flexible hinge region located in the middle of the protein. This hinge region contains a ‘‘zinc-finger-like hook’’ domain, which mediates the dimerization with the hinge region of another Rad50 molecule. Given the DNA-binding capability of Mre11, which interacts with the region proximal to the catalytic domain of Rad50, these molecules seem to form a flexible structure that can bridge DNA ends (D’Amours et al., 2002; de Jager et al., 2001; Hopfner et al., 2001; van den Bosch et al., 2003). NBS1 might stabilize the complex via its reported interaction with H2AX (Kobayashi et al., 2002) and is required for damage-induced phosphorylation and accumulation of Mre11 (Dong et al., 1999). Apart from holding DNA ends in close proximity, the MRN complex is also implicated in the initial processing of DSBs due to its enzymatic activity. Mre11 has endonuclease and 30 –50 exonuclease activities as well as stranddissociation and strand-annealing activities that are stimulated to diVerent degrees by Rad50, NBS1, and ATP (reviewed in D’Amours et al., 2002). However, because Mre11 possesses no 50 –30 exonuclease activity in vitro, it is probably not the nuclease responsible for the resection of DSBs early during HR, as previously thought. Instead, the endonuclease activity of Mre11 might be critical for the resolution of DNA hairpins that block DNA replication in S-phase cells (D’Amours et al., 2002). A role in S-phase has been suggested by the observation that the MRN complex colocalizes with


11

proliferating cell nuclear antigen (PCNA) at replication forks throughout S-phase (Maser et al., 2001; Mirzoeva et al., 2003). Moreover, immunodepletion of the MRN complex from Xenopus extracts resulted in the accumulation of DSBs in the replicated genomic DNA (Costanzo et al., 2001). Another function of the MRN complex involves the ability of NBS1 to facilitate ATM-mediated phosphorylation of other ATM substrates, such as Chk2 (Buscemi et al., 2001; Girard et al., 2002; Lee et al., 2003b), SMC1 (Kim et al., 2002; Yazdi et al., 2002), and Chk1 (Gatei et al., 2003). Interestingly, mutation of NBS1 on Ser 343, an ATM phosphorylation site, prevents this function (Buscemi et al., 2001; Yazdi et al., 2002), suggesting that ATMdependent phosphorylation of the Mre11 complex initiates this activity. However, findings have demonstrated that ATM autophosphorylation on Ser 1981 and retention of ATM on chromatin are impaired in NBS and A-TLD cells (Uziel et al., 2003). Moreover, viral-induced degradation of Mre11 has been shown to block the activation of ATM and ATR upon infection of cells with adenovirus (Carson et al., 2003), suggesting that the MRN complex may also act upstream of ATM. This is consistent with findings in the budding yeast, in which the Mre11 complex appears to activate the ATM ortholog Tel1, which in turn enhances the functions of the Mre11 complex (Usui et al., 2001).

C. MDC1/NFBD1 MDC1 (mediator of DNA damage checkpoint 1) is another ATM substrate that has been implicated in modulating ATM activity (Mochan et al., 2003). It is also termed NFBD1 (nuclear factor with BRCT domains protein 1) based on its two BRCA1 carboxy-terminal (BRCT) motifs. BRCT domains are often found in proteins involved in the DNA damage response (Bork et al., 1997; Callebaut and Mornon, 1997; Koonin et al., 1996), and work in our laboratory and others demonstrated that BRCT domains preferentially bind phosphopeptides (Manke et al., 2003; Rodriguez et al., 2003; Yu et al., 2003). At the amino terminus of MDC1 resides a forkhead-associated (FHA) domain, another conserved phosphopeptide-binding module that is often associated with proteins involved in DNA repair, cell cycle arrest, or pre-mRNA processing (Li et al., 2000). MDC1 interacts with all three constituents of the Mre11 complex in untreated and irradiated cells. Upon exposure of cells to IR, MDC1 rapidly localizes to sites of DNA strand breaks and is required for eYcient phosphorylation of H2AX and the subsequent formation of damage-induced NBS1 and BRCA1 foci (Goldberg et al., 2003; Lou et al., 2003a; Shang et al., 2003; Stewart et al., 2003; Xu et al., 2003). One study also reported defective 53BP1 foci formation in the absence

12

Ward and Chen

of MDC1 (Stewart et al., 2003), while others found no eVect on 53BP1 formation (Goldberg et al., 2003; Mochan et al., 2003; Peng and Chen, 2003). The association of MDC1 with phosphorylated H2AX seems required to keep MDC1 in the vicinity of DNA lesions because MDC1 accumulation is absent in H2AX-deficient cells (Stewart et al., 2003). Furthermore, MDC1 associates with activated Chk2 in response to DNA damage, and this interaction is mediated by the FHA domain of MDC1 and phosphorylated Thr 68 of Chk2. Chk2, in turn, controls the phosphorylation of MDC1 by ATM (Lou et al., 2003b). On the other hand, MDC1 seems not to be required to phosphorylate Chk2 (Goldberg et al., 2003; Lou et al., 2003b), although Chk2-mediated phosphorylation of Ser 20 on p53 and subsequent p53 stabilization is impaired in MDC1-deficient cells (Lou et al., 2003b). Similarly, phosphorylation of NBS1 and SMC1 occurs independent of MDC1 (Goldberg et al., 2003), but downregulation of MDC1 expression by small interfering RNA (siRNA) appears to aVect its function in the intra-S-phase checkpoint (Goldberg et al., 2003; Lou et al., 2003b; Stewart et al., 2003). This defect in the intra-S-phase checkpoint is unlikely to be mediated by the Chk2–cdc25A pathway because MDC1 is not required for degradation of Cdc25A in irradiated cells (Goldberg et al., 2003). Thus, the emerging picture suggests that MDC1 may work as an adaptor protein that links various proteins, thereby enabling them to interact in a timely manner. Such an adaptor function could account for the complex phenotype observed in cells with downregulated MDC1 expression, such as increased radiation sensitivity, impaired NHEJ repair, reduced apoptosis, and a defect in the early G2 checkpoint, in addition to the previously mentioned defect in the intra-S-phase checkpoint (Lou et al., 2003b; Stewart et al., 2003). The defect in the G2 checkpoint has been linked to impaired BRCA1 function and reduced Chk1 phosphorylation in MDC1-deficient cells (Lou et al., 2003a,b; Stewart et al., 2003). However, because all these observations have been made on cancer cell lines using the siRNA technique, an area of uncertainty remains. It will be necessary and very interesting to study murine MDC1 knockout models.

D. 53BP1 53BP1 is another ATM substrate that participates very early in the DNA damage response. It was originally identified in a yeast two-hybrid screen as a protein that binds to the central DNA-binding domain of p53 and activates p53-dependent gene transcription (Iwabuchi et al., 1994, 1998). Subsequent studies in Xenopus confirmed that X53BP1 interacts with Xp53 throughout the cell cycle in embryonic extracts. However, this interaction may depend on the developmental stage, as no association between X53BP1


13

and Xp53 was detected in somatic cells (Xia et al., 2001). Like MDC1, 53BP1 contains two carboxy-teminal BRCT domains and has been proposed to be the human ortholog of budding yeast Rad9 due to a limited sequence homology of this tandem BRCT motif. scRad9 was the first checkpoint gene identified (Weinert et al., 1988) and is required for G1, S, and G2 cell cycle arrests and transcriptional induction of DNA repair genes in response to DNA damage. DNA damage induces hyperphosphorylation of scRad9 in a Mec1p (budding yeast homolog of ATR)/Tel1 (homolog of ATM)– dependent manner. Hyperphosphorylated scRad9, in turn, physically interacts with Rad53, the budding yeast homolog of Chk2, resulting in the phosphorylation and activation of the checkpoint kinase (Emili, 1998; Schwartz et al., 2002; Sun et al., 1998; Vialard et al., 1998). 53BP1 relocates rapidly to sites of DSBs upon exposure of cells to IR (Rappold et al., 2001; Schultz et al., 2000; Xia et al., 2001). The accumulation of 53BP1 at these sites depends on phosphorylated H2AX (Bassing et al., 2002; Celeste et al., 2002) and seems to be mediated by a direct interaction between 53BP1 and -H2AX (Ward et al., 2003a). In addition, 53BP1 relocation may depend on histone deacetylase 4 (HDAC4), which is reported to constitutively interact with 53BP1 (Kao et al., 2003). 53BP1 is a substrate of ATM that becomes phopshorylated in response to IR on several sites, including Ser 25 and Ser 29 (Fernandez-Capetillo et al., 2002; Rappold et al., 2001; Ward et al., 2003a). The two proteins can also be co-immunoprecipitated from extracts of IR-treated, but not undamaged, cells (DiTullio et al., 2002). ATM-mediated 53BP1 phosphorylation is not required for the recruitment of 53BP1 to sites of DNA damage (Ward et al., 2003a), a finding that is similar to what has been reported for MRN and MDC1 foci formation (Goldberg et al., 2003; Mirzoeva et al., 2001; Stewart et al., 2003). Indeed, 53BP1 phosphorylation might preceed 53BP1 relocation. This is inferred from the observation that mutant 53BP1 that lacks the ability to form damage-induced foci can still be phosphorylated (Ward et al., 2003a) and that phosphorylation of 53BP1 on Ser 25 is only slightly impaired in H2AX-deficient cells (Fernandez-Capetillo et al., 2002). The delayed 53BP1 foci formation observed in AT cells after low-dose radiation (Rappold et al., 2001) is probably secondary to impaired H2AX phosphorylation. 53BP1 is activated not only in response to IR but also upon exposure to UV irradiation and hydroxyurea, treatments that induce replication arrest, the former by introducing helix-distorting nucleotide dimers, the latter by inhibiting ribonucleotide reductase that blocks replication through nucleotide deprivation (Rappold et al., 2001). Co-immunoprecipitation with ATR suggests that ATR is the kinase that phosphorylates 53BP1 in response to such replicational stress (Morales et al., 2003). Spindle disruption with colcemid also results in 53BP1 phosphorylation. Moreover, 53BP1 localizes

14

Ward and Chen

to the corona of kinetochores in mitosis, suggesting a potential function in mitotic checkpoint signaling (Jullien et al., 2002). A role of 53BP1 in cell cycle checkpoint control has also been postulated based on findings in 53BP1-deficient cells. siRNA-mediated downregulation of 53BP1 in diVerent cancer cell lines results in an RDS phenotype (Wang et al., 2002) and a defect in the early G2 checkpoint in response to low-dose radiation (DiTullio et al., 2002; Wang et al., 2002). In addition, p53 accumulation was decreased, and phosphorylation of Chk2 on Thr 68 as well as phosphorylation of SMC1 on Ser 966 was reduced in extracts from IR-treated cells (DiTullio et al., 2002; Wang et al., 2002). Moreover, IR-induced foci recognized by an antiserum raised against the phospho-Ser/Thr-Gln epitope of ATM/ATR substrates were abolished in 53BP1-silenced cells, suggesting that 53BP1 facilitates the phosphorylation of ATM substrates, perhaps by recruiting them to the sites of strand breaks (DiTullio et al., 2002). This would be consistent with a model in the budding yeast in which scRad9 recruits Rad53 to the Mec1 complex. However, findings have suggested that scRad9 does not function by bringing Rad53 in the proximity of Mec1; instead, Mec1/Tel1-dependent phosphorylation of Rad9 multimers may provide docking sites for the FHA domains of Rad53 molecules and bring them close enough together to allow transautophosphorylation (Gilbert et al., 2001). Moreover, unlike the checkpoint defects in scRad9 mutants, cells from mice with a targeted disruption of 53BP1 show only a very modest defect in the intra-S-phase and early G2 checkpoints (Fernandez-Capetillo et al., 2002; Morales et al., 2003; Ward et al., 2003b), suggesting a minor or redundant contribution of 53BP1 to cell cycle checkpoint control. The prolonged G2 accumulation observed in irradiated 53BP1-null cells (Ward et al., 2003b) is not a checkpoint defect but could be the result of a repair deficiency that leads to prolonged checkpoint activation. Consistent with a potential defect in DNA repair, 53BP1deficient mice are hypersensitive to IR and mouse embryonic fibroblasts (MEFs) derived from these animals show an increased rate of intrinsic and IR-induced chromosomal abnormalities (Morales et al., 2003; Ward et al., 2003b). However, evidence for an involvement of 53BP1 in DNA DSB repair is sparse. 53BP1-deficient embryonic cells appear to have normal rates of homologous recombination, and 53BP1 knockout mice sustain normal levels of NHEJ-dependent V(D)J recombination (Ward et al., 2004b). The only clear repair defect is seen in a special type of recombination—immunoglobulin CSR, which, like V(D)J recombination, involves the end-joining of DNA DSBs (Ward et al., in press). Similar observations have been made in ATM-deficient and H2AX-deficient mice, which are hypersensitive to IR and feature impaired CSR but sustain normal V(D)J recombination and show no gross repair defects on a cellular level (Bassing et al., 2002; Celeste et al., 2002; Reina-San-Martin et al., 2003). Other phenotypic similarities among ATM-, H2AX-, and 53BP1-deficient mice include (albeit to diVerent


15

degrees) growth retardation, immunodeficiency, predisposition to lymphoid malignancies, and genomic instability (Barlow et al., 1996; Bassing et al., 2003; Celeste et al., 2002; Elson et al., 1996; Morales et al., 2003; Ward et al., 2003b).

E. H2AX The significant overlap in the phenotypes of ATM, H2AX, and 53BP1 knockout mice, with ATM deficiency leading to the most severe defects, is consistent with a model in which activated ATM phosphorylates H2AX on Ser 139 at megabase regions flanking the sides of DSBs (Rogakou et al., 1998), thus generating the signal for the subsequent recruitment of 53BP1 and other repair and/or checkpoint proteins (Paull et al., 2000). Indeed, damage-induced H2AX foci formation is severely compromised in ATMdeficient fibroblasts (Burma et al., 2001) and 53BP1, MRN, and BRCA1 foci could not be detected in H2AX-deficient MEFs analyzed 45 min or more after IR (Bassing et al., 2002; Celeste et al., 2002, 2003b). However, using ‘‘laser scissors’’ to introduce DSBs along a defined path, Nussenzweig and colleagues (Celeste et al., 2003b) showed that H2AX-deficient cells sustain normal recruitment of these proteins to the sites of DSBs but fail to retain and concentrate them. The fact that H2AX-null mice are viable and appear to have a normal lifespan suggests that this initial recruitment of checkpoint and/or repair proteins is suYcient to elicit a damage response. Nonetheless, H2AX does play a critical role in protecting genomic integrity. Two studies showed that H2AX deficiency and, interestingly, also H2AX haploinsuYciency markedly enhance the susceptibility to cancer in the absence of p53. The onset of cancer was earlier in the homozygous null background than in the heterozygous background, suggesting that the amount of H2AX inversely correlates with the accumulation of oncogenic translocations. Furthermore, both groups concluded that the observed chromosomal aberrations in H2AX-deficient lymphomas arose from failure to correctly repair DSBs that occurred either spontaneously or in the context of V(D)J recombination (Bassing et al., 2003; Celeste et al., 2003a). Noteworthy is the fact that human H2AX maps together with ATM to chromosome 11q23.3, a region that is deleted in a large number of tumors. How H2AX contributes to the repair of DSBs remains to be defined. H2AX-deficient cells are sensitive to the DNA cross-linking agent mitomycin C and show impaired homologous integration of targeting constructs as well as increased levels of sister chromatid exchanges (SCE), suggesting that HR is less eVective in the absence of H2AX (Celeste et al., 2002). In addition, H2AX-deficient mice show defective CSR (Celeste et al., 2002; Reina-SanMartin et al., 2003), a process that is thought to require NHEJ (Manis et al.,

16

Ward and Chen

2002). On the other hand, NHEJ-dependent V(D)J recombination appears to be normal in the absence of H2AX (Bassing et al., 2002; Celeste et al., 2002). In S. cerevisiae, phosphorylation of the conserved carboxy terminus of H2A, the homolog of H2AX, seems to be required for eYcient DSB repair by NHEJ but not HR (Downs et al., 2000). It is thought that phosphorylation of H2AX causes alterations of higher order chromatin structure that could facilitate the access or function of repair factors (Bassing et al., 2002; Celeste et al., 2002; Downs et al., 2000; Fernandez-Capetillo et al., 2003; Hacques et al., 1990; Paull et al., 2000; Reina-San-Martin et al., 2003). Apart from facilitating DNA repair, H2AX may also aVect cell cycle checkpoint responses. B cells, and, to a lesser degree, MEFs from H2AXdeficient mice show a defect in the early G2 checkpoint in response to lowdose irradiation (Fernandez-Capetillo et al., 2002). A similar dose-dependent defect has been observed in 53BP1-deficient B cells (Fernandez-Capetillo et al., 2002). The rapid phosphorylation of H2AX at regions flanking DSBs has made H2AX a great marker for DSBs. While ATM is the predominant kinase that phosphorylates H2AX following exposure of cells to IR, ATR activates H2AX at replication-associated DSBs that arise, for instance, upon UV irradiation (Ward and Chen, 2001). DNA-dependent kinase (DNA-PK), like ATM and ATR a member of the PI-3 kinase-related kinases family, has also been implicated in H2AX phosphorylation based on severely reduced H2AX foci formation in irradiated DNA-PK-deficient tumor cells (Paull et al., 2000). Moreover, IR-induced H2AX phosphorylation was abrogated by pretreatment of cells with high concentrations of wortmannin, a potent inhibitor of DNA-PK (Paull et al., 2000). However, because ATM and, to a lesser degree, ATR are also sensitive to this fungal inhibitor (Sarkaria et al., 1998) and because the DNA-PK-deficient tumor line studied expresses extremely low levels of ATM (Chan et al., 1998), the relative contribution of DNA-PK to H2AX phosphorylation remains unclear.

F. DNA-PK DNA-PK is a nuclear enzyme composed of a large catalytic subunit, DNAPKcs, and a DNA-binding and regulatory subunit, the Ku70/Ku80 heterodimer, a preformed ring that threads onto duplex DNA (Dvir et al., 1993; Gottlieb et al., 1993; Walker et al., 2001). Whereas Ku70 and Ku80 are evolutionarily conserved in yeast, DNA-PKcs seems to have evolved more recently. Like ATM and ATR, DNA-PK plays an important role in the maintenance of genomic stability. While ATM and ATR are central players in checkpoint activation, DNA-PK appears not to signal to the checkpoint machinery but is instead central to NHEJ repair. The Ku heterodimer binds


17

tightly to a variety of DNA end structures, and it has been proposed that DNA-bound Ku recruits DNA-PKcs to the extreme end of DNA and then translocates approximately 10 bp inward (Yoo and Dynan, 1999). DNAPKcs, in turn, protects DNA ends from degradation and brings them together by mediating the formation of a synaptic complex between two DNA-PKcs molecules (Chen et al., 2001; DeFazio et al., 2002). DNAassociated DNA-PKcs is subsequently activated, perhaps through the induction of conformational changes in the enzyme, and undergoes autophosphorylation. This autophosphorylation at multiple residues, including Thr 2609, Ser 2612, Thr 2638, and Thr 2647, results in the loss of protein kinase activity and seems to induce the dissociation of DNA-PKcs from the extreme end of DNA (Chan and Lees-Miller, 1996; Chan et al., 2002; Douglas et al., 2002; Merkle et al., 2002). Phosphorylated DNA-PKcs can be detected as early as 10 min after IR and colocalizes with -H2AX and 53BP1 (Chan et al., 2002). Apart from itself, DNA-PKcs also phosphorylates other DNAbinding proteins. One of them, Artemis, an endonuclease, forms a complex with DNA-PKcs and is probably recruited together with DNA-PKcs to a subset of DSBs that requires endonucleolytic processing. Once activated, DNA-PKcs phosphorylates Artemis and stimulates its endonuclease activity, inducing it to process 50 and 30 DNA overhangs or hairpins to create double-strand ends that can be ligated by the DNA ligase IV–XRCC4 complex (Ma et al., 2002). As expected, Artemis-deficient mice have an overall phenotype similar to that of DNA-PKcs-deficient mice, including severe combined immune deficiency (SCID) due to severely impaired NHEJdependent V(D)J recombination as well as hypersensitivity to IR (Rooney et al., 2002). The latter appears to be more pronounced in DNA-PKcsdeficient cells, suggesting that Artemis may play a lesser role for NHEJ of IR-induced DSBs. Another protein that becomes phosphorylated by activated DNA-PKcs is Werner protein (WRN) (Karmakar et al., 2002; Yannone et al., 2001), a member of the recQ gene family that possesses both 30 ! 50 helicase and 30 ! 50 exonuclease activities (Gray et al., 1997; Shen et al., 1998). The gene that encodes WRN is defective in patients with Werner syndrome, a human autosomal recessive disorder characterized by premature aging and cancer predisposition (Yu et al., 1996). Interaction of WRN with DNA-bound Ku functionally stimulates WRN exonuclease activity, whereas phosphorylation of WRN by DNA-PKcs seems to inhibit WRN exonuclease and helicase activities (Karmakar et al., 2002; Yannone et al., 2001) and results in the displacement of DNA-PKcs from DNA-bound Ku (Li and Comai, 2002). Thus, although WRN is thought to be involved in HR repair, it may also participate in DNA end-processing during NHEJ repair. The MRN complex has also been associated with DNA-PK and NHEJ. S. cerevisiae strains lacking Mre11, Rad50, or Xrs2 (the yeast functional

18

Ward and Chen

homolog of NBS1) show 100-fold decreased levels of NHEJ in an HRinhibited background but, unlike cells defective in core components of the NHEJ, do not generate deletions in the few ends that are joined (Boulton and Jackson, 1998). In humans, the requirement for the MRN complex in NHEJ is less clear, although the complex was identified as a fraction that restores eYcient end-joining in the presence of Ku and DNA ligase IV/XRCC4 in a cell-free system (Huang and Dynan, 2002). In contrast, there is good evidence that p53 is a target of DNA-PKcs. Immediately following IR, DNA-PKcs seems to form a DNase-sensitive complex with the pre-existing, latent form of p53 and to phosphorylate mouse p53 on Ser 18 (which correspondings to Ser 15 in human cells) (Woo et al., 2002). DNA-PKcs-deficient MEFs that were growth-deregulated by expression of the adenovirus E1A protein failed to undergo damage-induced apoptosis. Mutation of p53 on Ser 18 resulted in a decreased apoptotic response, suggesting that DNA-PKc-mediated phsophorylation of this site contributes to DNA damage-induced apoptosis (Woo et al., 2002). In support of a role of DNA-PKcs in regulating the p53-dependent apoptosis pathway, induction of apoptosis and Bax expression were shown to be significantly suppressed in the thymocytes of DNA-PKcs-null mice exposed to whole-body IR (Wang et al., 2000).

G. Rad51/BRCA1/BRCA2 While DNA-PK is central to NHEJ, Rad51 plays a key role in HR repair. As a homolog of Escherichia coli RecA recombinase, Rad51 is required for strand invasion in eukaryotic cells (Baumann et al., 1996; Shinohara et al., 1993). After displacing the single-strand binding protein replication protein A (RPA), it forms a nucleoprotein filament on resected 30 ssDNA overhangs and catalyzes homologous DNA pairing and strand exchange (Baumann and West, 1998). Rad52 seems to assist Rad51 in displacing RPA and facilitates filament assembly (Benson et al., 1998; McIlwraith et al., 2000). A similar role has been proposed for the five Rad51 paralogs RAD51B, RAD51C, RAD51D, XRCC2, and XRCC3, which share 20–30% sequence identity with Rad51 (Schild et al., 2000; Sigurdsson et al., 2001; Thompson and Schild, 2001). Strand invasion of the Rad51-ssDNA nucleoprotein filament is stimulated by Rad54, an SWI2/SNF2-related protein that possesses double-stranded DNA-dependent ATPase activity. Rad54 binds to Rad51 and catalyzes bidirectional nucleosome redistribution by removing nucleosomes and other DNA-binding proteins from the DNA target site during the homology search (Alexeev et al., 2003). Upon invasion of the sister chromatid or homologous chromosome, DNA synthesis is initiated by a not-yet-identified DNA polymerase. The holiday junction (HJ) formed at


19

the side of invasion can branch migrate and release the invading end. Alternatively, the other DNA end may invade the same homologous DNA and form a double HJ that appears to be resolved primarily by non-crossing over resulting in gene conversion (Helleday, 2003; Johnson and Jasin, 2000). It is thought that WRN and the related RecQ-like helicase BLM are involved in the resolution of HJ structures (Mohaghegh et al., 2001; Saintigny et al., 2002). BLM, which is associated with Bloom’s syndrome, is, like Werner syndrome, a genetic recessive disorder that leads to early onset of aging and cancer predisposition. Consistent with a role in HR repair, BLM interacts with Rad51 following IR (Wu et al., 2001). Two other cancer susceptibility genes involved in HR repair are BRCA1 and BRCA2. BRCA2, which is, like BRCA1, linked to familial breast and ovarian cancer, interacts with the RecA-homology domain of RAD51 and is thought to control the loading of Rad51 onto ssDNA and Rad51 filament assembly (Chen et al., 1998; Kowalczykowski et al., 2002; Pellegrini et al., 2002; Powell et al., 2002; Yang et al., 2002). Consistent with this model, cells with mutations in BRCA2 show reduced Rad51 foci formation in response to IR (Yuan et al., 1999) and impaired homology-directed repair (Moynahan et al., 2001). BRCA2 was identified as a synonym of FANCB and FANCD1, two of the at least eight Fanconi anemia (FA) complementation groups associated with FA, a rare genetic cancer susceptibility syndrome (D’Andrea and Grompe, 2003; Howlett et al., 2002; Stewart et al., 2002). FANCA, -C, -E, -F, and -G form a constitutive nuclear complex, which mediates the monoubiquination of FANCD2 in response to DNA damage or during S-phase. Activated FANCD2, in turn, relocates to FANCD1/BRCA2 containing repair foci (D’Andrea and Grompe, 2003). Apart from Rad51, these foci also contain other proteins, including BRCA1 (Garcia-Higuera et al., 2001). BRCA1 contains an amino-terminal RINGfinger domain that interacts with BRCA1-associated RING domain 1 (BARD1) and forms a heterodimeric RING-finger complex with E3 ubiquitin ligase activity (Hashizume et al., 2001). It has been speculated that BRCA1 may monoubiquinate FANCD2, although findings suggest that FANCL, but not BRCA1, is the likely ligase (Meetei et al., 2004). The association of BRCA1 with Rad51 during S-phase (Chen et al., 1998; Scully et al., 1997) and the impaired recombinational repair of DSBs observed in Brca1-deficient mouse embryonic stem cells (Moynahan et al., 1999) provides strong evidence for a role of BRCA1 in HR repair, although the molecular mechanism remains to be determined. Moreover, one study proposed that Chk2-mediated phosphorylation of BRCA1 on Ser 988 might regulate the selectivity of repair by promoting Rad51-dependent HR repair and suppressing MRN-dependent NHEJ (Zhang et al., 2004). Apart from a function in DNA repair and checkpoint control (as discussed earlier), BRCA1 is also implicated in transcriptional control by transactivating genes

20

Ward and Chen

Figure 2 Formation of multiprotein complexes at sites of stalled replication forks. Multiple proteins localize to arrested replication forks, here caused by a UV-induced thymidine dimer. In the presence of TopBP1 (topoisomerase II -binding protein) Rad17 loads the Rad9-Hus1-Rad1 (9-1-1) complex on chromatin and becomes subsequently phosphorylated by ATR, which is independently recruited to replication protein A-coated single-strand DNA by its binding partner ATRIP. ATR-mediated phosphorylation of Chk1 triggers a checkpoint signaling cascade and induces transient cell cycle arrest. Claspin, a Chk1-binding protein, is also required to elicit a full checkpoint response. Recruitment of special translesion polymerases (t pol) allows translesion synthesis and can prevent replication fork collapse.

through interaction with p53, acting as a sequence-specific co-repressor of transcription, and interacting with chromatin remodeling proteins (reviewed in Thompson and Schild, 2002).

H. Rad17, the 9-1-1 Complex, and Claspin DNA damage that interferes with the replication process initiates the rapid recruitment of the 9-1-1 complex to chromatin (Burtelow et al., 2000) (Fig. 2). These proteins form a PCNA-like, heterotrimeric ring, which is loaded onto chromatin by a complex of Rad17 and the four small replication factor C (RFC) subunits (Rad17-RFC) (Burtelow et al., 2001; Lindsey-Boltz et al., 2001; Zou et al., 2002). Rad17-dependent association of the 9-1-1 complex with chromatin seems to be required for ATR-mediated phosphorylation of Rad17 and full activation of Chk1 in response to DNA damage (Weiss et al., 2002; You et al., 2002; Zou et al., 2002). Rad9 also interacts with TopBP1 (topoisomerase II -binding protein), a protein with eight BRCT domains (Makiniemi et al., 2001). Using the Xenopus egg extract system, it has been shown that Cut5, the ortholog of human TopBP1, is essential for the binding of Pol to arrested replication forks, a prerequisite for the loading of the 9–1–1 complex. Moreover, the recruitment of ATR to genotoxin-damaged


21

chromatin seems to require Cut5 as well (Parrilla-Castellar and Karnitz, 2003). It is thought that the Rad17-RFC complex senses abnormalities of the DNA structure and, by loading the 9-1-1 complex on chromatin, activates—together with ATR—the checkpoint pathway. Evidence in yeast suggests that upon replication arrest, the 9-1-1 complex also recruits the translesion polymerase DinB (human pol) onto chromatin, leading to the replacement of the stalled polymerase. Mutagenic translesion synthesis can then restart the replication fork and prevent replication fork collapse (Kai and Wang, 2003). Claspin, which was isolated from Xenopus extracts as a Chk1-binding protein, also interacts with Rad9 and ATR. Immunodepletion of Claspin from Xenopus egg extracts or downregulation of Claspin expression by siRNA inhibits ATR-dependent Chk1 activation and compromises the checkpoint response upon replication stress (Chini and Chen, 2003; Kumagai and Dunphy, 2000). Moreover, Claspin binds to chromatin around the time of initial DNA unwinding at replication origins, this binding occurs independently of ATR and Rad17 (Lee et al., 2003a). Interestingly, Claspin is related to fission and budding yeast Mrc1 (mediator of the replication checkpoint), which has been shown to be a component of the replication fork complex (Osborn and Elledge, 2003). Phosphorylation of Mrc1 following replication arrest is required for activation of Rad53 and initiation of the checkpoint response (Alcasabas et al., 2001; Osborn and Elledge, 2003; Tanaka and Russell, 2001).

III. Concluding Remarks Despite the impressive progress in our understanding of various components of the DNA damage signaling pathways, the mechanisms of how cells sense certain types of lesions and amplify the signal remain speculative. The hierarchical concept of a primary sensor that relays the signal downstream to eVectors via an intrinsic web of mediators, adaptors, and transducers is becoming replaced by a more complex model in which single molecules or protein complexes can have multiple functions in sensing as well as responding to DNA damage. The future challenge will be to identify all the partially redundant participants and to decipher the mutual crosstalk among them. It will also be important to fathom how stressinduced changes in the higher order chromatin structure relate to this complex network. Understanding how these components work together will not only increase our knowledge about tumorigenesis but may also lead to new clinical approaches.

22

Ward and Chen

Acknowledgments We thank Zhenkun Lou and Katherine MinterDykhouse for valuable comments on this manuscript. This work was supported by grants from the National Institutes of Health and the Breast Cancer Research Foundation. J. C. is a recipient of a Department of Defence (DOD) Breast Cancer Career Development Award (DAMD17-02-1-0472). I. W. is supported by a postdoctoral fellowship from the DOD Breast Cancer Research program (DAMD17-01-1-0317).

References Abraham, R. T. (2001). Cell cycle checkpoint signaling through the ATM and ATR kinases. Genes Dev. 15, 2177–2196. Ahn, J., Urist, M., Prives, C., Banin, S., Moyal, L., Shieh, S., Taya, Y., Anderson, C. W., Chessa, L., Smorodinsky, N. I., Reiss, Y., Shiloh, Y., and Ziv, Y. (2003). Questioning the role of checkpoint kinase 2 in the p53 DNA damage response. J. Biol. Chem. 278, 20480–20489. Ahn, J. Y., Schwarz, J. K., Piwnica-Worms, H., and Canman, C. E. (2000). Threonine 68 phosphorylation by ataxia telangiectasia mutated is required for eYcient activation of Chk2 in response to ionizing radiation. Cancer Res. 60, 5934–5936. Alcasabas, A. A., Osborn, A. J., Bachant, J., Hu, F., Werler, P. J., Bousset, K., Furuya, K., DiZey, J. F., Carr, A. M., and Elledge, S. J. (2001). Mrc1 transduces signals of DNA replication stress to activate Rad53. Nat. Cell Biol. 3, 958–965. Alexeev, A., Mazin, A., and Kowalczykowski, S. C. (2003). Rad54 protein possesses chromatinremodeling activity stimulated by the Rad51-ssDNA nucleoprotein filament. Nat. Struct. Biol. 10, 182–186. Andegeko, Y., Moyal, L., Mittelman, L., Tsarfaty, I., Shiloh, Y., and Rotman, G. (2001). Nuclear retention of ATM at sites of DNA double strand breaks. J. Biol. Chem. 276, 38224–38230. Antoccia, A., Stumm, M., Saar, K., Ricordy, R., Maraschio, P., Tanzarella, C., Banin, S., Moyal, L., Shieh, S., Taya, Y., Anderson, C. W., Chessa, L., Smorodinsky, N. I., Prives, C., Reiss, Y., Shiloh, Y., and Ziv, Y. (1999). Impaired p53-mediated DNA damage response, cell-cycle disturbance and chromosome aberrations in Nijmegen breakage syndrome lymphoblastoid cell lines. Int. J. Radiat. Biol. 75, 583–591. Bakkenist, C. J., and Kastan, M. B. (2003). DNA damage activates ATM through intermolecular autophosphorylation and dimer dissociation. Nature 421, 499–506. Banin, S., Moyal, L., Shieh, S., Taya, Y., Anderson, C. W., Chessa, L., Smorodinsky, N. I., Prives, C., Reiss, Y., Shiloh, Y., and Ziv, Y. (1998). Enhanced phosphorylation of p53 by ATM in response to DNA damage. Science 281, 1674–1677. Barlow, C., Hirotsune, S., Paylor, R., Liyanage, M., Eckhaus, M., Collins, F., Shiloh, Y., Crawley, J. N., Ried, T., Tagle, D., and Wynshaw-Boris, A. (1996). Atm-deficient mice: A paradigm of ataxia telangiectasia. Cell 86, 159–171. Barr, S. M., Leung, C. G., Chang, E. E., and Cimprich, K. A. (2003). ATR kinase activity regulates the intranuclear translocation of ATR and RPA following ionizing radiation. Curr. Biol. 13, 1047–1051. Bartek, J., Lukas, J., Banin, S., Moyal, L., Shieh, S., Taya, Y., Anderson, C. W., Chessa, L., Smorodinsky, N. I., Prives, C., Reiss, Y., Shiloh, Y., and Ziv, Y. (2001). Mammalian G1and S-phase checkpoints in response to DNA damage. Curr. Opin. Cell Biol. 13, 738–747. Bassing, C. H., Chua, K. F., Sekiguchi, J., Suh, H., Whitlow, S. R., Fleming, J. C., Monroe, B. C., Ciccone, D. N., Yan, C., Vlasakova, K., Livingston, D. M., Ferguson, D. O., Scully, R., and


23

Alt, F. W. (2002). Increased ionizing radiation sensitivity and genomic instability in the absence of histone H2AX. Proc. Natl. Acad. Sci. USA 99, 8173–8178. Bassing, C. H., Suh, H., Ferguson, D. O., Chua, K. F., Manis, J., EckersdorV, M., Gleason, M., Bronson, R., Lee, C., and Alt, F. W. (2003). Histone H2AX: A dosage-dependent suppressor of oncogenic translocations and tumors. Cell 114, 359–370. Baumann, P., Benson, F. E., and West, S. C. (1996). Human Rad51 protein promotes ATPdependent homologous pairing and strand transfer reactions in vitro. Cell 87, 757–766. Baumann, P., and West, S. C. (1998). Role of the human RAD51 protein in homologous recombination and double-stranded-break repair. Trends Biochem. Sci. 23, 247–251. Bell, D. W., Varley, J. M., Szydlo, T. E., Kang, D. H., Wahrer, D. C., Shannon, K. E., Lubratovich, M., Verselis, S. J., Isselbacher, K. J., Fraumeni, J. F., Birch, J. M., Li, F. P., Garber, J. E., and Haber, D. A. (1999). Heterozygous germ line hCHK2 mutations in Li-Fraumeni syndrome. Science 286, 2528–2531. Benson, F. E., Baumann, P., and West, S. C. (1998). Synergistic actions of Rad51 and Rad52 in recombination and DNA repair. Nature 391, 401–404. Bork, P., Hofmann, K., Bucher, P., Neuwald, A. F., Altschul, S. F., and Koonin, E. V. (1997). A superfamily of conserved domains in DNA damage-responsive cell cycle checkpoint proteins. Faseb J. 11, 68–76. Boulton, S. J., and Jackson, S. P. (1998). Components of the Ku-dependent non-homologous end-joining pathway are involved in telomeric length maintenance and telomeric silencing. EMBO J. 17, 1819–1828. Brown, E. J., and Baltimore, D. (2000). ATR disruption leads to chromosomal fragmentation and early embryonic lethality. Genes Dev. 14, 397–402. Brown, E. J., and Baltimore, D. (2003). Essential and dispensable roles of ATR in cell cycle arrest and genome maintenance. Genes Dev. 17, 615–628. Burma, S., Chen, B. P., Murphy, M., Kurimasa, A., and Chen, D. J. (2001). ATM phosphorylates histone H2AX in response to DNA double-strand breaks. J. Biol. Chem. 276, 42462–42467. Burtelow, M. A., Kaufmann, S. H., and Karnitz, L. M. (2000). Retention of the human Rad9 checkpoint complex in extraction-resistant nuclear complexes after DNA damage. J. Biol. Chem. 275, 26343–26348. Burtelow, M. A., Roos-Mattjus, P. M., Rauen, M., Babendure, J. R., and Karnitz, L. M. (2001). Reconstitution and molecular analysis of the hRad9-hHus1-hRad1 (9-1-1) DNA damage responsive checkpoint complex. J. Biol. Chem. 276, 25903–25909. Buscemi, G., Savio, C., Zannini, L., Micciche, F., Masnada, D., Nakanishi, M., Tauchi, H., Komatsu, K., Mizutani, S., Khanna, K., Chen, P., Concannon, P., Chessa, L., and Delia, D. (2001). Chk2 activation dependence on nbs1 after dna damage. Mol. Cell Biol. 21, 5214–5222. Callebaut, I., and Mornon, J. P. (1997). From BRCA1 to RAP1: A widespread BRCT module closely associated with DNA repair. FEBS Lett. 400, 25–30. Canman, C. E., Lim, D. S., Cimprich, K. A., Taya, Y., Tamai, K., Sakaguchi, K., Appella, E., Kastan, M. B., and Siliciano, J. D. (1998). Activation of the ATM kinase by ionizing radiation and phosphorylation of p53. Science 281, 1677–1679. Carney, J. P., Maser, R. S., Olivares, H., Davis, E. M., Le Beau, M., Yates, J. R., 3rd, Hays, L., Morgan, W. F., and Petrini, J. H. (1998). The hMre11/hRad50 protein complex and Nijmegen breakage syndrome: Linkage of double-strand break repair to the cellular DNA damage response. Cell 93, 477–486. Carson, C. T., Schwartz, R. A., Stracker, T. H., Lilley, C. E., Lee, D. V., Weitzman, M. D., Banin, S., Moyal, L., Shieh, S., Taya, Y., Anderson, C. W., Chessa, L., Smorodinsky, N. I., Prives, C., Reiss, Y., Shiloh, Y., and Ziv, Y. (2003). The Mre11 complex is required for ATM activation and the G2/M checkpoint. EMBO J. 22, 6610–6620.

24

Ward and Chen

Celeste, A., Difilippantonio, S., Difilippantonio, M. J., Fernandez-Capetillo, O., Pilch, D. R., Sedelnikova, O. A., Eckhaus, M., Ried, T., Bonner, W. M., and Nussenzweig, A. (2003a). H2AX haploinsuYciency modifies genomic stability and tumor susceptibility. Cell 114, 371–383. Celeste, A., Fernandez-Capetillo, O., Kruhlak, M. J., Pilch, D. R., Staudt, D. W., Lee, A., Bonner, R. F., Bonner, W. M., and Nussenzweig, A. (2003b). Histone H2AX phosphorylation is dispensable for the initial recognition of DNA breaks. Nat. Cell Biol. 5, 675–679. Celeste, A., Petersen, S., Romanienko, P. J., Fernandez-Capetillo, O., Chen, H. T., Sedelnikova, O. A., Reina-San-Martin, B., Coppola, V., MeVre, E., Difilippantonio, M. J., Redon, C., Pilch, D. R., Olaru, A., Eckhaus, M., Camerini-Otero, R. D., Tessarollo, L., Livak, F., Manova, K., Bonner, W. M., Nussenzweig, M. C., and Nussenzweig, A. (2002). Genomic instability in mice lacking histone H2AX. Science 296, 922–927. Chan, D. W., Chen, B. P., Prithivirajsingh, S., Kurimasa, A., Story, M. D., Qin, J., and Chen, D. J. (2002). Autophosphorylation of the DNA-dependent protein kinase catalytic subunit is required for rejoining of DNA double-strand breaks. Genes Dev. 16, 2333–2338. Chan, D. W., Gately, D. P., Urban, S., Galloway, A. M., Lees-Miller, S. P., Yen, T., and Allalunis-Turner, J. (1998). Lack of correlation between ATM protein expression and tumour cell radiosensitivity. Int. J. Radiat. Biol. 74, 217–224. Chan, D. W., and Lees-Miller, S. P. (1996). The DNA-dependent protein kinase is inactivated by autophosphorylation of the catalytic subunit. J. Biol. Chem. 271, 8936–8941. Chehab, N. H., Malikzay, A., Appel, M., and Halazonetis, T. D. (2000). Chk2/hCds1 functions as a DNA damage checkpoint in G(1) by stabilizing p53. Genes Dev. 14, 278–288. Chen, J., Silver, D. P., Walpita, D., Cantor, S. B., Gazdar, A. F., Tomlinson, G., Couch, F. J., Weber, B. L., Ashley, T., Livingston, D. M., and Scully, R. (1998). Stable interaction between the products of the BRCA1 and BRCA2 tumor suppressor genes in mitotic and meiotic cells. Mol. Cell 2, 317–328. Chen, S., Inamdar, K. V., PfeiVer, P., Feldmann, E., Hannah, M. F., Yu, Y., Lee, J. W., Zhou, T., Lees-Miller, S. P., and Povirk, L. F. (2001). Accurate in vitro end joining of a DNA double strand break with partially cohesive 30 -overhangs and 30 -phosphoglycolate termini: EVect of Ku on repair fidelity. J. Biol. Chem. 276, 24323–24330. Chini, C. C., and Chen, J. (2003). Human claspin is required for replication checkpoint control. J. Biol. Chem. 278, 30057–30062. Cleaver, J. E., Karplus, K., Kashani-Sabet, M., and Limoli, C. L. (2001). Nucleotide excision repair ‘‘a legacy of creativity’’. Mutat. Res. 485, 23–36. Cliby, W. A., Roberts, C. J., Cimprich, K. A., Stringer, C. M., Lamb, J. R., Schreiber, S. L., and Friend, S. H. (1998). Overexpression of a kinase-inactive ATR protein causes sensitivity to DNA-damaging agents and defects in cell cycle checkpoints. EMBO J. 17, 159–169. Cortez, D., Guntuku, S., Qin, J., and Elledge, S. J. (2001). ATR and ATRIP: Partners in checkpoint signaling. Science 294, 1713–1716. Cortez, D., Wang, Y., Qin, J., and Elledge, S. J. (1999). Requirement of ATM-dependent phosphorylation of brcal in the DNA damage response to double-strand breaks. Science 286, 1162–1166. Costanzo, V., Robertson, K., Bibikova, M., Kim, E., Grieco, D., Gottesman, M., Carroll, D., Gautier, J., Banin, S., Moyal, L., Shieh, S., Taya, Y., Anderson, C. W., Chessa, L., Smorodinsky, N. I., Prives, C., Reiss, Y., Shiloh, Y., and Ziv, Y. (2001). Mre11 protein complex prevents double-strand break accumulation during chromosomal DNA replication. Mol. Cell 8, 137–147. Costanzo, V., Robertson, K., Ying, C. Y., Kim, E., Avvedimento, E., Gottesman, M., Grieco, D., Gautier, J., Banin, S., Moyal, L., Shieh, S., Taya, Y., Anderson, C. W., Chessa, L., Smorodinsky, N. I., Prives, C., Reiss, Y., Shiloh, Y., and Ziv, Y. (2000). Reconstitution of


25

an ATM-dependent checkpoint that inhibits chromosomal DNA replication following DNA damage. Mol. Cell 6, 649–659. D’Amours, D., Jackson, S. P., Banin, S., Moyal, L., Shieh, S., Taya, Y., Anderson, C. W., Chessa, L., Smorodinsky, N. I., Prives, C., Reiss, Y., Shiloh, Y., and Ziv, Y. (2002). The Mre11 complex: At the crossroads of dna repair and checkpoint signalling. Nat. Rev. Mol. Cell Biol. 3, 317–327. D’Andrea, A. D., and Grompe, M. (2003). The Fanconi anaemia/BRCA pathway. Nat. Rev. Cancer. 3, 23–34. DeFazio, L. G., Stansel, R. M., GriYth, J. D., and Chu, G. (2002). Synapsis of DNA ends by DNA-dependent protein kinase. EMBO J. 21, 3192–3200. de Jager, M., van Noort, J., van Gent, D. C., Dekker, C., Kanaar, R., Wyman, C., Banin, S., Moyal, L., Shieh, S., Taya, Y., Anderson, C. W., Chessa, L., Smorodinsky, N. I., Prives, C., Reiss, Y., Shiloh, Y., and Ziv, Y. (2001). Human Rad50/Mre11 is a flexible complex that can tether DNA ends. Mol. Cell 8, 1129–1135. Deng, C., Zhang, P., Harper, J. W., Elledge, S. J., Leder, P., Banin, S., Moyal, L., Shieh, S., Taya, Y., Anderson, C. W., Chessa, L., Smorodinsky, N. I., Prives, C., Reiss, Y., Shiloh, Y., and Ziv, Y. (1995). Mice lacking p21 CIP1/WAF1 undergo normal development, but are defective in G1 checkpoint control. Cell 82, 675–684. Difilippantonio, M. J., Petersen, S., Chen, H. T., Johnson, R., Jasin, M., Kanaar, R., Ried, T., and Nussenzweig, A. (2002). Evidence for replicative repair of DNA double-strand breaks leading to oncogenic translocation and gene amplification. J. Exp. Med. 196, 469–480. DiTullio, R. A., Mochan, T. A., Venere, M., Bartkova, J., Sehested, M., Bartek, J., and Halazonetis, T. D. (2002). 53BP1 functions in an ATM-dependent checkpoint pathway that is constitutively activated in human cancer. Nat. Cell Biol. 4, 998–1002. Dong, Z., Zhong, Q., Chen, P. L., Banin, S., Moyal, L., Shieh, S., Taya, Y., Anderson, C. W., Chessa, L., Smorodinsky, N. I., Prives, C., Reiss, Y., Shiloh, Y., and Ziv, Y. (1999). The Nijmegen breakage syndrome protein is essential for Mre11 phosphorylation upon DNA damage. J. Biol. Chem. 274, 19513–19516. Douglas, P., Sapkota, G. P., Morrice, N., Yu, Y., Goodarzi, A. A., Merkle, D., Meek, K., Alessi, D. R., and Lees-Miller, S. P. (2002). Identification of in vitro and in vivo phosphorylation sites in the catalytic subunit of the DNA-dependent protein kinase. Biochem. J. 368, 243–251. Downs, J. A., Lowndes, N. F., and Jackson, S. P. (2000). A role for Saccharomyces cerevisiae histone H2A in DNA repair. Nature 408, 1001–1004. Dumaz, N., and Meek, D. W. (1999). Serine 15 phosphorylation stimulates p53 transactivation but does not directly influence interaction with HDM2. EMBO J. 18, 7002–7010. Dvir, A., Stein, L. Y., Calore, B. L., and Dynan, W. S. (1993). Purification and characterization of a template-associated protein kinase that phosphorylates RNA polymerase II. J. Biol. Chem. 268, 10440–10447. Elson, A., Wang, Y., Daugherty, C. J., Morton, C. C., Zhou, F., Campos-Torres, J., and Leder, P. (1996). Pleiotropic defects in ataxia-telangiectasia protein-deficient mice. Proc. Natl. Acad. Sci. USA 93, 13084–13089. Emili, A. (1998). MEC1-dependent phosphorylation of Rad9p in response to DNA damage. Mol. Cell 2, 183–189. Essers, J., van Steeg, H., de Wit, J., Swagemakers, S. M., Vermeij, M., Hoeijmakers, J. H., and Kanaar, R. (2000). Homologous and non-homologous recombination diVerentially aVect DNA damage repair in mice. EMBO J. 19, 1703–1710. Falck, J., Mailand, N., Syljuasen, R. G., Bartek, J., and Lukas, J. (2001). The ATM-Chk2Cdc25A checkpoint pathway guards against radioresistant DNA synthesis. Nature 410, 842–847.

26

Ward and Chen

Falck, J., Petrini, J. H., Williams, B. R., Lukas, J., and Bartek, J. (2002). The DNA damage-dependent intra-S phase checkpoint is regulated by parallel pathways. Nat. Genet. 30, 290–294. Fernandez-Capetillo, O., Chen, H. T., Celeste, A., Ward, I., Romanienko, P. J., Morales, J. C., Naka, K., Xia, Z., Camerini-Otero, R. D., Motoyama, N., Carpenter, P. B., Bonner, W. M., Chen, J., and Nussenzweig, A. (2002). DNA damage-induced G(2)-M checkpoint activation by histone H2AX and 53BP1. Nat. Cell Biol. 4, 993–997. Fernandez-Capetillo, O., Mahadevaiah, S. K., Celeste, A., Romanienko, P. J., Camerini-Otero, R. D., Bonner, W. M., Manova, K., Burgoyne, P., and Nussenzweig, A. (2003). H2AX is required for chromatin remodeling and inactivation of sex chromosomes in male mouse meiosis. Dev. Cell 4, 497–508. Garcia-Higuera, I., Taniguchi, T., Ganesan, S., Meyn, M. S., Timmers, C., Hejna, J., Grompe, M., and D’Andrea, A. D. (2001). Interaction of the Fanconi anemia proteins and BRCA1 in a common pathway. Mol. Cell 7, 249–262. Gatei, M., Sloper, K., Sorensen, C., Syljuasen, R., Falck, J., Hobson, K., Savage, K., Lukas, J., Zhou, B. B., Bartek, J., and Khanna, K. K. (2003). Ataxia-telangiectasia-mutated (ATM) and NBS1-dependent phosphorylation of Chk1 on Ser-317 in response to ionizing radiation. J. Biol. Chem. 278, 14806–14811. Gatei, M., Young, D., Cerosaletti, K. M., Desai-Mehta, A., Spring, K., Kozlov, S., Lavin, M. F., Gatti, R. A., Concannon, P., and Khanna, K. (2000). ATM-dependent phosphorylation of nibrin in response to radiation exposure. Nat. Genet. 25, 115–119. Gatti, R. A., Becker-Catania, S., Chun, H. H., Sun, X., Mitui, M., Lai, C. H., Khanlou, N., Babaei, M., Cheng, R., Clark, C., Huo, Y., Udar, N. C., and Iyer, R. K. (2001). The pathogenesis of ataxia-telangiectasia. Learning from a Rosetta Stone. Clin. Rev. Allergy Immunol. 20, 87–108. Gellert, M., Nakajima, P. B., and Bosma, M. J. (1992). V(D)J recombination gets a break. Trends Genet. 8, 408–412. Gilbert, C. S., Green, C. M., and Lowndes, N. F. (2001). Budding yeast Rad9 is an ATP-dependent Rad53 activating machine. Mol. Cell 8, 129–136. Girard, P. M., Riballo, E., Begg, A. C., Waugh, A., and Jeggo, P. A. (2002). Nbs1 promotes ATM dependent phosphorylation events including those required for G1/S arrest. Oncogene 21, 4191–4199. Goldberg, M., Stucki, M., Falck, J., D’Amours, D., Rahman, D., Pappin, D., Bartek, J., and Jackson, S. P. (2003). MDC1 is required for the intra-S-phase DNA damage checkpoint. Nature 421, 952–956. Goodman, M. F., and Tippin, B. (2000). Sloppier copier DNA polymerases involved in genome repair. Curr. Opin. Genet. Dev. 10, 162–168. Gottlich, B., Reichenberger, S., Feldmann, E., and PfeiVer, P. (1998). Rejoining of DNA double-strand breaks in vitro by single-strand annealing. Eur. J. Biochem. 258, 387–395. Gottlieb, T. M., and Jackson, S. P. (1993). The DNA-dependent protein kinase: Requirement for DNA ends and association with Ku antigen. Cell 72, 131–142. Gray, M. D., Shen, J. C., Kamath-Loeb, A. S., Blank, A., Sopher, B. L., Martin, G. M., Oshima, J., and Loeb, L. A. (1997). The Werner syndrome protein is a DNA helicase. Nat. Genet. 17, 100–103. Haber, J. E. (1992). Mating-type gene switching in Saccharomyces cerevisiae. Trends Genet. 8, 446–452. Hacques, M. F., Muller, S., De Murcia, G., Van Regenmortel, M. H., and Marion, C. (1990). Accessibility and structural role of histone domains in chromatin. biophysical and immunochemical studies of progressive digestion with immobilized proteases. J. Biomol. Struct. Dyn. 8, 619–641.


27

Hashizume, R., Fukuda, M., Maeda, I., Nishikawa, H., Oyake, D., Yabuki, Y., Ogata, H., and Ohta, T. (2001). The RING heterodimer BRCA1-BARD1 is a ubiquitin ligase inactivated by a breast cancer-derived mutation. J. Biol. Chem. 276, 14537–14540. HeVernan, T. P., Simpson, D. A., Frank, A. R., Heinloth, A. N., Paules, R. S., Cordeiro-Stone, M., and Kaufmann, W. K. (2002). An ATR- and Chk1-dependent S checkpoint inhibits replicon initiation following UVC-induced DNA damage. Mol. Cell Biol. 22, 8552–8561. Helleday, T. (2003). Pathways for mitotic homologous recombination in mammalian cells. Mutat. Res. 532, 103–115. Hirano, T., Banin, S., Moyal, L., Shieh, S., Taya, Y., Anderson, C. W., Chessa, L., Smorodinsky, N. I., Prives, C., Reiss, Y., Shiloh, Y., and Ziv, Y. (2002). The ABCs of SMC proteins: Two-armed ATPases for chromosome condensation, cohesion, and repair. Genes Dev. 16, 399–414. Hirao, A., Kong, Y. Y., Matsuoka, S., Wakeham, A., Ruland, J., Yoshida, H., Liu, D., Elledge, S. J., and Mak, T. W. (2000). DNA damage-induced activation of p53 by the checkpoint kinase Chk2. Science 287, 1824–1827. Honjo, T., Kinoshita, K., and Muramatsu, M. (2002). Molecular mechanism of class switch recombination: Linkage with somatic hypermutation. Annu. Rev. Immunol. 20, 165–196. Hopfner, K. P., Karcher, A., Craig, L., Woo, T. T., Carney, J. P., and Tainer, J. A. (2001). Structural biochemistry and interaction architecture of the DNA double-strand break repair Mre11 nuclease and Rad50-ATPase. Cell 105, 473–485. Howlett, N. G., Taniguchi, T., Olson, S., Cox, B., Waisfisz, Q., De Die-Smulders, C., Persky, N., Grompe, M., Joenje, H., Pals, G., Ikeda, H., Fox, E. A., and D’Andrea, A. D. (2002). Biallelic inactivation of BRCA2 in Fanconi anemia. Science 297, 606–609. Huang, J., and Dynan, W. S. (2002). Reconstitution of the mammalian DNA double-strand break end-joining reaction reveals a requirement for an Mre11/Rad50/NBS1-containing fraction. Nucleic Acids Res. 30, 667–674. Iwabuchi, K., Bartel, P. L., Li, B., Marraccino, R., and Fields, S. (1994). Two cellular proteins that bind to wild-type but not mutant p53. Proc. Natl. Acad. Sci. USA 91, 6098–6102. Iwabuchi, K., Li, B., Massa, H. F., Trask, B. J., Date, T., and Fields, S. (1998). Stimulation of p53-mediated transcriptional activation by the p53-binding proteins, 53BP1 and 53BP2. J. Biol. Chem. 273, 26061–26068. Johnson, R. D., and Jasin, M. (2000). Sister chromatid gene conversion is a prominent double-strand break repair pathway in mammalian cells. EMBO J. 19, 3398–3407. Jongmans, W., Vuillaume, M., Chrzanowska, K., Smeets, D., Sperling, K., and Hall, J. (1997). Nijmegen breakage syndrome cells fail to induce the p53-mediated DNA damage response following exposure to ionizing radiation. Mol. Cell Biol. 17, 5016–5022. Jullien, D., Vagnarelli, P., Earnshaw, W. C., and Adachi, Y. (2002). Kinetochore localisation of the DNA damage response component 53BP1 during mitosis. J. Cell Sci. 115, 71–79. Kai, M., and Wang, T. S. (2003). Checkpoint activation regulates mutagenic translesion synthesis. Genes Dev. 17, 64–76. Kao, G. D., McKenna, W. G., Guenther, M. G., Muschel, R. J., Lazar, M. A., and Yen, T. J. (2003). Histone deacetylase 4 interacts with 53BP1 to mediate the DNA damage response. J. Cell Biol. 160, 1017–1027. Karmakar, P., Piotrowski, J., Brosh, R. M., Jr., Sommers, J. A., Miller, S. P., Cheng, W. H., Snowden, C. M., Ramsden, D. A., and Bohr, V. A. (2002). Werner protein is a target of DNA-dependent protein kinase in vivo and in vitro, and its catalytic activities are regulated by phosphorylation. J. Biol. Chem. 277, 18291–18302. Kastan, M. B., Zhan, Q., el-Deiry, W. S., Carrier, F., Jacks, T., Walsh, W. V., Plunkett, B. S., Vogelstein, B., and Fornace, A. J. (1992). A mammalian cell cycle checkpoint pathway utilizing p53 and GADD45 is defective in ataxia-telangiectasia. Cell 71, 587–597.

28

Ward and Chen

Khanna, K. K., Keating, K. E., Kozlov, S., Scott, S., Gatei, M., Hobson, K., Taya, Y., Gabrielli, B., Chan, D., Lees-Miller, S. P., and Lavin, M. F. (1998). ATM associates with and phosphorylates p53: Mapping the region of interaction. Nat. Genet. 20, 398–400. Kim, S. T., Lim, D. S., Canman, C. E., and Kastan, M. B. (1999). Substrate specificities and identification of putative substrates of ATM kinase family members. J. Biol. Chem. 274, 37538–37543. Kim, S. T., Xu, B., and Kastan, M. B. (2002). Involvement of the cohesin protein, Smc1, in Atm-dependent and independent responses to DNA damage. Genes Dev. 16, 560–570. Kobayashi, J., Tauchi, H., Sakamoto, S., Nakamura, A., Morishima, K., Matsuura, S., Kobayashi, T., Tamai, K., Tanimoto, K., and Komatsu, K. (2002). NBS1 localizes to gamma-H2AX foci through interaction with the FHA/BRCT domain. Curr. Biol. 12, 1846–1851. Koonin, E. V., Altschul, S. F., and Bork, P. (1996). BRCA1 protein products Functional motifs [letter]. Nat. Genet. 13, 266–268. Kowalczykowski, S. C., Yang, H., JeVrey, P. D., Miller, J., Kinnucan, E., Sun, Y., Thoma, N. H., Zheng, N., Chen, P. L., Lee, W. H., and Pavletich, N. P. (2002). Molecular mimicry connects BRCA2 to Rad51 and recombinational DNA repair. Nat. Struct. Biol. 9, 897–899. Kozlov, S., Gueven, N., Keating, K., Ramsay, J., and Lavin, M. F. (2003). ATP activates ataxia-telangiectasia mutated (ATM) in vitro. Importance of autophosphorylation. J. Biol. Chem. 278, 9309–9317. Kumagai, A., and Dunphy, W. G. (2000). Claspin, a novel protein required for the activation of Chk1 during a DNA replication checkpoint response in Xenopus egg extracts. Mol. Cell 6, 839–849. Larner, J. M., Lee, H., Little, R. D., Dijkwel, P. A., Schildkraut, C. L., and Hamlin, J. L. (1999). Radiation down-regulates replication origin activity throughout the S phase in mammalian cells. Nucleic Acids Res. 27, 803–809. Lee, J., Kumagai, A., and Dunphy, W. G. (2003a). Claspin, a Chk1-regulatory protein, monitors DNA replication on chromatin independently of RPA, ATR, and Rad17. Mol. Cell 11, 329–340. Lee, J. H., Xu, B., Lee, C. H., Ahn, J. Y., Song, M. S., Lee, H., Canman, C. E., Lee, J. S., Kastan, M. B., and Lim, D. S. (2003b). Distinct functions of Nijmegen breakage syndrome in ataxia telangiectasia mutated-dependent responses to DNA damage. Mol. Cancer Res. 1, 674–681. Lee, S. E., Mitchell, R. A., Cheng, A., and Hendrickson, E. A. (1997). Evidence for DNA-PKdependent and -independent DNA double-strand break repair pathways in mammalian cells as a function of the cell cycle. Mol. Cell Biol. 17, 1425–1433. Li, B., and Comai, L. (2002). Displacement of DNA-PKcs from DNA ends by the Werner syndrome protein. Nucleic Acids Res. 30, 3653–3661. Li, J., Lee, G. I., Van Doren, S. R., and Walker, J. C. (2000). The FHA domain mediates phosphoprotein interactions. J. Cell Sci. 113, 4143–4149. Liang, F., Han, M., Romanienko, P. J., and Jasin, M. (1998). Homology-directed repair is a major double-strand break repair pathway in mammalian cells. Proc. Natl. Acad. Sci. USA 95, 5172–5177. Lim, D. S., Kim, S. T., Xu, B., Maser, R. S., Lin, J., Petrini, J. H., and Kastan, M. B. (2000). ATM phosphorylates p95/nbs1 in an S-phase checkpoint pathway. Nature 404, 613–617. Lin, F. L., Sperle, K., and Sternberg, N. (1984). Model for homologous recombination during transfer of DNA into mouse L cells: Role for DNA ends in the recombination process. Mol. Cell Biol. 4, 1020–1034. Lindsey-Boltz, L. A., Bermudez, V. P., Hurwitz, J., and Sancar, A. (2001). Purification and characterization of human DNA damage checkpoint Rad complexes. Proc. Natl. Acad. Sci. USA 98, 11236–11241.


29

Lou, Z., Chini, C. C., Minter-Dykhouse, K., and Chen, J. (2003a). Mediator of DNA damage checkpoint protein 1 regulates BRCA1 localization and phosphorylation in DNA damage checkpoint control. J. Biol. Chem. 278, 13599–13602. Lou, Z., Minter-Dykhouse, K., Wu, X., and Chen, J. (2003b). MDC1 is coupled to activated CHK2 in mammalian DNA damage response pathways. Nature 421, 957–961. Lukas, C., Falck, J., Bartkova, J., Bartek, J., and Lukas, J. (2003). Distinct spatiotemporal dynamics of mammalian checkpoint regulators induced by DNA damage. Nat. Cell Biol. 5, 255–260. Luo, G., Yao, M. S., Bender, C. F., Mills, M., Bladl, A. R., Bradley, A., and Petrini, J. H. (1999). Disruption of mRad50 causes embryonic stem cell lethality, abnormal embryonic development, and sensitivity to ionizing radiation. Proc. Natl. Acad. Sci. USA 96, 7376–7381. Lupardus, P. J., Byun, T., Yee, M. C., Hekmat-Nejad, M., and Cimprich, K. A. (2002). A requirement for replication in activation of the ATR-dependent DNA damage checkpoint. Genes Dev. 16, 2327–2332. Ma, Y., Pannicke, U., Schwarz, K., and Lieber, M. R. (2002). Hairpin opening and overhang processing by an Artemis/DNA-dependent protein kinase complex in nonhomologous end joining and V(D)J recombination. Cell 108, 781–794. Mahadevaiah, S. K., Turner, J. M., Baudat, F., Rogakou, E. P., de Boer, P., Blanco-Rodriguez, J., Jasin, M., Keeney, S., Bonner, W. M., and Burgoyne, P. S. (2001). Recombinational DNA double-strand breaks in mice precede synapsis. Nat. Genet. 27, 271–276. Makiniemi, M., Hillukkala, T., Tuusa, J., Reini, K., Vaara, M., Huang, D., Pospiech, H., Majuri, I., Westerling, T., Makela, T. P., and Syvaoja, J. E. (2001). BRCT domaincontaining protein TopBP1 functions in DNA replication and damage response. J. Biol. Chem. 276, 30399–30406, Epub 32001 Jun 30396. Manis, J. P., Tian, M., and Alt, F. W. (2002). Mechanism and control of class-switch recombination. Trends Immunol. 23, 31–39. Manke, I. A., Lowery, D. M., Nguyen, A., and YaVe, M. B. (2003). BRCT repeats as phosphopeptide-binding modules involved in protein targeting. Science 302, 636–639. Maser, R. S., Mirzoeva, O. K., Wells, J., Olivares, H., Williams, B. R., Zinkel, R. A., Farnham, P. J., and Petrini, J. H. (2001). Mre11 complex and DNA replication: Linkage to E2F and sites of DNA synthesis. Mol. Cell Biol. 21, 6006–6016. Matsuoka, S., Rotman, G., Ogawa, A., Shiloh, Y., Tamai, K., and Elledge, S. J. (2000). Ataxia telangiectasia-mutated phosphorylates Chk2 in vivo and in vitro. Proc. Natl. Acad. Sci. USA 97, 10389–10394. Matsuura, K., Balmukhanov, T., Tauchi, H., Weemaes, C., Smeets, D., Chrzanowska, K., Endou, S., Matsuura, S., and Komatsu, K. (1998). Radiation induction of p53 in cells from Nijmegen breakage syndrome is defective but not similar to ataxia-telangiectasia. Biochem. Biophys. Res. Commun. 242, 602–607. Maya, R., Balass, M., Kim, S. T., Shkedy, D., Leal, J. F., Shifman, O., Moas, M., Buschmann, T., Ronai, Z., Shiloh, Y., Kastan, M. B., Katzir, E., and Oren, M. (2001). ATM-dependent phosphorylation of Mdm2 on serine 395: Role in p53 activation by DNA damage. Genes Dev. 15, 1067–1077. McIlwraith, M. J., Van Dyck, E., Masson, J. Y., Stasiak, A. Z., Stasiak, A., and West, S. C. (2000). Reconstitution of the strand invasion step of double-strand break repair using human Rad51 Rad52 and RPA proteins. J. Mol. Biol. 304, 151–164. Meetei, A. R., Yan, Z., Wang, W., Johnson, R. D., and Jasin, M. (2004). FANCL replaces BRCA1 as the likely ubiquitin ligase responsible for FANCD2 monoubiquitination. Cell Cycle 3, 179–181. Melchionna, R., Chen, X. B., Blasina, A., and McGowan, C. H. (2000). Threonine 68 is required for radiation-induced phosphorylation and activation of Cds1. Nat. Cell Biol. 2, 762–765.

30

Ward and Chen

Memisoglu, A., and Samson, L. (2000). Base excision repair in yeast and mammals. Mutat. Res. 451, 39–51. Merkle, D., Douglas, P., Moorhead, G. B., Leonenko, Z., Yu, Y., Cramb, D., Bazett-Jones, D. P., and Lees-Miller, S. P. (2002). The DNA-dependent protein kinase interacts with DNA to form a protein-DNA complex that is disrupted by phosphorylation. Biochemistry 41, 12706–12714. Mirzoeva, O. K., and Petrini, J. H. (2001). DNA damage-dependent nuclear dynamics of the Mre11 complex. Mol. Cell Biol. 21, 281–288. Mirzoeva, O. K., and Petrini, J. H. (2003). DNA replication-dependent nuclear dynamics of the Mre11 complex. Mol. Cancer Res. 1, 207–218. Mochan, T. A., Venere, M., DiTullio, R. A., Jr., and Halazonetis, T. D. (2003). 53BP1 and NFBD1/MDC1-Nbs1 function in parallel interacting pathways activating ataxia-telangiectasia mutated (ATM) in response to DNA damage. Cancer Res. 63, 8586–8591. Mohaghegh, P., Karow, J. K., Brosh Jr, R. M., Jr., Bohr, V. A., and Hickson, I. D. (2001). The Bloom’s and Werner’s syndrome proteins are DNA structure-specific helicases. Nucleic Acids Res. 29, 2843–2849. Morales, J. C., Xia, Z., Lu, T., Aldrich, M. B., Wang, B., Rosales, C., Kellems, R. E., Hittelman, W. N., Elledge, S. J., and Carpenter, P. B. (2003). Role for the BRCT protein 53BP1 in maintaining genomic stability. J. Biol. Chem. 10, 10. Moynahan, M. E., Chiu, J. W., Koller, B. H., and Jasin, M. (1999). Brca1 controls homologydirected DNA repair. Mol. Cell 4, 511–518. Moynahan, M. E., Pierce, A. J., and Jasin, M. (2001). BRCA2 is required for homologydirected repair of chromosomal breaks. Mol. Cell 7, 263–272. Nelms, B. E., Maser, R. S., MacKay, J. F., Lagally, M. G., and Petrini, J. H. (1998). In situ visualization of DNA double-strand break repair in human fibroblasts. Science 280, 590–592. O’Driscoll, M., Ruiz-Perez, V. L., Woods, C. G., Jeggo, P. A., and Goodship, J. A. (2003). A splicing mutation aVecting expression of ataxia-telangiectasia and Rad3-related protein (ATR) results in Seckel syndrome. Nat. Genet. 33, 497–501. Osborn, A. J., and Elledges, S. J. (2003). Mrc1 is a replication fork component whose phosphorylation in response to DNA replication stress activates Rad53. Genes Dev. 17, 1755–1767. Painter, R. B., Young, B. R., Larner, J. M., Lee, H., Little, R. D., Dijkwel, P. A., Schildkraut, C. L., and Hamlin, J. L. (1980). Radiosensitivity in ataxia-telangiectasia: A new explanation. Proc. Natl. Acad. Sci. USA 77, 7315–7317. Parrilla-Castellar, E. R., and Karnitz, L. M. (2003). Cut5 is required for the binding of Atr and DNA polymerase alpha to genotoxin-damaged chromatin. J. Biol. Chem. 278, 45507–45511. Paull, T. T., Rogakou, E. P., Yamazaki, V., Kirchgessner, C. U., Gellert, M., and Bonner, W. M. (2000). A critical role for histone H2AX in recruitment of repair factors to nuclear foci after DNA damage. Curr. Biol. 10, 886–895. Pellegrini, L., Yu, D. S., Lo, T., Anand, S., Lee, M., Blundell, T. L., and Venkitaraman, A. R. (2002). Insights into DNA recombination from the structure of a RAD51-BRCA2 complex. Nature 420, 287–293. Peng, A., and Chen, P. L. (2003). NFBD1, like 53BP1, is an early and redundant transducer mediating Chk2 phosphorylation in response to DNA damage. J. Biol. Chem. 278, 8873–8876. Peng, C. Y., Graves, P. R., Thoma, R. S., Wu, Z., Shaw, A. S., and Piwnica-Worms, H. (1997). Mitotic and G2 checkpoint control: Regulation of 14-3-3 protein binding by phosphorylation of Cdc25C on serine-216. Science 277, 1501–1505.


31

PfeiVer, P., Goedecke, W., Obe, G., Thompson, L. H., and Schild, D. (2000). Mechanisms of DNA double-strand break repair and their potential to induce chromosomal aberrations. Mutagenesis 15, 289–302. Powell, S. N., Willers, H., Xia, F., Thompson, L. H., and Schild, D. (2002). BRCA2 keeps Rad51 in line. High-fidelity homologous recombination prevents breast and ovarian cancer? Mol. Cell 10, 1262–1263. Rappold, I., Iwabuchi, K., Date, T., and Chen, J. (2001). Tumor suppressor p53 binding protein 1 (53BP1) is involved in DNA damage-signaling pathways. J. Cell Biol. 153, 613–620. Reina-San-Martin, B., Difilippantonio, S., Hanitsch, L., Masilamani, R. F., Nussenzweig, A., and Nussenzweig, M. C. (2003). H2AX is required for recombination between immunoglobulin switch regions but not for intra-switch region recombination or somatic hypermutation. J. Exp. Med. 197, 1767–1778. Rodriguez, M., Yu, X., Chen, J., and Songyang, Z. (2003). Phosphopeptide binding specificities of BRCA1 COOH-terminal (BRCT) domains. J. Biol. Chem. 278, 52914–52918. Rogakou, E. P., Boon, C., Redon, C., and Bonner, W. M. (1999). Megabase chromatin domains involved in DNA double-strand breaks in vivo. J. Cell Biol. 146, 905–916. Rogakou, E. P., Pilch, D. R., Orr, A. H., Ivanova, V. S., and Bonner, W. M. (1998). DNA double-stranded breaks induce histone H2AX phosphorylation on serine 139. J. Biol. Chem. 273, 5858–5868. Rooney, S., Sekiguchi, J., Zhu, C., Cheng, H. L., Manis, J., Whitlow, S., DeVido, J., Foy, D., Chaudhuri, J., Lombard, D., and Alt, F. W. (2002). Leaky Scid phenotype associated with defective V(D)J coding end processing in Artemis-deficient mice. Mol. Cell 10, 1379–1390. Saintigny, Y., Makienko, K., Swanson, C., Emond, M. J., Monnat, R. J., Jr., and Hickson, I. D. (2002). Homologous recombination resolution defect in werner syndrome. Mol. Cell Biol. 22, 6971–6978. Sanchez, Y., Wong, C., Thoma, R. S., Richman, R., Wu, Z., Piwnica-Worms, H., and Elledge, S. J. (1997). Conservation of the Chk1 checkpoint pathway in mammals: Linkage of DNA damage to Cdk regulation through Cdc25. Science 277, 1497–1501. Sarkaria, J. N., Tibbetts, R. S., Busby, E. C., Kennedy, A. P., Hill, D. E., and Abraham, R. T. (1998). Inhibition of phosphoinositide 3-kinase related kinases by the radiosensitizing agent wortmannin. Cancer Res. 58, 4375–4382. Schild, D., Lio, Y. C., Collins, D. W., Tsomondo, T., and Chen, D. J. (2000). Evidence for simultaneous protein interactions between human Rad51 paralogs. J. Biol. Chem. 275, 16443–16449. Schofield, M. J., and Hsieh, P. (2003). DNA mismatch repair: Molecular mechanisms and biological function. Annu. Rev. Microbiol. 57, 579–608. Schultz, L. B., Chehab, N. H., Malikzay, A., and Halazonetis, T. D. (2000). p53 binding protein 1 (53BP1) is an early participant in the cellular response to DNA double-strand breaks. J. Cell Biol. 151, 1381–1390. Schwartz, M. F., Duong, J. K., Sun, Z., Morrow, J. S., Pradhan, D., and Stern, D. F. (2002). Rad9 phosphorylation sites couple Rad53 to the Saccharomyces cerevisiae DNA damage checkpoint. Mol. Cell 9, 1055–1065. Scully, R., Chen, J., Plug, A., Xiao, Y., Weaver, D., Feunteun, J., Ashley, T., and Livingston, D. M. (1997). Association of BRCA1 with Rad51 in mitotic and meiotic cells. Cell 88, 265–275. Shang, Y. L., Bodero, A. J., and Chen, P. L. (2003). NFBD1, a novel nuclear protein with signature motifs of FHA and BRCT, and an internal 41-amino acid repeat sequence, is an early participant in DNA damage response. J. Biol. Chem. 278, 6323–6329. Shen, J. C., Gray, M. D., Oshima, J., Kamath-Loeb, A. S., Fry, M., and Loeb, L. A. (1998). Werner syndrome protein. I. DNA helicase and dna exonuclease reside on the same polypeptide. J. Biol. Chem. 273, 34139–34144.

32

Ward and Chen

Shieh, S. Y., Ahn, J., Tamai, K., Taya, Y., and Prives, C. (2000). The human homologs of checkpoint kinases Chk1 and Cds1 (Chk2) phosphorylate p53 at multiple DNA damageinducible sites. Genes Dev. 14, 289–300. Shiloh, Y. (2003). ATM and related protein kinases: Safeguarding genome integrity. Nat. Rev. Cancer 3, 155–168. Shinohara, A., Ogawa, H., Matsuda, Y., Ushio, N., Ikeo, K., and Ogawa, T. (1993). Cloning of human, mouse and fission yeast recombination genes homologous to RAD51 and recA. Nat. Genet. 4, 239–243. Sigurdsson, S., Van Komen, S., Bussen, W., Schild, D., Albala, J. S., and Sung, P. (2001). Mediator function of the human Rad51B-Rad51C complex in Rad51/RPA-catalyzed DNA strand exchange. Genes Dev. 15, 3308–3318. Skibbens, R. V., Corson, L. B., Koshland, D., and Hieter, P. (1999). Ctf7p is essential for sister chromatid cohesion and links mitotic chromosome structure to the DNA replication machinery. Genes Dev. 13, 307–319. Smith, G. C., Cary, R. B., Lakin, N. D., Hann, B. C., Teo, S. H., Chen, D. J., and Jackson, S. P. (1999). Purification and DNA binding properties of the ataxia-telangiectasia gene product ATM. Proc. Natl. Acad. Sci. USA 96, 11134–11139. Stewart, G., Elledge, S. J., Weber, B. L., Ashley, T., Livingston, D. M., and Scully, R. (2002). The two faces of BRCA2, a FANCtastic discovery. Mol. Cell 10, 2–4. Stewart, G. S., Maser, R. S., Stankovic, T., Bressan, D. A., Kaplan, M. I., Jaspers, N. G., Raams, A., Byrd, P. J., Petrini, J. H., and Taylor, A. M. (1999). The DNA double-strand break repair gene hMRE11 is mutated in individuals with an ataxia-telangiectasia-like disorder. Cell 99, 577–587. Stewart, G. S., Wang, B., Bignell, C. R., Taylor, A. M., and Elledge, S. J. (2003). MDC1 is a mediator of the mammalian DNA damage checkpoint. Nature 421, 961–966. Sun, H., Treco, D., Schultes, N. P., and Szostak, J. W. (1989). Double-strand breaks at an initiation site for meiotic gene conversion. Nature 338, 87–90. Sun, Z., Hsiao, J., Fay, D. S., and Stern, D. F. (1998). Rad53 FHA domain associated with phosphorylated Rad9 in the DNA damage checkpoint. Science 281, 272–274. Sutherland, B. M., Bennett, P. V., Cintron, N. S., Guida, P., and Laval, J. (2003). Low levels of endogenous oxidative damage cluster levels in unirradiated viral and human DNAs. Free. Radic. Biol. Med. 35, 495–503. Suzuki, K., Kodama, S., and Watanabe, M. (1999). Recruitment of ATM protein to double strand DNA irradiated with ionizing radiation. J. Biol. Chem. 274, 25571–25575. Takata, M., Sasaki, M. S., Sonoda, E., Morrison, C., Hashimoto, M., Utsumi, H., YamaguchiIwai, Y., Shinohara, A., and Takeda, S. (1998). Homologous recombination and nonhomologous end-joining pathways of DNA double-strand break repair have overlapping roles in the maintenance of chromosomal integrity in vertebrate cells. EMBO J. 17, 5497–5508. Tanaka, K., and Russell, P. (2001). Mrc1 channels the DNA replication arrest signal to checkpoint kinase Cds1. Nat. Cell Biol. 3, 966–972. Thacker, J., Thompson, L. H., and Schild, D. (1999). Repair of ionizing radiation damage in mammalian cells. Alternative pathways and their fidelity. C R Acad. Sci. III. 322, 103–108. Thompson, L. H., and Schild, D. (2001). Homologous recombinational repair of DNA ensures mammalian chromosome stability. Mutat. Res. 477, 131–153. Thompson, L. H., and Schild, D. (2002). Recombinational DNA repair and human disease. Mutat. Res. 509, 49–78. Tibbetts, R. S., Brumbaugh, K. M., Williams, J. M., Sarkaria, J. N., Cliby, W. A., Shieh, S. Y., Taya, Y., Prives, C., and Abraham, R. T. (1999). A role for ATR in the DNA damageinduced phosphorylation of p53. Genes Dev. 13, 152–157.


33

Tibbetts, R. S., Cortez, D., Brumbaugh, K. M., Scully, R., Livingston, D., Elledge, S. J., and Abraham, R. T. (2000). Functional interactions between BRCA1 and the checkpoint kinase ATR during genotoxic stress. Genes Dev. 14, 2989–3002. Unsal-Kacmaz, K., Makhov, A. M., GriYth, J. D., and Sancar, A. (2002). Preferential binding of ATR protein to UV-damaged DNA. Proc. Natl. Acad. Sci. USA 99, 6673–6678. Usui, T., Ogawa, H., and Petrini, J. H. (2001). A DNA damage response pathway controlled by Tel1 and the Mre11 complex. Mol. Cell 7, 1255–1266. Uziel, T., Lerenthal, Y., Moyal, L., Andegeko, Y., Mittelman, L., and Shiloh, Y. (2003). Requirement of the MRN complex for ATM activation by DNA damage. EMBO J. 22, 5612–5621. van den Bosch, M., Bree, R. T., and Lowndes, N. F. (2003). The MRN complex: Coordinating and mediating the response to broken chromosomes. EMBO Rep. 4, 844–849. Vialard, J. E., Gilbert, C. S., Green, C. M., and Lowndes, N. F. (1998). The budding yeast Rad9 checkpoint protein is subjected to Mec1/Tel1-dependent hyperphosphorylation and interacts with Rad53 after DNA damage. EMBO J. 17, 5679–5688. Walker, J. R., Corpina, R. A., and Goldberg, J. (2001). Structure of the Ku heterodimer bound to DNA and its implications for double-strand break repair. Nature 412, 607–614. Wallace, S. S. (1998). Enzymatic processing of radiation-induced free radical damage in DNA. Radiat. Res. 150, S60–S79. Wang, B., Matsuoka, S., Carpenter, P. B., and Elledge, S. J. (2002). 53BP1, a Mediator of the DNA Damage Checkpoint. Science 3, 3. Wang, S., Guo, M., Ouyang, H., Li, X., Cordon-Cardo, C., Kurimasa, A., Chen, D. J., Fuks, Z., Ling, C. C., and Li, G. C. (2000). The catalytic subunit of DNA-dependent protein kinase selectively regulates p53-dependent apoptosis but not cell-cycle arrest. Proc. Natl. Acad. Sci. USA 97, 1584–1588. Ward, I. M., and Chen, J. (2001). Histone H2AX is phosphorylated in an ATR-dependent manner in response to replicational stress. J. Biol. Chem. 276, 47759–47762. Ward, I. M., Minn, K., and Chen, J. (2004a). UV-induced ATR activation requires replication stress. J. Biol. Chem. 23, 23. Ward, I. M., Minn, K., Jorda, K. G., and Chen, J. (2003a). Accumulation of checkpoint protein 53BP1 at DNA breaks involves its binding to phosphorylated histone H2AX. J. Biol. Chem. 278, 19579–19582. Ward, I. M., Minn, K., Van Deursen, J., and Chen, J. (2003b). p53 binding protein 53BP1 is required for DNA damage responses and tumor suppression in mice. Mol. Cell Biol. 23, 2556–2563. Ward, I. M., Reina-San-Martin, B., Olaru, A., Minn, K., Tamada, K., Lau, J. S., Cascalho, M., Chen, L., Nussenzweig, A., Livak, F., Nussenzweig, M. C., and Chen, J. (2004b). 53BP1 is required for class switch recombination. J. Cell Biol. 165, 459–464. Weinert, T. A., and Hartwell, L. H. (1988). The RAD9 gene controls the cell cycle response to DNA damage in Saccharomyces cerevisiae. Science 241, 317–322. Weiss, R. S., Matsuoka, S., Elledge, S. J., and Leder, P. (2002). Hus1 acts upstream of chk1 in a mammalian DNA damage response pathway. Curr. Biol. 12, 73–77. Woo, R. A., Jack, M. T., Xu, Y., Burma, S., Chen, D. J., and Lee, P. W. (2002). DNA damageinduced apoptosis requires the DNA-dependent protein kinase, and is mediated by the latent population of p53. EMBO J. 21, 3000–3008. Wu, L., Davies, S. L., Levitt, N. C., and Hickson, I. D. (2001). Potential role for the BLM helicase in recombinational repair via a conserved interaction with RAD51. J. Biol. Chem. 276, 19375–19381. Wu, X., Ranganathan, V., Weisman, D. S., Heine, W. F., Ciccone, D. N., O’Neill, T. B., Crick, K. E., Pierce, K. A., Lane, W. S., Rathbun, G., Livingston, D. M., and Weaver, D. T. (2000).

34

Ward and Chen

ATM phosphorylation of Nijmegen breakage syndrome protein is required in a DNA damage response. Nature 405, 477–482. Xia, Z., Morales, J. C., Dunphy, W. G., and Carpenter, P. B. (2001). Negative cell cycle regulation and DNA-damage inducible phosphorylation of the BRCT protein 53BP1. J. Biol. Chem. 276, 2708–2718. Xiao, Y., Weaver, D. T., Banin, S., Moyal, L., Shieh, S., Taya, Y., Anderson, C. W., Chessa, L., Smorodinsky, N. I., Prives, C., Reiss, Y., Shiloh, Y., and Ziv, Y. (1997). Conditional gene targeted deletion by Cre recombinase demonstrates the requirement for the double-strand break repair Mre11 protein in murine embryonic stem cells. Nucleic Acids Res. 25, 2985–2991. Xu, B., Kim, S. T., Lim, D. S., and Kastan, M. B. (2002). Two molecularly distinct G(2)/M checkpoints are induced by ionizing irradiation. Mol. Cell Biol. 22, 1049–1059. Xu, X., and Stern, D. F. (2003). NFBD1/MDC1 regulates ionizing radiation-induced focus formation by DNA checkpoint signaling and repair factors. FASEB J. 17, 1842–1848. Yamazaki, V., Wegner, R. D., and Kirchgessner, C. U. (1998). Characterization of cell cycle checkpoint responses after ionizing radiation in Nijmegen breakage syndrome cells. Cancer Res. 58, 2316–2322. Yang, H., JeVrey, P. D., Miller, J., Kinnucan, E., Sun, Y., Thoma, N. H., Zheng, N., Chen, P. L., Lee, W. H., and Pavletich, N. P. (2002). BRCA2 function in DNA binding and recombination from a BRCA2-DSS1-ssDNA structure. Science 297, 1837–1848. Yannone, S. M., Roy, S., Chan, D. W., Murphy, M. B., Huang, S., Campisi, J., and Chen, D. J. (2001). Werner syndrome protein is regulated and phosphorylated by DNA-dependent protein kinase. J. Biol. Chem. 276, 38242–38248. Yarden, R. I., Pardo-Reoyo, S., Sgagias, M., Cowan, K. H., and Brody, L. C. (2002). BRCA1 regulates the G2/M checkpoint by activating Chk1 kinase upon DNA damage. Nat. Genet. 30, 285–289. Yazdi, P. T., Wang, Y., Zhao, S., Patel, N., Lee, E. Y., and Qin, J. (2002). SMC1 is a downstream eVector in the ATM/NBS1 branch of the human S-phase checkpoint. Genes Dev. 16, 571–582. Yoo, S., and Dynan, W. S. (1999). Geometry of a complex formed by double strand break repair proteins at a single DNA end: Recruitment of DNA-PKcs induces inward translocation of Ku protein. Nucleic Acids Res. 27, 4679–4686. You, Z., Kong, L., and Newport, J. (2002). The role of single-stranded DNA and polymerase alpha in establishing the ATR, Hus1 DNA replication checkpoint. J. Biol. Chem. 277, 27088–27093. Yu, C. E., Oshima, J., Fu, Y. H., Wijsman, E. M., Hisama, F., Alisch, R., Matthews, S., Nakura, J., Miki, T., Ouais, S., Martin, G. M., Mulligan, J., and Schellenberg, G. D. (1996). Positional cloning of the Werner’s syndrome gene. Science 272, 258–262. Yu, X., Chini, C. C., He, M., Mer, G., and Chen, J. (2003). The BRCT domain is a phosphoprotein binding domain. Science 302, 639–642. Yuan, S. S., Lee, S. Y., Chen, G., Song, M., Tomlinson, G. E., and Lee, E. Y. (1999). BRCA2 is required for ionizing radiation-induced assembly of Rad51 complex in vivo. Cancer Res. 59, 3547–3551. Zhang, J., Willers, H., Feng, Z., Ghosh, J. C., Kim, S., Weaver, D. T., Chung, J. H., Powell, S. N., and Xia, F. (2004). Chk2 phosphorylation of BRCA1 regulates DNA double-strand break repair. Mol. Cell Biol. 24, 708–718. Zhao, S., Weng, Y. C., Yuan, S. S., Lin, Y. T., Hsu, H. C., Lin, S. C., Gerbino, E., Song, M. H., Zdzienicka, M. Z., Gatti, R. A., Shay, J. W., Ziv, Y., Shiloh, Y., and Lee, E. Y. (2000). Functional link between ataxia-telangiectasia and Nijmegen breakage syndrome gene products. Nature 405, 473–477.


35

Zhu, C., Mills, K. D., Ferguson, D. O., Lee, C., Manis, J., Fleming, J., Gao, Y., Morton, C. C., and Alt, F. W. (2002). Unrepaired DNA breaks in p53-deficient cells lead to oncogenic gene amplification subsequent to translocations. Cell 109, 811–821. Zhu, J., Petersen, S., Tessarollo, L., and Nussenzweig, A. (2001). Targeted disruption of the Nijmegen breakage syndrome gene NBS1 leads to early embryonic lethality in mice. Curr. Biol. 11, 105–109. Zou, L., Cortez, D., and Elledge, S. J. (2002). Regulation of ATR substrate selection by Rad17dependent loading of Rad9 complexes onto chromatin. Genes Dev. 16, 198–208. Zou, L., and Elledge, S. J. (2003). Sensing DNA damage through ATRIP recognition of RPA-ssDNA complexes. Science 300, 1542–1548.

2

____________________________________________________________________________

Afrotherian Origins and Interrelationships: New Views and Future Prospects Terence J. Robinson* and Erik R. Seiffert{ *Evolutionary Genomics Group, Department of Zoology University of Stellenbosch Matieland 7602, South Africa { Department of Earth Sciences Oxford University Oxford OX1 3PK, United Kingdom

I. II. III. IV. V.

Introduction Morphological and Molecular Evidence Continental Drift, Cladistic Biogeography, and Afrotherian Origins Rare Genomic Changes Conclusions References

I. Introduction Few areas of evolutionary investigation have attracted greater attention than higher level mammalian relationships and the geographic origins of major mammalian clades. Comprehensive molecular studies (e.g., AmrineMadsen et al., 2003; Madsen et al., 2001; Murphy et al., 2001a,b; Scally et al., 2001; Waddell et al., 1999) have produced highly congruent and relatively well-resolved higher level phylogenies that partition the extant placental (or eutherian) mammals into four major clades. These are the Neotropical Xenarthra (containing sloths, anteaters, and armadillos), the Laurasiatheria (containing bats, shrews, moles, hedgehogs, even- and odd-toed ungulates, carnivorans, and pangolins), the Euarchontoglires (containing primates, tree shrews, flying lemurs, rodents, and rabbits) and an endemic Afro-Arabian group called Afrotheria (Figs. 1 and 2). Afrotheria is a remarkable assemblage that includes forms as morphologically diverse as golden moles (Chrysochloridae), tenrecs (Tenrecidae), elephant-shrews or sengis (order Macroscelidea), aardvarks (order Tubulidentata) and Paenungulata—the clade containing hyraxes (order Hyracoidea), elephants (order Proboscidea), and the dugongs and manatees (order Sirenia) (Fig. 1). Although the past decade witnessed a brief resurgence in the popularity of morphological arguments favoring a hyracoid-perissodactyl clade over Paenungulata Current Topics in Developmental Biology, Vol. 63 Copyright 2004, Elsevier Inc. All rights reserved. 0070-2153/04 $35.00

37

38

Robinson and SeiVert

Figure 1 Representative species of the six orders of mammals comprising the Afrotheria: (A) elephant (order Proboscidea), (B) manatee (order Sirenia), (C) hyrax (order Hyracoidea), (D) golden mole (order Afrosoricida), (E) elephant-shrews (order Macroscelidea), and (F) aardvark (order Tubulidentata). Branching sequence represents the prevailing hypothesis of relationships among extant placentals derived from phylogenetic analyses of DNA sequence data (see also Fig. 2). (Images A and F courtesy of R. J. van Aarde; B, R. K. Bonde; C, T. Jackson; D, G. Bronner; E, G. Rathbun.)

(e.g., Court, 1992; Fischer and Tassy, 1993), an association of hyaxes, elephants, and sirenians has nevertheless long been recognized (Gregory, 1910; Novacek, 1986, 1992a,b; Novacek and Wyss, 1986; Shoshani, 1986, 1992, 1993; Simpson, 1945). In contrast, the nonpaenungulate afrotherians have been phylogenetically enigmatic. While some morphologists had noted unique anatomical features shared by aardvarks and paenungulates (e.g., Le Gros Clark and Sonntag, 1926), later studies of the remaining afrotherians were unanimous in concluding that their phylogenetic aYnities lay with a variety of other, non-paenungulate placental orders (e.g., Butler, 1972, 1988; Gregory, 1910; MacPhee and Novacek, 1993; McDowell, 1958; McKenna,

2. Afrotherian Origins and Interrelationships

39

Figure 2 Hypothesis of interordinal relationships among extant placental mammals derived from Amrine-Madsen and colleagues’ (2003) maximum likelihood and Bayesian phylogenetic analyses of >17 kb of DNA sequence data from three mitochondrial and 20 nuclear genes.

1975; Novacek, 1986; Szalay, 1977). Despite the lack of unambiguous morphological evidence linking extant afrotherians to the exclusion of other living placentals, this assemblage has, nonetheless, remained one of the most robust mammalian clades since its formal description. In this chapter, we address a number of issues relevant to our understanding of afrotherian origins and interrelationships. In so doing, we pay particular attention to the apparent conflict between molecular and morphological evidence for afrotherian monophyly, and the evolutionary and/or interpretive bases underlying the apparent absence of morphological characters that unite this group. We also deal with the phylogenetic position and time of origin of Afrotheria, and discuss the implications that these data hold for competing views of a southern (Gondwanan) versus northern (Laurasian) origin for afrotherians and other extant placental mammals (Archibald, 2003; Asher et al., 2003; Eizirik et al., 2001; Murphy et al., 2001b; Springer et al., 2003). Finally, we conclude with our views on the possible inability of sequence data to conclusively resolve many of the evolutionary relationships within Afrotheria and propose that an emphasis on the identification

40


of so-called rare genomic changes (RGCs) may ultimately provide the most convincing resolution of afrotherian supraordinal phylogeny.

II. Morphological and Molecular Evidence The controversy surrounding the issue of afrotherian monophyly stems from the fact that acceptance of Afrotheria requires the rejection of a number of alternative supraordinal hypotheses that had previously been proposed on the basis of morphological data. Foremost among these are the Ungulata, the Lipotyphla, and the Anagalida hypotheses. Ungulata is a grouping that includes paenungulates with perissodactyls and cetartiodactyls and, according to some workers, aardvarks (Court, 1992; Fischer, 1986, 1989; Fischer and Tassy, 1993; Novacek, 1986, 1992b; Novacek and Wyss, 1986; Prothero, 1993; Prothero et al., 1988). Lipotyphla aligns tenrecs and golden moles with shrews, moles, hedgehogs, and Solenodon (Butler, 1972, 1988; MacPhee and Novacek, 1993; McDowell, 1958). Finally, Anagalida places elephantshrews closest to Glires—the rodents and lagomorphs (Novacek, 1986, 1992b; Novacek and Wyss, 1986). However, these and other phylogenetic hypotheses have themselves been matters of perennial debate, and so it would be a misrepresentation to claim that the conflict surrounding placental supraordinal phylogeny is limited to one between the molecular and morphological data. In reality, the available morphological data have always provided weak, or at least consistently debatable, supraordinal phylogenetic signal. Afrotheria is, however, truly remarkable for its inclusion of tenrecs and golden moles. The craniodental morphology of these taxa is so radically diVerent from that of ‘‘ungulates’’ that it has been argued that there is little or no compelling morphological support for the monophyly of Afrotheria (Asher, 1999; Novacek, 2001). In fact, it is highly unlikely that an afrotherian clade that includes golden moles and tenrecs would have been recognized without genetic evidence (Asher, 2001). In retrospect, it can now be seen that there were clues that hinted at the possibility of an afrotherian clade. The first of these is the past geographic distribution of this group. The earliest unequivocal fossil record of each terrestrial afrotherian order is of Afro-Arabian origin (Gheerbrant et al., 1998, 2002, 2003; Hartenberger, 1986; Patterson, 1975; SeiVert, 2003). Additionally, most of these records derive from a time period during which Afro-Arabia was a continental island, well isolated from other major landmasses. Although Paenungulata has produced a number of aquatic and semi-aquatic forms whose Eocene-to-Recent fossil remains have been found outside of the Afro-Arabian domain (Domning et al., 1986; Radulesco et al., 1976; Savage et al., 1994), these records of highly specialized paenungulate


41

taxa have no relevance for our understanding of the geographic origin of either Paenungulata or Afrotheria. The second clue is found in the male reproductive system: afrotherians appear to have descended from a common ancestor that had completely intra-abdominal testes (the rare ‘‘primary testicond’’ condition that is also seen in monotremes). Werdelin and Nilsonne (1999) have argued that this is a derived condition among therians and, if this is correct, this feature would provide compelling support for afrotherian monophyly. Several other soft-tissue features that were hypothesized to be possible afrotherian synapomorphies (e.g., a mobile proboscis [Hedges, 2001], fetal membranes) were subsequently found to not support afrotherian monophyly (Carter, 2001; Carter et al., 2004; Whidden, 2002), but these features do provide some support for supraordinal relationships within Afrotheria. The third clue comes from recent fossil discoveries. As it turns out, the teeth of the earliest alleged fossil macroscelideans show little resemblance to rodents or lagomorphs and instead closely resemble those of paenungulates and a variety of so-called ‘‘condylarths’’ (i.e., primitive fossil ‘‘ungulate’’ progenitors) (Hartenberger, 1986; SeiVert, 2003; Simons et al., 1991; Tabuce et al., 2001). By the time the implications of these fossils had been fully digested, however, molecular systematists were already well on their way to finding genetic evidence that linked elephant-shrews and paenungulates to the exclusion of living perissodactyls and artiodactyls. De Jong et al. (1981) were the first to marshall genetic evidence (from alphaA-lens crystallin protein sequences) in support of the hypothesis that paenungulates were more closely related to aardvarks than to perissodactyls and artiodactyls. Later comparison of homologous alphaA-lens crystallin sequences from an elephant-shrew (de Jong et al., 1993) also revealed remarkable similarity to paenungulates and indicated that macroscelideans may be more closely related to aardvarks and paenungulates than to rodents or rabbits. This hypothesis was subsequently supported by an analysis of the nuclear vWF gene (Porter et al., 1996). Phylogenetic analysis of 12S rRNA sequences by Lavergne et al. (1996) provided the first surprising evidence indicating that paenungulates might be more closely related to golden moles than to other ‘‘ungulates.’’ The monophyly of a clade containing paenungulates, golden moles, elephant-shrews, and aardvarks was later upheld by analyses of both mitochondrial (valine transfer RNA, 12S and 16S rRNA) and nuclear (vWF and A2AB) gene sequences (Springer et al., 1997). Finally, using these same sequences, Stanhope et al. (1998) convincingly demonstrated that the Afro-Malagasy tenrecs also fell within this clade, which they named Afrotheria to reflect the clade’s presumed Afro-Arabian origin. Afrotherian monophyly has since been tested and strongly supported multiple times, with additional sequences published by other authors (AmrineMadsen et al., 2003; Madsen et al., 2001; Malia et al., 2002; Murata et al., 2003; Murphy et al., 2001a,b; Waddell and Shelley, 2003). It now finds

42


perfect statistical support in phylogenetic analyses of the longest (>17 kb) available concatenation of placental mammalian nuclear and mitochondrial genes (Amrine-Madsen et al., 2003). This clade has also come to be supported by a variety of protein sequence signatures (van Dijk et al., 2001), deletions (Madsen et al., 2001; Scally et al., 2001), SINEs (short interspersed nuclear elements) (Nikaido et al., 2003), and cytogenetic markers (Fro¨ nicke et al., 2003; Robinson et al., 2004; Svartman et al., 2004; Yang et al., 2003). Given the overwhelming molecular support for afrotherian monophyly, it is peculiar that this clade has not yet come to be clearly delineated by a suite of easily recognizable morphological features. It is possible that the last common ancestor of extant afrotherians simply had not acquired any distinctive morphological apomorphies since its divergence from other extant placentals. However, this seems highly unlikely given that stem afrotherians are estimated to have had 25 million years of independent evolution before the appearance of the afrotherian crown group at approximately 80 million years ago (Ma) (Springer et al., 2003). Another possible explanation for this phenomenon is that the supraordinal diversification of crown (extant) afrotherians was to some extent ‘‘explosive.’’ That is, extant afrotherian lineages were derived from groups that rapidly entered into highly specialized niches in the later Cretaceous or early Paleogene, and afrotherian synapomorphies may have been ‘‘erased’’ in the more specialized afrotherian clades over the course of the past 60–70 million years (Madsen et al., 2001). In this regard, it is clear that similar selection pressures favored the evolution of detailed morphological convergences in the laurasiatherian and afrotherian radiations (Madsen et al., 2001; Scally et al., 2001; SeiVert, 2002). For instance, these clades independently produced habitually aquatic taxa (cetaceans vs sirenians), dedicated large-bodied herbivores (perissodactyls vs extinct Paleogene hyracoids), fossorial insectivores (true moles vs golden moles), zalambdodont insectivores (Solenodon vs golden moles and tenrecs), and large-bodied myrmecophagous forms (pangolins vs aardvarks), among others. In some cases, these adaptations were occurring almost simultaneously on the diVerent landmasses. For instance, an expanding fossil record (e.g., Domning, 2001; Gingerich et al., 2001) now allows us to infer that sirenians and cetaceans were probably making independent transitions from semiterrestrial to habitual aquatic behavior during the early Eocene. If strong selection pressures have helped to conceal deep morphological phylogenetic signal in the afrotherian radiation, then our understanding of this group would benefit greatly if we could identify which taxa might have had their afrotherian synapomorphies erased, and which taxa might preserve the morphological apomorphies that were present in the afrotherian ancestor. The most pressing interpretive problem in this regard is the uncertain phylogenetic position of aardvarks and elephant-shrews with respect to


43

other afrotherians (SeiVert, 2002). Aardvarks have highly modified craniodental and postcranial morphology due to their myrmecophagous and fossorialhabits, but it is clear that elephant-shrews preserve a number of specialized paenungulate-like features in their postcranial, cranial, and dental morphology. The greater similarity of macroscelidean teeth to those of primitive fossil ‘‘ungulates’’ (rather than to those of primitive lagomorphs and rodents) was first noted by Hartenberger (1986) and Simons et al. (1991) following the discovery and analysis of the alleged Eocene macroscelideans Chambius and Herodotius. These taxa share a suite of dental features with primitive paenungulates that are not seen in tenrecs and golden moles. These features include low-crowned upper and lower molars, lower molars with an anterior (trigonid) portion of the tooth only slightly taller than the posterior (talonid), loss of the paraconid (anterior) cusp on lower molars, upper molars with a large hypocone cusp, and, in some cases, incipient bilophodonty (i.e., incipient development of two lophs or crests connecting the inner and outer cusps of the upper and lower molars). All of these dental features are absent in primitive Cretaceous placentals (e.g., Cifelli, 1999; Ji et al., 2002; Kielan-Jaworowska and Dashzeveg, 1989) and are almost certainly derived within Placentalia. The dental similarities shared by early fossil macroscelideans and fossil paenungulates are so striking that a phylogenetic analysis of dental features by Tabuce et al. (2001) found elephant-shrews to be nested within Paenungulata alongside proboscideans to the exclusion of hyracoids and many Laurasian ‘‘condylarths.’’ A more recent phylogenetic analysis that included 378 craniodental, postcranial, and soft-tissue characters scored across a diverse sample of 53 living and extinct afrotherians recovered a well-supported macroscelidean-paenungulate clade and a weaker tubulidentate-macroscelidean-paenungulate clade, to the exclusion of tenrecs and golden moles (SeiVert, 2003). That study demonstrated that apomorphic morphological similarities shared by paenungulates and elephant-shrews can also be found in the cranial and postcranial anatomy of these groups. With the sole exception of a phylogenetic analysis of combined 12S rRNA, tRNA-valine, 16S rRNA, vWF, and A2AB sequences published by Stanhope et al. (1998), which provided very weak support for a Tubulidentata-Macroscelidea-Paenungulata clade within Afrotheria, all major maximum likelihood and Bayesian phylogenetic analyses of concatenated sequence data have provided very diVerent results. The longest concatenations of nuclear and mitochondrial sequences now provide strong support for a placement of elephant-shrews and aardvarks with tenrecs and golden moles to the exclusion of paenungulates in a clade that Waddell et al. (2001) have named ‘‘Afroinsectiphillia’’ (Amrine-Madsen et al., 2003; Murphy et al., 2001b; Waddell and Shelley, 2003) (Fig. 1). Murphy et al. (2001b) and Amrine-Madsen et al. (2003) have also found support

44


for a macroscelidean-tenrec-golden mole clade (Afroinsectivora) within Afroinsectiphillia, although this is the weakest grouping within that clade. Waddell and Shelley’s (2003) analysis of a diVerent concatenation instead found strong support for an aardvark-tenrec clade and a weaker grouping of these taxa and golden moles to the exclusion of elephant-shrews within a well-supported Afroinsectiphillia. If these molecular data are correct, then a close relationship of macroscelideans, tenrecs, golden moles, and aardvarks would indicate that the morphological features shared by elephant-shrews and paenungulates must be either homoplasious or plesiomorphic within Afrotheria. The former scenario would do little to aid our understanding of afrotherian origins, for it would seem to solidify the hypothesis that there is little or no morphological evidence for either afrotherian monophyly or most supraordinal relationships within Afrotheria. Intriguingly, however, if macroscelidean-paenungulate morphological similarities are plesiomorphic within Afrotheria, then they would, given all available fossil evidence, nevertheless appear to be apomorphic within Placentalia, and those apomorphies could then provide compelling morphological support for afrotherian monophyly (SeiVert, 2002, 2003). This scenario would, in turn, imply that it was the tenrecs and golden moles that had their afrotherian synapomorphies erased at some point in their evolutionary history. This would have occurred either during their period of presumed shared ancestry or through convergence and would suggest that these groups had undergone a number of detailed reversals to the primitive placental morphotype (and/or convergences with other members of ‘‘Lipotyphla’’) in their craniodental and postcranial morphology.

III. Continental Drift, Cladistic Biogeography, and Afrotherian Origins Phylogenetic analyses of mitochondrial genomes often support a nested placement for Afrotheria within Placentalia (e.g., Arnason et al., 2002). In contrast, the longest concatenations of combined nuclear and mitochondrial sequences suggest a basal placement for Afrotheria (Fig. 2) as the sister group of all other extant placentals (i.e., a clade containing ‘‘Boreoeutheria’’ [Laurasiatheria þ Euarchontoglires] and Xenarthra) (Amrine-Madsen et al., 2003; Murphy et al., 2001b; Waddell and Shelley, 2003). Importantly, these data imply that the time and place of origin of stem Afrotheria is coincident with the time and place of origin of crown Placentalia. Murphy et al. (2001b) and Eizirik et al. (2001) interpreted the basal placement of both Afrotheria and Xenarthra and the probable paraphyly of these Gondwanan clades with respect to Boreoeutheria as supporting a Gondwanan origin for crown placentals. Murphy et al. (2001b) specifically argued that the ancestral stock


45

of crown Placentalia was present on the Gondwanan landmass in the late Cretaceous. They maintain that continental break-up during this interval was responsible for the isolation of Afrotheria in Afro-Arabia and Xenarthra þ Boreoeutheria in South America at approximately 105 Ma, which agrees with the estimate for the break-up of Africa and South America at 100–120 Ma (Smith et al., 1994). A stem boreoeutherian is hypothesized to have subsequently dispersed from South America into Laurasia between 88 and 100 Ma, ultimately giving rise to the boreoeutherian crown group. As noted by Amrine-Madsen et al. (2003), however, it is still impossible to statistically reject the alternative phylogenetic scenarios of an AfrotheriaBoreoeutheria clade, or an Afrotheria-Xenarthra (‘‘Atlantogenata’’) clade with the largest published molecular data set, and so a number of alternative biogeographic hypotheses also cannot yet be rejected. In fact, given the strong fossil evidence for a Laurasian origin of stem Placentalia (Averianov and Skutschas, 2001; Cifelli, 1999; Ji et al., 2002; Kielan-Jaworowska and Dashzeveg, 1989; Wible et al., 2001) and stem Marsupialia (Averianov and Kielan-Jaworowska, 1999; Cifelli, 1993; Luo et al., 2003; Rougier et al., 1998), as well as the relatively few lineage segments on which the Gondwanan hypothesis currently hinges, should the Afrotheria þ Boreoeutheria or Atlantogenata hypotheses ultimately be statistically rejected, a Gondwanan versus Laurasian origin for crown placentals would remain highly speculative in the absence of positive fossil evidence from Afro-Arabia or South America. Indeed, it is this very absence of fossil evidence—specifically, the absence of fossil evidence from the late Cretaceous of Afro-Arabia (where, with the exception of a single late Cretaceous mammalian caudal vertebra [Nessov et al., 1998b; Rage and Cappetta, 2002], no late Cretaceous or early Paleocene mammals have been discovered)—that presently leaves the Gondwanan hypothesis particularly unsatisfying. Other problems concern the interpretation of some alleged Cretaceous crown placentals from the northern and southern landmasses. The hypothesis of a Gondwanan origin for crown placentals appears to have been influenced in part by the initial identification of an early Cretaceous tribosphenic mammal from Australia named Ausktribosphenos, as a placental mammal possibly related to hedgehogs (Rich et al., 1997) (see e.g., Madsen et al., 2001). The later discovery of the tribosphenic mammal Ambondro in Madagascar led to the proposal that tribosphenic stem therians were present in the southern hemisphere by the middle Jurassic (Flynn et al., 1999). Although it has been claimed that all of these Mesozoic Gondwanan tribosphenic mammals, as well as the more recently described Australian form Bishops (Rich et al., 2001) and the South American form Asfaltomylos (Rauhut et al., 2002), are placental mammals (Woodburne et al., 2003), a reanalysis of Ausktribosphenos and Ambondro (Luo et al., 2001) and later Bishops (Luo et al., 2002) and Asfaltomylos (Luo et al., 2003) suggested that

46


these peculiar forms are actually all more closely related to living monotremes than to therians. This implies that the tribosphenic molar evolved convergently in a northern (‘‘boreosphenidan’’) group that gave rise to placentals and marsupials and a southern (‘‘australosphenidan’’) group that gave rise to monotremes. Interestingly, Luo et al. (2001) largely dismissed the fact that some of the earliest tribosphenic mammals are African (Sigogneau-Russell, 1991) by stating that ‘‘the earliest tribosphenic mammals (earliest Cretaceous) were originally known only from Laurasia, or an adjacent and biogeographically related part of northwestern Africa’’ (p. 55). While acknowledging a probable relationship of Ausktribosphenos to monotremes, Sigogneau-Russell et al. (2001) have argued that Malagasy Ambondro is nevertheless more likely to be related to Laurasian Cretaceous tribosphenic mammals and leave open the possibility that the tribosphenic molar evolved in Gondwana. Ultimately, these possible Gondwanan ‘‘boreosphenidans’’ may not be particularly relevant to the issue of crown therian (or placental) origins, for in addition to the aforementioned stem placentals and stem marsupials known from northern continents, there is also an increasing abundance of placentals from the late Cretaceous of Asia. Of these, the most notable are the zhelestids and zalambdalestids, which have been interpreted as probable members of the placental crown group (Archibald, 1996, 2003; Archibald et al., 2001; Nessov et al., 1998a). Zhelestids were initially aligned with the artificial ‘‘Ungulata’’ assemblage (Archibald, 1996) and more recently the ‘‘Fereungulata’’ clade (Fig. 1) containing Cetartiodactyla, Perissodactyla, Carnivora, and Pholidota (Archibald, 2003). Interestingly, zalambdalestids have been interpreted by Archibald et al. (2001) as being stem members of the Glires (rodent and lagomorph) clade. If these assessments are correct, then these taxa would provide evidence for continuity of placental evolution through the early and late Cretaceous in Asia with no clear indication of an invasion of crown placentals or replacement of stem placentals in the later Cretaceous. In constrast, evidence from the zhelestid ear region (Ekdale et al., 2004) and the zalambdalestid cranium (Wible et al., 2004) and postcranium (Kielan-Jaworowska, 1975; Novacek et al., 1997) has demonstrated that these taxa lack apparent morphological synapomorphies of all extant placentals. This fossil evidence indicates either that zhelestids and zalambdalestids are not members of the placental crown group or that there has been considerable morphological homoplasy within crown Placentalia (Archibald et al., 2001; Ekdale et al., 2004). Interestingly, the age of the earliest zhelestids (85–90 Ma forms from Asia) (Nessov et al., 1998a) approximately matches recent Bayesian molecular estimates for the origin of Laurasiatheria (between 80 and 90 Ma) (Springer et al., 2003), suggesting that if zhelestids are crown placentals, it is perhaps more likely that these taxa are basal


47

laurasiatherians or boreoeutherians. Such a basal position for zhelestids could lend support to the hypothesis of a ‘‘condylarth’’ origin for these supraordinal clades (as is also possible for Afrotheria and, as such, potentially all of crown Placentalia). Another interpretation of afrotherian biogeography developed by Asher et al. (2003) derives from a parsimony analysis of morphological data collected from 49 extant and 11 extinct taxa and the molecular data published by Murphy et al. (2001b). This analysis recovered a host of extinct North American ‘‘condylarths,’’ specifically the late Paleocene genera Hyopsodus and Meniscotherium and possibly Phenacodus as the earliest known members of an afrotherian clade that was found to be nested deep within Placentalia. As these North American condylarths were found to form a paraphyletic group that were aligned with paenungulates to the exclusion of aardvarks, elephant-shrews, tenrecs, and golden moles, Asher et al. (2003) argued that their study called into question the validity of a hypothesized Afro-Arabian origin for Afrotheria. These authors did, however, acknowledge that ‘‘the existence of ancient, Laurasian crown members of Afrotheria does not necessarily imply a non-African origin for the clade’’ (p. 155). Asher and colleagues’ (2003) eVorts to resolve the phylogenetic position of these enigmatic taxa represent a major methodological advance over previous such attempts, but in our opinion a critical question is whether their study’s character and taxon sampling is suYcient to confidently draw the conclusion that Meniscotherium, Hyopsodus, and possibly Phenacodus are indeed members of the afrotherian clade. The problem is that if we accept that clades such as Afrotheria, Laurasiatheria, and Euarchontoglires exist, then we must also accept the fact that the widespread accumulation of homoplasious morphological features in various placental clades since the late Cretaceous seems to have left it practically impossible to reconstruct placental supraordinal relationships using a limited sample of morphological characters scored across a limited sample of living and extinct taxa. For instance, in their analyses that exclude molecular data, Asher et al. (2003) recovered a number of clades that are likely to be bound together in large part by convergent morphological adaptations facilitating a specialized lifestyle. Examples include a ‘‘gliding/flying clade’’ containing euarchontan flying lemurs and laurasiatherian bats and a ‘‘myrmecophagous clade’’ containing xenarthrans, laurasiatherian pangolins, afrotherian aardvarks, and the enigmatic Malagasy subfossil Plesiorycteropus. In light of such results, why should we expect our phylogenetic answers to be any less misleading when we turn to the fossil record? If it is so diYcult to place extant taxa in their ‘‘correct’’ phylogenetic position using morphological data, then it is worth asking what hope we might have for recognizing their even more enigmatic and incompletely known extinct relatives for what they are, given that they cannot, in almost

48


all cases, be sampled for genetic material. This problem is perhaps the most daunting issue currently facing paleomammalogy. Indeed, part of the importance of Asher et al.’s (2003) study is that it has exposed the enormity of the task required to systematically synthesize the molecular and morphological evidence necessary for assessing relationships among living and extinct placentals using modern phylogenetic techniques. Given the incalculable evolutionary complexity that characterized the last 105 million years of crown placental evolution, we suggest that convincing answers to the sorts of questions now being asked by Asher et al. (2003) and others are attainable, but they will only come through vastly increased character and taxon sampling in combined analyses of molecular (sensu lato) and morphological data. By way of example, Asher et al. (2003) employed 196 morphological characters to resolve relationships among the 58 relatively complete taxa that represented the entirety of Placentalia. However, in a more recent phylogenetic analysis of the afrotherian radiation alone, SeiVert (2003) required almost twice as many morphological characters (378) to partially resolve afrotherian interrelationships, and he sampled not only complete taxa but also highly incomplete, but no less important, extinct taxa. In our opinion, if we are to do the fossil record justice, this sort of morphological character sampling and increased sampling of both highly complete and highly incomplete (but nevertheless potentially pivotal) taxa must be undertaken. This is, admittedly, an enormous task that cannot move forward without the use of supercomputers and a great deal of patience. These reasons explain why we remain highly skeptical of the alleged afrotherian aYnities of Hyopsodus and Meniscotherium, and why we would not be at all surprised if additional sampling of morphological characters (or of living or extinct taxa) eventually came to support an alignment of Hyopsodus, Meniscotherium, and Phenacodus with some historically Laurasian group such as perissodactyls or cetartiodactyls. These are, after all, the other extant ‘‘ungulates’’ that share numerous derived morphological similarities with paenungulates and extinct condylarths, but their early members (unlike early undoubted paenungulates) actually lived alongside these condylarths in the Laurasian Paleocene and Eocene. While the picture will hopefully become clearer as the early Afro-Arabian record improves, the goal of recovering later Cretaceous mammals from Africa remains an elusive one. The discovery of a possible gondwanatherian mammal from the Cretaceous of Tanzania (Krause et al., 2003) represents a very positive development, but unfortunately this highly derived mammal is probably of no relevance to placental mammalian evolution. Moreover, if the site that produced the fossil is early Cretaceous in age, then these sediments would not be expected to yield fossils of early afrotherians. Asher et al. (2003) argued that discoveries in the latest Cretaceous of Madagascar (e.g., Krause, 2001; Krause et al., 1997) ‘‘have improved the African record


49

and show no sign of modern African mammal lineages in an otherwise very diverse vertebrate fauna’’ (p. 156). This is, however, an incorrect interpretation of the relevance of Madagascar for understanding Cretaceous AfroArabian faunas. Geophysical evidence indicates that Madagascar has had no connection with Afro-Arabia since at least the Jurassic (CoYn and Rabinowitz, 1987), and Malagasy late Cretaceous faunas would be predicted to have greater similarity with late Cretaceous faunas of India, Antarctica, and South America, rather than with those of Afro-Arabia (Hay et al., 1999; Smith et al., 1994). Regardless, it should be noted that given such predicted patterns, the possibility that Madagascar may have been home to placentals during the late Cretaceous should not be excluded based on that continent’s notoriously poor fossil record. This is especially so given that late Cretaceous placentals were present on the Indian landmass prior to that plate’s collision with Asia (Prasad and Sahni, 1988; Prasad et al., 1994; Rana and Wilson, 2003), and an enigmatic late Cretaceous therian has only recently been discovered in Madagascar (Averianov et al., 2003; Krause, 2001). These records are, nevertheless, still unlikely to tell us much about what mammals were in Africa at the time that crown placentals are estimated to have been diversifying (Springer et al., 2003).

IV. Rare Genomic Changes The long length of the molecular concatenations used in recent studies and associated problems such as the presence of long edges, short internodes, saturation, nonindependent substitution, presence of base composition shifts, and functional constraints leave it unlikely that sequence models hold all the answers to the problems of afrotherian interordinal relationships. Thus, both testing of the sequence-based assemblages and conclusive resolution within Afrotheria await genetic markers that are not plagued by excessive homoplasy, such as so-called rare genomic changes or RGCs (reviewed in Rokas and Holland, 2000). Characters such as long deletions or indels (Amrine-Madsen et al., 2003; Madsen et al., 2001), SINE (Nikaido et al., 2003), protein sequence signatures (van Dijk et al., 2001), and chromosomal rearrangements have the advantage of being similar to conventional morphological synapomorphies in that they are either present or absent, and do not require complex models to explain their evolution (Amrine-Madsen et al., 2003; Waddell et al., 2001). The usefulness of RGCs is no more apparent than in the recognition of afrotherian monophyly. For example, afrotherian monophyly is supported by the detection of a unique 9-bp deletion in exon 11 of the BRCA1 gene (Madsen et al., 2001), the protein sequence signatures (proteinmorphological synapomorphies) detected by van Dijk et al. (2001), the 50

50


30

and deletions present in exon 26 of apolipoprotein B (APOB) (AmrineMadsen et al., 2003), and the so-called AfroSINEs reported by Nikaido et al. (2003). It may be, however, that RGC’s greatest potential lies in providing resolution of afrotherian supraordinal relationships. For example, a distinct AfroSINE subfamily identified by Nikaido and colleagues is restricted to the genomes of the hyraxes, elephants, and sirenians, supporting the monophyly of Paenungulata suggested by many diverse lines of evidence. In contrast to the overwhelming support for paenungulate monophyly, diVerent sister-group relationships have been suggested for hyraxes, elephants, and sirenians. Asher et al. (2003), Lavergne et al. (1996), Murata et al. (2003), and Murphy et al. (2001a) have found evidence of a proboscideansirenian association (Tethytheria), whereas the data sets of Amrine and Springer (1999) and Murphy et al. (2001b), among others, suggest a Sirenia-Hyracoidea clade, admittedly with variable and often weak bootstrap probability. Most recently, Amrine-Madsen et al. (2003) retrieved a preferred hyrax þ elephant grouping, which was also supported by a unique protein signature in the APOB gene. While this finding is encouraging, additional RGCs will be critical to unequivocally resolving the paenungulate trichotomy. Recent cross-species chromosome painting studies (Fro¨ nicke et al., 2003; Robinson et al., 2004; Svartman et al., 2004; Yang et al., 2003) provide another such example of the power of RGCs. By using fluorescence in situ hybridization (FISH) of human chromosome painting probes that are isolated by bivariate flow sorting (Telenius et al., 1992), several shared segmental syntenies have been identified that provide support for the Afrotheria and some of the problematic associations within the superorder. In this procedure, the individual chromosomes of a given species are physically isolated using fluorescence-activated cell sorting. Characterization of flow-sorted chromosomes occurs by degenerate oligonucleotide-primed (DOP) PCR amplification of DNA from the chromosome pools, labeling with a particular fluorescence dye, and subsequent hybridization of the labeled DNA to metaphase chromosomes of the donor species. Speciesspecific painting probes generated in this manner can be used in cross-species FISH experiments to identify regions of chromosome conservation (Fig. 3). Using a comparative FISH approach that relied on human painting probes, Fro¨ nicke et al. (2003) identified two segmental combinations in the elephant (HSA 5/21, HSA 1/19p) that were hypothesized to be synapomorphies uniting Afrotheria. Robinson et al. (2004) confirmed and extended the observations, particularly with respect to the former synteny. They suggested that the proposed ancestral eutherian association of HSA 3/21 (Yang et al., 2003) be expanded to include segments homologous to human chromosome 5, forming an HSA 3/21/5 segmental combination defining


51

Figure 3 (A) Computer generated flow karyotype of aardvark. (B) Characterization of chromosome pools by labeling DNA with fluorescent dyes and in situ hybridization to aardvark chromosomes 1 (red), 2 (blue), and 3 (green). (C) Cross-species painting using human (HSA) chromosome painting probes 1 and 19 showing DNA sequence homology to chromosomal regions in the aardvark. The HSA 1/19p segmental combination is a synapomorphy uniting the Afrotheria. (Images courtesy of F. Yang.)

Afrotheria. Arguably of more interest, given the enigmatic relationships within Afrotheria, was the detection by Robinson and colleagues of one segmental configuration (HSA 2/8p/4) in the golden mole, elephant-shrew, and aardvark that is absent in the elephant. This provides the first nonsequence-based evidence in support of the Afroinsectiphillia as well as additional genomic evidence that contradicts the hypothesis that the morphological similarities shared by elephant-shrews and paenungulates are derived within crown Afrotheria. Additionally, the painting data support the recognition of elephant-shrew and aardvark as sister taxa (HSA 10q/17 and HSA 3/20 are present in both the elephant-shrew and aardvark but are absent in the golden mole and elephant). This phylogenetic association has been detected (although with relatively weak bootstrap support) by Stanhope et al. (1998) in some trees based on protein residues by van Dijk et al. (2001) and Waddell et al. (2001). Importantly, Robinson et al. (2004) found no support for a closer relationship of elephant-shrews and golden moles within Afroinsectivora (tenrecs were not included in their study). While this is in conflict with Amrine-Madsen et al. (2003), Madsen et al. (2001), and Murphy et al. (2001b), among others, it is more in keeping with the results of, for instance, Waddell and Shelley’s (2003) phylogenetic analyses. Despite the apparent usefulness of RGCs such as LINEs (long and short interspersed nuclear elements) and SINEs in phylogenetic inference and conserved chromosomal segmental combinations similar to those discussed previously, they are not exempt from problems of convergence (Slattery et al., 2000) or lineage sorting of ancestral polymorphisms. Lineage sorting can introduce homoplasy in RGCs when a polymorphism becomes fixed in

52


some but not all descendants of the polymorphic ancestor (Hillis, 1999). This is particularly pertinent to chromosomal data because there are usually very limited numbers of characters that define clades. Statistical analyses need to be developed as a matter of priority in order to draw robust phylogenetic inferences from these and other RGC data (Waddell et al., 2001). An equally important caveat concerns the question of true homology among segmental combinations shared by diVerent lineages and the need to determine this by reciprocal painting experiments. The defining character in a conserved segmental association, for example, HSA1/19p, is the presence of the 1/19p breakpoint. While this would be expected to be strongly conserved, it should, ideally, be verified by comparative sequencing. In contrast, however, gene order within the synenic block may be altered by intrachromosomal rearrangement, and the size of segments by subsequent translocations to other regions in the genome. Although cross-species FISH using chromosome-specific painting probes does not give insight to changes in gene order in the regions of homology detected by the probe, this does not detract from use of the presence or absence of the evolutionary breakpoint for inferring phylogenetic relationships. Of concern, however, are instances in which multiple translocations of the same chromosome have occurred. In these instances, the use of a single whole chromosome painting probe can readily lead to erroneous conclusions. By way of example, an HSA1q/19q syntenty identified in the prosimians (Nie et al., unpublished) would be indistinguishable from the HSA1/19p found in the Afrotheria when using unidirectional painting with human 1 and 19 paints. This underscores the importance of using reciprocal painting schemes when determining homology among putatively conserved segmental chromosome associations.

V. Conclusions New molecular studies and fossil discoveries are rapidly altering our perceptions of the origin and radiation of living placental mammals, and perhaps the most profound change is the recognition of afrotherian monophyly. The identification of this group presents a major challenge to molecular biologists, evolutionary morphologists, and paleontologists alike, for this enigmatic assemblage demands an evolutionary explanation that can only come through the integration of new genomic data, reanalysis of the poorly documented gross anatomy of the living afrotherians, and novel fossil discoveries from the later Cretaceous and early Paleogene of Afro-Arabia (and possibly Laurasia). Our understanding of Afrotheria is still in its infancy, but the recognition of afrotherian monophyly has exposed a number of unique phylogenetic patterns in the morphology and physiology of these peculiar mammals that had previously gone unrecognized. As noted


53

previously, afrotherians evidently evolved from a common ancestor with a primary testicond male reproductive system that represents either a primitive retention from the ancestral therian or perhaps a reversal from an ancestral from that had evolved testicular descent. Similarly, some afrotherians are unlike many other extant placentals in exhibiting low core body temperatures (indeed, some of the lowest yet recorded among therians [Lovegrove et al., 2001]), daily heterothermy, and, in some cases, poor thermoregulatory capabilities. These types of physiological characteristics can constrain both activity cycles and geographic expansion given changing ecological conditions, and it is possible that these constraints limit the ecological flexibility of afrotherians, particularly the smaller members of Afroinsectiphillia (SeiVert, 2002). These physiological patterns may provide an explanation for why ‘‘afroinsectivorans’’ appear to have never left Africa. As with most of the other recently recognized superordinal clades within Placentalia, the origin of Afrotheria will remain enigmatic until substantial improvements are made to the group’s fossil record. Based on available evidence, it appears that the common ancestor of crown afrotherians was either ‘‘condylarth’’-like or ‘‘insectivore’’-like in its overall morphology. If the former scenario is correct, then such ‘‘condylarth’’-like features were subsequently modified by evolutionary reversals to the crown placental morphotype and/or convergences with eulipotyphlans in tenrecs and golden moles. On the other hand, should the common ancestor of afrotherians have been ‘‘insectivore’’-like, the ‘‘condylarth’’-like features observable in paenungulates, aardvarks, and elephant-shrews must have evolved convergently in all three lineages. Molecular support for the paraphyly of the ‘‘condylarth’’-like taxa compared to the ‘‘insectivore’’-like taxa (e.g., AmrineMadsen et al., 2003) appears to favor the first scenario. However, the long terminal branches evident among non-paenungulate afrotherians would have left plenty of time for morphological autapomorphies and convergences to accumulate, and this consideration could lend support to the second scenario. Without additional fossil evidence it is impossible to predict which of these alternatives (or if some other, intermediate version) is correct. After all, now that we have good reason to believe that whales can be highly derived artiodactyls (e.g., Amrine-Madsen et al., 2003; Gingerich et al., 2001) and that megabats can be highly derived microbats (Springer et al., 2001; Teeling et al., 2000, 2002), it is obvious that placental supraordinal relationships can no longer be evaluated using simple directional notions about what can and cannot happen in mammalian evolution. Fortunately, the phylogenetic framework provided by molecular data has forced a major paradigm shift that is now redirecting our eVorts to new phylogenetic, biogeographic, and adaptive questions, but the historical biological information provided by the last remaining twigs of extant mammalian diversity pales in comparison to the infinitely more complex story that will ultimately

54


be illuminated as extinct placentals come to be analyzed within this new molecular framework. Much remains to be done in resolving the relationships of extant afrotherians using molecular techniques. It is, however, somewhat ironic that in the age of genomics the colossal task of making sense of the emerging afrotherian molecular phylogeny now falls, in large part, on the shoulders of the paleontologists, for in many cases, it is only through the recovery of new fossil evidence that we will come to understand how, when, where, and why these remarkable evolutionary patterns came to be.

References Amrine, H., and Springer, M. S. (1999). Maximum-likelihood analysis of the tethythere hypothesis based on a multigene data set and a comparison of diVerent models of sequence evolution. J. Mammal. Evol. 6, 161–176. Amrine-Madsen, H., Koepfli, K.-P., Wayne, R. K., and Springer, M. S. (2003). A new phylogenetic marker, apolipoprotein B, provides compelling evidence for eutherian relationships. Mol. Phylogenet. Evol. 28, 225–240. Archibald, J. D. (1996). Fossil evidence for a late Cretaceous origin of ‘‘hoofed’’ mammals. Science 272, 1150–1153. Archibald, J. D. (2003). Timing and biogeography of the eutherian radiation: Fossils and molecules compared. Mol. Phylogenet. Evol. 28, 350–359. Archibald, J. D., Averianov, A. O., and Ekdale, E. G. (2001). Late Cretaceous relatives of rabbits, rodents, and other extant eutherian mammals. Nature 414, 62–65. Arnason, U., Adegoke, J. A., Bodin, K., Born, E. W., Esa, Y. B., Gullberg, A., Nilsson, M., Short, R. V., Xu, X., and Janke, A. (2002). Mammalian mitogenomic relationships and the root of the eutherian tree. Proc. Natl. Acad. Sci. USA 99, 8151–8156. Asher, R. J. (1999). A morphological basis for assessing the phylogeny of the ‘‘Tenrecoidea’’ (Mammalia, Lipotyphla). Cladistics 15, 231–252. Asher, R. J. (2001). Cranial anatomy in tenrecid insectivorans: Character evolution across competing phylogenies. Am. Mus. Nov. 3352, 1–54. Asher, R. J., Novacek, M. J., and Geisler, J. H. (2003). Relationships of endemic African mammals and their fossil relatives based on morphological and molecular evidence. J. Mammal. Evol. 10, 131–194. Averianov, A., and Kielan-Jaworowska, Z. (1999). Marsupials from the Late Cretaceous of Uzbekistan. Acta Palaeontol. Pol. 44, 71–81. Averianov, A. O., Archibald, J. D., and Martin, T. (2003). Placental nature of the alleged marsupial from the Cretaceous of Madagascar. Acta Palaeontol. Pol. 48, 149–151. Averianov, A. O., and Skutschas, P. (2001). A new genus of eutherian mammal from the Early Cretaceous of Transbaikalia, Russia. Acta Palaeontol. Pol. 46, 431–436. Butler, P. M. (1972). The problem of insectivore classification. In ‘‘Studies in Vertebrate Evolution’’ (T. S. Kemp, Ed.), pp. 253–265. Oliver and Boyd, Edinburgh. Butler, P. M. (1988). Phylogeny of the insectivores. In ‘‘The Phylogeny and Classification of the Tetrapods, Volume 2: Mammals’’ (M. J. Benton, Ed.), pp. 117–141. Clarendon Press, Oxford. Carter, A. M. (2001). Evolution of the placenta and fetal membranes seen in the light of molecular phylogenetics. Placenta 22, 800–807. Carter, A. M., Blankenship, T. N., Kunzle, H., and Enders, A. C. (2004). Structure of the definitive placenta of the tenrec, Echinops telfairi. Placenta 25, 218–232.


55

Cifelli, R. L. (1993). Early Cretaceous mammal from North America and the evolution of marsupial dental characters. Proc. Natl. Acad. Sci. USA 90, 9413–9416. Cifelli, R. L. (1999). Tribosphenic mammal from the North American Early Cretaceous. Nature 401, 363–366. CoYn, M. F., and Rabinowitz, P. D. (1987). Reconstruction of Madagascar and Africa: Evidence from the Davie Fracture Zone and western Somali Basin. J. Geophys. Res. 92, 9385–9406. Court, N. (1992). The skull of Arsinoitherium (Mammalia, Embrithopoda) and the higher order interrelationships of ungulates. Palaeovertebrata 22, 1–43. de Jong, W. W., Leunissen, J. A. M., and Wistow, G. J. (1993). Eye lens crystallins and the phylogeny of placental orders: Evidence for a macroscelid-paenungulate clade? In ‘‘Mammal Phylogeny: Placentals’’ (F. S. Szalay, M. J. Novacek, and M. C. McKenna, Eds.), pp. 5–12. Springer-Verlag, Berlin. de Jong, W. W., Zweers, A., and Goodman, M. (1981). Relationship of aardvark to elephants, hyraxes and sea cows from alpha-crystallin sequences. Nature 292, 538–540. Domning, D. P. (2001). The earliest known fully quadrupedal sirenian. Nature 413, 625–627. Domning, D., Ray, C. E., and McKenna, M. C. (1986). Two new Oligocene desmostylians and a discussion of tethytherian systematics. Smithson. Contrib. Paleobiol. 59, 1–56. Eizirik, E., Murphy, W. J., and O’Brien, S. J. (2001). Molecular dating and biogeography of the early placental mammal radiation. J. Hered. 92, 212–219. Ekdale, E. G., Archibald, J. D., and Averianov, A. O. (2004). Petrosal bones of placental mammals from the late Cretaceous of Uzbekistan. Acta Palaeontol. Pol. 49, 161–176. Fischer, M. S. (1986). Die Stellung der Schliefer (Hyracoidea) im phylogenetischen System der Eutheria. Cour. Forschingsinst. Senckenberg 84, 1–132. Fischer, M. S. (1989). Hyracoids, the sister-group of perissodactyls. In ‘‘The Evolution of Perissodactyls’’ (D. R. Prothero and R. M. Schoch, Eds.), pp. 37–56. Oxford University Press, New York. Fischer, M. S., and Tassy, P. (1993). The interrelations between Proboscidea, Sirenia, Hyracoidea, and Mesaxonia: The morphological evidence. In ‘‘Mammal Phylogeny: Placentals’’ (F. S. Szalay, M. J. Novacek, and M. C. McKenna, Eds.), pp. 217–243. Springer Verlag, New York. Flynn, J. J., Parrish, J. M., Rakotosamimanana, B., Simpson, W. F., and Wyss, A. R. (1999). A middle Jurassic mammal from Madagascar. Nature 401, 57–60. Fro¨ nicke, L., Wienberg, J., Stone, G., Adams, L., and Stanyon, R. (2003). Towards the delineation of the ancestral eutherian genome organization: Comparative genome maps of human and the African elephant (Loxodonta africana) generated by chromosome painting. Proc. R. Soc. Lond. B 270, 1331–1340. Gheerbrant, E., Sudre, J., Cappetta, H., and Bignot, G. (1998). Phosphatherium escuilliei du Thane´ tien du Bassin des Ouled Abdoun (Maroc), plus ancien proboscidien (Mammalia) d’Afrique. Geobios 30, 247–269. Gheerbrant, E., Sudre, J., Capetta, H., Iaroche`ne, M., Amaghzaz, M., and Bouya, B. (2002). A new large mammal from the Ypresian of Morocco: Evidence of surprising diversity of early proboscideans. Acta Palaeontol. Pol. 47, 493–506. Gheerbrant, E., Sudre, J., Capetta, H., Mourer-Chauvire´ , C., Bourdon, E., Iarochene, M., Amaghzaz, M., and Bouya, B. (2003). Les localite´ s a` mammife`res des carrie`res de Grand Daoui, bassin des Ouled Abdoun, Maroc, Ypre´ sien: Premier e´ tat des lieux. Bull. Soc. Geól. Fr. 174, 279–293. Gingerich, P. D., ul Haq, M., Zalmout, I. S., Khan, I. H., and Malkani, M. S. (2001). Origin of whales from early artiodactyls: Hands and feet of Eocene Protocetidae from Pakistan. Science 293, 2239–2243. Gregory, W. K. (1910). The orders of mammals. Bull. Am. Mus. Nat. Hist. 27, 1–525.

56


Hartenberger, J.-L. (1986). Hypothe`se pale´ ontologique sur l’origine des Macroscelidea (Mammalia). C. R. Acad. Sci. Paris, Ser. II 302, 247–249. Hay, W. W., DeConto, R. M., Wold, C. N., Wilson, K. M., Voigt, S., Schulz, M., Rossby Wold, A., Dullo, W.-C., Ronov, A. B., Balukhovsky, A. N., and Soding, E. (1999). Alternative global Cretaceous paleogeography. In ‘‘Evolution of the Cretaceous OceanClimate System (Geological Society of America Special Paper)’’ (E. Barrera and C. C. Johnson, Eds.), pp. 1–47. Geological Society of America. Boulder, Colorado. Hedges, S. B. (2001). Afrotheria: Plate tectonics meets genomics. Proc. Natl. Acad. Sci. USA 98, 1–2. Hillis, D. (1999). SINEs of the perfect character. Proc. Natl. Acad. Sci. USA 96, 9979–9981. Ji, Q., Luo, Z.-X., Yuan, C.-X., Wible, J. R., Zhang, J.-P., and Georgi, J. A. (2002). The earliest known eutherian mammal. Nature 416, 816–822. Kielan-Jaworowska, Z. (1975). Possible occurrence of marsupial bones in Cretaceous eutherian mammals. Nature 255, 698–699. Kielan-Jaworowska, Z., and Dashzeveg, D. (1989). Eutherian mammals from the early Cretaceous of Mongolia. Zool. Scripta 18, 347–355. Krause, D. W. (2001). Fossil molar from a Madagascan marsupial. Nature 412, 497–498. Krause, D. W., Gottfried, M. D., O’Connor, P. M., and Roberts, E. M. (2003). A Cretaceous mammal from Tanzania. Acta Palaeontol. Pol. 48, 321–330. Krause, D. W., Prasad, G. V. R., Koenigswald, W. V., Sahni, A., and Grine, F. E. (1997). Cosmopolitanism among Gondwanan Late Cretaceous mammals. Nature 390, 504–507. Lavergne, A., Douzery, E., Stichler, T., Catzeflis, F. M., and Springer, M. S. (1996). Interordinal mammalian relationships: Evidence for paenungulate monophyly is provided by complete mitochondrial 12S rRNA sequences. Mol. Phylogenet. Evol. 6, 245–258. Le Gros Clark, W. E., and Sonntag, C. F. (1926). A monograph of Orycteropus after - III. The skull. The skeleton of the trunk and limbs. General summary. Proc. R. Soc. Lond. B 30, 445–485. Lovegrove, B. G., Raman, J., and Perrin, M. R. (2001). Heterothermy in elephant shrews, Elephantulus spp. (Macroscelidea): Daily torpor or hibernation? J. Comp. Physiol. B 171, 1–10. Luo, Z.-X., Cifelli, R. L., and Kielan-Jaworowska, Z. (2001). Dual origin of tribosphenic mammals. Nature 409, 53–57. Luo, Z.-X., Ji, Q., Wible, J. R., and Yuan, C.-X. (2003). An early Cretaceous tribosphenic mammal and metatherian evolution. Science 302, 1934–1940. Luo, Z.-X., Kielan-Jaworowska, Z., and Cifelli, R. L. (2002). In quest for a phylogeny of Mesozoic mammals. Acta Palaeontol. Pol. 47, 1–78. MacPhee, R. D. E., and Novacek, M. J. (1993). Definition and relationships of Lipotyphla. In ‘‘Mammal Phylogeny: Placentals’’ (F. S. Szalay, M. J. Novacek, and M. C. McKenna, Eds.), pp. 13–31. Springer-Verlag, Berlin. Madsen, O., Scally, M., Douady, C. J., Kao, D. J., DeBry, R. W., Adkins, R., Amrine, H. M., Stanhope, M. J., de Jong, W. W., and Springer, M. S. (2001). Parallel adaptive radiations in two major clades of placental mammals. Nature 409, 610–614. Malia, M. J., Jr., Adkins, R. M., and Allard, M. W. (2002). Molecular support for Afrotheria and the polyphyly of Lipotyphla based on analyses of the growth hormone receptor gene. Mol. Phylogenet. Evol. 24, 91–101. McDowell, S. B. (1958). The Greater Antillean insectivores. Bull. Am. Mus. Nat. Hist. 115, 115–214. McKenna, M. (1975). Toward a phylogenetic classification of the Mammalia. In ‘‘Phylogeny of the Primates’’ (W. P. Luckett and F. S. Szalay, Eds.), pp. 21–46. Plenum Press, New York. Murata, Y., Nikaido, M., Sasaki, T., Cao, Y., Fukumoto, Y., Hasegawa, M., and Okada, N. (2003). Afrotherian phylogeny inferred from complete mitochondrial genomes. Mol. Phylogenet. Evol. 28, 253–260.


57

Murphy, W. J., Eizirik, E., Johnson, W. E., Zhang, Y. P., Ryder, O. A., and O’Brien, S. J. (2001a). Molecular phylogenetics and the origins of placental mammals. Nature 409, 614–618. Murphy, W. J., Eizirik, E., O’Brien, S. J., Madsen, O., Scally, M., Douady, C. J., Teeling, E., Ryder, O. A., Stanhope, M. J., de Jong, W. W., and Springer, M. S. (2001b). Resolution of the early placental mammal radiation using Bayesian phylogenetics. Science 294, 2348–2351. Nessov, L. A., Archibald, J. D., and Kielan-Jaworowska, Z. (1998a). Ungulate-like mammals from the late Cretaceous of Uzbekistan and a phylogenetic analysis of Ungulatomorpha. Bull. Carnegie Mus. Nat. Hist. 34, 40–88. Nessov, L. A., Zhegallo, V. I., and Averianov, A. O. (1998b). A new locality of Late Cretaceous snakes, mammals and other vertebrates in Africa (western Libya). Annales Pale´ ontol. 84, 265–274. Nie, W., O’Brien, C. M., Fu, B., Wang, J., Su, W., Robinson, T. J., and Yang, F. Chromosome painting between human and lorisform prosimians suggests that the HSA7/16 syntenic association is an ancestral karyotypic character for primates. Cytogenet. Genome Res. (unpublished). Nikaido, M., Nishihara, H., Hukumoto, Y., and Okada, N. (2003). Ancient SINEs from African endemic mammals. Mol. Biol. Evol. 20, 522–527. Novacek, M. J. (1986). The skull of leptictid insectivorans and the higher-level classification of eutherian mammals. Bull. Am. Mus. Nat. Hist. 183, 1–111. Novacek, M. J. (1992a). Fossils, topologies, missing data, and the higher level phylogeny of eutherian mammals. Syst. Biol. 41, 58–73. Novacek, M. J. (1992b). Mammalian phylogeny: Shaking the tree. Nature 356, 121–125. Novacek, M. J. (2001). Mammalian phylogeny: Genes and supertrees. Current Biol. 11, R573–R575. Novacek, M. J., Rougier, G. W., Wible, J. R., McKenna, M. C., Dashzeveg, D., and Horovitz, I. (1997). Epipubic bones in eutherian mammals from the Late Cretaceous of Mongolia. Nature 389, 483–486. Novacek, M. J., and Wyss, A. R. (1986). Higher-level relationships of the recent eutherian orders: Morphological evidence. Cladistics 2, 257–287. Patterson, B. (1975). The fossil aardvarks (Mammalia: Tubulidentata). Bull. Mus. Comp. Zool. 147, 185–237. Porter, C. A., Goodman, M., and Stanhope, M. J. (1996). Evidence on mammalian evolution from sequences of exon 28 of the von Willebrand factor gene. Mol. Phylogenet. Evol. 5, 89–101. Prasad, G. V. R., Jaeger, J. J., Sahni, A., Gheerbrant, E., and Khajuria, C. K. (1994). Eutherian mammals from the upper Cretaceous (Maastrichtian) intertrappean beds of Naskal, Andhra Pradesh, India. J. Vert. Paleo. 14, 260–277. Prasad, G. V. R., and Sahni, A. (1988). First Cretaceous mammal from India. Nature 332, 638–640. Prothero, D. R. (1993). Ungulate phylogeny: Molecular vs. morphological evidence. In ‘‘Mammal Phylogeny: Placentals’’ (F. S. Szalay, M. J. Novacek, and M. C. McKenna, Eds.), pp. 173–181. Springer-Verlag, Berlin. Prothero, D. R., Manning, E. M., and Fischer, M. (1988). The phylogeny of the ungulates. In ‘‘Phylogeny and Classification of the Tetrapods’’ (M. J. Benton, Ed.), pp. 201–234. Clarendon Press, Oxford. Radulesco, C., Iliesco, G., and Iliesco, M. (1976). Decouverte d’un Embrithopode nouveau (Mammalia) dans la Pale´ oge`ne de la de´ pression de Hateg (Roumanie) et conside´ ration ge´ ne´ rales sur la ge´ ologie de la re´ gion. N. Jb. Geol. Pala¨ on. Monat. 11, 690–698. Rage, J.-C., and Cappetta, H. (2002). Vertebrates from the Cenomanian, and the geological age of the Draa Ubari fauna (Libya). Annales Pale´ ontol. 88, 79–84.

58


Rana, R. S., and Wilson, G. P. (2003). New Late Cretaceous mammals from the Intertrappean beds of Rangapur, India and paleobiogeographic framework. Acta Palaeontol. Pol. 48, 331–348. Rauhut, O. W. M., Martin, T., Ortiz-Jaureguizar, E., and Puerta, P. (2002). A Jurassic mammal from South America. Nature 416, 165–168. Rich, T. H., Flannery, T. F., Trusler, P., Kool, L., van Klaveren, N. A., and Vickers-Rich, P. (2001). A second tribosphenic mammal from the Mesozoic of Australia. Rec. Queen Victoria Mus. 110, 1–9. Rich, T. H., Vickers-Rich, P., Constantine, A., Flannery, T. F., Kool, L., and van Klaveren, N. (1997). A tribosphenic mammal from the Mesozoic of Australia. Science 278, 1438–1442. Robinson, T. J., Fu, B., Ferguson-Smith, M. A., and Yang, F. (2004). Cross-species chromosome painting in the golden mole and elephant shrew: Support for the mammalian clades Afrotheria and Afroinsectiphillia but not Afroinsectivora. Proc. R. Soc. Lond. B. 271, 1477–1484. Rokas, A., and Holland, P. W. H. (2000). Rare genomic changes as a tool for phylogenetics. Trends Ecol. Evol. 15, 454–459. Rougier, G. W., Wible, J. R., and Novacek, M. J. (1998). Implications of Deltatheridium for early marsupial history. Nature 396, 459–463. Savage, R. J. G., Domning, D. P., and Thewissen, J. G. M. (1994). Fossil Sirenia of the west Atlantic and Caribbean region. V. The most primitive known sirenian, Prorastomus sirenoides Owen, 1855. J. Vert. Paleo. 14, 427–449. Scally, M., Madsen, O., Douady, C. J., de Jong, W. W., Stanhope, M. J., and Springer, M. S. (2001). Molecular evidence for the major clades of placental mammals. J. Mammal. Evol. 8, 239–277. SeiVert, E. R. (2002). The reality of afrotherian monophyly, and some of its implications for the evolution and conservation of Afro-Arabia’s endemic placental mammals. Afrotherian Conserv. 1, 3–6. SeiVert, E. R. (2003). ‘‘A Phylogenetic Analysis of Living and Extinct Afrotherian Placentals.’’ Ph.D. thesis, Duke University, Durham, NC. Shoshani, J. (1986). Mammalian phylogeny: Comparison of morphological and molecular results. Mol. Biol. Evol. 3, 222–242. Shoshani, J. (1992). The controversy continues: An overview of the evidence for HyracoideaTethytheria aYnity. Israel J. Zool. 38, 233–244. Shoshani, J. (1993). Hyracoidea-Tethytheria aYnity based on myological data. In ‘‘Mammal Phylogeny: Placentals’’ (F. S. Szalay, M. J. Novacek, and M. C. McKenna, Eds.), pp. 235–256. Springer Verlag, New York. Sigogneau-Russell, D. (1991). De´ couverte du premier Mammife`re tribosphe´ nique du Me´ sozoı¨que africain. C. R. Acad. Sci. Paris Ser. II 313, 1635–1640. Sigogneau-Russell, D., Hooker, J. J., and Ensom, P. C. (2001). The oldest tribosphenic mammal from Laurasia (Purbeck Limestone Group, Berriasian, Cretaceous, UK) and its bearing on the ‘dual origin’ of Tribosphenida. C. R. Acad. Sci. Paris Ser. IIA 333, 141–147. Simons, E. L., Holroyd, P. A., and Bown, T. M. (1991). Early Tertiary elephant shrews from Egypt and the origin of the Macroscelidea. Proc. Natl. Acad. Sci. USA 58, 9734–9737. Simpson, G. G. (1945). The principles of classification and a classification of mammals. Bull. Am. Mus. Nat. Hist. 85, 1–350. Slattery, J. C., Murphy, W. C., and O’Brien, S. J. (2000). Patterns of diversity among SINE elements isolated from three Y-chromsome genes in carnivores. Mol. Biol. Evol. 17, 825–829. Smith, A. G., Smith, D. G., and Funnell, B. M. (1994). ‘‘Atlas of Mesozoic and Cenozoic Coastlines.’’ Cambridge University Press, Cambridge, UK.


59

Springer, M. S., Cleven, G. C., Madsen, O., de Jong, W. W., Waddell, V. G., Amrine, H. M., and Stanhope, M. J. (1997). Endemic African mammals shake the phylogenetic tree. Nature 388, 61–64. Springer, M. S., Murphy, W. J., Eizirik, E., and O’Brien, S. J. (2003). Placental mammal diversification and the Cretaceous-Tertiary boundary. Proc. Natl. Acad. Sci. USA 100, 1056–1061. Springer, M. S., Teeling, E. C., Madsen, O., Stanhope, M. J., and de Jong, W. W. (2001). Integrated fossil and molecular data reconstruct bat echolocation. Proc. Natl. Acad. Sci. USA 98, 6241–6246. Stanhope, M. J., Waddell, V. G., Madsen, O., de Jong, W., Hedges, S. B., Cleven, G. C., Kao, D., and Springer, M. S. (1998). Molecular evidence for multiple origins of Insectivora and for a new order of endemic African insectivore mammals. Proc. Natl. Acad. Sci. USA 95, 9967–9972. Svartman, M., Stone, G., and Stanyon, R. (2004). A chromosome painting test of the basal Eutherian karyotype. Chromosome Res. 12, 45–53. Szalay, F. S. (1977). Phylogenetic relationships and a classification of the eutherian Mammalia. In ‘‘Major Patterns of Vertebrate Evolution’’ (M. K. Hecht, P. C. Goody, and B. M. Hecht, Eds.), pp. 375–394. Plenum Press, New York. Tabuce, R., CoiVait, B., CoiVait, P.-E., Mahboubi, M., and Jaeger, J.-J. (2001). A new genus of Macroscelidea (Mammalia) from the Eocene of Algeria: A possible origin for elephantshrews. J. Vert. Paleo. 21, 535–546. Teeling, E. C., Madsen, O., van den Bussche, R. A., de Jong, W. W., Stanhope, M. J., and Springer, M. S. (2002). Microbat paraphyly and the convergent evolution of a key innovation in Old World rhinolophoid microbats. Proc. Natl. Acad. Sci. USA 99, 1431–1436. Teeling, E. C., Scally, M., Kao, D. J., Romagnoli, M. L., Springer, M. S., and Stanhope, M. J. (2000). Molecular evidence regarding the origin of echolocation and flight in bats. Nature 403, 188–192. Telenius, H., Carter, N. P., Bebb, C. E., Nordenskjold, M., Ponder, B. A. J., and TunnacliVe, A. (1992). Degenerate oligonucleotide-primed PCR: General amplification of target DNA by a single degenerate primer. Genomics 13, 718–725. van Dijk, M. A. M., Madsen, O., Catzeflis, F., Stanhope, M. J., de Jong, W. W., and Pagel, M. (2001). Protein sequence signatures support the African clade of mammals. Proc. Natl. Acad. Sci. USA 98, 188–193. Waddell, P. J., and Shelley, S. (2003). Evaluating placental inter-ordinal phylogenies with novel sequences including RAG1, -fibrinogen, ND6, and mt-tRNA, plus MCMC-driven nucleotide, amino acid, and codon models. Mol. Phylogenet. Evol. 28, 197–224. Waddell, P. J., Kishino, H., and Ota, R. (2001). A phylogenetic foundation for comparative mammalian genomics. Genome Inform. 12, 141–154. Waddell, P. J., Okada, N., and Hasegawa, M. (1999). Towards resolving the interordinal relationships of placental mammals. Syst. Biol. 48, 1–5. ˚ (1999). The evolution of the scrotum and testicular descent in Werdelin, L., and Nilsonne, A mammals: A phylogenetic view. J. Theor. Biol. 196, 61–72. Whidden, H. P. (2002). Extrinsic snout musculature in Afrotheria and Lipotyphla. J. Mammal. Evol. 9, 161–184. Wible, J. R., Novacek, M. J., and Rougier, G. W. (2004). New data on the skull and dentition in the Mongolian late Cretaceous eutherian mammal Zalambdalestes. Bull. Am. Mus. Nat. Hist. 281, 1–144. Wible, J. R., Rougier, G. W., Novacek, M. J., and McKenna, M. C. (2001). Earliest eutherian ear region: A petrosal referred to Prokennalestes from the early Cretaceous of Mongolia. Am. Mus. Nov. 3322, 1–44.

60


Woodburne, M. O., Rich, T. H., and Springer, M. S. (2003). The evolution of tribospheny and the antiquity of mammalian clades. Mol. Phylogenet. Evol. 28, 360–385. Yang, F., Alkalaeva, E. Z., Perelman, P. L., Pardini, A. T., Harrison, W. R., O’Brian, P. C. M., Fu, B., Graphodatsky, A. S., Ferguson-Smith, M. A., and Robinson, T. J. (2003). Reciprocal chromosome painting among human, aardvark, and elephant (superorder Afrotheria) reveals the likely eutherian ancestral karyotype. Proc. Natl. Acad. Sci. USA 100, 1062–1066.

3

____________________________________________________________________________

The Role of Antisense Transcription in the Regulation of X-Inactivation Claire Rougeulle and Philip Avner Unite´ de Geńe´tique Molećulaire Murine Institut Pasteur 75015 Paris, France

I. II. III. IV. V. VI. VII. VIII.

Introduction Early Signs of Antisense Transcription within Xist Is There One Xist Antisense Transcript of Several? Tsix Is Involved in Imprinted X-Inactivation Tsix Is Involved in Neither Counting nor Silencing Is Tsix Involved in Choice? A Role for Tsix in Xist Metabolism Mechanistic Insights into Tsix Function A. At the Level of Xist Regulation B. At the Level of Chromatin Structure C. At the Level of Choice

IX. Regulation of Tsix Transcription X. Is Tsix Functional Only in the Mouse? XI. Concluding Remarks References

I. Introduction The past 10 years have seen the emergence of new and fascinating concepts emphasizing the central role of noncoding RNAs in the regulation of a wide variety of biological systems and processes, including parental imprinting in mammals, dosage compensation in Drosophila melanogaster and mammals, cosuppression in plants, and the establishment and maintenance of heterochromatin in plants and fungi. In many cases, the noncoding RNA regulates the expression of one or more genes, either noncoding or ‘‘classical’’ protein-coding genes. In certain cases, the locus encoding the noncoding RNA is physically linked to the gene it regulates and partially or totally overlaps its target, in an antisense orientation. X chromosome inactivation (XCI), the dosage compensation mechanism encountered in mammals, is dependent on noncoding RNAs that exhibit Current Topics in Developmental Biology, Vol. 63 Copyright 2004, Elsevier Inc. All rights reserved. 0070-2153/04 $35.00

61

62

Rougeulle and Avner

both of these characteristics. Xist, the key player in the X-inactivation process, generates a noncoding nuclear transcript that is crucial for the cis inactivation of the genes located on the inactive X chromosome: an X chromosome that carries a nonfunctional copy of Xist is unable to undergo X-inactivation (Marahrens et al., 1997; Penny et al., 1996). Xist RNA has the unique property of coating the chromosome from which it is expressed, forming an Xist RNA domain that can be detected by RNA-FISH (fluorescence in situ hybridization) (Clemson et al., 1996). This chromosome coating is thought to be critical to the induction of a cascade of events associated with the X-inactivation process, which include chromatin remodeling and transcriptional silencing. Xist is not the only noncoding RNA involved in XCI, as its own regulation is controlled by a second noncoding transcript antisense to Xist, known for this reason as Tsix. All eutherians have adopted XCI as their dosage compensation mechanism, although the form that the XCI takes varies from species to species. A biased form of XCI, with preferential inactivation of the paternal X chromosome, is observed in marsupials (Cooper et al., 1994) and in the trophectoderm and the extraembryonic endoderm of the mouse embryo. Imprinted XCI is believed to depend on the presence of an imprint on the maternal X, which makes it resistant to XCI, and/or an imprint on the paternal X marking it for inactivation. In the mouse, imprinted X-inactivation was initially thought to be put in place in the early blastocysts, around 3.5 days post coitum (dpc); however, data suggest that XCI is already initiated by the 4- to 8-cell stage (Huynh and Lee, 2003; Mak et al., 2004; Okamoto et al., 2004). This initial form of imprinted XCI appears to be highly labile as it is reversed, or reprogrammed, in cells of the inner cell mass, where random Xinactivation is the rule (Mak et al., 2004; Okamoto et al., 2004). The presence of unstable imprinted X-inactivation in marsupials, which diverged from eutherians approximately 150 million years ago, has led to the postulate that imprinted XCI is the initial, primitive form of XCI from which the second, random form of XCI occuring in the mouse embryonic tissues and in humans has evolved. Random XCI, through a stochastic process, results in mosaic individuals in which a more or less equal proportion of cells will have inactivated either the paternal or the maternal X chromosome. Random XCI occurs in the ex vivo system provided by diVerentiating embryonic stem (ES) cells, which are totipotent cells derived from the inner cell mass of the blastocyst. Female ES cells possess two active X chromosomes, one of which is randomly inactivated when the cells are induced to diVerentiate. ES cells are an excellent model system for the study of XCI because they appear to recapitulate most of the known steps that occur in vivo. XCI must be tightly controlled in order to avoid the inactivation of the single X chromosome in male cells and inactivation of either none or both Xs in female cells. Indeed, cells appear able to eYciently sense or count their

3. The Role of Antisense Transcription

63

number of X chromosomes relative to the number of autosomes, so that they keep active one X chromosome per diploid set of autosomes. When more than one X chromosome is present in a diploid cell, the cell must also be able to choose which chromosome to inactivate and which chromosome to keep active. Counting and choice are the very earliest steps in the initiation of XCI, which is under the control of a cis-acting region called the X-inactivation center (Xic). The notion of such a master control region came from the analysis of chromosome rearrangements involving the X chromosome (Rastan, 1983; Therman and Sarto, 1983). Although the analyses of several such rearrangements in humans and mouse (Brown et al., 1991; Cattanach et al., 1991; Leppig et al., 1993) and of large genomic X-derived transgenes inserted on autosomes (Heard et al., 1996, 1999a; Lee et al., 1996, 1999b) clearly indicate that a unique region on the X chromosome is necessary for XCI, these analyses have not allowed the precise determination of the minimal region suYcient to induce XCI (Rougeulle and Avner, 2003). The candidate Xic is currently viewed as a region of 700 kilobase (kb) in mouse and 1200-kb in humans, that contains several transcription units without apparent XCI function (Chureau et al., 2002; Rougeulle and Avner, 2003), as well as the several loci involved and cooperating in the many aspects of XCI. In addition to the crucial Xist gene and its antisense Tsix, which are discussed later, other subdomains of the Xic have been found to play roles in counting and choice. A 65-kb region 30 to Xist appears to carry a complex of elements important for both counting and choice (Clerc and Avner, 1998; Morey et al., 2004). Similarly, Xce (X controlling element), a genetically defined element believed to aVect the probability of inactivation (Cattanach and Williams, 1972), has been mapped genetically distal to Xist (Simmler et al., 1993). Xce aVects choice, insofar as diVerent Xce alleles, when present in a heterozygous configuration, confer on their chromosome of origin diVerent probabilities of being inactivated. In an Xcea/Xcec heterozygous strain, for instance, the chromosome carrying the Xcea allele has a higher probability of being inactivated than the chromosome carrying the Xcec allele. In strains homozygous for a given Xce allele, there is an equal probability of each X being inactivated. The molecular nature of the Xce element remains unknown. Even though counting and choice are clearly key aspects of the regulation of XCI, it is still unclear whether each involves discrete and separate elements, or whether the processes are interlinked and depend on complex interactions between multiple loci. Recent results tend to support this latter point of view. Although counting and possibly choice appear to be mostly independent of Xist, Xist is required in cis for the inactivation of the X chromosome. It is therefore no surprise that regulating XCI implies in part regulating Xist. The critical and tight regulation of Xist during the onset of X-inactivation has


65

been extensively studied in ES cells. In undiVerentiated ES cells, whether female or male, Xist is expressed from each X chromosome as an unstable transcript that can be visualized by RNA-FISH as a punctate signal, or pinpoint, at its site of transcription. When XCI is induced, the first known event is the stabilization of the Xist transcript from the presumptive inactive X and its downregulation on the active X. The stabilized form of Xist appears to spread out in cis from its site of transcription to cover the entire X chromosome. At least three regions for Xist transcription initiation have been identified in vivo (Fig. 1) (Johnston et al., 1998; Warshawsky et al., 1999), although the putative roles of these alternative promoters remain elusive. One of these, P0, described as responsible for the initiation of the unstable Xist transcript, seems to be an artifact whose detection was due to the presence of both antisense transcription within Xist (see later) and a ribosomal protein S12 pseudogene (Warshawsky et al., 1999). The other two Xist promoters, P1 and P2, have been touted as being responsible for the production of the stabilized form of Xist, with P2 being the major somatic promoter (Johnston et al., 1998). The unstable form of Xist expressed in undiVerentiated ES cells is now thought to originate also from P1/P2 (Warshawsky et al., 1999). In vitro characterization of the Xist P1 promoter region suggests that, while it contains a conserved consensus binding site for YY1, whose mutation has been shown in vitro to reduce the expression of a reporter gene (Hendrich et al., 1997), it does not bind TBP (TATA-binding protein), or only with low aYnity (Pillet et al., 1995). Whether either of these factors is involved in the in vivo regulation of Xist remains to be determined. Most of the data available on the in vivo regulation of Xist expression at the promoter level concern CpG methylation. In female somatic cells, the active Xist promoter on the inactive X chromosome is unmethylated, whereas the inactive Xist promoter associated with the active X chromosome is methylated (Norris et al., 1994). A protein was found that in vitro binds

Figure 1 A schematic map of the Xist/Tsix region. Xist putative promoters P0, P1, and P2 are represented by red arrows and the Xist exons by red boxes. The repeat A (rA), which is involved in the silencing function of Xist (Wutz et al., 2002), is indicated. Tsix multiple initiation sites are shown as green arrows (solid arrow for the major initiation sites, dotted arrows for the minor ones) and Tsix termination sites by ‘‘A’’. Tsix exons (green boxes) and alternative splicing are indicated in the lower part of the figure. Blue asterisks represent CpG islands, and black arrows, DNase hypersensitive sites (Ogawa and Lee, 2003). The DXPas34 locus is indicated (in blue). In the upper part of the figure, most of the mutations described in the text are represented. Mutations that will lead to skewed inactivation toward the mutated allele appear in red; mutations that lead to inactivation of the unmutated allele are in green. The double lines indicate deletions, the stop symbols the truncation of transcription through the insertion of an acceptor splice site and a poly-adenylation signal, and the large horizontal arrows the insertion of ectopic promoters.

66

Rougeulle and Avner

specifically to a CpG-rich sequence within the Xist promoter only when it is methylated, and that is able to repress the expression of a reporter gene (Huntriss et al., 1997). In contrast to what had been observed using restriction enzymes sensitive to methylation (Ariel et al., 1995; Zuccotti and Monk, 1995), bisulfite analysis has failed to reveal diVerential methylation of the Xist promoter in gametes and preimplantation stages, suggesting that DNA methylation is a consequence rather than a cause of early diVerential Xist expression (MacDonald et al., 1998). This notion is supported by findings indicating that the Xist diVerential expression is correctly established in the absence of the de novo DNA methyltransferases Dnmt3a and Dnmt3b (Sado et al., 2004). Histone H4 acetylation of the Xist promoter correlates with Xist expression in somatic cell hybrids (Gilbert and Sharp, 1999), but the function, if any, of such modification(s) in the precocious regulation of Xist expression remains to be defined. The only data currently available concern the presence, in undiVerentiated female ES cells, of moderately elevated levels of H4 acetylation in an extended region lying 50 to Xist compared to the rest of the X chromosome (O’Neill et al., 1999). Important insights into the regulation of Xist expression during the early stages of XCI were brought about by the discovery, less than 5 years ago, of an antisense transcription within the Xist locus (Debrand et al., 1999; Lee et al., 1999a) similar to that described a year before for two imprinted genes, Igf2r and Ube3A, that had been found to be associated with the paternally expressed antisense transcripts, Air and Ube3A-AS, respectively (Rougeulle et al., 1998; Wutz et al., 1997). In this chapter, after (i) a description of the nature of antisense transcription within Xist that favors (ii) the notion that there are multiple antisense transcripts rather than a single unique transcript, we discuss the role of Tsix in (iii) imprinted inactivation and in random inactivation, by dissociating its potential eVects on (iv) counting from that on (v) choice. We then discuss (vi) how the eVects of Tsix on X-inactivation are likely mediated through the regulation of Xist metabolism and (vii) the mechanistical aspects underlying this antisense regulation of Xist. Current knowledge, albeit sparse, of (viii) the regulation of Tsix transcription itself is outlined and (ix) evolutionary aspects of Tsix function presented.

II. Early Signs of Antisense Transcription within Xist The major traditional approaches used to study gene expression, namely, Northern blot hybridization and RT-PCR (reverse transcription-polymerase chain reaction) are, or were, most often performed using double-strand material (probes in the case of Northern, cDNAs for RT-PCR) that did not allow the orientation of the detected transcript to be assigned. In the case of Xist, the first evidence for the presence of an overlapping antisense


67

transcription came from the analysis of Xist expression in ES cells using strand-specific probes in RNA-FISH experiments (Lee et al., 1999a). Antisense-specific probes (i.e., probes from the Xist region that are able to specifically detect transcripts antisense to Xist) were able to detect a signal very similar to the pinpoint that had been found in male and female ES cells using double-strand probes and assigned to Xist. The existence of Xist antisense transcription was also detected by Northern analysis using riboprobes (Debrand et al., 1999) and has been confirmed using strand-specific RT-PCR (Debrand et al., 1999; Lee et al., 1999a; Mise et al., 1999) and cDNA cloning (Mise et al., 1999). Comparison of signals obtained with sense and antisense probes revealed that the antisense signal is stronger than that of the sense, suggesting a higher abundance of antisense transcripts compared to sense, a conclusion that has been confirmed by more rigorous quantitation (Shibata and Lee, 2003). Strand-specific RT-PCR analysis revealed extensive antisense transcription extending over a 40-kb span, over and well beyond the Xist gene (Debrand et al., 1999; Lee et al., 1999a; Mise et al., 1999) (Fig. 1). The major transcription initiation sites of this antisense transcription have been mapped by primer extension and 50 RACE (rapid amplification of cDNA ends) to a CpG island lying approximately 15-kb downstream of Xist (Lee et al., 1999a), associated with the DXPas34 locus. The DXPas34 locus was initially identified through its unusual CpG methylation pattern, with methylation being specifically associated with the active X chromosome and the transcriptionally silent Xist gene in both somatic tissues and diVerentiated ES cells (Avner et al., 1998; Courtier et al., 1995). On its 30 side, the antisense transcription continues up to 1.5-kb beyond the 50 end of Xist (Mise et al., 1999). Xist antisense transcription can be detected in early (3.5 and 4.5 dpc) embryos (Mise et al., 1999) but not at the 8-cell stage (Sado et al., 2001), suggesting that this transcription begins at the late morula stage. Antisense transcription is also present in undiVerentiated male and female ES cells (Debrand et al., 1999; Lee et al., 1999a; Mise et al., 1999). Upon diVerentiation of female ES cells, antisense transcription is downregulated, first on the future inactive X and then on the active X, although the exact kinetics varies according to the diVerent reports (Debrand et al., 1999; Lee et al., 1999a). Lee et al., have suggested that silencing of antisense transcription on the future inactive X precedes the formation of Xist RNA domains, whereas Debrand et al., were able to detect, albeit in a low proportion (6%) of early diVerentiating ES cells, an antisense pinpoint that was associated with an Xist RNA domain. Such diVerences could be due to the use of diVerent probes. Irrespective of whether extinction of Tsix precedes Xist upregulation, the tight correlation between antisense downregulation and Xist RNA accumulation during the onset of XCI is clearly established.

68

Rougeulle and Avner

Low levels of antisense transcription can also be detected in diVerentiated ES cells and adult tissues but appear to be more restricted in size and intensity (Debrand et al., 1999). The function of such somatic antisense transcription, if any, is unknown.

III. Is There One Xist Antisense Transcript or Several? In her initial description of Xist antisense transcription, J. Lee concluded that Tsix antisense activity was due to the existence of a unique, continuous 40-kb antisense transcript (Lee et al., 1999a). This conclusion was mainly based on strand-specific RT-PCR experiments in which PCR products were amplified using primers located up to 9-kb downstream of the primer used in the RT reaction. Although technical factors have made it diYcult to demonstrate unambiguously that Tsix is a single transcript, there are several lines of evidence, involving multiple levels of heterogeneity in the antisense transcription, that argue strongly against this hypothesis. One level of heterogeneity concerns the site of Tsix transcription initiation: Lee was able to detect, within the CpG island associated with the DXPas34 locus, two initiation sites distant from each other by some 50 base pairs (bp), although these diVerences could be due to incomplete reverse transcription during the 50 RACE procedure (Lee et al., 1999a). When a 10-kb region upstream of these major initiation sites was examined by strand-specific RTPCR, however, low levels of antisense transcription were detected specifically in undiVerentiated ES cells (Debrand et al., 1999). Moreover, weak antisense transcription persists in cells in which the major initiation sites have been deleted (Lee and Lu, 1999), confirming that there must be additional initiation sites upstream of the major ones. The existence of other clusters of minor initiation sites, mapping upstream of those described by Lee et al. (clusters A and B, Fig. 1) has since been confirmed (Ogawa and Lee, 2003; Sado et al., 2002). More recent studies suggest that additional initiation sites downstream of the major ones, lying within the DXPas34 locus itself, are also likely to exist (Shibata and Lee, 2003; Morey, in preparation). A second level of heterogeneity concerns the splicing of Tsix. The initial, extensive strand-specific RT-PCR analysis used in ES cells to characterize Xist antisense transcription strongly supported the notion of an unspliced transcript (Debrand et al., 1999; Lee et al., 1999a). Antisense transcription was detected at all genomic positions tested within the 40-kb span corresponding to Tsix. However, spliced forms of Xist antisense transcripts have since been detected in both placenta and ES cells using 50 RACE (Sado et al., 2002; Shibata and Lee, 2003) (Fig. 1). Among the exons described, only one (exon 4, Fig. 1) maps to within an exon of Xist. Interestingly, this would limit the complementarity between Xist RNA and spliced Tsix


69

antisense transcripts to a 1.9-kb region that includes the Xist promoter and the repeat A region that is involved in the silencing function of Xist (Wutz et al., 2002). The most complex patterns of alternative splicing of the Tsix antisense involve the region between the major Tsix initiation site and the DXPas34 locus (Shibata and Lee, 2003). By quantifying the ratio of spliced versus unspliced forms, Shibata and Lee showed that spliced variants account for 30 to 60% of total antisense transcripts. Intriguingly, the extent of splicing was found to vary along the length of the antisense RNA, with the most 30 exon being more eYciently spliced than the 50 exons. Such variation persisted during ES cell diVerentiation. A third level of heterogeneity concerns the termination of Tsix antisense transcription. In addition to the previously described termination sites 1.5-kb upstream of the Xist promoter (Mise et al., 1999), a termination site not far from DXPas34 was detected by 30 RACE (Shibata and Lee, 2003) (Fig. 1). This finding, which suggested that premature termination of antisense transcription may occur, was confirmed by extensive quantitation of both spliced and unspliced antisense trancripts at several positions, which revealed a gradient of Tsix abundance, with 10 times more transcripts present at 50 positions than at the 30 end (Shibata and Lee, 2003). This indicates that a significant proportion of antisense transcription must be terminated before it reaches the Xist gene. Taken together, these data strongly favor the notion of multiple antisense transcripts within the Xist region, involving diVerent promoters, alternative and incomplete splicing, and termination variants. The emerging picture of antisense function within the Xist locus has therefore clearly become very diVerent from our original ideas. For clarity, we will continue, however, to refer to all of these variant transcripts as Tsix. A question that immediately arises concerns the reason for this transcript diversity. Are all the transcripts equal in terms of function, or do they each play specific roles? To answer this question, one must first understand the general function and role of Tsix which is discussed in the following sections. The role of Tsix in X-inactivation has been investigated in several ways, initially by targeted deletions into the Tsix initiation sites and later by artificially promoting or terminating Tsix transcription.

IV. Tsix Is Involved in Imprinted X-Inactivation Because antisense genes are commonly found in imprinted regions (Reik and Walter, 2001), and because X-inactivation is imprinted in the murine preimplantation embryos, an immediate question that arose was whether Tsix could be critical to the imprinted X-inactivation process. Could Tsix, for instance, be the protective, maternal factor that is responsible for the maternal X chromosome being resistant to XCI in the early development?

70

Rougeulle and Avner

Initial evidence in favor of this hypothesis was the demonstration that expression of Tsix is imprinted in preimplantation blastocysts, albeit in a manner opposite that of Xist, with early expression of Tsix being from the maternal allele (Lee, 2000; Sado et al., 2001). In contrast to Xist, Tsix is expressed in male and female embryos, as they both carry an X chromosome of maternal origin (Lee, 2000). The imprinted expression of Tsix is strictly controlled, since deleting the maternal allele of Tsix does not result in expression of the paternal allele (Lee, 2000). Several distinct mutated Tsix alleles were generated by deleting either the major transcription start sites located within the CpG island associated with DXPas34 (3.7 [Lee, 2000]; 1.7 [Sado et al., 2001]) or the alternative promoter described by Sado et al. (2.7 [Sado et al., 2001]), located 16-kb upstream (cluster A, Fig. 1). Importantly, deletion of the latter had no eVect on X-inactivation, suggesting for the first time that the multiple Tsix transcripts might not be equivalent in terms of function. This could, however, be due in part to quantitative eVects, as the most upstream promoter is associated with only low levels of transcription. Maternal-specific expression of Tsix would suggest that transmitting a deleted Tsix allele will have an eVect only when inherited from the mother. In agreement with this, while paternal transmission of a deleted Tsix allele had no detectable eVect, very few mutant animals were born after maternal transmission. Such animals were lost as a result of postimplantation lethality, at approximately 8.5 dpc, linked to the abnormal formation of extraembryonic tissues (Lee, 2000; Sado et al., 2001). Sex ratio distortion was not observed upon maternal transmission, implying that the absence of maternal Tsix aVected both males and females equally (Lee, 2000; Sado et al., 2001). At the molecular level, maternal transmission of a deleted Tsix results in the disruption of imprinted Xist expression: a proportion of cells from both male and female mutant embryos (3.5 and 7.5 dpc) shows ectopic Xist expression from every X chromosome, as monitored by allele-specific RT-PCR and RNA-FISH (Lee, 2000; Sado et al., 2001). The equally detrimental eVect of maternal transmission of the mutant Tsix allele in both sexes is thus thought to result from ectopic inactivation of the single X in males and of both Xs in females. The inactivation status of these chromosomes, however, is yet to be investigated other than by analysis of Xist RNA domain formation. Surprisingly, 15% of mutants carrying the deletion generated by Lee (3.7) was able to reach term, which coincided with the detection of normal Xist expression in a significant proportion of mutant male and female trophoblast cells (Lee, 2000). This led Lee to suggest that imprinted X-inactivation might not be absolute, a conclusion superficially in agreement with older observations that paternal X-inactivation aVects ‘‘only’’ 85–90% of extraembryonic cells (Takagi and Sasaki, 1975; West et al., 1977). However, as underlined by Takagi in a review, the approximately 21% of cells from the extraembryonic region of the 7.5 dpc embryo having an inactive


71

maternal X may correspond to mesoderm components undergoing random XCI (Takagi, 2003). Cytogenetic analysis of early blastocysts and postimplantation embryos have highlighted the degree of inflexibility in the nonrandom choice that characterizes XCI in extraembryonic tissues (Takagi et al., 1982; reviewed in (Takagi, 2003). In agreement with the notion of strict control of imprinted X-inactivation, another deletion of the major Tsix initiation sites (1.7) resulted in a very much lower frequency (2%) of newborn mutants (Sado et al., 2001). The phenotypic diVerences between the mutations are most likely due to the nature of the targeting constructs, which, in the case of Sado et al., contained a strong splicing acceptor site and a poly-A signal that would result essentially in an absence of mature Tsix RNA. In contrast, a minor Tsix transcript originating from the upstream promoter could still be produced in the mutant generated by Lee (3.7); this might account for the substantial level of viable progeny produced after maternal transmission of the mutation. In agreement with this, weak antisense activity was detected in mutant cells when the number of PCR cycles in the RT-PCR experiments was increased (Lee and Lu, 1999). Although these data superficially appear to confirm that Tsix is the protective factor that makes the maternal X resistant to XCI, as originally suggested by Lee (2000), several observations support a more complex picture of imprinted X-inactivation regulation. Indeed, because Xist is expressed very early during preimplantation development, from the 4-cell stage onward (Kay et al., 1994) and exclusively from the paternal allele, if Tsix were the protective maternal factor, it would be expected to be expressed from the maternal allele at the equivalent 4-cell stage. The failure of Sado et al. to detect Tsix expression by RT-PCR even at the 8-cell stage (Sado et al., 2001) and of Debrand et al. to detect by RNA-FISH a maternal Tsix signal in other than a minor fraction (5%) of early morula (8 to 16 cells) (Debrand et al., 1999) argues against such a role. In addition, if Tsix were functioning as the maternal imprint, the maternally transmitted Tsix knockout might be expected to result in biallelic Xist accumulation in the early embryo. Surprisingly, such biallelic Xist accumulation has been reported only in blastocysts. We conclude that, although Tsix is clearly implicated in imprinted X inactivation, its role as the protective maternal imprint remains purely hypothetical and even unlikely.

V. Tsix Is Involved in Neither Counting nor Silencing While the role of Tsix in random XCI has been investigated both in ES cells and in vivo, its involvement in counting has mainly been deduced using ex vivo approaches. If Tsix were to aVect counting, deleting Tsix from the unique X chromosome in male ES cells or from one X in female ES cells

72

Rougeulle and Avner

might be expected to result, upon diVerentiation, in induction of inactivation in male cells and in either inactivation of both Xs or a complete absence of inactivation in female cells that would lead to massive cell death. Deleting a 3.7-kb region (3.7, Fig. 1) that includes the major Tsix promoter and the DXPas34 locus in male and female ES cells did not, however, induce aberrant X-inactivation, as monitored by RNA-FISH analysis of Xist expression (Lee and Lu, 1999). Despite transient and ectopic Xist accumulation in a small proportion (10%) of mutant male ES cells (Sado et al., 2002), no massive upregulation of Xist was observed. Mutated female cells, moreover, appropriately upregulated a single allele of Xist upon diVerentiation without massive cell death being detected. These data indicate that dosage compensation can occur in the absence of Tsix. In other words, neither the counting nor the silencing functions of the X-inactivation process appear to be critically dependant on Tsix antisense transcription.

VI. Is Tsix Involved in Choice? As mentioned previously, female ES cells that carry a deletion at the major initiation sites of Tsix transcription (3.7) are still able to inactivate correctly one of the two X chromosomes present in the cell when diVerentiation is induced. An eVect of Tsix deletion on random XCI emerged, however, when allelic analysis of the inactivation was undertaken. In contrast to normal XX ES cells, which stochastically inactivate either the paternally or the maternally inherited X chromosome, heterozyygous deletion of Tsix leads to the systematic inactivation of the chromosome carrying the deletion. Allelic RT-PCR and DNA–RNA FISH show a complete skewing of Xist upregulation toward the deleted allele, which correlates with the in cis downregulation of X-linked genes (Lee and Lu, 1999). Biased X-inactivation toward the deleted allele was also observed in vivo (Lee, 2000; Sado et al., 2001). In theory, this failure of random inactivation can be attributed to either a primary or a secondary eVect on the inactivation process. In the case of a secondary eVect, the deleted chromosome could still be chosen to remain active but would upregulate Xist due to Tsix deficiency. This would result in cells having both Xs inactive, and their death. Primary eVects would be due to a direct alteration aVecting the process of choice and would imply that the Tsix deleted chromosome is automatically and systematically chosen to be inactivated. Because neither biallelic upregulation of Xist nor excess cell death was observed during early diVerentiation, Tsix deficiency is believed to have a primary eVect on the randomness of inactivation (Lee and Lu, 1999). The question that immediately arises, however, is whether the


73

biased inactivation results from the absence of antisense transcription itself or from the removal of a critical DNA element. To get insight into this, Luikenhuis and colleagues created a truncated transcript of Tsix (2 lox) by inserting a strong acceptor splice site and a transcriptional stop signal approximately 4-kb downstream of the major Tsix transcriptional start sites. Similarly to the phenotype of the deletion that removed the major Tsix initiation sites (3.7) (Lee and Lu, 1999), the X chromosome carrying the truncated allele of Tsix was always inactivated when female ES cells heterozygous for the mutation were allowed to diVerentiate (Luikenhuis et al., 2001). This clearly indicates that an X chromosome that is unable to synthesize a full-length Tsix transcript cannot remain active once XCI is induced. However, because inactivation was only analyzed in cells diVerentiated for 6 days, a time when diVerentiation is normally well advanced, it is diYcult to conclude whether the mutation acts at the primary or secondary level. An alternative approach to clarifying the role of Tsix in primary choice relies on the creation, by the insertion of ectopic promoters, of Tsix alleles that maintain expression when diVerentiation is induced (TsixEF-1, Fig. 1). These modifications result in nonrandom inactivation, with the chromosome bearing the modified allele being systematically the active one (Luikenhuis et al., 2001; Stavropoulos et al., 2001). Persistant Tsix expression from an X chromosome therefore prevents in cis inactivation of the chromosome. Careful examination of embryoid bodies obtained after 9 to 15 days of in vitro diVerentiation of the mutated ES cells led Stavropoulos and collegues to conclude, however, that the decision process regarding choice had not been aVected by the constitutive expression of Tsix (Stavropoulos et al., 2001) and that antisense transcription itself is most likely a downstream eVector in the choice process. The sequences deleted in the Tsix knockout approach (3.7) (Lee and Lu, 1999) would be responsible for the primary eVect of Tsix deletion in choice. Further insight into the role of Tsix in the choice process can be drawn from experiments involving the creation of large Cre-Lox deletions in the region 30 to Xist including Tsix, followed by complementation leading to reestablishment of part of the originally deleted region. The deletion of a 65-kb region extending 30 to Xist exon 6 in female ES cells results in obligatory inactivation of the deleted chromosome, an event that is believed to depend on interference with primary choice (Clerc and Avner, 1998). Restoration of Tsix transcription by adding back 16-kb of sequence that includes the major Tsix transcription start sites did not, however, revert X-inactivation back to normal (Morey et al., 2001). Despite the presence of antisense transcription from both Xs, inactivation remained biased toward the rearranged chromosome. The most parsimonious explanation of these

74

Rougeulle and Avner

data is that the region added back does not contain a regulatory element important or at least suYcient for ensuring random choosing and that Tsix antisense transcription by itself is insuYcient to ensure random choice. In this context, it is interesting to note that an approximately 850-bp sequence that is deleted in the Tsix 3.7 knockout (Lee and Lu, 1999) is also absent from the complemented line (Morey et al., 2001). It is therefore likely that the choice process requires a sequential series of events that includes antisense transcription but that also necessitates other functional elements that may act upstream of Tsix. Other elements aVecting primary choice, albeit moderately, such as Xce (Cattanach and Williams, 1972) and Xite (Ogawa and Lee, 2003), have been described. Xite (X-inactivation intergenic transcription element) is a region upstream of the Tsix major initiation sites that is able to promote minor antisense transcription and that contains a series of DNA hypersensitive sites that are developmentally regulated (Fig. 1). DNase hypersensitivity of one of these sites was found to vary in diVerent mouse strains, with the degree of sensitivity correlating with the levels of Xite transcription. Deletion of this region (L, Fig. 1) has a mild eVect on choice, with the deletion slightly increasing the probability that the mutated chromosome will be inactivated (Ogawa and Lee, 2003). This eVect was thought unlikely to be mediated by the intergenic RNA in itself because its truncation by the insertion of a splice acceptor site together with polyadenylation signals (XiteSTIL, Fig. 1) failed to give rise to skewing (Ogawa and Lee, 2003). It has been proposed that Xite is the long-searched-for Xce locus (Ogawa and Lee, 2003), based on its genomic localization, the existence of sequence and DNase hypersensitivity polymorphisms between strains carrying diVerent Xce alleles, and its Xce-like eVect on choice. Because polymorphisms in other parts of the region 30 to Xist are extensive (P. Avner, unpublished results), the presence of polymorphisms per se can hardly be considered determinant. Further functional analysis, involving replacement of the Xite region from a strain carrying a given Xce allele with that of another Xce strain in order to see whether the randomness of inactivation is aVected, will be necessary to clarify this issue further.

VII. A Role for Tsix in Xist Metabolism The kinetics of Tsix expression during the course of early X-inactivation suggested an antinomic relationship between antisense transcription and Xist upregulation (Lee et al., 1999a). This notion was further supported by the phenotype of the Tsix deletions in random X-inactivation, in which the chromosome deficient for antisense transcription systematically upregulated Xist, leading to its inactivation (Lee and Lu, 1999). It therefore appears that,


75

in the course of XCI, Tsix acts as a negative regulator of Xist. This repressive activity seems to be mediated by the antisense transcription (or the transcript) itself rather than through a genomic element: truncation of Tsix by the insertion, a few kilobases downstream of the promoter, of an acceptor splice site and a polyadenylation signal resulted in the systematic Xist upregulation from the modified chromosome (Luikenhuis et al., 2001). Artificially extending Tsix expression through the use of a constitutive or inducible promoter symmetrically prevented Xist upregulation in cis (Luikenhuis et al., 2001; Stavropoulos et al., 2001). Xist upregulation therefore appears to be incompatible with Tsix expression. However, the absence of antisense transcription does not of itself necessarily result in Xist RNA accumulation. Tsix eventually becomes extinguished from the active X chromosome in both male and female cells without resulting in Xist RNA accumulation. Similarly, primary fibroblasts from mice homozygous for the deletion in the major Tsix start site (CpG) never display two Xist RNA domains, but only one (Lee, 2002). Taken together, these observations suggest the existence of additional elements capable of downregulating Xist expression on the active X chromosome during the onset of X inactivation. In addition to its role in regulating Xist expression during X-inactivation, Tsix also appears to control Xist metabolism in undiVerentiated cells, prior to the initiation of X-inactivation. Distinct and sometimes contradictory results, however, have been obtained by diVerent groups. With the 3.7 mutation, Lee and colleagues were able to detect, in undiVerentiated female ES cells, an increase in the steady state levels of Xist specifically on the mutated allele by quantitative allelic PCR. RNA-FISH did not, however, reveal the formation of an Xist RNA domain. Surprisingly, no diVerence in Xist expression was detected in mutant male ES cells compared to wild-type (WT) cells (Lee and Lu, 1999). The male mutant ES cell line generated by Sado (1.7), on the other hand, in which all antisense transcription was virtually eliminated by the insertion of an acceptor splice site and a polyadenylation signal, displayed very high levels of Xist RNA as detected by Northern analysis (Sado et al., 2001). This diVerence may, as already suggested, be due to the amount of residual antisense transcription present in the two mutants. Residual antisense transcription in the 3.7 mutant however, would have to be hypothesized to allow correction of Xist regulation in male but not female ES cells. If true, this would indicate that the regulation of Xist expression diVers in male and female cells and/or that distinct Tsix transcripts have distinct roles in both sexes. The necessary exhaustive comparative analysis of Tsix isoforms has yet to be performed in parallel in male and female cells. Truncation of Tsix expression (2 lox) also resulted in elevated Xist RNA levels in cis in female ES cells (Luikenhuis et al., 2001).

76

Rougeulle and Avner

Unfortunately, Xist expression in the undiVerentiated mutant male cells was not reported, although these cells have been generated and should be available. An additional role for Tsix in regulating the localization of Xist RNA in the nucleus was suggested by the Cre-Lox deletion/complementation approach described previously (Morey et al., 2001). While increased steady state levels of Xist RNA transcribed from the chromosome carrying a large (65-kb) deletion 30 to Xist encompassing all the Tsix initiation sites described so far were detected by quantitative allele-specific RT-PCR, the corresponding Xist RNA pinpoint detected by RNA-FISH was reduced compared to the signal from the WT allele. Instead, faint scattered dots could be seen surrounding the mutated allele in a proportion of cells. Restoration of the antisense transcription by adding back 16-kb of sequence that included the major Tsix start sites completely reverted this phenotype to normal, as judged by both RT-PCR and RNA-FISH (Morey et al., 2001). These observations strongly underline the importance of Tsix sequences in the regulation of Xist metabolism and suggest that Tsix functions not only as a repressor of steady state Xist RNA levels but also as a regulator of Xist RNA distribution within the nucleus.

VIII. Mechanistic Insights into Tsix Function A. At the Level of Xist Regulation The data discussed so far reinforce the notion of an intimate relationship between sense and antisense transcription at the Xist locus. The mode of action of Tsix, however, remains largely unresolved. For instance, it is unclear whether it is antisense transcription per se or the transcript itself that is necessary to control Xist. This question is likely to remain unanswered, as it is formally impossible to alter transcription without aVecting the transcript and vice versa in the case of a sense–antisense configuration. Whether Tsix modulates the level of Xist transcription or acts on the level of Xist RNA through its stabilization is also an open question. While applying models relevant to other cases of sense–antisense regulation to what is known about X-inactivation is an enticing approach to escaping from this dilemma, the intrinsic diVerences between antisense transcripts that are involved in genomic imprinting and Tsix must be borne in mind. Antisense transcripts involved in imprinting, such as Air and Ube3a-AS, for example, are never expressed simultaneously from the same allele as their sense counterpart (Rougeulle et al., 1998; Yamasaki et al., 2003; Zwart et al., 2001). When imprinted expression is functional, the sense transcript is expressed from one parental allele and the antisense from the other, whereas


77

when the sense transcript is expressed biallelically, as is the case for Ube3A in most cell types, antisense expression is extinguished. While such an alternative expression profile is also observed for Tsix early in development, when its expression, like that of Xist, is imprinted (Lee, 2000; Sado et al., 2001), in undiVerentiated ES cells Xist and Tsix are expressed from both X chromosomes, meaning that a given X chromosome is able to express both a sense transcript and its antisense at the same time. This raises important questions concerning the potential mechanistic impact on the transcription machinery. Can an RNA polymerase complex on one strand cross a second RNA polymerase complex on the other strand without aVecting the transcriptional elongation of either or both? One possibility is that Xist and Tsix transcription does not actually take place simultaneously, but rather alternatively. In this model, once the polymerase has reached the termination site for Tsix transcription, an initiation complex can be recruited at the Xist promoter and vice versa. This might imply a role for Tsix in regulating Xist at the level of transcriptional initiation/elongation. It should be noted that the polymerase responsible for Xist and Tsix transcription has yet to be determined with certainty, although chromatin immunoprecipitation data suggest that RNA pol II is responsible for both Xist and Tsix transcription (Navarro and Rougeulle, in preparation). Another attractive hypothesis relies on the physical interaction between sense and antisense RNAs. This could result in the degradation of both (or only the sense?) transcripts through RNA interference. However, because both Xist and Tsix transcripts can be detected in undiVerentiated ES cells, any such degradation must be incomplete. Moreover, despite extensive investigation, short interfering RNAs (siRNAs) corresponding to Xist or Tsix have yet to be identified. In another hypothesis, Tsix would act by modulating Xist RNA stability. In agreement with this hypothesis, it is known that Xist upregulation is, at least partially, mediated by stabilization of the transcript (Panning et al., 1997; Sheardown et al., 1997). Regulation of Xist stability is clearly a critical step in the inactivation process, and one in which Tsix could be involved. Confirmation may come from specific analysis of ongoing transcription through run-on experiments in Tsix mutant versus WT cells. Whatever the mode of action of Tsix, the 10- to 100-fold excess of Tsix RNA over Xist RNA is consistent with a model in which Tsix RNA could titrate out Xist RNA (Shibata and Lee, 2003). The existence of a gradient of Tsix transcripts running from its 50 end through to its 30 end, with the majority of transcripts having no homology with Xist, might allow fine-tuning of the titration, providing that the gradient is subject to changes during the course of X-inactivation—which remains to be demonstrated. An additional mode of action of Tsix that has been mentioned (Brown and Chow, 2003) concerns the blocking of the Xist silencing domains by interaction between the Xist and Tsix RNAs. It is interesting to note that the spliced forms of Tsix, whose specific function relative to

78

Rougeulle and Avner

the unspliced form is as yet unknown, carry regions of homology with the 50 region of the Xist gene that contains the repeat A, known to be involved in its silencing function. While it has been shown that the coating and silencing functions of Xist are mediated by distinct domains (Wutz et al., 2002), it is unclear, however, whether such blocking of the repeat A motif would aVect the ability of the transcript to be upregulated.

B. At the Level of Chromatin Structure One intriguing observation is the drastic eVect of the region 30 to Xist on the chromatin structure within the Xist locus. The deletion of 65-kb 30 to Xist has been shown to dramatically reduce levels of histone H3-Lys4 methylation within Xist without aVecting H3-Lys 9 methylation, as compared to WT cells. Adding back a 37-kb region that includes Tsix initiation sites and likely an element responsible for counting restores, at least partially, the H3-Lys4 methylation levels (Morey et al., 2004). Given that partial restoration of H3-Lys4 methylation levels can also be observed when 16-kb containing the Tsix major transcription sites is added back (C. Rougeulle, unpublished results), this strongly suggests a role for Tsix in regulating the chromatin structure within Xist. This eVect could be mediated by the antisense transcription in itself, as a link between H3-Lys4 methylation, and transcription elongation has been observed in other systems (Krogan et al., 2003; Ng et al., 2003). In agreement with this hypothesis, significant levels of H3-Lys4 methylation are observed over the entire region corresponding to Tsix transcription, including the region located between the 30 of Xist and the Tsix major promoter (Navarro and Rougeulle, in preparation). Definitive proof for a direct link between H3-Lys4 methylation and Tsix transcription requires the analysis of such methylation profiles in cell lines in which Tsix transcription is truncated, in the absence of genomic sequence deletion, by the addition of a strong acceptor splice site and a poly-A signal (2 lox) (Luikenhuis et al., 2001). The function of such putative Tsix-induced H3-Lys4 methylation is currently unknown, but it is of interest to note that the dramatic reduction of H3-Lys4 methylation resulting from the absence of Tsix does not seem to perturb ongoing Xist transcription.

C. At the Level of Choice An attractive model for the choice function of XCI proposes that the decision as to which chromosome will be inactivated depends on the relative abundance of Xist and Tsix transcripts (Plath et al., 2002). Several


50

79

mutations in the region to Xist that result in skewed X-inactivation favoring the mutated alleles support this hypothesis (Nesterova et al., 2003; Newall et al., 2001). In such mutants, the levels of Xist transcripts produced from the mutated allele were increased, most probably due to ectopic sense transcription initiating upstream of Xist. Intriguingly, the degree to which levels of Xist RNA are increased in undiVerentiated ES cells correlates with the extent of X-inactivation skewing seen in female animals heterozygous for the corresponding mutation. This skewing is believed to occur at the primary level. The apparent similarity between the eVect of modifying Xist levels on choice and that observed for the Tsix mutations described in the previous sections has led several investigators to propose that the X chromosome choice is modulated by the balance between sense and antisense transcripts (or transcription) at the onset of X-inactivation (Nesterova et al., 2003; Plath et al., 2002). Moreover, the observation of normal Tsix RNA levels from the allele displaying elevated Xist expression (Nesterova et al., 2003) suggests that in WT cells, the balance is controlled primarily by Tsix. Although modifying Tsix clearly has an impact on Xist RNA metabolism, this also indicates that the converse may not be true. Data relevant to these ideas can be found in the phenotype of partial complementations of the 65-kb deletion previously described. Careful examination of Tsix RNA levels revealed that the antisense restoration obtained with the add-back of the major Tsix initiation sites was only partial, with the level of antisense transcripts from the complemented allele remaining at only 50% of that of WT, most likely due to the absence of regulatory sequences further upstream (Morey et al., in preparation). The biased inactivation toward the modified allele would, in this interpretative framework, result from the skew remaining in the balance between sense and antisense transcripts in favor of Xist. Although these data support the idea that the balance between sense and antisense transcripts can influence the likelihood of a given chromosome to be inactivated, as already described, contradictions and uncertainties persist as to whether this balance aVects primary choice or is a downstream eVector of this process. Some of the mutations of Xist and of Tsix discussed have been described as having a primary eVect on choice, whereas for others the eVect is supposed to be secondary. This contradiction might be due to the criteria used by diVerent groups to evaluate the phenotype of their mutations. Conclusions regarding the primary or secondary eVect(s) of a given mutation on choice need to be judged with extreme care. The notion of ‘‘buyer beware’’ applies in a wider sense to the entire area of defining the meaning of choice in the X-inactivation process.

80

Rougeulle and Avner

IX. Regulation of Tsix Transcription Given the importance of Tsix in the initiation of X-inactivation, the regulation of Tsix transcription is likely to be critical for the correct initiation of X-inactivation. Very little is known, however, concerning the Tsix promoter function. Although several Tsix transcription initiation sites exist, the ones located 15-kb downstream of the 30 end of Xist are responsible for the majority of antisense transcription. This region strongly binds the classical RNA pol II machinery (Navarro and Rougeulle, in preparation), although more specific transcription factors have yet to be identified. Bisulfite analysis of CpG dinucleotides within the major Tsix initation sites revealed an almost complete absence of methylation in pre- and postimplantation embryos and in adult male cells (Prissette et al., 2001), even though Tsix is almost always silent in adult tissues. This observation strongly suggests that DNA methylation of the Tsix initiation sites is unlikely to control Tsix expression at the onset of inactivation. Could other modifications, such as histone modifications, be involved in Tsix regulation? The region around the Tsix initiation sites appears to be enriched for acetylated histone H3 and H4 in male ES cells but not in somatic cells, consistent with Tsix being active in the former but not in the latter (Kimura et al., 2002). The acetylation status of this region on the inactive X in female ES cells has yet to be investigated. In contrast, the major Tsix initiation site on the active X chromosome is characterized by constitutive H3-Lys4 methylation in both ES and somatic male cells (Kimura et al., 2002; Rougeulle, in preparation), suggesting that this domain is an active configuration, irrespective of the transcription state of Tsix. This is in agreement with the absence of CpG methylation from this region (Prissette et al., 2001). The DXPas34 locus, located very close to the major Tsix start sites, has long been thought to be an important element in the regulation of X-inactivation. DXPas34 shares striking similarities with diVerentially methylated regions (DMRs) frequently associated with imprinted domains. Like the direct repeats that are often found at DMRs, DXPas34 is constituted of a minisatellite, in this case a 34-mer minisatellite, and is diVerentially methylated on the active versus the inactive chromosome in somatic cells. Based on these properties and on the phenotype of DXPas34 deletions, this locus remains an excellent candidate for regulating Tsix expression. To explore the possible role of DNA methylation in DXPas34 function, the methylation status of the CpG dinucleotides contained within the locus was systematically examined by bisulfite sequencing (Prissette et al., 2001). The region was shown to be equally unmethylated in both sperm and oocytes, with the diVerential methylation of the region appearing to be a late event, taking place mostly at the postimplantation


81

stage (Prissette et al., 2001), similar to what has been found for the Xist 50 region (MacDonald et al., 1998). This argues against a role for DXPas34 DNA methylation in the regulation of imprinted Tsix expression. In the postimplantation embryo, CpG methylation mainly aVects the active X chromosome (as in the adult), compatible with a role for DNA methylation at the DXPas34 locus in random X-inactivation, although not necessarily through the control of Tsix expression. Indeed, because Tsix expression is believed to first be extinguished on the future inactive X and only then on the active X, if methylation were involved in this regulation, it should first be detected in the postimplantation embryo on the inactive X rather than on the active one. The role of the DXPas34 locus in regulating Tsix in particular and X-inactivation in general, therefore, remains to be confirmed. This is likely to require the removal of the DXPas34 locus in the absence of alteration of the major Tsix start sites. CTCF (CCCTC-binding factor), a factor that has been implicated in boundary element function in other epigenetic systems such as imprinting at the H19/Igf2 locus, where it may prevent interaction between an enhancer and the Igf2 promoter depending on the methylation status of its binding sites (Bell and Felsenfeld, 2000; Hark et al., 2000; Lewis and Murrell, 2004), has been suggested to bind at the DXPas34 locus (Chao et al., 2002). Several CTCF-binding sites are present within DXPas34 that are capable of binding CTCF in vitro and possibly in vivo, although the in vivo data appear rather weak (Chao et al., 2002). Based on their data, Lee and colleagues have suggested a role for CTCF in designating the future active X. As the binding of CTCF was only moderately impaired by CpG methylation in an in vitro assay, the authors suggested that rather than CpG, non-CpG methylation could regulate CTCF binding at DXPas34 in vivo. Despite the attractive nature of the hypothesis, at the moment, little concrete evidence supports a role for CTCF in DXPas34 function and, more largely, in X chromosome inactivation. Xite, a region located upstream of the major Tsix initation sites, is the only element for which there is clear evidence for a role in regulating Tsix expression. It contains two clusters of initiation sites (A and B, Fig. 1), whose function has been explored through classical deletion approaches (Ogawa and Lee, 2003; Sado et al., 2001). Although deleting the most upstream cluster (A, Fig. 1) had no visible eVect on X-inactivation (Ogawa and Lee, 2003; Sado et al., 2001), the inclusion in the deletion of moreproximal sequences together with the second initiation cluster (B, Fig. 1) had an impact, albeit slight, on inactivation (Ogawa and Lee, 2003). The Xite locus has been postulated to act genetically upstream of Tsix, based on reduced Tsix expression from the deleted allele compared to the WT at the onset of inactivation. On these bases, the authors postulated a role for Xite in promoting persistence of Tsix expression on the future active X.

82

Rougeulle and Avner

X. Is Tsix Functional Only in the Mouse? Opinions diVer as to the evolutionary conservation of Tsix in humans; a related issue is the question of what can be considered a homolog of Tsix. If a human Tsix homolog does exist, as expression studies suggest (Migeon et al., 2001), then not only does it show poor sequence conservation compared to mouse (Chureau et al., 2002; Migeon et al., 2001), but it also diVers from its murine counterpart in several important ways. These include an absence of a CpG island at its site of initiation, a much shorter transcript length (the transcript does not span the entire XIST gene but rather seems to end within intron 4 of XIST), and its expression from the inactive X, like the XIST gene, in several embryonic cell lines (Migeon et al., 2001, 2002). Given these dissimilarities, the first question that arises is whether the human XIST antisense transcription should be refered to as TSIX (Migeon et al., 2001, 2002). The second concerns the function, if any, of the XIST antisense transcription in humans. Migeon, for these reasons, concluded that ‘‘TSIX’’ is not functional in humans and cannot be a regulator of XIST expression or X-inactivation (Migeon, 2003; Migeon et al., 2001, 2002). In agreement with this idea, the lack of imprinted X-inactivation in human preimplantation embryos (Ray et al., 1997) could correlate with the apparent absence of imprinted TSIX expression in human placenta (Migeon et al., 2002), although the placental cells that have been studied by Migeon et al., were derived from newborns and not early embryos. Imprinted TSIX expression therefore remains to be assessed in early placenta. The lack of a functional TSIX could explain the diVerent behavior of human transgenes, as compared to mouse ones, that has been observed in murine ES cells. Whereas multicopy murine YAC transgenes upregulate Xist appropriately upon diVerentiation (Heard et al., 1999b; Migeon et al., 1999), XIST expression from multicopy human transgenes was found to be upregulated even in undiVerentiated ES cells (Heard et al., 1999b; Migeon et al., 1999). This suggests that the human XIST gene cannot be regulated like its murine counterpart in such cells, due to either the lack of a functional repressor, such as TSIX, or the incapacity of this repressor to function in a genetically heterologous context. Species-specific control mechanisms might involve the recognition, by a trans-acting regulating factor, of the murine but not the human Tsix 50 region, due to the lack of CpG enrichment in the human homolog.

XI. Concluding Remarks Tsix, a label that refers to a group of transcripts that are antisense to Xist, is clearly involved in X-inactivation, at least in the mouse. Although Tsix plays no part in the counting or silencing functions of X-inactivation, by


83

modulating the metabolism of Xist RNA before and at the onset of X-inactivation, Tsix aVects the ability of a given chromosome to be inactivated. It remains unclear, however, whether Tsix is a primary element of the choice process. A generally accepted and noncontroversial definition of Tsix function would be as a negative regulator of Xist expression. Tsix is not alone, however, in preventing Xist upregulation. Removing Tsix transcription in undiVerentiated ES cells, for instance, does not result in the formation of Xist RNA domains. Similarly, the active X chromosome remains active despite the downregulation of Tsix that occurs in the course of diVerentiation. Therefore, other elements that also control Xist expression/accumulation must exist, and they may, at least partly, be linked to the diVerentiation process. Antisense RNAs are found associated with a significant proportion (15%) of imprinted genes (Reik and Walter, 2001), and a crucial role in regulating the imprinting of imprinted domains has been demonstrated for some of them (Ube3A-AS, Air, Kcnq1ot1 [Chamberlain and Brannan, 2001; Horike et al., 2000; Wutz et al., 1997]). Because Tsix is antisense to Xist, and because its expression is imprinted early in development, it has been tempting to extrapolate the knowledge of one field to the other (Lee, 2003). This type of transposition and crossfertilization has proven helpful in understanding many biological processes. However, the specificity of each system must be carefully assessed. One surprising observation, for instance, is that all the autosomal imprinted antisense transcripts discovered so far are paternally expressed (Reik and Walter, 2001), whereas the imprinted expression of Tsix in early development is of maternal origin. A prominent feature of imprinted domains is the existence of DMRs. DMRs are CpG islands that are diVerentially methylated according to their parental origin; several observations suggest that this diVerential methylation may be directly involved in the control of imprinting. First, the diVerential methylation at DMRs is set up very early in development, often in male and female gametes, and resists the extensive waves of demethylation and remethylation that characterize the early embryo. Second, mutations of DNA methyltransferases, such as Dnmt1 and Dnmt3L, that modify the methylation status of these DMRs result in the disregulation of imprinted gene expression. Third, such disregulation is also observed upon direct deletion of the DMRs, which are therefore viewed as imprinting centers (IC). Interestingly, many antisense RNAs involved in autosomal imprinting, such as Ube3A-as, Kcnqlot1, and Air, have been found to originate at DMRs. Because deletions of the Tsix CpG island residing within the DXPas34 locus alter the imprinted expression of Tsix and Xist, this region has been postulated to be an imprinting center (Lee, 2000). However, there is, for the moment, little or no evidence to support this idea. Another way of

84

Rougeulle and Avner

thinking about Xist and Tsix imprinting is to consider these two genes as belonging to a cluster of oppositely imprinted genes on the X chromosome, similar to the numerous clusters found on the autosomes, whose imprinting would be regulated in a coordinated manner by an as-yet-unidentified IC. In this model, Tsix would not have the primary responsibility for regulating Xist imprinting. Candidate ICs for coordinately controlling Xist and Tsix imprinting might be looked for among CpG islands that lie close to the Xist/ Tsix locus or, alternatively, because ICs are known to control the expression of genes located several hundreds of kilobases away, in more remote locations. Because many autosomal ICs correspond to promoter regions of noncoding RNAs, focusing on the CpG islands associated with other noncoding RNAs lying within the Xic (Chureau et al., 2002) could prove profitable. The occurrence of antisense RNAs in the mammalian genome is not restricted to imprinted regions; in fact, they seem to be much more common than initially thought. Indeed, computational analysis of the FANTOM2 clone set, which contains 60,770 full-length enriched mouse cDNAs, revealed that as many as 8% of these cDNAs are organized into sense– antisense transcript pairs, implying that 4% of the cDNAs correspond to antisense RNAs (Kiyosawa et al., 2003). The authors made a distinction between sense–antisense transcript pairs, in which the cDNAs share sequence complementarity with each other, and the ‘‘nonantisense bidirectional transcription pair,’’ in which cDNA sequences do not overlap but derive from the same locus (as a result, for instance, of the antisense transcript being fully included in an intron of the sense). Interestingly, when the genomic distribution of sense–antisense pairs was analyzed, a bias against the X chromosome could be observed specifically for sense–antisense transcript pairs, as compared to nonantisense bidirectional pairs. This could be related to X-inactivation. Indeed, if the sense–antisense configuration is important for the allelic regulation of either or both transcripts, as in imprinting, loci encoding sense–antisense transcript pairs may have been excluded from the X chromosome during evolution (Kiyosawa et al., 2003). The occurrence of such a high number of antisense transcripts raises the question of whether they all function through identical mechanisms. Interestingly, in contrast to nonantisense bidirectional pairs, sense–antisense transcript pairs tend to have been isolated from the same cDNA library. Although this does not necessarily imply that the sense and the antisense are expressed in the same cell, it could suggest common biological pathways of regulation for sense–antisense pairs. One of these pathways could involve RNA interference, provided that the sense and antisense transcripts are expressed in the same cell. Among the sense–antisense transcript pairs, in most (73%), at least one member of the pair has no coding potential (Kiyosawa et al., 2003). This


85

indicates that antisense transcripts tend to be noncoding RNAs. They (mostly) belong to the class of long, mRNA-like noncoding RNAs, which were found, using highly stringent parameters, to represent 7% of the FANTOM2 clone set (Numata et al., 2003). It therefore becomes obvious that noncoding RNAs represent a fairly important part of the genome. The simplistic view, according to which any piece of DNA with no protein-coding information was considered as junk DNA, clearly has to be reevaluated.

References Ariel, M., Robinson, E., McCarrey, J. R., and Cedar, H. (1995). Gamete-specific methylation correlates with imprinting of the murine Xist gene. Nat. Genet. 9, 312–315. Avner, P., Prissette, M., Arnaud, D., Courtier, B., Cecchi, C., and Heard, E. (1998). Molecular correlates of the murine Xce locus. Genet. Res. 72, 217–224. Bell, A. C., and Felsenfeld, G. (2000). Methylation of a CTCF-dependent boundary controls imprinted expression of the Igf2 gene. Nature 405, 482–485. Brown, C. J., and Chow, J. C. (2003). Beyond sense: The role of antisense RNA in controlling Xist expression. Sem. Cell Dev. Biol. 14, 341–347. Brown, C. J., Lafreniere, R. G., Powers, V. E., Sebastio, G., Ballabio, A., Pettigrew, A. L., Ledbetter, D. H., Levy, E., Craig, I. W., and Willard, H. F. (1991). Localization of the X inactivation centre on the human X chromosome in Xq13. Nature 349, 82–84. Cattanach, B. M., Rasberry, C., Evans, E. P., Dandolo, L., Simmler, M. C., and Avner, P. (1991). Genetic and molecular evidence of an X-chromosome deletion spanning the tabby (Ta) and testicular feminization (Tfm) loci in the mouse. Cytogenet. Cell Genet. 56, 79–84. Cattanach, B. M., and Williams, C. E. (1972). Evidence of non-random X-chromosome activity in the mouse. Genet. Res. 19, 229–240. Chamberlain, S. J., and Brannan, C. I. (2001). The Prader-Willi syndrome imprinting center activates the paternally expressed murine Ube3a antisense transcript but represses paternal Ube3a. Genomics 73, 316–322. Chao, W., Huynh, K. D., Spencer, R. J., Davidow, L. S., and Lee, J. T. (2002). CTCF, a candidate trans-acting factor for X-inactivation choice. Science 295, 345–347. Chureau, C., Prissette, M., Bourdet, A., Barbe, V., Cattolico, L., Jones, L., Eggen, A., Avner, P., and Duret, L. (2002). Comparative sequence analysis of the X-inactivation center region in mouse, human, and bovine. Genome Res. 12, 894–908. Clemson, C. M., Mc Neil, J. A., Willard, H. F., and Lawrence, J. B. (1996). XIST RNA paints the inactive X chromosome at interphase: Evidence for a novel RNA involved in nuclear/ chromosome structure. J. Cell. Biol. 132, 259–275. Clerc, P., and Avner, P. (1998). Role of the region 30 to Xist exon 6 in the counting process of X-chromosome inactivation. Nat. Genet. 19, 249–253. Cooper, D. J., Johnston, P. G., Watson, J. M., and Graves, J. A. M. (1994). X-inactivation in marsupials and monotremes. Sem. Dev. Biol. 4, 117–128. Courtier, B., Heard, E., and Avner, P. (1995). Xce haplotypes show modified methylation in a region of the active X chromosome lying 30 to Xist. Proc. Natl. Acad. Sci. USA 92, 3531–3535. Debrand, E., Chureau, C., Arnaud, D., Avner, P., and Heard, E. (1999). Functional analysis of the DXPas34 locus, a 30 regulator of Xist expression. Mol. Cell. Biol. 19, 8513–8525.

86

Rougeulle and Avner

Gilbert, S. L., and Sharp, P. A. (1999). Promoter-specific hypoacetylation of X-inactivated genes. Proc. Natl. Acad. Sci. USA 96, 13825–13830. Hark, A. T., Schoenherr, C. J., Katz, D. J., Ingram, R. S., Levorse, J. M., and Tilghman, S. M. (2000). CTCF mediates methylation-sensitive enhancer-blocking activity at the H19/Igf2 locus. Nature 405, 486–489. Heard, E., Kress, C., Mongelard, F., Courtier, B., Rougeulle, C., Ashworth, A., Vourc’h, C., Babinet, C., and Avner, P. (1996). Transgenic mice carrying an Xist-containing YAC. Hum. Mol. Genet. 5, 441–450. Heard, E., Mongelard, F., Arnaud, D., and Avner, P. (1999a). Xist yeast artificial chromosome transgenes function as X-inactivation centers only in multicopy arrays and not as single copies. Mol. Cell. Biol. 19, 3156–3166. Heard, E., Mongelard, F., Arnaud, D., Chureau, C., Vourc’h, C., and Avner, P. (1999b). Human XIST yeast artificial chromosome transgenes show partial X inactivation center function in mouse embryonic stem cells. Proc. Natl. Acad. Sci. USA 96, 6841–6846. Hendrich, B. D., Plenge, R. M., and Willard, H. F. (1997). Identification and characterization of the human XIST gene promoter: Implications for models of X chromosome inactivation. Hum. Mol. Genet. 25, 2661–2671. Horike, S., Mitsuya, K., Meguro, M., Kotobuki, N., Kashiwagi, A., Notsu, T., Schulz, T. C., Shirayoshi, Y., and Oshimura, M. (2000). Targeted disruption of the human LIT1 locus defines a putative imprinting control element playing an essential role in BeckwithWiedemann syndrome. Hum. Mol. Genet. 9, 2075–2083. Huntriss, J., Lorenzi, R., Purewal, A., and Monk, M. (1997). A methylation-dependent DNAbinding activity recognizing the methylated promoter region of the mouse Xist gene. Biochem. Biophys. Res. Comm. 235, 730–738. Huynh, K. D., and Lee, J. T. (2003). Inheritance of a pre-inactivated paternal X chromosome in early mouse embryos. Nature 426, 857–862. Johnston, C. M., Nesterova, T. B., Formstone, E. J., Newall, A. E., Duthie, S. M., Sheardown, S. A., and BrockdorV, N. (1998). Developmentally regulated Xist promoter switch mediates initiation of X inactivation. Cell 94, 809–817. Kay, G. F., Barton, S. C., Surani, M. A., and Rastan, S. (1994). Imprinting and X chromosome counting mechanisms determine Xist expression in early mouse development. Cell 77, 639–650. Kimura, H., Tada, M., Hatano, S., Yamazaki, M., Nakatsuji, N., and Tada, T. (2002). Chromatin reprogramming of male somatic cell-derived Xist and Tsix in ES hybrid cells. Cytogenet. Genome Res. 99, 106–114. Kiyosawa, H., Yamanaka, I., Osato, N., Kondo, S., and Hayashizaki, Y. (2003). Antisense transcripts with FANTOM2 clone set and their implications for gene regulation. Genome Res. 13, 1324–1334. Krogan, N. V., Dover, J., Wood, A., Schneider, J., Heidt, J., Boateng, M. A., Dean, K., Ryan, O. W., Golshani, A., Johnston, M., Greenblatt, J. F., and Shilatifard, A. (2003). The Paf1 complex is required for histone H3 methylation by COMPASS and Dot1p: Linking transcriptional elongation to histone methylation. Mol. Cell 11, 721–729. Lee, J. T. (2000). Disruption of imprinted X inactivation by parent-of-origin eVects at Tsix. Cell 103, 17–27. Lee, J. T. (2002). Homozygous Tsix mutant mice reveal a sex ration distortion and revert to random X-inactivation. Nat. Genet. 32, 195–200. Lee, J. T. (2003). Molecular links between X-inactivation and autosomal imprinting: Xinactivation as a driving force for the evolution of imprinting? Curr. Biol. 13, R242–R254. Lee, J. T., Davidow, L. S., and Warshawsky, D. (1999a). Tsix, a gene antisense to Xist at the Xinactivation centre. Nat. Genet. 21, 400–404.


87

Lee, J. T., and Lu, N. (1999). Targeted mutagenesis of Tsix leads to nonrandom X inactivation. Cell 99, 47–57. Lee, J. T., Lu, N., and Han, Y. (1999b). Genetic analysis of the mouse X inactivation center defines an 80-kb multifunction domain. Proc. Natl. Acad. Sci. USA 96, 3836–3841. Lee, J. T., Strauss, W. M., Dausman, J. A., and Jaenisch, R. (1996). A 450 kb transgene displays properties of the mammalian X-inactivation center. Cell 86, 83–94. Leppig, K. A., Brown, C. J., Bressler, S. L., Gustashaw, K., Pagon, R. A., Willard, H. F., and Disteche, C. M. (1993). Mapping of the distal boundary of the X-inactivation center in a rearranged X chromosome from a female expressing XIST. Hum. Mol. Genet. 2, 883–887. Lewis, A., and Murrell, A. (2004). Genomic imprinting: CTCF protects the boundaries. Curr. Biol. 14, R284–R286. Luikenhuis, S., Wutz, A., and Jaenisch, R. (2001). Antisense transcription through the Xist locus mediates Tsix function in embryonic stem cells. Mol. Cell. Biol. 21, 8512–8520. MacDonald, L. E., Paterson, C. A., and Kay, G. F. (1998). Bisulfite genomic sequencingderived methylation profile of the Xist gene throughout early mouse development. Genomics 54, 379–386. Mak, W., Nesterova, T. B., de Napoles, M., Appanah, R., Yamanaka, S., Otte, A. P., and BrockdorV, N. (2004). Reactivation of the paternal X chromosome in early mouse embryos. Science 303, 666–669. Marahrens, Y., Panning, B., Dausman, J., Strauss, W., Jaenisch, R., Keohane, A. M., O’Neill, L. P., Belyaev, N. D., Lavender, J. S., and Turner, B. M. (1997). Xist-deficient mice are defective in dosage compensation but not spermatogenesis. Genes Dev. 11, 156–166. Migeon, B. R. (2003). Is Tsix repression of Xist specific to mouse? Nat. Genet. 33, 337. Migeon, B. R., Chowdhury, A. K., Dunston, J. A., and McIntosh, I. (2001). Identification of TSIX, encoding an RNA antisense to human XIST, reveals diVerences from its murine counterpart: Implications for X inactivation. Am. J. Hum. Genet. 69, 951–960. Migeon, B. R., Kazi, E., Haisley-Royster, C., Hu, J., Reeves, R., Call, L., Lawler, A., Moore, C. S., Morrison, H., and Jeppesen, P. (1999). Human X inactivation center induces random X chromosome inactivation in male transgenic mice. Genomics 59, 113–121. Migeon, B. R., Lee, C. H., Chowdhury, A. K., and Carpenter, H. (2002). Species diVerences in TSIX/Tsix reveal the roles of these genes in X-chromosome inactivation. Am. J. Hum. Genet. 71, 286–293. Mise, N., Goto, Y., Nakajima, N., and Takagi, N. (1999). Molecular cloning of antisense transcripts of the mouse Xist gene. Biochem. Biophys. Res. Comm. 258, 537–541. Morey, C., Arnaud, D., Avner, P., and Clerc, P. (2001). Tsix-mediated repression of Xist accumulation is not suYcient for normal random X inactivation. Hum. Mol. Genet. 10, 1403–1411. Morey, C., Navarro, P., Debrand, E., Avner, P., Rougeulle, C., and Clerc, P. (2004). The region 30 to Xist mediates X chromosome counting and H3 Lys-4 dimethylation within the Xist gene. EMBO J. 23, 594–604. Nesterova, T., Johnston, C. M., Appanah, R., Newall, A. E., Godwin, J., Alexiou, M., and BrockdorV, N. (2003). Skewing X chromosome choice by modulating sense transcription across the Xist locus. Genes Dev. 17, 2177–2190. Newall, A. E., Duthie, S., Formstone, E., Nesterova, T., Alexiou, M., Johnston, C., Caparros, M. L., and BrockdorV, N. (2001). Primary non-random X inactivation associated with disruption of Xist promoter regulation. Hum. Mol. Genet. 10, 581–589. Ng, H. H., Robert, F., Young, R. A., and Struhl, K. (2003). Targeted recruitment of Set1 histone methylase by elongating pol II provides a localized mark and memory of recent transcriptional activity. Mol. Cell 11, 709–719.

88

Rougeulle and Avner

Norris, D. P., Patel, D., Kay, G. F., Penny, G. D., BrockdorV, N., Sheardown, S. A., and Rastan, S. (1994). Evidence that random and imprinted Xist expression is controlled by preemptive methylation. Cell 77, 41–51. Numata, K., Kanai, A., Saito, R., Kondo, S., Adachi, J., Wilming, L. G., Hume, D. A., RIKEN GER Group, GSL Members, Hayashizaki, Y., and Tomita, M. (2003). Identification of putative noncoding RNAs among the RIKEN mouse full-length cDNA collection. Genome Res. 13, 1301–1306. Ogawa, Y., and Lee, J. T. (2003). Xite, X-inactivation intergenic transcription elements that regulate the probability of choice. Mol. Cell 11, 731–743. Okamoto, I., Otte, A. P., Allis, C. D., Reinberg, D., and Heard, E. (2004). Epigenetic dynamics of imprinted X inactivation during early mouse development. Science 303, 644–649. O’Neill, L. P., Keohane, A. M., Lavender, J. S., McCabe, V., Heard, E., Avner, P., BrockdorV, N., and Turner, B. M. (1999). A developmental switch in H4 acetylation upstream of Xist plays a role in X chromosome inactivation. EMBO J. 18, 2897–2907. Panning, B., Dausman, J., and Jaenisch, R. (1997). X chromosome inactivation is mediated by Xist RNA stabilization. Cell 90, 907–916. Penny, G. D., Kay, G. F., Sheardown, S. A., Rastan, S., BrockdorV, N., Avner, P., and Turner, B. M. (1996). Requirement for Xist in X chromosome inactivation. Nature 379, 131–137. Pillet, N., Bonny, C., and Schorderet, D. F. (1995). Characterization of the promoter of the mouse Xist gene. Proc. Natl. Acad. Sci. USA 92, 12515–12519. Plath, K., Mlynarczyk-Evans, S. K., Nusinow, D. A., and Panning, B. (2002). Xist RNA and the mechanism of X chromosome inactivation. Ann. Rev. Genet. 36, 233–278. Prissette, M., El-Maarri, O., Arnaud, D., Walter, J., and Avner, P. (2001). Methylation profiles of DXPas34 during the onset of X-inactivation. Hum. Mol. Genet. 10, 31–38. Rastan, S. (1983). Non-random X-chromosome inactivation in mouse X-autosome translocation embryos—Location of the inactivation centre. J. Embryol. Exp. Morph. 78, 1–22. Ray, P. F., Winston, R. M., and Handyside, A. H. (1997). XIST expression from the maternal X chromosome in human male preimplantation embryos at the blastocyst stage. Hum. Mol. Genet. 6, 1323–1327. Reik, W., and Walter, J. (2001). Genomic imprinting: Parental influence on the genome. Nat. Rev. Genet. 2, 21–32. Rougeulle, C., and Avner, P. (2003). Controlling X-inactivation in mammals: What does the centre hold? Sem. Cell Dev. Biol. 14, 331–340. Rougeulle, C., Cardoso, C., Fontes, M., Colleaux, L., and Lalande, M. (1998). An imprinted antisense RNA overlaps UBE3A and a second maternally expressed transcript. Nat. Genet. 19, 15–16. Sado, T., Li, E., and Sasaki, H. (2002). EVect of Tsix disruption on Xist expression in male ES cells. Cytogenet. Genome Res. 99, 115–118. Sado, T., Okano, M., Li, E., and Sasaki, H. (2004). De novo DNA methylation is dispensable for the initiation and propagation of X chromosome inactivation. Development 131, 975–982. Sado, T., Wang, Z., Sasaki, H., and Li, E. (2001). Regulation of imprinted X-chromosome inactivation in mice by Tsix. Development 128, 1275–1286. Sheardown, S. A., Duthie, S. M., Johnston, C. M., Newall, A. E., Formstone, E. J., Arkell, R. M., Nesterova, T. B., Alghisi, G. C., Rastan, S., and BrockdorV, N. (1997). Stabilization of Xist RNA mediates initiation of X chromosome inactivation. Cell 91, 99–107. Shibata, S., and Lee, J. T. (2003). Characterization and quantitation of diVerential Tsix transcripts: Implications for Tsix function. Hum. Mol. Genet. 12, 125–136. Simmler, M. C., Cattanach, B. M., Rasberry, C., Rougeulle, C., and Avner, P. (1993). Mapping the murine Xce locus with (CA)n repeats. Mamm. Genome 4, 523–530.


89

Stavropoulos, N., Lu, N., and Lee, J. T. (2001). A functional role for Tsix transcription in blocking Xist RNA accumulation but not in X-chromosome choice. Proc. Natl. Acad. Sci. USA 98, 10232–10237. Takagi, N. (2003). Imprinted X-chromosome inactivation: Enlightenment from embryos in vivo. Sem. Cell Dev. Biol. 14, 319–329. Takagi, N., and Sasaki, M. (1975). Preferential inactivation of the paternally derived X chromosome in the extraembryonic membranes of the mouse. Nature 256, 640–642. Takagi, N., Sugawara, O., and Sasaki, M. (1982). Regional and temporal changes in the pattern of X-chromosome replication during the early post-implantation development of the female mouse. Chromosoma 85, 275–286. Therman, E., and Sarto, G. E. (1983). Inactivation center on the human X chromosome. In ‘‘Cytogenetics of the Mammalian X Chromosome, Part A: Basic Mechanisms of X Chromosome Behavior’’ (A. A. Sandberg, Ed.), pp. 315–325. Liss, A. R., New York. Warshawsky, D., Stavropoulos, N., and Lee, J. T. (1999). Further examination of the Xist promoter-switch hypothesis in X inactivation: Evidence against the existence and function of a P0 promoter. Proc. Natl. Acad. Sci. USA 96, 14424–14429. West, J. D., Frels, W. I., and Chapman, V. M. (1977). Preferential expression of the Xm in the mouse yolk sac. Cell 12, 873–882. Wutz, A., Rasmussen, T. P., and Jaenisch, R. (2002). Chromosomal silencing and localization are mediated by diVerent domains of Xist RNA. Natl. Genet. 30, 167–174. Wutz, A., Smrzka, O. W., Schweifer, N., Schellander, K., Wagner, E. F., and Barlow, D. P. (1997). Imprinted expression of the Igf2r gene depends on an intronic CpG island. Nature 389, 745–749. Yamasaki, K., Joh, K., Ohta, T., Masuzaki, H., Ishimaru, T., Mukai, T., Niikawa, N., Ogawa, M., WagstaV, J., and Kishino, T. (2003). Neurons but not glial cells show reciprocal imprinting of sense and antisense transcripts of Ube3a. Hum. Mol. Genet. 12, 837–847. Zuccotti, M., and Monk, M. (1995). Methylation of the mouse Xist gene in sperm and eggs correlates with imprinted Xist expression and paternal X-inactivation. Nat. Genet. 9, 316–320. Zwart, R., Sleutels, F., Wutz, A., Schinkel, A. H., and Barlow, D. P. (2001). Bidirectional action of the Igf2r imprint control element on upstream and downstream imprinted genes. Genes Dev. 15, 2361–2366.

4

____________________________________________________________________________

The Genetics of Hiding the Corpse: Engulfment and Degradation of Apoptotic Cells in C. elegans and D. melanogaster Zheng Zhou,*,{ Paolo M. Mangahas,{ and Xiaomeng Yu* *Verna and Marrs McLean Department of Biochemistry and Molecular Biology { Program in Developmental Biology Baylor College of Medicine Houston, Texas 77030

I. Introduction II. Studies of Cell Corpse Engulfment in C. elegans A. Morphological and Physiological Studies of the Engulfment Process B. The Genetics of Cell Corpse Engulfment C. Molecular Studies of the Engulfment Genes and Identification of the Signaling Pathways D. The Engulfment of Cells That Die by Necrosis Requires the Same Genes That Act to Engulf Apoptotic Cells E. The Process of Engulfment Assists in the Execution of Apoptosis F. The Cases of Murder: Programmed Cell Deaths That Are Critically Dependent on Engulfing Cells III. The Degradation of Nuclear DNA During Programmed Cell Death in C. elegans A. nuc-1 Encodes a DNaseII Homolog Critical for DNA Degradation During Cell Death B. cps-6 Regulates Both DNA Degradation and the Timing of Cell Death C. Multiple Pathways Are Involved in DNA Degradation in C. elegans IV. Study of Engulfment and DNA Degradation in Drosophila A. Hemocytes Act as Engulfing Cells B. croquemort Encodes a Phagocytic Receptor in Drosophila C. Engulfment of Apoptotic Cells Is Required for Proper Patterning of the CNS D. Caspase-Independent Engulfment E. DNA Degradation in Drosophila Proceeds in Two Steps V. Concluding Remarks Acknowledgments References

I. Introduction During an animal’s life, a large number of unwanted cells undergo programmed cell death, or apoptosis, a genetically controlled cell suicide process. These dying cells are rapidly removed from the body: they are Current Topics in Developmental Biology, Vol. 63 Copyright 2004, Elsevier Inc. All rights reserved. 0070-2153/04 $35.00

91

92

Zhou et al.

Figure 1 A cell that undergoes apoptosis is rapidly removed through the engulfment by phagocytic cells and is degraded inside the phagocyte.

internalized by other cells through the process of phagocytosis and then are degraded inside the phagocytes (Fig. 1) (reviewed in Platt et al., 1998). Phagocytic removal is a common fate for cells undergoing apoptosis in metazoans. It has been observed and characterized in hydra, nematodes, insects including the fruit fly Drosophila melanogaster and the tobacco hornworm Manduca sexta, the zebrafish Danio rerio, the African clawed toad Xenopus laevis, the chicken Gallus gallus, and mammals (reviewed in Ellis et al., 1991b; Golstein et al., 2003; Tepass et al., 1994). The process of engulfing apoptotic cells is strikingly similar in organisms ranging from the very simple to the most complex. In diVerent organisms, diVerent types of cells act as engulfing cells. In simpler organisms such as hydra or the nematode Caenorhabditis elegans, there are no designated phagocytes; rather, neighboring cells engulf and degrade apoptotic cells (reviewed in Golstein et al., 2003 and Section IIA.2). Vertebrates develop specialized cells called macrophages that are mobile and that act as ‘‘professional phagocytes’’ for most apoptotic cells (Platt et al., 1998). In addition, a number of other cell types can act, possibly less eYciently, as ‘‘nonprofessional phagocytes’’ in vertebrates (Platt et al., 1998). The fruit fly uses blood cells, or hemocytes, as its major phagocytes for the removal of apoptotic cells (Tepass et al., 1994). After hemocytes ingest apoptotic cells and/or cell fragments, they are defined as macrophages (Tepass et al., 1994). Because it eliminates cells generated in excess, apoptosis is important for the establishment and maintenance of tissue architecture (reviewed in Jacobson et al., 1997). Apoptosis also acts as part of a quality-control mechanism by getting rid of cells that are infected, abnormal, tumorigenic, or in any way harmful to the body (Jacobson et al., 1997). Phagocytic

4. Engulfment and Degradation of Apoptotic Cells

93

removal of apoptotic cells ensures the elimination of dying cells before they release harmful cellular contents. This process actively prevents tissue injury, inflammatory response and autoimmunity, and facilitates organ sculpture and tissue remodeling (reviewed in Savill and Fadok, 2000). Furthermore, studies have shown that macrophages, after ingestion of apoptotic cells, secrete antiinflammatory cytokines and thus actively participate in the resolution of inflammation and the repair of tissue injury (Savill and Fadok, 2000). IneYcient removal of apoptotic cells has been related to a number of human inflammatory and autoimmune diseases. The studies of one autoimmune disease, systemic lupus erythematosus (SLE), both in humans and in animal models, strongly indicate that lack of phagocytosis of apoptotic cells is closely linked to the appearance of autoantibodies in the body and the development of SLE (Botto et al., 1998; Mevorach et al., 1998). Studying the phagocytic removal of apoptotic cells is thus important for understanding animal development and homeostasis and will have an important impact in medicine. Programmed cell death has been extensively studied in the nematode C. elegans, a small free-living round-worm, due to its simple anatomy, described invariant cell lineage, well-established genetics, optical transparency, and easily distinguishable apoptotic cell morphology. Genetic studies in C. elegans have identified genes essential for the control of cell death, including the first genes required for programmed cell death (reviewed in Horvitz, 2003). Studies in C. elegans and other organisms have demonstrated that the mechanisms that regulate apoptosis are conserved throughout the animal kingdom (reviewed in Metzstein et al., 1998). Similarly, genetic analyses probing the mechanisms controlling the engulfment of apoptotic cells were pioneered in C. elegans. This line of research has identified at least seven genes required for the recognition and engulfment of apoptotic cells, ordered these genes in two partially redundant genetic pathways, and led to the finding that similar genes act to control the same process in other organisms, including mammals. Another genetic system well suited for the study of programmed cell death and the fate of the dying cells is the fruit fly Drosophila melanogaster (reviewed in Song and Steller, 1999). Studies of how apoptotic cells are engulfed and degraded in Drosophila, which have begun to shed light on the molecular mechanisms underlying this process, are described in later sections. In this chapter, we review the logic and methods employed to study how apoptotic cells are recognized, internalized, and degraded by engulfing cells in C. elegans and Drosophila and describe exciting findings in these two systems resulting from a combination of genetic and molecular studies of gene function. We also discuss the implications of these results for the understanding of the clearance of apoptotic cells in mammals.

94

Zhou et al.

II. Studies of Cell Corpse Engulfment in C. elegans A. Morphological and Physiological Studies of the Engulfment Process 1. Cell Corpse Engulfment Is a Highly Efficient Process Of the 1090 somatic cells generated during the development of a C. elegans hermaphrodite, 131 undergo programmed cell death (Kimble and Hirsh, 1979; Sulston and Horvitz, 1977; Sulston et al., 1983). Of the cells destined to die, 113 die during embryogenesis, mainly during midembryogenesis, between 250 and 450 min after fertilization (Sulston et al., 1983); the remaining 18 cells are generated, and subsequently die, during larval development (Sulston and Horvitz, 1977). Cells undergoing programmed cell death can be distinguished in live animals, using Nomarski diVerential interference contrast microscopy, by their highly refractile, button-like appearance and are commonly referred to as ‘‘cell corpses’’ (Fig. 2A) (Sulston, 1976; Sulston and Horvitz, 1977; Sulston et al., 1983). Lineage studies showed that, like most other C. elegans development events, these somatic cell deaths are essentially invariant among individuals regarding the identities of the cells that die and the timing of their deaths (Sulston, 1976; Sulston and Horvitz, 1977; Sulston et al., 1983). In C. elegans, like in other metazoan organisms, cells undergoing programmed cell death proceed through four stages: (1) specification of the dying fate; (2) execution of cell death, when the killing machinery is activated inside the cells that are determined to die and multiple morphological changes specific to apoptosis are observed; (3) engulfment by neighboring cells; and (4) degradation inside engulfing cells (Fig. 1) (reviewed in Ellis et al., 1991b). Engulfment and degradation of cells undergoing programmed cell death is a swift process, so swift that very few cell corpses can be seen in wild-type animals in the late embryonic stage and thereafter (Ellis et al., 1991a). The majority of embryonic deaths occur within 30 min after the cell divisions that generate the cells destined to die (Sulston et al., 1983). Observed using Nomarski microscopy, the progression from the first increase in refractility of the dying cell to the disappearance of the dying cell takes about an hour (Sulston and Horvitz, 1977; Sulston et al., 1983). Electron microscopic studies detected pseudopods extended from engulfing cells to surround dying cells, which are highly electron dense, at an early stage of cell death (Robertson and Thomson, 1982). In some cases, engulfing cells recognize and extend pseudopods around the cells destined to die even before the mitotic divisions that generate those cells have completed (Robertson and Thomson, 1982). This phenomenon indicates that dying cells present on their surfaces at least one specific signal that distinguishes them from living cells, the so-called ‘‘eat me’’ signal, at an early stage of death.


95

Figure 2 Engulfment-defective ced mutants have persistent cell corpses. (A, C) Images of late embryos prior to hatching under Nomarski microscope. Cell corpses appear as refractile objects and are indicated by arrows. (A) In the ced-1 mutant, many cell corpses persist through embryogenesis, whereas in wild-type, most dying cells have been eliminated by the late embryonic stage. (B) Transmission electron micrographs (EMs) of cell corpses. The membranes of neighboring cells surrounding the cell corpse are indicated by arrows. Cell corpses are internalized by their neighboring cells in wild-type (a) but are left unengulfed in ced-1 mutant (b, c). (d–f ) Traces of membranes in the corresponding upper EM picture. Reproduced from Zhou et al., 2001b. (C) Some unengulfed cell corpses drift away from the embryo soma into surrounding fluid, as shown in the ced-1 mutant.

In addition to the cell deaths that occur in the soma, a large number of germ cells (estimated at between 300 and 500 germ cells per animal) undergo programmed cell death in the hermaphrodite gonad (Gumienny et al., 1999). During the adulthood of hermaphrodites, dying nuclei, which are at the pachytene stage of meiosis I and are restricted to a particular region of the germ line syncytium, quickly cellularize and are rapidly engulfed by the gonadal sheath cells (Gumienny et al., 1999). Like dying somatic cells, dying germ cells can also be distinguished, using Nomarski microscopy, by their highly refractile, disklike appearance. Due to the high eYciency of

96

Zhou et al.

engulfment and degradation, fewer than 10 germ cell corpses are found per gonad arm in the average wild-type adult hermaphrodite at any given time, despite the large number of germ cell deaths that are occurring (Gumienny et al., 1999). 2. Several Types of Cells Can Act as Engulfing Cells for Apoptotic Cells There are no ‘‘macrophage-like’’ specialized phagocytes for dying cells in C. elegans. Most cell corpses are engulfed by neighboring cells. At early stages of embryogenesis, dying cells are mostly engulfed by their sisters, which usually will go through at least one more round of cell division and thus have not become fully diVerentiated (Hoeppner et al., 2001; Sulston et al., 1983). At later stages of embryonic development and during larval development, several types of cells play major roles in engulfment. Among them are hypodermal cells, which fuse and form an outer monolayer of the animal’s body and which are the predominant engulfing cells for somatic cell corpses. Pharyngeal muscle cells and intestinal cells have also been observed to engulf cell corpses (Sulston and Horvitz, 1977; Sulston et al., 1983; Zhou et al., 2001b). Unlike the identities of the cells fated to die, the identity of the cell engulfing a particular dying cell during embryogenesis is not absolutely invariant. Hoeppner and co-workers (2001) reported that although some dying cells were observed to be repeatedly engulfed by the same engulfing cell in multiple embryos, other dying cells were observed to be engulfed by one of the several diVerent neighboring cells in diVerent embryos. Zhou and co-workers (2001b) captured transmission electron microscopy images in which two neighboring cells residing on the opposite sides of one dying cell both extend pseudopods to the same dying cell. These observations indicate that multiple neighboring cells are capable of recognizing the dying cells and initiating engulfment. What determines which cell will win as an engulfing cell remains unclear. Dying germ cells in the germ line are exclusively engulfed by gonadal sheath cells, which are part of the somatic gonad and which form a tube wrapping around the germ line (Gumienny et al., 1999; Zhou et al., 2001b). In addition to the engulfment of dying cells, gonadal sheath cells also play important roles in germ line development, oocyte maturation, and ovulation (Hubbard and Greenstein, 2000). 3. The Fate of the Engulfed Cell Corpses Electron microscopic observations showed that the entire engulfed dying cell is confined in the phagosome structure, an internalized plasma membrane structure derived from the engulfing cell, and isolated from the engulfing cell


97

cytoplasm (Fig. 2B) (Ellis et al., 1991a; Robertson and Thomson, 1982; Zhou et al., 2001b). Within an engulfing cell, the dying cell shrinks in size, a process accompanied by the breakdown of the nuclear membrane and the persistent existence of condensed chromatin-like fragments (Robertson and Thomson, 1982). The last recognizable appearance of the dying cell is one or more whorls of membranes within vacuoles of the engulfing cells (Robertson and Thomson, 1982). These observations clearly indicate that the engulfed cell corpses are degraded inside the phagosomes, although very little is known about what triggers and regulates the degradation events. 4. The Fate of Unengulfed Cell Corpses Some dying cells may disappear without engulfment. In mutant embryos in which engulfment was perturbed (see Section IIB.1), a number of cell corpses were observed to have separated from the developing embryos and drifted into the extraembryonic fluid within the eggshell (Fig. 2C) (Hedgecock et al., 1983). Some embryonic cell corpses that persist for many hours to late larval stages in engulfment-defective mutants were shown to contain vacuoles and seemed to be slowly disintegrating, suggesting that these dying cells undergo secondary necrosis (Ellis et al., 1991a). In adult hermaphrodites with mutations in engulfment genes, many unengulfed germ cell corpses also seem to undergo secondary necrosis: they swell and eventually break open (Gumienny et al., 1999). This phenomenon is similar to what has been observed for mammalian apoptotic cells that escape phagocytosis.

B. The Genetics of Cell Corpse Engulfment 1. Multiple Genetic Screens Led to the Identification of at Least Seven Engulfment Genes Genetic screens have been performed to systematically identify genes whose functions are required for the engulfment of cells undergoing programmed cell death. In those screens, chemical mutagens were used to induce mutations, and second- or third-generation progeny descended from the mutated gametes that bore a large number of persistent cell corpses in the bodies were identified using Nomarski microscopy (Ellis et al., 1991a; Hedgecock et al., 1983; Zhou et al., 2001a). The lineage and morphological studies described previously laid the groundwork for these genetic screens. Three features made it possible to identify engulfment defects this way: (1) most somatic cell deaths occur within a relatively short period in embryogenesis, (2) cell corpses are distinguishable in living animals using Nomarski microscopy, and (3) engulfment and degradation is a swift process. Additional mutants in

98

Zhou et al.

engulfment genes were recovered from screens for defects in embryogenesis, in cell migration, in the removal of necrotic cells, and for enhancers of weak defects in the execution of programmed cell death (Chung et al., 2000; Gumienny et al., 2001; Kodama et al., 2002; Nishiwaki, 1999; Reddien et al., 2001; Wu et al., 2001). Hedgecock and co-workers (1983) isolated the first eight recessive mutants containing persistent cell corpses. In the mutants, cells that undergo apoptosis during embryogenesis persist for many hours, sometimes throughout larval development. These eight mutations define two genetic loci, ced-1 and ced-2 (cell death abnormal). The fact that the cells that are destined to die still die as programmed in ced-1 and ced-2 mutants despite a lack of eYcient engulfment indicated, for the first time, that engulfment is not necessary for most programmed cell deaths (Hedgecock et al., 1983). This finding is consistent with the results from mosaic analysis and transgenic expression studies for the cell death execution genes ced-3 and ced-4, which demonstrate that the key process leading to the execution of programmed cell death takes place within the dying cell (Shaham and Horvitz, 1996; Yuan and Horvitz, 1990). Programmed cell death thus is considered a cell suicide process. Although engulfment is definitely not required for most cell deaths in C. elegans, recent discoveries indicate that engulfing cells do enhance the eYciency of cell death execution. Furthermore, in C. elegans, there are a few documented examples in which cell deaths are caused by engulfment (murder) (see later). The ced-2 mutant phenotype is subject to maternal-eVect rescue: homozygous ced-2 progeny of homozygous ced-2 mothers have persistent cell corpses, but homozygous ced-2 progeny of heterozygous ced-2/+ mothers appear wild-type for engulfment (Ellis et al., 1991a). Based on this observation, Ellis and co-workers (1991a) designed a visual screen strategy that allowed the isolation of both maternal-eVect and zygotic mutant animals with engulfment defects (Fig. 3). To obtain mutants in the absence of possible maternal-eVect rescue, Ellis and co-workers (1991a) screened F3 (3rd generation from mutagenesis) progeny instead of F2 progeny for the presence of persistent cell corpses. This screen was performed in a sem-4 (sex muscles defective) mutant background in which worms could not lay eggs (Basson and Horvitz, 1996). The eggs retained in the mothers then develop and hatch internally. F2 mothers, half of which were homozygous for a mutagenized allele at any given locus, formed transparent ‘‘bags-of-worms’’ that held a brood of F3 self-progeny consisting of embryos and young larvae suitable for screening using Nomarski microscopy and that were easy to recover from the microscopic slides. Using this strategy, 24 additional recessive mutants bearing persistent cell corpses were isolated, among which were new alleles of ced-1 and ced-2. In addition, five new genes, ced-5, ced-6, ced-7, ced-8, and ced-10, were identified from mutants recovered from this screen


99

Figure 3 The design for the two screens for mutations aVecting the removal of cell corpses in C. elegans as performed by Ellis et al. (1991a) and Zhou and Horvitz (unpublished). Three generations after sem-4 animals (P0) were mutagenized with EMS, F3 animals, which were held within each F2 mother, were examined under a Nomarski microscope for persistent cell corpses. F2 animals are generated either by þ/þ F1 mothers or by rare ced/þ F1 animals carrying a mutation in one of the ced genes. F2 generated in the latter case have genotypes of ced/ced, ced/þ, or þ/þ. The phenotypes of F3 animals that are held within these F2 mothers will depend on not only the F2 mother genotype but also whether the mutation is zygotic or subject to maternal-eVect rescue. As illustrated in the table, if the mutation is subject to maternal-eVect rescue, the ced/ced F2 animals will generate 100% Ced F3 animals, whereas ced/þ F2 animals will generate 0% Ced F3 animals. If the mutation is zygotic, both ced/ced and ced/þ F2 will generate Ced F3, in proportions of 100% and 25%, respectively.

(Ellis et al., 1991a). Of these genes, ced-8 was later found to be a gene controlling the timing of cell death execution rather than regulating engulfment (Stanfield and Horvitz, 2000). ced-5, ced-6, ced-7, and ced-10 represent new genes genuinely required for eYcient engulfment (see later discussion).

100

Zhou et al.

Maternal eVect rescue appears to be the rule rather than the exception among the engulfment genes. Loss-of-function mutations of six out of seven engulfment genes (including ced-12, an engulfment gene identified later) are subject to maternal eVect rescue for the engulfment of somatic cell corpses generated during embryonic and larval development (Table I) (Ellis et al., 1991a). The only zygotically required gene is ced-1 (Table I) (Ellis et al., 1991a). This phenomenon demonstrates that the maternally provided wild-type products of these six genes must be suYcient for the engulfment of dying cells. The screen performed by Ellis and co-workers (1991a) did not reach saturation. Zhou and Horvitz later performed an additional screen for engulfment

Table I Genes Functioning in the Engulfment of Cell Corpses in C. elegans

Gene

Mammalian Homologs

ced-1

1. SREC 2. CD91/LRP 3. mEGF10

ced-2 ced-5

Crk-II DOCK180

ced-6 ced-7

hCED-6/hGULP ABC transporters

ced-10 ced-12

Rac1 1. ELMO1 2. ELMO2/hCED-12A 3. Human BAB14712

a

Hedgecock et al., 1983. Zhou et al., 2001b. c Ellis et al., 1991a. d Reddien and Horvitz, 2000. e Wu and Horvitz, 1998b. f Liu and Hengartner, 1998. g Wu and Horvitz, 1998a. h Lundquist et al., 2001. i Chung et al., 2000. j Gumienny et al., 2001. k Zhou et al., 2001a. l Wu et al., 2001. b

Mutant Alleles e1735 a,b, e1754 a,b, e1797 a,b, e1798 a,b, e1799 a,b, e1800b, e1801 a,b, e1814 a,b, n691 b,n1951 b, n1995 b,c, n2000 b,c, n2089 b,c, n2091 b,c, n2092 b,c e1752 a,d, n1994 c,d, n3238 d mu57 e, n1812 c,e, n2002 c,e, n2098 e, n2099 e, n2691 e n1813 c, f, n2095 c, f n1892 c,g, n1996 c,g, n1997 c,g, n1998 c,g, n2001 c,g, n2094 c,g, n2690 g, n3072 g, n3073 g n1993 c,d, n3246 d, n3417 h bz187 i,j, k145 j, k149 j, k156 j, k158 j, n3261 k, oz167 i, j, tp2 l

Subject to Maternal-EVect Rescue? NO

Yes Yes Yes Yes

Yes Yes


101

mutants using similar strategy but at a larger scale (Zhou and Horvitz, unpublished results). In addition to new alleles of ced-1, -2, -5, -6, -7, and -10, a new engulfment gene, ced-12, was identified from this screen (Zhou et al., 2001a). ced-12 was independently identified by other groups in screens seeking defects other than the failure to remove cells dying by programmed cell death (Chung et al., 2000; Nishiwaki, 1999; Wu et al., 2001). Mutants strongly defective in the engulfment of cell corpses, including null mutants of several of the engulfment genes, are fully viable, suggesting that engulfment per se is not essential for C. elegans development (Wu and Horvitz, 1998a, Zhou et al., 2001b). However, in the screens performed by Zhou and Horvitz, a new phenotypic class of mutants that displays both persistent cell corpses and embryonic lethality in late-stage embryos has been isolated (Zhou and Horvitz, unpublished results). Genetic analyses have shown that both phenotypes result from a single gene mutation. Because mutants strongly defective in engulfment are fully viable, the existence of these mutations that cause both engulfment defects and embryonic lethality suggests that the corresponding genes control processes other than engulfment and are required for viability. Thus, these mutations are likely to define a new class of genes missed in previous screens that focused on viable engulfment mutants. 2. Loss-of-Function Mutants of the Engulfment Genes Are Truly Defective in Cell Corpse Engulfment The presence of persistent cell corpses in the engulfment ced mutants listed previously could result from a lack of engulfment or from perturbation in the degradation of the engulfed cell corpses. To distinguish between these two possibilities, transmission electron microscopy was used to examine cell corpses in these mutants. In animals bearing representative mutant alleles of each of the seven engulfment genes, the observed cell corpses remained unengulfed (for example, see Fig. 2B [b, c, e, f]) (Ellis et al., 1991a; Hedgecock et al., 1983; Zhou et al., 2001a,b). In contrast, the corpses found in wild-type animals were engulfed (for example, see Fig. 2B [a, d]) (Ellis et al., 1991a; Zhou et al., 2001b). These results indicate that the seven ced genes control the process of engulfment rather than a later step of degradation. 3. Two Partially Redundant Genetic Pathways Contribute to Efficient Engulfment Ellis and co-workers (1991a) and others observed that none of the single engulfment mutants completely abolished the engulfment of cell corpses. For example, in the head region of a wild-type hermaphrodite, 93 cells undergo programmed cell death. Of these, 92 die during embryogenesis and one dies during the early L1 larval stage (Sulston et al., 1983). However,

102

Zhou et al.

in ced-1(e1735) or ced-5(n1812) mutant animals, which represent null mutants of each corresponding gene, an average of 27.4 and 33.3 corpses, respectively, were observed in the heads of L1 larvae (Zhou et al., 2001a). These persistent cell corpses represent less than 36% of the total number of dying cells in this region. This phenomenon suggests that many cell corpses have probably been engulfed in addition to those that have been shed into the extraembryonic fluid and raises the possibility that the known engulfment genes might have mutually redundant functions. To test this idea, Ellis and co-workers (1991a) generated double mutant combinations among the engulfment mutants. By scoring the number of persistent cell corpses in the pharynges of the double mutant L1 larvae, they identified two functional groups of genes, the ced-1, -6, and -7 group and the ced-2, -5, and -10 group (Fig. 4). Double mutants built within each group contain no more persistent cell corpses than the strongest single mutant in the group. However, double mutants built between the two groups contain many more persistent cell corpses than seen in any single mutant in either group (Ellis et al., 1991a). For example, the ced-1(e1735); ced-5(n1812) double mutant L1 larvae have 44 persistent cell corpses in the head. Furthermore, none of the triple mutants built among the engulfment mutants display a stronger phenotype than that seen for the strongest double mutants. Thus, none of the six genes seems to act in a third pathway (Ellis et al., 1991a). ced-12 was later placed in the ced-2, -5, -10 group by similar analyses (Gumienny et al., 2001; Wu et al., 2001; Zhou et al., 2001a). These two groups do not seem to act on distinct groups of dying cells. Mutations in all seven genes aVect the eYciency of the engulfment of somatic cell corpses generated in embryos and in larvae and of germ-cell corpses generated in the adult hermaphrodite gonad (Ellis et al., 1991a; Gumienny et al., 1999, 2001; Wu et al., 2001; Zhou et al., 2001a,b). However, in some cases it was observed that mutations in diVerent engulfment genes appear to have diVerent relative eVects on diVerent cell corpses (Ellis et al., 1991a; Gumienny et al., 1999). Further molecular and biochemical analyses revealed that these two distinct functional groups represent two

Figure 4 The known components of the two parallel and partially redundant genetic pathways that control the engulfment of cell corpses in C. elegans.


103

partially redundant signaling pathways that control the initiation and execution of engulfment. Still, none of the double mutants retain all 93 cell corpses in the head. There are three probable explanations: the shedding of unengulfed cell corpses from the developing embryo (see earlier discussion), the degeneration and disappearance of unengulfed cell corpses (see earlier discussion), and possibly a third, unidentified pathway for the engulfment of cell corpses (Ellis et al., 1991a). C. Molecular Studies of the Engulfment Genes and Identification of the Signaling Pathways 1. How Are Dying Cells Recognized by Engulfing Cells? The Story of CED-1 and Its Partners a. CED-1 Is a Phagocytic Receptor. It was proposed that cells undergoing programmed cell death expose or release special substances on their outer surfaces to activate neighboring cells to engulf them through a receptor-mediated signaling pathway (Ellis et al., 1991a). Prior to the cloning of CED-1, no such receptor had been identified in C. elegans. However, a number of mammalian cell surface proteins have been implicated, mainly by in vitro studies, in the recognition of apoptotic cells and the promotion of their phagocytosis. These include members of the integrin family, members of the scavenger receptor family, lectins, MER (novel human kinase expressed in monocytes, epithelial, and reproductive tissues), CD14, and PSR (phosphatidylserine receptor) (reviewed in Henson et al., 2001). At the time of the identification of CED-1, however, the in vivo contribution of any of these proteins to the clearance of apoptotic cells was not clear. Zhou and co-workers (2001b) cloned C. elegans ced-1 using standard positional cloning techniques and found that ced-1 encodes a single-pass transmembrane protein with a long N-terminal domain predicted to be an extracellular domain based on protein topology (Fig. 5). The predicted extracellular domain of CED-1 contains an N-terminal signal peptide and 16 tandem copies of an atypical form of EGF-like repeat. An EGF-like repeat is a cysteine-rich motif found in the extracellular domains of many proteins functioning in adhesive or ligand-receptor interactions (reviewed in Campbell and Bork, 1993). N-terminal to the EGF-like repeats there is an EMI domain, another cysteine-rich domain found in many extracellular proteins including emilins and multimerin (Callebaut et al., 2003). The C-terminal, presumable intracellular domain contains two small motifs known for their functions in signal transduction pathways: putative binding motifs (NPXY and YXXL) for a PTB (phosphotyrosine-binding) domain and an SH2 (Src homology 2) domain, respectively (Songyang and Cantley,

Figure 5 The domain structures of the seven proteins involved in the engulfment of cell corpses in C. elegans (Brugnera et al., 2002; Callebaut et al., 2003; Reddien et al., 2000; Wu et al., 1998a,b; Zhou et al., 2001a,b).


105

1995). The overall structure and sequence features of CED-1 suggest that it could act as a transmembrane receptor. The extracellular domain of CED-1, CED-1Ex, is similar in sequence to a number of proteins in the database that contain the CED-1 type of atypical EGF-like repeats. These include proteins with unknown functions and a protein called SREC (scavenger receptor from endothelial cells) (Adachi et al., 1997). The similarity between CED-1Ex and SREC is particularly interesting, as scavenger receptors mediate the endocytosis of a variety of anionic substrates, including lipoproteins and phospholipids (Krieger and Herz, 1994). Several scavenger receptors, including SR-A, CLA-1, CD36, SR-BI, and CD68, have been implicated in mediating the phagocytosis of apoptotic cells in mammalian cell culture studies (Platt et al., 1998). CED-1Ex may possess ligand-binding specificity scavenger receptors. Several lines of evidence indicate that CED-1 recognizes apoptotic cells in vivo (Zhou et al., 2001b). By following the expression pattern of a CED1::GFP (green fluorescence protein) fusion protein produced under the control of the ced-1 promoter, CED-1 was found to be expressed at high levels in cell types that can function as engulfing cells, such as hypodermal cells, intestinal cells, and gonadal sheath cells, but is not expressed in cells destined to die (Zhou et al., 2001b). Consistent with its expression pattern, CED-1’s function is required only in engulfing cells, but not in dying cells, for the engulfment of dying cells. Furthermore, the CED-1::GFP fusion protein is present on cell surfaces, and the CED-1::GFP on the surface of an engulfing cell accumulates at a higher concentration within the region of the plasma membrane that is contacting the dying cell, the so-called phagocytic cup (Zhou et al., 2001b). In some engulfment-mutant (e.g., ced-6 mutant) larvae, pseudopodia from engulfing cells were observed to extend around dying cells but did not fully enclose. In those mutants, CED-1::GFP was observed to form a partial green circle around the dying cells (Fig. 6A) (Zhou et al., 2001b). These observations indicate that CED-1 clusters in response to dying cells, which presumably expose extracellular signals, and that the clustering of CED-1 is not a consequence of the completion of engulfment but could actually precede the initiation of engulfment. Deletion analysis of CED-1 showed that a truncated form of CED-1 missing its intracellular domain could still localize to the plasma membrane of engulfing cells and cluster around dying cells. However, its engulfment activity, which could be scored as the ability to rescue the engulfment defects of a ced-1 null mutant, was lost (Zhou et al., 2001b). Thus, the intracellular domain of CED-1 is not needed for CED-1 to recognize dying cells; rather, it is needed for a downstream step, probably the activation of engulfing cells. This observation reinforces the hypothesis that CED-1 acts as a receptor; CED-1 not only receives the signal, but also relays the signal.

106

Zhou et al.

Figure 6 The signaling pathway led by CED-1, CED-6, and CED-7 promotes engulfing cells to recognize apoptotic cells. (A) CED-1::GFP clusters around cell corpse in ced-6 but not ced-7 mutant animals. ced-6(n2095) (upper panels) and ced-7(n1996) (lower panels) mutant larvae were induced to express Phsp ced-1::gfp. CED-1::GFP green circles around cell corpses are indicated with white arrows. Cell corpses are indicated with black arrows. Reproduced from Zhou et al., 2001b. (B) The current model elucidating the signaling pathway led by CED-1, CED-6, and CED-7. See text for details.

The clustering of CED-1 could be a consequence of a CED-1–ligand interaction and may play an important role in activating downstream signaling. Receptor clustering in response to extracellular signals has been observed for many transmembrane receptors. For example, Fc receptor (Fc R), the mammalian phagocytic receptor for opsonized foreign particles, clusters around opsonized particles and triggers downstream phosphorylation events that lead to polarized cell surface extension (reviewed in Kwiatkowska and Sobota, 1999). Overexpression of a truncated form of CED-1 that retains only its extracellular and transmembrane domains in wild-type C. elegans embryos results in the presence of a few persistent cell corpses, despite the fact that this truncated protein retains the ability to


107

cluster around cell corpses (Zhou et al., 2001b). This weak dominantnegative eVect suggests that this truncated form of CED-1 may act to compete for extracellular ligand(s) and/or to disrupt the assembly or function of a putative multisubunit protein complex organized by wildtype CED-1. These results are consistent with a model in which clustering of CED-1 in response to ligand(s) is important for its signaling function. b. Tyrosine Phosphorylation Events and Adaptor Molecules May Participate in CED-1 Signaling. How does CED-1 initiate downstream signaling inside engulfing cells? Mutational analyses have shown that the two putative adaptor-binding motifs (NPLY and YASL) located within the intracellular domain of CED-1 are important for the engulfment activity of CED-1, and that they are partially redundant with each other for CED-1 function (Zhou et al., 2001b). Furthermore, mutating the tyrosine in the SH2-binding motif (YASL) to phenylalanine, a nonphosphorylatable residue structurally similar to tyrosine, completely abolishes the function of this motif in regards to engulfment activity (Z. Zhou, unpublished observations), suggesting that phosphorylation of this tyrosine is essential for function. This is consistent with previous observations that SH2 domains bind to the YXXL motif in a tyrosine phosphorylation-dependent manner (Songyang and Cantley, 1995). Mutating the tyrosine residue in the PTB-binding motif (NPLY) to phenylalanine only weakly decreases the function of this motif, although its mutation to alanine strongly aVects CED-1 function, indicating that tyrosine phosphorylation at this site is not essential (Z. Zhou, unpublished observations). This is also consistent with the knowledge that the recognition by certain PTB domains of their binding motifs is not dependent on tyrosine phosphorylation (Pawson and Scott, 1997). The identities of the putative tyrosine kinase(s) that phosphorylate the YASL motif and the SH2-containing adaptor(s) that bind to this motif remain unknown. However, CED-6, one of the two other engulfment proteins acting in the CED-1 functional group, stands out as a candidate adaptor for CED-1 that may bind to the NPLY motif with its PTB domain. ced-6 was cloned by Liu and Hengartner (1998). Its protein structure is composed of an N-terminal PTB domain with similarity to that of Shc, Numb, and Disabled, followed by a leucine zipper domain and a prolinerich C-terminal region (Fig. 5) (Liu and Hengartner, 1998; Su et al., 2000). Like the function of ced-1, the function of ced-6 is required only in engulfing but not in dying cells, based on genetic mosaic analysis (Liu and Hengartner, 1998). In addition, the CED-1::GFP clustering around persistent cell corpses is not aVected in ced-6 mutant embryo or larvae (Fig. 6A), indicating that mutations in ced-6 do not block CED-1’s recognition of the ‘‘eat me’’ signal; rather, they block a signaling pathway downstream of CED-1 (Fig. 6B) (Zhou et al., 2001b). Furthermore, Su and co-workers (2002) reported that

108

Zhou et al.

a GST-CED-1C (C-terminal intracellular domain) fusion protein could pull down the PTB domain of CED-6 when both were expressed in COS-7 cells, indicating a possible interaction between CED-1C and CED-6-PTB. This interaction was reproduced in a Sos-based yeast two-hybrid protein interaction assay and was shown to require the NPLY motif of CED-1C (Su et al., 2002). This interaction, as detected in heterologous systems, suggests that in engulfing cells in C. elegans CED-6 might be recruited to bind to the intracellular domain of CED-1. Because the interaction between CED-1C and full-length CED-6 has not been detected in vitro or in vivo, it is unclear if CED-6 might associate with CED-1 constitutively or might be recruited to CED-1 only when CED-1 is associated with its extracellular ligand(s). What are the events regulated by the CED-1 signaling pathway? Multiple events need to be initiated and coordinated inside engulfing cells for engulfment to occur. These events include the reorganization of the actin cytoskeleton, which is the driving force for cell surface extension, membrane remodeling and extension, membrane closure, and the initiation of cell corpse degradation inside the phagosomes. The CED-1 pathway could control one or several of these events. Identifying downstream targets of this pathway would help resolve this question. c. What Substance(s) Do Apoptotic Cells Expose on Their Outer Surfaces and What Does CED-1 Recognize? A number of changes have been detected on the surfaces of mammalian cells when they undergo apoptosis. One such change is the exposure of phosphatidylserine (PS), a phospholipid normally kept in the inner leaflet of the lipid bilayer in living cells (reviewed in Henson et al., 2001). This event has been observed in Drosophila, X. laevis, the chicken G. gallus, and mammals and thus represents an evolutionarily conserved feature of apoptosis (Nera et al., 2000; Schlegel and Williamson, 2001; van den Eijnde et al., 1998). Changes in cell surface carbohydrates and ionic charge have also been observed (reviewed in Savill et al., 1993). These changes may result in distinct features of apoptotic cells recognizable by phagocytes. Indeed, PS exposure on the outer surface of apoptotic mammalian cells has been implicated as a predominant ‘‘eat me’’ signal recognizable by phagocytic cells (Henson et al., 2001). In C. elegans, the events of programmed cell death and engulfment have thus far been examined only in living animals covered with either cuticles or chitin-based egg shells. Consequently, very little is known about the surfaces of apoptotic cells in C. elegans. The identity of CED-1 has provided a hint. As previously proposed, CED-1 may possess a ligand-binding specificity similar to that of scavenger receptors and, in particular, may be able to bind anionic phospholipids such as PS (Zhou et al., 2001b). The proof of this hypothesis and the identification of the C. elegans ‘‘eat me’’ signal thus await the identification of CED-1 ligand(s).


109

The studies of ced-7 have provided supportive evidence that the ‘‘eat me’’ signal detected by CED-1 is likely to be PS. In ced-7 mutant animals, although CED-1::GFP is expressed and is localized to cell surfaces as in wild-type animals, GFP-positive cell corpses are nearly absent (Fig. 6A) (Zhou et al., 2001b). This result indicates that the function of CED-7 is required for CED-1 to cluster around cell corpses. In contrast, although engulfment is defective in ced-2, -5, -6, -10, and -12 mutants, CED-1::GFP still clusters around cell corpses in these mutants, indicating the block in CED-1 clustering is a defect specific to ced-7 mutants and not a result of failure to engulf cell corpses (Zhou et al., 2001b). It is unclear how CED-7 acts to promote CED-1 clustering around dying cells. ced-7 was cloned and characterized by Wu and Horvitz (1998a) and was shown to encode a member of the family of ABC (ATP-binding cassette) transporters (Fig. 5). ABC transporters have been known to actively transport a variety of substances, including sugars, ions, lipids, peptides, proteins, and lipoproteins, into and out of cells (reviewed in Klein et al., 1999). Some members of the ABC transporter family have been shown to influence the distribution of lipid species across the membrane bilayer. Of potential relevance to CED-7 function is the ability of ABC1, the closest mammalian homolog of CED-7, to promote the exposure of PS onto the outer layer of plasma membrane (Hamon et al., 2000). It is thus possible that CED-7 promotes the exposure of PS or some other ‘‘eat me’’ signal on the surface of apoptotic cells for CED-1 to recognize. CED-7 is broadly expressed in C. elegans embryos and is localized to cell surfaces (Wu and Horvitz, 1998a). Genetic mosaic analysis indicates that the function of CED-7 is needed in both dying and engulfing cells for engulfment (Wu and Horvitz, 1998a). This model predicts CED-7 functions in dying cells but is not suYcient to explain its function in engulfing cells. CED-7 may perform multiple functions: in dying cells, it may act to present the ‘‘eat me’’ signal(s); in engulfing cells, it may act to assist CED-1 to recognize the signal(s) or it may act downstream of CED-1. It is equally possible that CED-7 may act to present an adhesive molecule or molecules on the surface of both cell types to facilitate their adhesion. CED-7 is the only protein known to be required in apoptotic cells for their engulfment, and it may provide a link between the cell death execution pathway and phagocytosis pathway. It is also possible that the direct ligand of CED-1 is not the ‘‘eat me’’ signal itself. Studies in mammalian cell cultures indicated that in many cases phagocytic receptors may require ‘‘bridging molecules’’ to interact with apoptotic cells. Known bridging molecules include complement pathway components, the serum adhesive protein thrombospondin (TSP), and the secreted glycoprotein MFG-E8 (Hanayama et al., 2002; Mevorach et al., 1998; Savill et al., 1992). If the association of engulfing and dying cells also requires bridging molecules in C. elegans, identifying the ligand(s) of CED-1

110

Zhou et al.

will lead to the identification of the bridging molecule(s). Human annexin I, a calcium-dependent PS-binding protein, has been implicated as a new bridging molecule between apoptotic and engulfing cells (Arur et al., 2003). Arur and co-workers (2003) also suggested that nex-1, one of the C. elegans homologs of human annexin I, may provide a bridging function that contributes to engulfment in C. elegans. In summary, the molecular characterization of CED-1, CED-6, and CED7 indicates that this functional group actually represents a signal transduction pathway leading engulfing cells to recognize dying cells for their engulfment (Fig. 6B). In this pathway, CED-7 acts upstream of CED-1 and may be involved in the presentation and/or recognition of the ‘‘eat me’’ signals. CED-1 acts as the phagocytic receptor that activates engulfing cells in response to neighboring dying cells, and CED-6, a candidate adaptor, is a strong candidate to relay CED-1 signal to downstream cellular machinery. It is unknown whether nex-1 is part of this pathway. d. The Functions of Mammalian Homologs of CED-1, CED-6, and CED-7 in the Engulfment of Apoptotic Cells. Humans and mice have one close homolog of CED-6 named hCED-6 or hGULP (engulfment adapter protein) and mGULP, respectively (Liu and Hengartner, 1999; Smits et al., 1999; Su et al., 2002). These proteins have structural features similar to those of CED-6, although they are approximately 150 amino acids shorter at the C termini (Su et al., 2002). hGULP partially rescued the engulfment defect of ced-6 mutants when expressed in C. elegans, suggesting that hGULP and CED-6 might be conserved in function (Liu and Hengartner, 1999). In addition, overexpression of hGULP in a macrophage cell line promotes phagocytosis (Smits et al., 1999), and the PTB domain of mGULP can interact with the intracellular domain of CED-1 when both are transiently expressed in mammalian cells (Su et al., 2002), again suggesting that GULP may be involved in phagocytosis in mammals. Similarly, ABC1, the closest mouse homolog of CED-7, acts in macrophages to promote the engulfment of apoptotic cells (Hamon et al., 2000). As discussed previously, ABC1 promotes Ca2+-induced exposure of phosphatidylserine to the outer membrane, an event believed to act as a trigger for the engulfment of the PS-presenting cell (Hamon et al., 2000). As mentioned previously, CED-1 and the scavenger receptor SREC display strong structural and sequence similarity to each other in their extracellular domains and therefore may possess similar ligand-binding specificity (Zhou et al., 2001b). However, whether SREC and CED-1 can replace each other’s function is not known. SREC lacks the NPXY or YXXL motifs that are critical for CED-1 function (Adachi et al., 1997; Zhou et al., 2001b). CD91/LRP, a human protein reported to play a role in the phagocytosis of apoptotic cells, was proposed to be a functional ortholog of CED-1 (Su et al.,


111

2002). The intracellular domain of CD91/LRP contains both the NPXY and YXXL motifs. However, its extracellular domain, although cysteine rich, possesses an architecture very diVerent from CED-1, and it is unknown whether CD91 shares CED-1’s ligand-binding specificity (Callebaut et al., 2003). MEGF10, a protein predicted from a human cDNA isolated from brain (Nagase et al., 2001), has been shown to display extensive structural and sequence similarity to CED-1 throughout its entire length (Callebaut et al., 2003). In particular, human MEGF10 contains the same number of tandem copies (16) of the CED-1 type of atypical EGF-like repeats in its putative extracellular domain and an NPXY motif in its C-terminal putative intracellular domain. In addition, both human MEGF10 and CED-1 were shown to bear an EMI domain at the very N terminus, N-terminal to the EGF-like repeats (Callebaut et al., 2003). MEGF10 is a protein of unknown function (Callebaut et al., 2003; Nagase et al., 2001). It will be interesting to find out whether MEGF10, CD91/LRP, and SREC have any functional relationship with CED-1. 2. What Promotes the Polarized Extension of Engulfing Cell Surfaces? CED-10 and Its Upstream Regulators a. CED-10 Is a Rac GTPase That Regulates Cytoskeletal Reorganization. It has long been established that polarized cell surface extension, which is observed during many cellular processes such as cell migration, axonal outgrowth, and phagocytosis of foreign objects, requires a driving force that comes from the reorganization of the actin cytoskeleton underneath the plasma membrane (reviewed in Hall, 1998). The Rho/Rac/Cdc42 family of small GTPases has been found to act as molecular switches that regulate cytoskeleton reorganization (reviewed in Hall and Nobes, 2000). Reddien and Horvitz (2000) cloned ced-10 and found that it encodes a C. elegans homolog of human Rac1, a member of the Rho/Rac/Cdc42 family (Fig. 5). Two recessive viable alleles of ced-10, n3246 and n1993, bear missense mutations in the conserved residues required for GTP binding and membrane targeting, respectively, suggesting that, like other members of this family, both GTPase activity and membrane localization are necessary for CED-10 function (Reddien and Horvitz, 2000). Several lines of evidence support the notion that CED-10 is critical for the reorganization of the actin cytoskeleton. It was observed that in addition to engulfment defects, ced-10 mutant animals also suVer from multiple cell migration defects, in particular in the migration of the distal tip cells (DTCs) (Kishore and Sundaram, 2002; Reddien and Horvitz, 2000; Soto et al., 2002). Accumulating evidence also indicates that ced-10 mutants are defective in axon pathfinding (Gitai et al., 2003; Lundquist et al., 2001; Wu et al., 2002). As the processes of both cell migration and axon pathfinding require

112

Zhou et al.

polarized cell surface extension, the fact that ced-10 function is required for these processes in addition to engulfment strongly suggests that it regulates cytoskeletal reorganization in response to multiple extracellular cues (Lundquist et al., 2001; Reddien and Horvitz, 2000). Additionally, ced-10 controls multiple essential cellular and developmental processes because a complete loss-of-function mutation of ced-10 results in maternal-eVect embryonic lethality (Lundquist et al., 2001; Soto et al., 2002). The direct evidence that CED-10 acts on cytoskeletal reorganization came from experiments performed in Swiss 3T3 cells (Zhou et al., 2001a). Serumstarved Swiss 3T3 fibroblasts lost almost all recognizable actin structures, including actin stress fibers, lamellipodia, and membrane ruZes (Ridley et al., 1992). Injection of DNA constructs driving the expression of the active forms of Rho family GTPases can induce the formation of distinct polymerized actin in serum-starved fibroblasts (Ridley and Hall, 1992; Ridley et al., 1992). Injection of CED-10(G12V), a presumptive active form of CED-10, resulted in the formation of membrane ruZes similar to those resulting from the injection of active human Rac1 (Fig. 7A) (Zhou et al., 2001a). This result indicated that CED-10 and Rac1 could result in similar changes to the actin cytoskeleton, indicating that the in vivo activities of CED-10 and Rac1 must be similar. b. CED-10 Is Not Alone—CED-2, CED-5, and CED-12 Form a Protein Complex That Activates CED-10. Gene interaction studies placed ced-2, -5, and -12 and ced-10 in one functional group. Strikingly, like ced-10 mutants, but unlike ced-1, -6, or -7 mutants, loss-of-function mutants of ced-2, -5, and -12 all display similar DTC migration defects. The abnormal migration patterns of DTCs in these mutants are identical to those observed in ced10 mutants, suggesting that ced-2, -5, and -12 may act in the cytoskeletal reorganization events regulated by ced-10 (Gumienny et al., 2001; Reddien and Horvitz, 2000; Wu and Horvitz, 1998b; Wu et al., 2001; Zhou et al., 2001a). ced-2, -5, -10, and -12 all act in engulfing, but not dying, cells (Gumienny et al., 2001; Reddien and Horvitz, 2000; Wu and Horvitz, 1998b; Wu et al., 2001; Zhou et al., 2001a). Furthermore, overexpression of CED-10 under the control of C. elegans heat-shock promoters was suYcient to bypass the requirement for ced-2, ced-5, or ced-12 in the engulfment of cell corpses (Gumienny et al., 2001; Reddien and Horvitz, 2000; Wu et al., 2001; Zhou et al., 2001a). This result indicates that the main functions of CED-2, CED-5, and CED-12 are to activate CED-10 for the reorganization of actin cytoskeleton. These results support a model in which CED-2, -5, and -12 act as upstream regulators of CED-10, and all four proteins form a signaling pathway in engulfing cells to regulate the polarized extension of cell surfaces.


113

Figure 7 Activation of CED-10 leads to the rearrangement of the actin cytoskeleton, an event required for the extension of pseudopods during the engulfment of cell corpses. (A) Overexpression in Swiss 3T3 cells of active forms of CED-10 or Rac1, the mammalian homolog of ced-10, induces the formation of membrane ruZes (indicated by arrowheads). Nuclei of serum-starved Swiss 3T3 cells were microinjected with construct that express myc-ced10(G12V) or myc-Rac1G12V, a constitutively active form of CED-10 or Rac1. Biotin dextran was also injected as a negative control. Distribution of filamentous actin (F-actin) was examined using phalloidin staining. An arrow indicates an actin bundle, which results from a secondary eVect after the injection of myc-Rac1G12V. (Reproduced from Zhou et al., 2001a.) (B) Current model for the action of the signaling pathway composed of CED-2, CED-5, CED-10, and CED-12. See text for details. The thin double line indicates the plasma membrane of an engulfing cell. The question mark indicates that the transmembrane receptor(s) for the ‘‘eat me’’ signal exposed by apoptotic cells is(are) of unknown identity.

CED-2 and CED-5 are homologs of mammalian CrkII, an SH2 SH3 domain-containing adaptor protein, and Dock180, a CrkII-binding protein, respectively (Fig. 5) (Reddien and Horvitz, 2000; Wu and Horvitz, 1998b). CED-12 has one Drosophila homolog (DCED-12) and three human homologs (ELMO1, ELMO2/HCED-12A, and human BAB14712), all of which contain a pleckstrin homology (PH) domain, a lipid-binding domain, and a proline-rich motif for binding to SH3 domains (Fig. 5) (Gumienny et al., 2001; Wu et al., 2001; Zhou et al., 2001a). Assays performed in vitro or in the yeast two-hybrid protein interaction detection system have shown that CED-12 interacts with CED-5, which interacts with CED-2, and that all three interact simultaneously in vitro (Wu et al., 2001; Zhou et al., 2001a). Thus, CED-2, CED-5, and CED-12 may form a complex in vivo. How might this protein complex activate CED-10? The activity of Rho family GTPases is regulated by a number of proteins, including the guanine

114

Zhou et al.

nucleotide exchange factors (GEFs), which enhance the exchange of bound GDP for GTP and promote the conversion of the Rho family proteins to an active state to regulate downstream signaling events (reviewed in Van Aelst and D’Souza-Schorey, 1997). More than 50 GEFs capable of regulating the Rho family GTPases have been identified, all of which contain a Db1 homology (DH) domain required for the nucleotide exchange activity in tandem with a PH domain, which also makes important contributions to the nucleotide exchange activity (reviewed in Braga, 2002). However, no known C. elegans DH-containing GEF proteins are required for the engulfment of cell corpses and thus are unlikely to act as GEFs for CED-10 during phagocytosis (Lundquist et al., 2001; Wu et al., 2001). Although CED-2, -5, and -12 do not have DH domains, CED-12 does contain a PH domain. Biochemical studies suggest that DOCK180 and ELMO1, the mammalian homologs of CED-5 and CED-12, respectively, could act together as an unconventional bipartite GEF for Rac (Brugnera et al., 2002). DOCK180 interacts with Rac (Kiyokawa et al., 1998; Nolan et al., 1998). The CED-2/-5/-12 complex thus is likely to possess similar GEF activity for CED-10 (Fig. 7B). Not much is known about how the CED-2/-5/-12 protein complex activates CED-10 in response to extracellular cues. CED-10 is membrane bound and is distributed evenly over the plasma membrane of many cells (Lundquist et al., 2001). The recruitment and assembly of the CED-2/-5/-12 complex is likely to be a region-specific event induced by extracellular signals. The PH domain, which binds phopholipids, in particular, phosphatidylinositols (PIs), is proposed to recruit GEFs to membranes (Braga, 2002). CED-12, which contains a PH domain, was detected to be membrane-bound in C. elegans cells (Zhou et al., 2001a). In addition, CED-2 contains an SH2 domain likely to associate with phosphorylated tyrosine residues, possibly in the cytosplasmic domain of an unknown transmembrane receptor (Reddien and Horvitz, 2000). The recruitment of the CED-2/-5/-12 complex to the membrane may therefore occur according to one of two possible models: (1) the binding of the PH domain of CED-12 to particular forms of membrane PIs generated in response to an extracellular signal, or (2) the binding of the SH2 domain of CED-2 to the cytoplasmic domain of an activated transmembrane receptor. It is also possible that both models are correct and the processes act cooperatively, or the complex might be recruited and activated by an unforeseen mechanism. CrkII, Dock180, and ELMO1, the mammalian counterparts of CED-2, -5, and -12, respectively, function cooperatively to promote Rac activation and phagocytosis in assays performed using cultured mammalian cells (Albert et al., 2000; Gumienny et al., 2001). Integrin v 5 was implicated as one of the upstream receptors that links extracellular signals with this Rac-activating complex through the function of p130cas, which interacts with CrkII (Albert et al., 2000). These studies confirmed that the signaling


115

pathway led by CED-2, -5, -10, and -12 represents an evolutionarily conserved basic mechanism that regulates cytoskeletal reorganization. However, none of the three integrin subunits identified in C. elegans seem to be involved in engulfment (Gumienny et al., 2001; Wu et al., 2001). The receptor that directly responds to the presence of cell corpses and that activates the CED-2/-5/-12 complex remains to be identified. c. GEX-2 and GEX-3 Are Two Potential New Components of the CED-10Signaling Pathway. Animals homozygous for a ced-10 null allele (n3417) display a maternal-eVect embryonic lethal phenotype (Lundquist et al., 2001). Embryos produced by ced-10(n3417) homozygous mothers arrest development with apparently well-diVerentiated cell types that fail to become properly organized, indicating the existence of multiple defects in morphogenesis and cell migration (Soto et al., 2002). Inactivation of two genes, gex-2 and gex-3 (gut on the exterior), results in embryonic lethality and embryonic developmental defects (e.g., defects in epidermal enclosure) in many ways similar to or more severe than that observed in ced-10(n3417) arrested embryos (Soto et al., 2002). GEX-2 and GEX-3 are homologous to p140/Sra-1 and Hem-2, two mammalian Rac1-interacting proteins, respectively (Soto et al., 2002). Because mammalian p140/Sra-1 can directly interact with filamentous actin, and because the Drosophila homolog of Hem-2 is required for axon pathfinding and correct actin organization, p140/Sra-1 and Hem-2 were proposed to provide a link between the Rac1 GTPase and actin cytoskeleton (Hummel et al., 2000; Kobayashi et al., 1998). Interestingly, arrested embryos generated in gex-2 and gex-3 loss-of-function background contain persistent cell corpses in numbers similar to those found in ced-10(n3417) mutant embryos (Soto et al., 2002). Although it is unknown whether this observation reflects a potential defect in cell corpse engulfment or a secondary consequence of the morphogenesis defects, it is possible that GEX-2 and GEX-3 act in the signaling pathways regulated by CED-10, including cell corpse engulfment (Soto et al., 2002). 3. CDL-1, a Stem-Loop Binding Protein and a Regulator of Core Histone Expression, May Be Involved in the Efficient Engulfment of Cell Corpses In addition to the 10 C. elegans genes that play or may play roles in cell corpse engulfment described previously, cdl-1 (cell death lethal), which encodes a stem-loop binding protein, was found to be required for the progression of programmed cell death and other aspects of embryogenesis in C. elegans (Kodama et al., 2002). In cdl-1 homozygous embryos, which are zygotic lethal, excess cell corpses accumulate, in addition to the defects observed in body elongation, pharyngeal development, and mitotic chromosome condensation (Kodama et al., 2002). A close examination indicates

116

Zhou et al.

that both the appearance and the elimination of cell corpses are delayed and that cell corpses persist much longer in these embryos (Kodama et al., 2002). CDL-1 was shown to bind to the stem-loop structure in the 30 -UTR of C. elegans core histone mRNAs, an event thought to be critical in regulating histone expression (Kodama et al., 2002). This activity and the cdl-1 mutant phenotypes are consistent with a model in which chromosome condensation, global gene expression, or both were perturbed in cdl-1 mutant embryos (Kodama et al., 2002). However, it remains to be determined whether CDL-1 aVects the progression of cell death and the eYciency of cell corpse engulfment by, perhaps, aVecting chromosome condensation during programmed cell death, or by a more indirect action.

D. The Engulfment of Cells That Die by Necrosis Requires the Same Genes That Act to Engulf Apoptotic Cells Necrosis is a form of cell death that results from acute cell injury (Wyllie et al., 1980). Necrotic cells swell and burst open, releasing their cellular contents, rather than shrinking and remaining intact, as is seen in apoptotic death (Wyllie et al., 1980). In C. elegans, dominant mutations in genes encoding several ion channel subunits such as MEC-4 (a subunit of the degenerin Naþ channels) and DEG-3 (acetylcholine receptor Ca2þ channel) induce necrotic death of specific groups of neurons as a result of channel hyperactivity and increased ion influx (Fig. 8) (reviewed in Driscoll and Gerstbrein, 2003). Expression of a constitutively active form of a heterotrimeric G-protein subunit Gs can also induce similar necrotic neuronal death in C. elegans (Driscoll and Gerstbrein, 2003). The induction of necrosis is independent of the functions of ced-3, ced-4, and egl-1 (egg-laying defective), three genes required for programmed cell death in C. elegans, and is not protected by ced-9, which protects against programmed cell death, indicating that the execution of necrosis is mechanistically distinct from that of apoptosis (Chalfie and Wolinsky, 1990; Chung et al., 2000; Ellis and Horvitz, 1986). Like apoptotic cells, necrotic cells in C. elegans are removed through the process of engulfment, and the functions of all seven known engulfment genes are required for this removal (Chung et al., 2000; Hall et al., 1997). A genetic screen was carried out for genes required for the eYcient removal of necrotic cell corpses generated in a mec-4 dominant mutant background, and one allele of ced-12, which was one of the first ced-12 mutant alleles identified, was recovered from this screen (Chung et al., 2000). As observed in the engulfment of apoptotic cells, the seven engulfment genes seem to act in the same two partially redundant pathways for necrotic cell removal: ced-1, -6, -7 in one pathway and ced-2, -5, -10, and -12 in the other (Chung et al., 2000). The requirement for the same engulfment


117

Figure 8 Nomarski microscope image of a ced-1(e1735);mec-4(e1611dm) larva at L1 stage showing the morphology of cells undergoing necrotic and apoptotic deaths. In this genetic background, a dominant mutation in mec-4 results in the necrotic death of six mechanosensory neurons, three of which are visible and indicated with arrows. These necrotic cells appear swollen and bigger in size compared to apoptotic cells, which appear condensed and highly refractile (indicated by arrowheads). These apoptotic cells are not engulfed at this time (as they would be in wild-type) because of the ced-1 mutation.

genes suggests that the mechanisms behind the recognition and engulfment of necrotic and apoptotic cells are similar. In particular, the requirement for CED-1 and CED-7, two proteins implicated in the recognition of apoptotic cells, suggests that cells dying of apoptosis and cells dying of necrosis may present similar ‘‘eat me’’ signals to induce their engulfment by neighboring cells. Alternatively, CED-1 may possess the ability to recognize multiple types of ‘‘eat me’’ signals. The removal of necrotic cells occurs much more slowly than that of apoptotic cells (Chung et al., 2000). This raises the possibility that the ‘‘eat me’’ signals are actively exposed onto the surface of apoptotic cells in a process triggered by the cell death execution machinery but are only passively released to the surface of necrotic cells as the cells break apart. A comparison of the cell surface properties of apoptotic and necrotic cells may help to identify the ‘‘eat me’’ signal(s). E. The Process of Engulfment Assists in the Execution of Apoptosis As discussed in Section II.B, the notion that engulfing cells are solely responsible for the removal of dying cells but are not involved in the execution of death resulted from the observation that mutations that severely

118

Zhou et al.

impair engulfment do not prevent cells destined to die from undergoing apoptosis. However, recent progress has indicated that engulfing cells play an active role in assisting apoptosis (Hoeppner et al., 2001; Reddien et al., 2001). Genetic and molecular studies in C. elegans have identified the pathway that controls the execution of somatic programmed cell death (reviewed in Metzstein et al., 1998). This pathway is composed of ced-3, the most downstream gene that encodes a caspase (CED-3) crucial for the execution of most if not all cell deaths, and ced-4, ced-9, and egl-1, three genes encoding its positive and negative regulators. According to the current model established through genetic and biochemical studies of these genes (reviewed in Metzstein et al., 1998), in cells destined to die, CED-3 is activated by CED-4, an Apaf-1-like caspase activator. In cells that should normally live, CED-9, a Bcl-2-like survival factor, directly interacts with CED-4 and inhibits CED4’s ability to activate CED-3. CED-4, CED-9, and most likely CED-3 are present in most, if not all, somatic cells. EGL-1, the most upstream protein required for the execution of somatic programmed cell deaths, is a BH3-only protein and is specifically expressed in cells destined to die in which it frees CED-4 from CED-9-mediated interaction and interacts with CED-9 itself (Chen et al., 2000; Conradt and Horvitz, 1998, 1999; Thellmann et al., 2003). Active CED-3 is thought to induce a number of downstream events that together lead to cell death. These events include cell and nuclear shrinkage, nuclear DNA degradation, and the exposure of the ‘‘eat me’’ signal(s) (reviewed in Metzstein et al., 1998). Reddien and co-workers (2001) and Hoeppner and co-workers (2001) each separately observed that in animals carrying homozygous weak partial loss-of-function mutations in the downstream killer genes ced-3 or ced-4, the presence or absence of an intact engulfment system aVects the eYciency of the cell death execution process. For example, animals homozygous for ced3(n2427), a weak loss-of-function allele, have an average of 1.5 extra undead cells in the anterior pharynx of 16 cells that normally die during embryogenesis (Reddien et al., 2001). Although no extra cells were observed in this area in any engulfment mutant alone, animals carrying both ced3(n2427) and mutations in engulfment genes have a greater number of extra cells (4.5–6.2) than in ced-3(n2427) mutants alone (Reddien et al., 2001). This phenomenon is not limited to the anterior pharynx, as the enhancement was also observed in cells that underwent apoptosis postembryonically in the lateral ectoderm and in the posterior ventral nerve cord (Reddien et al., 2001). This enhancement by mutations aVecting engulfment is not limited to mutations of particular engulfment genes; rather, mutations in all seven engulfment genes have a similar eVect, indicating that it is the whole engulfment machinery that plays a role in death execution (Hoeppner et al., 2001; Reddien et al., 2001). Reddien and co-workers (2001) also found that


119

expressing CED-1 specifically in engulfing cell types was suYcient to rescue the defects of ced-1 loss-of-function mutants in engulfment-mediated enhancement of cell death execution. Therefore, the death-assisting activity acts from engulfing cells in a cell nonautonomous manner. On the other hand, a deletion of the entire caspase domain of ced-3 results in the presence of 12 extra cells in the anterior pharynx, a phenotype that is not enhanced by any engulfment mutants, suggesting that the death-promoting eVect of the engulfing cells is dependent on some remaining caspase activity (Reddien et al., 2001). In engulfment mutant animals, even when the genes controlling the execution of cell death are wild-type, there is a low penetrance incidence of failure of or delayed cell death. In the posterior ventral cord of ced-6 and ced-7 mutant larvae, Reddien and co-workers (2001) observed that in many cases cells destined to die initially displayed some morphological characteristics of dying cells, then fluctuated between a dying and a living appearance for a period of time, and eventually chose one of the two fates. The living cells that had visibly undergone morphological changes associated with death and then recovered could extend neuronal processes and express a lin-11::gfp reporter, indicating that they had fully recovered and properly diVerentiated (Reddien et al., 2001). Using four-dimensional microscope time course analysis, Hoeppner and co-workers (2001) observed a similar phenomenon in ced-7 mutant embryos. How do engulfing cells promote death and ensure that death is irreversible? Although the molecular mechanism is not clear, several models have been proposed to explain the action of the engulfing cells (reviewed in Conradt, 2002). Cells in which a low-level caspase activity is expressed are not necessarily committed to die. In weak ced-3 mutants, embryonic cell deaths occur but are apparently delayed (Hoeppner et al., 2001; Stanfield and Horvitz, 2000). Moreover, Hoeppner and co-workers (2001) showed that in a weak ced-3 mutant background, some cells that have begun the morphological process associated with programmed cell death have the ability to recover and escape death. Thus, the eVect of low caspase activity is reversible. When the caspase activity within a dying cell is low, an engulfing cell may internalize such a cell and start its degradation before it has time to recover and survive. Alternatively, the neighboring engulfing cells may secret digestive enzymes before a dying cell is fully engulfed, or they may initiate a positive feedback loop through the contact with the dying cells to reinforce the decision of the dying cells to complete apoptosis (reviewed in Conradt, 2002). Consistent with the second and third models, ced-1 and ced-7 have been found to be required for an initial step in the degradation of the nuclear DNA of apoptotic cells, a step that occurs eYciently even when the engulfment of the dying cell is prevented by other engulfment mutations (Wu et al., 2000). These data are consistent with an eVect of the neighboring cell to

120

Zhou et al.

promote an apoptotic process within the unengulfed dying cell (discussed later). The discovery that engulfment assists in the execution of apoptosis is important to further our understanding of the regulation of apoptosis. In mammals, the enhancement of death by phagocytes may ensure an eYcient clearance of apoptotic cells and thus will lead to better prevention of potentially harmful inflammatory and autoimmune responses. Furthermore, macrophages, which are the ‘‘professional phagocytes’’ in mammals, have been shown to promote the apoptosis of certain types of cells, an event important for tissue remodeling. For example, macrophages induce apoptosis in normal vascular endothelial cells in rat eyes during programmed capillary regression (Diez-Roux and Lang, 1997). In vitro, co-culture of interferon-stimulated macrophages with myofibroblast-like mesangial cells was found to induce the apoptosis of the mesangial cells, and this phenomenon has been implicated as a step of the resolution of inflammatory injury in vivo (DuYeld et al., 2000). It remains to be studied whether macrophages in mammals and engulfing cells in C. elegans use similar molecular mechanisms for inducing apoptosis. As discussed in the next section, in C. elegans, there are also several known examples of cell deaths that are solely dependent on engulfment.

F. The Cases of Murder: Programmed Cell Deaths That Are Critically Dependent on Engulfing Cells For a few of the described cases of programmed cell deaths in C. elegans, engulfing cells appear to act as killers, required for the deaths to occur. There are two notable cases in male development where a cell programmed to die must interact with its neighbors in order for its death to occur: the linker cell and the B.a(l/r)apaav equivalence group (Sulston et al., 1980). In the first example, the linker cell leads the extending male gonad from its starting point in the midbody to its final destination in the tail (Kimble and Hirsh, 1979). Once the gonad has reached its target, the linker cell is engulfed by one of two cells, U.lp or U.rp, neighboring to its final position. However, if these cells are ablated or if the linker cell is prevented from coming in contact with them, the linker cell survives (Sulston et al., 1980). In the second case, B.alapaav and B.arapaav are left-right homologs that comprise an equivalence group: one of these two cells fuses to form part of the vas deferens, while the other cell dies and is engulfed by its neighbor, P12.pa. Similar to linker cell death, both B.alapaav and B.arapaav survive when the cell that normally engulfs them is ablated (Sulston et al., 1980). In both of these cases, death occurs through murder instead of suicide (i.e., death required the participation of another cell). ced-1 or ced-2 loss-of-function


121

mutations can result in the survival of the linker cell and both B.alapaav and B.arapaav, indicating that engulfment is required for cell death in these two instances (Ellis and Horvitz, 1986; Hedgecock et al., 1983; Reddien et al., 2001). In the case of the linker cell, death can occur even in the absence of ced-3 or ced-4, suggesting that engulfment leads to the activation of a nonclassical killing pathway (Ellis and Horvitz, 1986). In contrast, while the death within the B.alapaav and B.arapaav equivalence group is completely dependent upon engulfment, it also requires the function of ced-3 (Reddien et al., 2001). In this case, the engulfing cell may be able to trigger caspase activation or may be necessary to respond to a sublethal level of caspase activation by promoting cell death execution (Reddien et al., 2001). Another example of cell death that is dependent on engulfment was observed in semidominant mutants of lin-24 and lin-33, two genes required for the specification of vulval cell lineages (Ferguson et al., 1987). In these mutants, up to six of 12 Pn.p cells, which are a group of epithelial precursor cells that should normally live, develop an abnormal morphology and often die. These dying cells undergo morphological changes that resemble necrotic cell deaths (Ferguson et al., 1987). While these deaths appear to be independent of ced-3 and ced-4 activities, they are suppressed by loss-of-function mutations in the engulfment genes ced-2, ced-5, and ced-10, which constitute one of two functional groups required for the engulfment of dying cells, but not by mutations in ced-1, ced-6, or ced-7 (S. C. Kim and H. R. Horvitz, unpublished results). This phenomenon suggests that the deaths of the abnormal Pn.p cells are dependent on the function of CED-2/CED-5/CED-10 and presumably involves aggressive engulfment by neighboring cells.

III. The Degradation of Nuclear DNA During Programmed Cell Death in C. elegans Wyllie observed that DNA from thymocytes induced to undergo apoptosis is processed into multimers of 180 base pair subunits (Wyllie, 1980). This nuclear DNA fragmentation is a general phenomenon observed in many types of cells undergoing apoptosis and was demonstrated to be dependent upon caspase activation during cell death (Enari et al., 1996; Liu et al., 1996). Pulse field gel electrophoresis reveals that DNA from apoptotic cells is initially cleaved at the scaVold regions to produce fragments of 50–300 kb, before they are subsequently processed into smaller fragments (Lagarkova et al., 1995; Oberhammer et al., 1993). TUNEL (terminal deoxynucleotidyl transferase-mediated dUTP nick-end labeling) (Gavrieli et al., 1992) and LM-PCR (linker-mediated-PCR) (Staley et al., 1997), two common assays for cell death, measure the extent of DNA fragmentation by detecting the

122

Zhou et al.

30 -hydroxyl

50 -phosphate

reactive and ends generated in the process, respectively. In this section, we discuss studies investigating nuclear DNA degradation during programmed cell death in C. elegans, and in Section IV, we compare them to what is known in mice and Drosophila.

A. nuc-1 Encodes a DNaseII Homolog Critical for DNA Degradation During Cell Death The first gene in C. elegans identified as playing a role in cell death was nuc-1 (nuclease abnormal). nuc-1 was isolated in a genetic screen seeking to recover mutants with lineage defects in the ventral nerve cord (Sulston, 1976). Sulston (1976) was visualizing cell nuclei by Fuelgen staining (a DNA dye) when he noticed a mutant, nuc-1, in which ingested bacterial DNA was not properly degraded. nuc-1 mutants also retained persistent DNA at positions consistent with the nuclei of cells that had undergone programmed cell death (Hedgecock et al., 1983; Sulston, 1976). An acid endonuclease activity is significantly reduced in protein extracts generated from nuc-1 mutants compared to those from wild-type animals (Hevelone and Hartman, 1988). Although DNA degradation is inhibited in nuc-1 mutants, neither the execution of cell death nor the engulfment of cell corpses appears to be aVected (Hedgecock et al., 1983; Parrish et al., 2001; Sulston, 1976; Wu et al., 2000). To examine the mechanism regulating DNA degradation during apoptosis, Wu and co-workers (2000) developed a TUNEL assay for use on C. elegans embryos. As previously mentioned, TUNEL specifically labels 30 hydroxyl ends present in DNA fragments (Gavrieli et al., 1992) and has been extensively used to label cells undergoing apoptosis. In wild-type C. elegans embryos, during midembryogenesis, at the stage when 14 cell corpses are visible using Nomarski microscopy, an average of only 1.7 nuclei are TUNEL positive (Wu et al., 2000). In contrast, approximately 48 TUNELreactive nuclei can be seen in nuc-1 mutants at the same stage. In addition, the generation of TUNEL-reactive nuclei follows kinetics similar to that of the appearance of cell corpses. NUC-1 must therefore act in an intermediate step for DNA degradation, converting TUNEL-positive ends to TUNEL-negative ends (Wu et al., 2000). Wu and co-workers (2000) cloned nuc-1 and determined that it encodes a DNaseII homolog. DNaseII is an endonuclease that functions at low pH and generates DNA fragments with 50 -hydroxyl and 30 -phosphate groups, which are not substrates for the terminal transferase used in TUNEL assays (Harosh et al., 1991). Thus, the disappearance of a TUNEL-positive signal in wild-type embryos is consistent with the predicted enzymatic properties of NUC-1. Engulfment does not appear to be required for the generation and degradation of TUNEL-reactive ends in cells undergoing programmed cell death.


123

Single mutant embryos of the engulfment genes ced-2, -5, -6, and -10 have similar numbers of TUNEL-positive nuclei compared to wild-type embryos, and double mutants between each of these engulfment mutants and nuc-1 produced approximately the same number of TUNEL-staining nuclei as a nuc-1 single mutant (Wu et al., 2000). However, engulfment is required for the completion of DNA degradation, as Feulgen-reactive spots persist in unengulfed cell corpses (Hedgecock et al., 1983). In addition, unengulfed cell corpses are positive for the staining of SYTO-11, another DNA dye, in ced-2, ced-5, ced-6, and ced-10 mutant animals (Wu et al., 2000). The fact that DNA is present in unengulfed cell corpses suggests that the completion of DNA degradation requires a contribution from either the engulfing cell or the engulfment process. In contrast to ced-2, -5, -6, and -10, ced-1 and ced-7 were observed to be required or partially required for the initial step of DNA degradation that results in the generation of TUNEL-reactive ends that are then degraded in a nuc-1-dependent manner. In ced-1;nuc-1, and ced-7;nuc-1 double mutant embryos, an average of only 1.0 and 20 TUNEL-reactive nuclei, respectively, were observed, compared to the 48 TUNEL-positive nuclei observed in nuc-1 embryos (Wu et al., 2000). This requirement for ced-1 and ced-7 might be explained if these two genes, in addition to their functions in promoting engulfment, are required to promote certain aspects of cell death such as nuclear DNA degradation within the dying cell (Wu et al., 2000). As discussed in Section II, CED-1, in particular, functions as a phagocytic receptor, while CED-7 appears to be required for the clustering of CED-1. Wu and co-workers proposed a three-step model for DNA degradation in C. elegans (Fig. 9) (Wu et al., 2000). Initially, intact DNA is cleaved by a CED-1/CED-7-dependent nuclease that produces TUNEL-reactive fragments. These fragments serve as substrates for NUC-1, which processes them into TUNEL-nonreactive fragments. Given that NUC-1 is secreted into the intestinal lumen, it is possible that NUC-1 is produced in the engulfing cell and then is transported to the dying cell (Wu et al., 2000). Alternatively, the observation that NUC-1 functions in unengulfed cell corpses to degrade TUNEL-reactive DNA ends raises the possibility that it acts, at least in part, cell autonomously (Wu et al., 2000). Complete digestion into nucleotides occurs through an unidentified nuclease activity provided in the engulfing cell.

B. cps-6 Regulates Both DNA Degradation and the Timing of Cell Death cps-6 (CED-3 protease suppressor) was identified by Parrish and co-workers (2001) from a genetic screen for suppressors of cell deaths induced by ectopially ced-3 expressed. cps-6 encodes the C. elegans homolog of mitochondrial

124

Zhou et al.

Figure 9 Current model for the process and pathway of nuclear DNA degradation during apoptosis in C. elegans. The process for DNA degradation involves at least three steps. In the first step, chromosomal DNA of dying cells is digested into TUNEL-positive fragments by an unknown endonuclease activity that requires CED-1 and CED-7 functions. In the second step, these fragments are further cleaved into TUNEL-nonreactive fragments. Endonucleases including NUC-1 mediate this step, and they form four partially functional redundant groups as described in the text. In the third step, which also requires the engulfment of the cell corpses, the TUNEL-negative DNA is further degraded into free nucleotides.

endonuclease G. Like NUC-1, CPS-6 is required for the eYcient processing of TUNEL-reactive DNA fragments to a TUNEL-nonreactive form. Inactivation of cps-6 by RNA interference (RNAi) in a nuc-1 mutant background causes the presence of TUNEL-positive nuclei in numbers larger than seen in either nuc-1 mutants or cps-6 (RNAi) animals, indicating that cps-6 and nuc-1 may have partially redundant functions (Parrish et al., 2001). In addition to its role in DNA degradation, cps-6 regulates the timing and execution of cell death, which might explain the finding that inactivating cps-6 could partially suppress ced-3-induced apoptosis (Parrish et al., 2001). cps-6 mutants display a delayed appearance of cell corpses during development similar to that observed in ced-8 mutants (Parrish et al., 2001). ced-8 encodes a transmembrane protein required for the proper kinetics of programmed cell death (Stanfield and Horvitz, 2000). The appearance of cell corpses is further delayed in cps-6;ced-8 double mutants compared to either single mutant. Although a mutation in cps-6 on its own has little eVect in blocking apoptosis, cps-6 loss of function enhances cell survival in both weak and strong mutants of ced-3 and ced-4. These observations support the hypothesis that cps-6 may normally function in promoting cell death (Parrish et al., 2001). Wang and co-workers (2002) have reported that cps-6 functions together with an apoptosis inducing factor (AIF) homolog in C. elegans. AIF is a mitochondrial oxidoreductase that is released into the cytoplasm to induce cell death in response to an apoptotic signal (Joza et al., 2001; LoeZer et al., 2001; Susin et al., 1999). RNAi inactivation of wah-1 (worm AIF homolog) results in an enhancement of ced-3, ced-4, and ced-8 single mutant phenotypes similar to that seen for cps-6 inactivation (Wang et al., 2002). wah-1 promotes DNA degradation: wah-1 (RNAi) embryos contain more TUNEL-reactive nuclei than do embryos treated with control RNAi. In addition, wah-1(RNAi) fails to delay death more than control RNAi in a cps-6;ced-8 double mutant background, and wah-1(RNAi) does not enhance the DNA degradation


125

defect seen in cps-6(sm116) animals (Wang et al., 2002). Like cps-6, wah1(RNAi) animals are not deficient in degrading bacterial DNA. Thus, inactivations of cps-6 and wah-1 cause similar phenotypes, and inactivation of both does not cause more severe defects than those seen for either one alone. Finally, recombinant WAH-1 and recombinant CPS-6 display synergetic ability to degrade DNA, a finding that could not be reiterated by sequential addition of CPS-6 and WAH-1 each in the absence of the other (Wang et al., 2002). These results are consistent with a model in which WAH-1 and CPS-6 act together to promote cell killing and degradation of nuclear DNA, a relationship that may be conserved in mammals (Wang et al., 2002).

C. Multiple Pathways Are Involved in DNA Degradation in C. elegans A paper by Parrish and Xue (2003) describes the identification of several cell death-related nucleases (crn genes) via a functional genomics approach. Seventy-seven predicted genes were selected for inactivation based on the predicted nuclease activities of their protein products. Animals treated with RNAi against each of these 77 candidates were screened for alterations in the pattern of TUNEL staining. Nine nucleases were identified, including two products of previously characterized genes nuc-1 and cps-6. The seven remaining nucleases were named crn-1 through crn-6 and cyp-13 (a cyclophilin E homolog); each of the identified genes has a mammalian homolog. Parrish and Xue (2003) proposed that these proteins normally perform housekeeping functions but when needed are recruited to execute cell death. The 10 genes that have been identified to function in DNA degradation (Table II) have been classified into four epistasis groups: (1) nuc-1; (2) wah-1, cps-6, crn-1, crn-4, crn-5, cyp-13; (3) crn-2, crn-3; and (4) crn-6 (Fig. 9) (Parrish and Xue, 2003; Parrish et al., 2001; Wang et al., 2002). Interestingly, animals mutant for cps-6 and inactivated for a gene in the crn-2 pathway may display a synthetic engulfment phenotype. Cell corpses persist 55% longer in cps-6(sml16);crn-2(RNAi) animals compared to corpses in the wild-type, cps-6(sm116) mutants, or crn-2(RNAi) animals. This observation hints of a possible link between DNA degradation, possibly as a checkpoint for the successful progression of apoptosis, and cell corpse engulfment (Parrish and Xue, 2003). Alternatively, it has been observed that cells fated to die in which the execution process has been impaired often persist longer than do dying cells in the wild-type, suggesting that the persistent cell corpses of cps-6(sm116);crn-2(RNAi) animals may result from a defect in the execution of death (Hoeppner et al., 2001; Stanfield and Horvitz, 2000). In addition, proteins encoded by members of the cps-6 pathway can interact in vitro, indicating that they may form a DNA degradation complex, named the degradeosome (Parrish and Xue, 2003). A more extensive

126

Zhou et al.

Table II Genes Whose Functions Are Involved in the Degradation of Nuclear DNA During Programmed Cell Death in C. elegans Identity of Mammalian Homologs

Gene Name cps-6

Endonuclease G

cm-1

Flap endonuclease I

cm-2 cm-3

cm-4

TatD 100 kDa polymyositis/ scleroderma autoantigen (PM/Scl-100) 30 to 50 exonuclease

cm-5

Rrp46

cm-6 cyp-13 nuc-1 wah-1

DNase II Cyclophilin E DNase II Apoptosis inducing factor (AIF)

Function of the Mammalian Homologs

Reference

Caspase-independent cell death DNA replication, damage repair No known function Ribonuclease component of the exosome

Parrish et al., 2001

tRNA processing, DNA replication Ribonuclease component of the exosome Lysomomal nuclease RNA splicing Lysosomal nuclease Caspase-independent cell death

Parrish and Xue, 2003

Parrish and Xue, 2003 Parrish and Xue, 2003 Parrish and Xue, 2003

Parrish and Xue, 2003 Parrish and Xue, 2003 Parrish and Xue, 2003 Wu et al., 2000 Wang et al., 2002

characterization of crn-1 function supports the model that the distinct components of the degradeosome provide diVerent enzymatic activities that cooperate in degrading DNA (Parrish and Xue, 2003; Parrish et al., 2003). Thus, the emerging picture of apoptotic DNA degradation appears to be far more complex than initially anticipated, with multiple pathways contributing to the digestion of apoptotic DNA and linking DNA degradation with the execution of cell death and, possibly as a consequence, the engulfment of dying cells (Fig. 9) (Parrish and Xue, 2003).

IV. Study of Engulfment and DNA Degradation in Drosophila A. Hemocytes Act as Engulfing Cells In Drosophila, the vast majority of apoptotic cells is cleared by hemocytes (blood cell), although engulfment by nonprofessional phagocytes (epidermal cells in the eye and glial cells in the central nervous system [CNS]) has also been reported (Sonnenfeld and Jacobs, 1995; WolV and Ready, 1991). Using an antibody against the extracellular matrix (ECM) protein peroxidasin as a hemocyte-specific marker, Tepass and co-workers (1994) traced


127

the embryonic origin of hemocytes and studied how cell death aVects the hemocyte-to-macrophage transition during development. They found that hemocytes arise from about 700 progenitors generated in the procephalic mesoderm. These cells migrate from their birthplace to populate the entire embryo, and their number remains constant during embryonic development. Earlier in development, hemocytes are typically small and round mesodermal cells. Coincident with their migration and the initiation of cell death in the embryo, they develop an extensive endoplasmic reticulum network and filipodia that extend from the cell body (Tepass et al., 1994). At this point, hemocytes near the brain and nerve cord become phagocytic. These phagocytic cells contain dense inclusions composed of apoptotic cells that have been engulfed (Tepass et al., 1994). Hemocytes containing ingested cells (or cell fragments) are defined as macrophages (Tepass et al., 1994). At later stages of development, approximately 90% of the peroxidasin-positive hemocytes have become phagocytic, with only the hemocytes in the gut remaining nonphagocytic, possibly due to the absence of cell death there (Tepass et al., 1994). Interestingly, complete conversion of hemocytes to macrophages was observed in embryos in which ectopic cell death had been genetically induced. This suggests that a cell death-dependent signal is suYcient to induce the hemocyte-to-macrophage transition and that peroxidasin-expressing cells define a homogenous population capable of undergoing this conversion (Tepass et al., 1994). On the other hand, programmed cell death is unaVected in embryos from which the hemocyte linage has been ablated (Tepass et al., 1994).

B. croquemort Encodes a Phagocytic Receptor in Drosophila Phagocytosis has not been studied as extensively in Drosophila as in C. elegans. The functions of myoblast city (mbc) and Rac1, homologs of the worm engulfment genes ced-5 and ced-10, respectively, have been characterized in flies. Although MBC and RAC1 have been demonstrated to regulate pseudopod extension and cytoskeletal reorganization (Erickson et al., 1997; Kiyokawa et al., 1998; Nolan et al., 1998)—roles consistent with those proposed for their C. elegans homologs—they have not been reported to function in the engulfment of apoptotic cells. However, Croquemort (French for pallbearer), a transmembrane receptor, has been demonstrated to be required for the engulfment of apoptotic cells (Franc et al., 1996, 1999). croquemort (crq) encodes a single-pass transmembrane protein that bears homology to human CD36, a scavenger receptor expressed in macrophages, and can mediate recognition of apoptotic cells (Franc et al., 1996). Expression of crq in COS7 cells promotes recognition and internalization of

128

Zhou et al.

apoptotic murine thymocytes by COS7 cells. Moreover, binding of apoptotic thymocytes is specifically inhibited by pre-incubation of transfected cells with antisera against the extracellular domain of Crq (Franc et al., 1996). To examine the in vivo function of crq, Franc and co-workers (1999) examined embryos homozygous for deficiencies that span the crq locus. Embryos homozygous for these deficiencies display a 20-fold decrease in the ability of hemocytes to engulf apoptotic cells compared to that seen in wild-type hemocytes, and this defect can be rescued with ubiquitous expression of crq (Franc et al., 1999). There appears to be a specific role for crq in clearing apoptotic cells, as other macrophage functions including phagocytosis of bacteria, endocytosis of acetylated low-density lipoprotein (AcLDL) particles, and production of ECM components are unaVected in crq-deficient embryos (Franc et al., 1999). Strong crq expression starts coincident with the initial wave of embryonic cell death and is detected in macrophages that contain ingested TUNELreactive apoptotic bodies (Franc et al., 1996). Cell death appears to promote crq expression, as a 74% decrease in crq levels was observed in H99 embryos, which contain a small deficiency that deletes the genes grim, reaper, and hid required for embryonic cell death (Franc et al., 1999; White et al., 1994). Moreover, increases in crq expression and phagocytic activity were detected in wild-type embryos and a hemocyte-derived cell line l(2)mbn induced to undergo ectopic apoptosis (Franc et al., 1999). crq expression thus seems to depend on the presence of apoptotic cells. The studies of crq raise multiple questions. First, the ligand for Crq has yet to be identified. In addition to chemically modified LDL, the homologous mammalian scavenger receptor CD36 was reported to bind other polyanionic ligands (Krieger and Herz, 1994). It is tempting to speculate that apoptotic cells present such a signal—perhaps phosphatidylserine—that mediates their recognition by Crq. Second, components of the signaling pathway downstream of crq need to be identified in order to understand how the phagocytic machinery is activated. In addition, loss-of-function alleles that specifically aVect crq have not been reported. The ability of hemocytes to phagocytose apoptotic cells was not totally abolished in crq-deficient embryos, indicating that other receptors may mediate recognition of dying cells (Franc et al., 1999). Other receptors known to be expressed by Drosophila macrophages, such as the scavenger receptor dSR-C1 (Pearson et al., 1995), Malvolio (Rodrigues et al., 1995), and the phosphatidylserine receptor (Fadok et al., 2000), may perform redundant functions (Fig. 10). Given the severe reduction in engulfment in crq-deficient embryos, Crq probably functions as the major phagocytic receptor in flies (Franc et al., 1999).


129

Figure 10 Model of Croquemort function as a major phagocytic receptor that mediates the engulfment of apoptotic cells in Drosophila. Recognition of apoptotic cells leads to multiple events, including cytoskeletal rearrangement, that are required for the engulfment of apoptotic cells. The recognition and/or engulfment of apoptotic cells by engulfing cells also promotes Crq expression and diVerentiation of hemocytes into macrophages.

C. Engulfment of Apoptotic Cells Is Required for Proper Patterning of the CNS Many nerve cells die during Drosophila development, and these apoptotic cells in the CNS are continuously cleared by phagocytes (Sonnenfeld and Jacobs, 1995). In a 2003 paper, Sears and co-workers suggested that the engulfment of apoptotic cells is required for normal CNS morphogenesis in Drosophila by studying mutants in PDGF- and VEGF-receptor related (Pvr), a receptor tyrosine kinase. One significant feature of Pvr loss-of-function mutants is that hemocytes fail to migrate, instead remaining at their birthplace in the head mesoderm (Sears et al., 2003). Additionally, although CNS architecture was largely normal in Pvr mutants, the precise ladder-like structure of the axonal scaVold was disrupted (Sears et al., 2003). The nerve cord in Drosophila consists of a set of two longitudinal axon tracts that run parallel to the ventral midline. In Pvr mutants, these two longitudinal axon tracts appear to be compressed much closer to each other. There was also an abnormal accumulation of glia in the midline, indicating that Pvr function is required for correct positioning of these cells. CNS axonal pathfinding appears to be normal, with no inappropriate crossing at the midline (Sears et al., 2003).

130

Zhou et al.

Several lines of evidence suggest that Pvr function is required in hemocytes for proper development of the CNS. Inducing the expression of a dominant negative form of Pvr(DN-Pvr) in the hemocyte linage results in migration defects comparable to those in Pvr-mutants. In addition, the CNS phenotypes observed in Pvr-mutants were recapitulated by hemocyte-specific expression of DN-Pvr, suggesting that hemocyte function is required for proper CNS patterning (Sears et al., 2003). To test this hypothesis, the hemocyte lineage was genetically ablated using a mutation in serpent (srp), a gene that encodes a GATA transcription factor required for hemocyte development (Rehorn et al., 1996). Axonal tract and glial abnormalities indistinguishable from Pvr mutants were observed in srpneo45 mutant embryos (Sears et al., 2003). Embryos in which crq expression had been knocked down using RNAi have compressed longitudinal axon tracts and glial positioning defects that are very similar to those found in Pvr and srp mutants (Sears et al., 2003). Given these observations, Sears et al. (2003) proposed that clearance of apoptotic cells plays a critical role in patterning the Drosophila CNS. Defects in phagocytosis may result in the accumulation of apoptotic cells in the CNS, which could constitute a physical barrier that disrupts glial and axonal positioning in mutants lacking hemocytes. However, the possibility that Pvr and crq could function in CNS regulation independent of phagocytosis cannot be ruled out.

D. Caspase-Independent Engulfment In order to follow the engulfment of apoptotic cells during development, Mergliano and Minden (2003) developed an in vivo assay in Drosophila that employs resorufin- -galactoside-polyethylene glycol1,900 (VGAL). VGAL is a fluorogenic substrate that is activated after being internalized into macrophages by lysosomal -galactosidase. VGAL is injected into the common cytoplasm of early Drosophila embryos so that it can be taken up by all cells. In wild-type embryos, the pattern of engulfment detected using the VGAL assay closely mirrors that of cell death as observed by acridine orange staining (Mergliano and Minden, 2003). Cell death-deficient H99 embryos have very few acridine orange-positive nuclei indicative of apoptotic cells and develop only a few macrophages (White et al., 1994). Observations made in H99 embryos earlier in development using the VGAL assay agree with the results generated by acridine orange staining. However, later in development, striped VGAL-positive spots in the epidermis were observed in H99 embryos in a pattern that resembles that seen in wild-type embryos, indicating the engulfment of these cells (Mergliano and Minden, 2003). Likewise, ubiquitous expression of the pan-caspase inhibitor Baculovirus


131

p35 protein in wild-type embryos does not aVect the production of these VGAL-positive stripes in the epidermis, despite the fact that it eYciently blocks apoptosis, suggesting that a form of cell engulfment proceeds even in the absence of caspase activity (Mergliano and Minden, 2003). This observation suggests a possibility that living cells are being engulfed or that certain cells undergo programmed cell death in a caspase-independent manner. Because dying epidermal cells are usually engulfed by neighboring cells instead of hemocytes, to test if the inhibition of cell death aVects engulfment by hemocytes, p35 was expressed in a patch of neurogenic region, in which dying cells are known to be engulfed by hemocytes (Mergliano and Minden, 2003). Neural precursors deficient for caspase activity and presumably remaining alive were nonetheless recognized and engulfed by hemocytes, again suggesting that certain signals can trigger phagocytosis in the absence of caspase activity in cells otherwise fated to die of caspase-dependent apoptosis (Mergliano and Minden, 2003). In C. elegans, blocking phagocytosis can weakly promote cell survival, an event that can be more strikingly seen as an enhancement of mild defects in cell death (Hoeppner et al., 2001; Reddien et al., 2001). This raises an interesting possibility that the observed engulfment of caspase-deficient cells in Drosophila could reflect an analogous mechanism for promoting cell death in cells that have begun the process of apoptosis. Alternatively, the engulfed cells may die of a very diVerent mechanism or may be engulfed alive (Mergliano and Minden, 2003).

E. DNA Degradation in Drosophila Proceeds in Two Steps 1. Cell-Autonomous DNA Degradation The major cell-autonomous mechanism by which the DNA of apoptotic cells is degraded was initially discovered through elegant biochemical studies performed in mammalian systems. Caspase-activated DNase (CAD) was purified from mouse T-cell lymphoma as the protein responsible for the nuclease activity in caspase-3-treated cytosol (Enari et al., 1998). CAD is normally kept inactive by binding to its inhibitor, ICAD (inhibitor of CAD), and is only de-repressed by active caspases. Caspase activation results in the cleavage of ICAD and the release of functional CAD (Enari et al., 1998; Sakahira et al., 1998). Protein refolding experiments suggest that in addition to inhibiting CAD activity, ICAD functions as a specific chaperone that is required for the proper folding of CAD (Sakahira et al., 2000). CAD and ICAD were also independently identified and characterized as DFF40/ CPAN and DFF45, respectively (Halenbeck et al., 1998; Liu et al., 1997).

132

Zhou et al.

Homologs of CAD and ICAD have been identified in Drosophila (Mukae et al., 2000; Yokoyama et al., 2000), but there are no identifiable homologs of CAD and ICAD in the C. elegans genome. In contrast to mouse and human CAD (mCAD and hCAD, respectively), which are not known to require protease processing, Drosophila CAD (dCAD) is cleaved by eVector caspases into large (p32) and small (p20) subunits during activation (Yokoyama et al., 2000). Activated dCAD behaves as a 100-kDa tetramer of (p32)2(p20)2 in a gel filtration column, while mCAD and hCAD do not seem to display any subunit structure. Additionally, dCAD lacks a C-terminal nuclear translocation signal found in mCAD and hCAD. While dCAD, like its mammalian homolog, has nuclease activity in vitro, it fails to degrade DNA in intact nuclei in standard fragmentation assays (Yokoyama et al., 2000). dICAD is a Drosophila protein approximately 17% identical in sequence to mouse and human ICAD (Yokoyama et al., 2000). Despite this low sequence resemblance, dICAD possesses the biochemical properties of a Drosophila ICAD homolog: it copurifies with the pro-form of dCAD, it inhibits dCAD activity, and it is required to produce functional dCAD (Fig. 11) (Yokoyama et al., 2000). To assess the contribution of dCAD to DNA degradation in apoptotic cells, Mukae and co-workers inactivated the dICAD gene by P element mutagenesis (Mukae et al., 2002). Because ICAD is essential for the production of functional CAD, dICAD-null flies are likewise inactive for CAD. dICAD-null flies are fertile and appear to have no discernable phenotype. However, DNA laddering was detected in wildtype but not ICAD-nulls, indicating that CAD is the major nuclease responsible for the degradation of DNA of apoptotic cells during fly oogenesis and embryogenesis (Fig. 11) (Mukae et al., 2002).

2. Cell Nonautonomous DNA Degradation McIlroy and co-workers generated transgenic mice that ubiquitously express a caspase-resistant form of ICAD (Sdm-ICAD), which should therefore prevent the caspase-induced activation of CAD (McIlroy et al., 2000). No DNA fragmentation was observed when thymocytes from transgenic mice were induced to undergo apoptosis in vitro. However, subsequent analyses in vivo demonstrated that lysosomal DNaseII from phagocytes provides an auxiliary mechanism by which DNA from apoptotic cells that have been engulfed is degraded (McIlroy et al., 2000). McIlroy and co-workers (2000) proposed a two-step model whereby DNA degradation proceeds initially in a cell-autonomous manner through the action of CAD and then proceeds further in a non-cell-autonomous manner through the action of lysosomal DNaseII in engulfing cells.


133

Figure 11 The model for the mechanism that activates dCAD in apoptotic cells. dICAD functions as both a molecular chaperone and an inhibitor of dCAD. dICAD is required for correct folding of the newly translated dCAD protein. However, dCAD activity is inhibited upon its binding with dICAD. During apoptosis, an activated caspase in the dying cell cleaves dICAD and releases dCAD, which is further processed by the caspase into two fragments, p32 and p20. These fragments form an active tetramer (p32)2(p20)2, which degrades nuclear DNA into LM-PCR-detectable fragments.

Mukae and co-workers (2002) report a Drosophila model that is deficient for apoptotic DNA degradation. As discussed previously, CAD is the major nuclease responsible for the fragmentation of nuclear DNA detected by LMPCR (Mukae et al., 2002). Oogenesis proceeds in egg chambers housed in the Drosophila ovary. At a late stage of oogenesis, nurse cells transfer their cytoplasm to provide maternal determinants to the oocyte. The nuclei of nurse cells are subsequently degraded in a caspase-dependent manner (Cavaliere et al., 1998; McCall and Steller, 1998). However, despite the lack of DNA fragmentation in dICAD-null flies, no observable DNA accumulation was detected in egg chambers (Mukae et al., 2002). This could be explained if a mode of non-cell-autonomous DNA degradation similar to that seen in mammals operates in flies to promote the degradation of DNA in the absence of cell-autonomous CAD activity. In the same report, Mukae and co-workers described the cloning and characterization of Dnase-1lo, a loss-of-function mutation in Drosophila DNaseII (Mukae et al., 2002). DNaseII loss-of-function mutations do not prevent the generation of DNA laddering in apoptotic cells, indicating that DNaseII is not responsible for the cell-autonomous DNA degradation occurring in apoptotic cells. No DNA laddering in dying nurse cells was observed in flies deficient for both dICAD and dDNaseII. Instead, acridine orange-reactive vesicles accumulate in ovaries of Dnase-1lo single mutants and dICAD, Dnase-1lo double mutants, indicating that DNA from the dying nurse cells is not degraded. Thus, at least in nurse cells, dCAD and dDNaseII define two independent mechanisms for DNA degradation (Fig. 12) (Mukae

134

Zhou et al.

Figure 12 Degradation of nuclear DNA of apoptotic cells is a multistep process in Drosophila. In wild-type animals, caspase-activated dCAD promotes the processing of nuclear DNA within apoptotic cells into fragments that can be detected by LM-PCR. DNaseII activity, likely provided by the engulfing cell, eventually degrades these DNA fragments into free nucleotides. In dCAD/ animals, however, DNaseII activity is suYcient to bypass the requirement for dCAD and provides an auxilliary mechanism of degrading the nuclear DNA of apoptotic cells.

et al., 2002). These observations are in agreement with the two-step model proposed in mice (McIlroy et al., 2000; Mukae et al., 2002). Comparisons made between the mechanism of apoptotic DNA degradation in C. elegans and Drosophila/mammalian systems led to the following notions. In both organisms, initial processing of DNA occurs through the action of a nuclease activity that functions in the dying cell: the CED-1, CED-7-dependent nuclease in worms and CAD in flies/mammals. In C. elegans, NUC-1, a DNaseII homolog, is required in the dying cell to process DNA intermediates to ensure the eYcient execution of downstream steps. In contrast, DNaseII activity has been demonstrated to function in engulfing cells in flies and mammals to direct the complete degradation of nuclear DNA from apoptotic cells, while in C. elegans, yet-unidentified nucleases accomplish this step. 3. Undegraded DNA Can Trigger an Immune Response Mukae and co-workers examined whether accumulation of undegraded DNA could activate innate immunity in flies (Mukae et al., 2002). dICAD deficiency had little or no impact on the expression of antibacterial and antifungal peptides compared to the expression seen in the wild-type. In


135

lo

contrast, the Dnase-1 mutation, which resulted in the accumulation of chromosomal DNA from apoptotic cells, resulted in the constitutive expression of antibacterial peptides diptericin and attacin but had no eVect on the expression of the antifungal peptide drosomycin (Mukae et al., 2002). The expression level of antibacterial peptides is enhanced in double mutant flies deficient for both dICAD and dDNAse II (Mukae et al., 2002). The specific upregulation of antibacterial peptides indicates that chromosomal DNA left undigested during apoptosis leads to the induction of innate immune response. Interestingly, mice deficient for both CAD and DNaseII suVer from a block in T cell development and display a strong upregulation of interferon- (Kawane et al., 2003). Kawane and co-workers (2003) hypothesized that DNA from apoptotic cells that remain undigested can activate innate immunity, leading to defects in thymic development.

V. Concluding Remarks Studies in C. elegans and Drosophila have contributed much to our understanding of how apoptotic cells are recognized, cleared, and degraded. Genetic analyses have made it possible to identify key components of the engulfment and degradation machinery, which include receptor molecules, downstream signaling components, regulators of polarized cell surface extension, and digestive enzymes. Accumulative evidence has shown that multiple emerging signaling pathways regulate the events of engulfment and degradation. Our future goal is to delineate these pathways by identifying and characterizing the functions of missing components. For example, it remains to be confirmed whether PS acts as one of the ‘‘eat me’’ signals to promote the recognition of apoptotic cells by engulfing cells. Whether there are any cell surface molecules that act as potential ‘‘eat me’’ signals other than PS remains unknown. One field of study that will associate with the identity of the ‘‘eat me’’ signal(s) is to identify the mechanisms that lead to the generation, presentation, and recognition of such signal(s) for the initiation of engulfment. The success of engulfment and degradation relies on the proper coordination of multiple cellular and subcellular events, including cytoskeletal reorganization, membrane remodeling, membrane fusion, the initiation of lysosome-phagosome fusion, and much more. A major question to answer is what molecules link the phagocytic receptor(s) to the downstream molecules responsible for each of these events. Current studies in C. elegans have revealed that CED-10, a member of the Rho-family small GTPases, acts to control cytoskeletal reorganization. However, this is only the tip of the iceberg. Identifying additional downstream components of the CED-1 and the CED-10 pathways using molecular genetic methods will lead to better understanding of what takes place inside the engulfing cells.

136

Zhou et al.

Furthermore, in C. elegans, two parallel, partially redundant signaling pathways (represented by ced-7, -1, -6 and ced-2, -5, -10, -12, respectively) act together to control cell corpse engulfment. It is important to know how these two pathways cooperate with each other and at what point they converge. Last but not least, the mechanism that leads to the degradation of apoptotic cells inside engulfing cells has so far been under-studied, and the only known protein acting in this process is DNaseII. However, this is an important process in mammals, as it is associated with antigen presentation, resolution of inflammatory responses, and development. Genetic screens specifically seeking mutants in which apoptotic cells are engulfed but not degraded, in both worms and flies, should help unravel the mechanism. While much still remains to be discovered about how these processes occur, past and current studies of the mechanisms behind the engulfment and degradation of apoptotic cells provide us a basic framework to be filled in. Given that many components of the engulfment machinery appear to be conserved during evolution, these invertebrate systems prove to be invaluable in vivo models and starting points for studying their mammalian counterparts. Although engulfment and subsequent degradation of apoptotic cells have traditionally been considered as downstream events to cell death, there is an emerging consensus that there is some cross-talk among these pathways. Unraveling the intricate relationship that exists among the execution of apoptosis, engulfment, and degradation will provide us, at least in the cellular sense, an understanding of the diVerence between life and death.

Acknowledgments We apologize to many authors whose papers are not cited here due to page limit. We thank Hillel Schwartz, Andreas Bergmann, Xiaohong Leng, and Kavita Oommen for critical reading of this manuscript. Z. Z. is supported by the National Institutes of Health (GM067848), the Cancer Research Institute, and the March of Dimes Birth Defects Foundation.

References Adachi, H., Tsujimoto, M., Arai, H., and Inoue, K. (1997). Expression cloning of a novel scavenger receptor from human endothelial cells. J. Biol. Chem. 272, 31217–31220. Albert, M. L., Kim, J. I., and Birge, R. B. (2000). v 5 integrin recruits the CrkIIDock180-Rac1 complex for phagocytosis of apoptotic cells. Nat. Cell Biol. 2, 899–905. Arur, S., Uche, U. E., Rezaul, K., Fong, M., Scranton, V., Cowan, A. E., Mohler, W., and Han, D. K. (2003). Annexin I is an endogenous ligand that mediates apoptotic cell engulfment. Dev. Cell 4, 587–598.


137

Basson, M., and Horvitz, H. R. (1996). The Caenorhabditis elegans gene sem-4 controls neuronal and mesodermal cell development and encodes a zinc finger protein. Genes Dev. 10, 1953–1965. Botto, M., Dell’Agnola, C., Bygrave, A. E., Thompson, E. M., Cook, H. T., Petry, F., Loos, M., Pandolfi, P. P., and Walport, M. J. (1998). Homozygous C1q deficiency causes glomerulonephritis associated with multiple apoptotic bodies. Nat. Genet. 19, 56–59. Braga, V. M. (2002). GEF without a Db1 domain? Nat. Cell Biol. 4, E188–190. Brugnera, E., Haney, L., Grimsley, C., Lu, M., Walk, S. F., Tosello-Trampont, A. C., Macara, I. G., Madhani, H., Fink, G. R., and Ravichandran, K. S. (2002). Unconventional Rac-GEF activity is mediated through the Dock180-ELMO complex. Nat. Cell Biol. 4, 574–582. Callebaut, I., Mignotte, V., Souchet, M., and Mornon, J. P. (2003). EMI domains are widespread and reveal the probable orthologs of the Caenorhabditis elegans CED-1 protein. Biochem. Biophys. Res. Commun. 300, 619–623. Campbell, I. D., and Bork, P. (1993). Epidermal growth factor-like modules. Curr. Opin. Struct. Biol. 3, 385–392. Cavaliere, V., Taddei, C., and Gargiulo, G. (1998). Apoptosis of nurse cells at the late stages of oogenesis of Drosophila melanogaster. Dev. Genes Evol. 208, 106–112. Chalfie, M., and Wolinsky, E. (1990). The identification and suppression of inherited neurodegeneration in Caenorhabditis elegans. Nature 345, 410–416. Chen, F., Hersh, B. M., Conradt, B., Zhou, Z., Riemer, D., Gruenbaum, Y., and Horvitz, H. R. (2000). Translocation of C. elegans CED-4 to nuclear membranes during programmed cell death. Science 287, 1485–1489. Chung, S., Gumienny, T. L., Hengartner, M. O., and Driscoll, M. (2000). A common set of engulfment genes mediates removal of both apoptotic and necrotic cell corpses in C. elegans. Nat. Cell Biol. 2, 931–937. Conradt, B. (2002). With a little help from your friends: Cells don’t die alone. Nat. Cell Biol. 4, E139–143. Conradt, B., and Horvitz, H. R. (1998). The C. elegans protein EGL-1 is required for programmed cell death and interacts with the Bcl-2-like protein CED-9. Cell 93, 519–529. Conradt, B., and Horvitz, H. R. (1999). The TRA-1A sex determination protein of C. elegans regulates sexually dimorphic cell deaths by repressing the egl-1 cell death activator gene. Cell 98, 317–327. Diez-Roux, G., and Lang, R. A. (1997). Macrophages induce apoptosis in normal cells in vivo. Development 124, 3633–3638. Driscoll, M., and Gerstbrein, B. (2003). Dying for a cause: Invertebrate genetics takes on human neurodegeneration. Nat. Rev. Genet. 4, 181–194. DuYeld, J. S., Erwig, L. P., Wei, X., Liew, F. Y., Rees, A. J., and Savill, J. S. (2000). Activated macrophages direct apoptosis and suppress mitosis of mesangial cells. J. Immunol. 164, 2110–2119. Ellis, H. M., and Horvitz, H. R. (1986). Genetic control of programmed cell death in the nematode C. elegans. Cell 44, 817–829. Ellis, R. E., Jacobson, D. M., and Horvitz, H. R. (1991a). Genes required for the engulfment of cell corpses during programmed cell death in Caenorhabditis elegans. Genetics 129, 79–94. Ellis, R. E., Yuan, J. Y., and Horvitz, H. R. (1991b). Mechanisms and functions of cell death. Annu. Rev. Cell Biol. 7, 663–698. Enari, M., Sakahira, H., Yokoyama, H., Okawa, K., Iwamatsu, A., and Nagata, S. (1998). A caspase-activated DNase that degrades DNA during apoptosis, and its inhibitor ICAD. Nature 391, 43–50. Enari, M., Talanian, R. V., Wong, W. W., and Nagata, S. (1996). Sequential activation of ICE-like and CPP32-like proteases during Fas-mediated apoptosis. Nature 380, 723–726.

138

Zhou et al.

Erickson, M. R., Galletta, B. J., and Abmayr, S. M. (1997). Drosophila myoblast city encodes a conserved protein that is essential for myoblast fusion, dorsal closure, and cytoskeletal organization. J. Cell Biol. 138, 589–603. Fadok, V. A., Bratton, D. L., Rose, D. M., Pearson, A., Ezekewitz, R. A., and Henson, P. M. (2000). A receptor for phosphatidylserine-specific clearance of apoptotic cells. Nature 405, 85–90. Ferguson, E. L., Sternberg, P. W., and Horvitz, H. R. (1987). A genetic pathway for the specification of the vulval cell lineages of Caenorhabditis elegans. Nature 326, 259–267. Franc, N. C., Dimarcq, J. L., Lagueux, M., HoVmann, J., and Ezekowitz, R. A. (1996). Croquemort, a novel Drosophila hemocyte/macrophage receptor that recognizes apoptotic cells. Immunity 4, 431–443. Franc, N. C., Heitzler, P., Ezekowitz, R. A., and White, K. (1999). Requirement for croquemort in phagocytosis of apoptotic cells in Drosophila. Science 284, 1991–1994. Gavrieli, Y., Sherman, Y., and Ben-Sasson, S. A. (1992). Identification of programmed cell death in situ via specific labeling of nuclear DNA fragmentation. J. Cell Biol. 119, 493–501. Gitai, Z., Yu, T. W., Lundquist, E. A., Tessier-Lavigne, M., and Bargmann, C. I. (2003). The netrin receptor UNC-40/DCC stimulates axon attraction and outgrowth through enabled and, in parallel, Rac and UNC-115/AbLIM. Neuron 37, 53–65. Golstein, P., Aubry, L., and Levraud, J. P. (2003). Cell-death alternative model organisms: Why and which? Nat. Rev. Mol. Cell Biol. 4, 798–807. Gumienny, T. L., Brugnera, E., Tosello-Trampont, A. C., Kinchen, J. M., Haney, L. B., Nishiwaki, K., Walk, S. F., Nemergut, M. E., Macara, I. G., Francis, R., Schedl, T., Qin, Y., Van Aelst, L., Hengartner, M. O., and Ravichandran, K. S. (2001). CED-12/ELMO, a novel member of the CrkII/Dock180/Rac pathway, is required for phagocytosis and cell migration. Cell 107, 27–41. Gumienny, T. L., Lambie, E., Hartwieg, E., Horvitz, H. R., and Hengartner, M. O. (1999). Genetic control of programmed cell death in the Caenorhabditis elegans hermaphrodite germline. Development 126, 1011–1022. Halenbeck, R., MacDonald, H., Roulston, A., Chen, T. T., Conroy, L., and Williams, L. T. (1998). CPAN, a human nuclease regulated by the caspase-sensitive inhibitor DFF45. Curr. Biol. 8, 537–540. Hall, A. (1998). Rho GTPases and the actin cytoskeleton. Science 279, 509–514. Hall, A., and Nobes, C. D. (2000). Rho GTPases: Molecular switches that control the organization and dynamics of the actin cytoskeleton. Philos. Trans. R. Soc. Lond. B Biol. Sci. 355, 965–970. Hall, D. H., Gu, G., Garcia-Anoveros, J., Gong, L., Chalfie, M., and Driscoll, M. (1997). Neuropathology of degenerative cell death in Caenorhabditis elegans. J. Neurosci. 17, 1033–1045. Hamon, Y., Broccardo, C., Chambenoit, O., Luciani, M. F., Toti, F., Chaslin, S., Freyssinet, J. M., Devaux, P. F., McNeish, J., Marguet, D., and Chimini, G. (2000). ABC1 promotes engulfment of apoptotic cells and transbilayer redistribution of phosphatidylserine. Nat. Cell Biol. 2, 399–406. Hanayama, R., Tanaka, M., Miwa, K., Shinohara, A., Iwamatsu, A., and Nagata, S. (2002). Identification of a factor that links apoptotic cells to phagocytes. Nature 417, 182–187. Harosh, I., Binninger, D. M., Harris, P. V., Mezzina, M., and Boyd, J. B. (1991). Mechanism of action of deoxyribonuclease II from human lymphoblasts. Eur. J. Biochem. 202, 479–484. Hedgecock, E. M., Sulston, J. E., and Thomson, J. N. (1983). Mutations aVecting programmed cell deaths in the nematode Caenorhabditis elegans. Science 220, 1277–1279. Henson, P. M., Bratton, D. L., and Fadok, V. A. (2001). Apoptotic cell removal. Curr. Biol. 11, R795–805.


139

Hevelone, J., and Hartman, P. S. (1988). An endonuclease from Caenorhabditis elegans: Partial purification and characterization. Biochem. Genet. 26, 447–461. Hoeppner, D. J., Hengartner, M. O., and Schnabel, R. (2001). Engulfment genes cooperate with ced-3 to promote cell death in Caenorhabditis elegans. Nature 412, 202–206. Horvitz, H. R. (2003). Worms, life, and death (Nobel lecture). Chembiochem. 4, 697–711. Hubbard, E. J., and Greenstein, D. (2000). The Caenorhabditis elegans gonad: A test tube for cell and developmental biology. Dev. Dyn. 218, 2–22. Hummel, T., Leifker, K., and Klambt, C. (2000). The Drosophila HEM-2/NAP1 homolog KETTE controls axonal pathfinding and cytoskeletal organization. Genes Dev. 14, 863–873. Jacobson, M. D., Weil, M., and RaV, M. C. (1997). Programmed cell death in animal development. Cell 88, 347–354. Joza, N., Susin, S. A., Daugas, E., Stanford, W. L., Cho, S. K., Li, C. Y., Sasaki, T., Elia, A. J., Cheng, H. Y., Ravagnan, L., et al. (2001). Essential role of the mitochondrial apoptosisinducing factor in programmed cell death. Nature 410, 549–554. Kawane, K., Fukuyama, H., Yoshida, H., Nagase, H., Ohsawa, Y., Uchiyama, Y., Okada, K., Iida, T., and Nagata, S. (2003). Impaired thymic development in mouse embryos deficient in apoptotic DNA degradation. Nat. Immunol. 4, 138–144. Kimble, J., and Hirsh, D. (1979). The postembryonic cell lineages of the hermaphrodite and male gonads in Caenorhabditis elegans. Dev. Biol. 70, 396–417. Kishore, R. S., and Sundaram, M. V. (2002). ced-10 Rac and mig-2 function redundantly and act with unc-73 Trio to control the orientation of vulval cell divisions and migrations in Caenorhabditis elegans. Dev. Biol. 241, 339–348. Kiyokawa, E., Hashimoto, Y., Kobayashi, S., Sugimura, H., Kurata, T., and Matsuda, M. (1998). Activation of Rac1 by a Crk SH3-binding protein, DOCK180. Genes Dev. 12, 3331–3336. Klein, I., Sarkadi, B., and Varadi, A. (1999). An inventory of the human ABC proteins. Biochim. Biophys. Acta 1461, 237–262. Kobayashi, K., Kuroda, S., Fukata, M., Nakamura, T., Nagase, T., Nomura, N., Matsuura, Y., Yoshida-Kubomura, N., Iwamatsu, A., and Kaibuchi, K. (1998). p140Sra-1 (specifically Rac1-associated protein) is a novel specific target for Rac1 small GTPase. J. Biol. Chem. 273, 291–295. Kodama, Y., Rothman, J. H., Sugimoto, A., and Yamamoto, M. (2002). The stem-loop binding protein CDL-1 is required for chromosome condensation, progression of cell death and morphogenesis in Caenorhabditis elegans. Development 129, 187–196. Krieger, M., and Herz, J. (1994). Structures and functions of multiligand lipoprotein receptors: Macrophage scavenger receptors and LDL receptor-related protein (LRP). Annu. Rev. Biochem. 63, 601–637. Kwiatkowska, K., and Sobota, A. (1999). Signaling pathways in phagocytosis. Bioessays 21, 422–431. Lagarkova, M. A., Iarovaia, O. V., and Razin, S. V. (1995). Large-scale fragmentation of mammalian DNA in the course of apoptosis proceeds via excision of chromosomal DNA loops and their oligomers. J. Biol. Chem. 270, 20239–20241. Liu, Q. A., and Hengartner, M. O. (1998). Candidate adaptor protein CED-6 promotes the engulfment of apoptotic cells in C. elegans. Cell 93, 961–972. Liu, Q. A., and Hengartner, M. O. (1999). Human CED-6 encodes a functional homologue of the Caenorhabditis elegans engulfment protein CED-6. Curr. Biol. 9, 1347–1350. Liu, X., Kim, C. N., Yang, J., Jemmerson, R., and Wang, X. (1996). Induction of apoptotic program in cell-free extracts: Requirement for dATP and cytochrome c. Cell 86, 147–157.

140

Zhou et al.

Liu, X., Zou, H., Slaughter, C., and Wang, X. (1997). DFF, a heterodimeric protein that functions downstream of caspase-3 to trigger DNA fragmentation during apoptosis. Cell 89, 175–184. LoeZer, M., Daugas, E., Susin, S. A., Zamzami, N., Metivier, D., Nieminen, A. L., Brothers, G., Penninger, J. M., and Kroemer, G. (2001). Dominant cell death induction by extramitochondrially targeted apoptosis-inducing factor. FASEB J. 15, 758–767. Lundquist, E. A., Reddien, P. W., Hartwieg, E., Horvitz, H. R., and Bargmann, C. I. (2001). Three C. elegans Rac proteins and several alternative Rac regulators control axon guidance, cell migration and apoptotic cell phagocytosis. Development 128, 4475–4488. McCall, K., and Steller, H. (1998). Requirement for DCP-1 caspase during Drosophila oogenesis. Science 279, 230–234. McIlroy, D., Tanaka, M., Sakahira, H., Fukuyama, H., Suzuki, M., Yamamura, K., Ohsawa, Y., Uchiyama, Y., and Nagata, S. (2000). An auxiliary mode of apoptotic DNA fragmentation provided by phagocytes. Genes Dev. 14, 549–558. Mergliano, J., and Minden, J. S. (2003). Caspase-independent cell engulfment mirrors cell death pattern in Drosophila embryos. Development 130, 5779–5789. Metzstein, M. M., Stanfield, G. M., and Horvitz, H. R. (1998). Genetics of programmed cell death in C. elegans: Past, present and future. Trends Genet. 14, 410–416. Mevorach, D., Mascarenhas, J. O., Gershov, D., and Elkon, K. B. (1998). Complementdependent clearance of apoptotic cells by human macrophages. J. Exp. Med. 188, 2313–2320. Mukae, N., Yokoyama, H., Yokokura, T., Sakoyama, Y., and Nagata, S. (2002). Activation of the innate immunity in Drosophila by endogenous chromosomal DNA that escaped apoptotic degradation. Genes Dev. 16, 2662–2671. Mukae, N., Yokoyama, H., Yokokura, T., Sakoyama, Y., Sakahira, H., and Nagata, S. (2000). Identification and developmental expression of inhibitor of caspase-activated DNase (ICAD) in Drosophila melanogaster. J. Biol. Chem. 275, 21402–21408. Nagase, T., Nakayama, M., Nakajima, D., Kikuno, R., and Ohara, O. (2001). Prediction of the coding sequences of unidentified human genes. XX. The complete sequences of 100 new cDNA clones from brain which code for large proteins in vitro. DNA Res. 8, 85–95. Nera, M. S., Vanderbeek, G., Johnson, R. O., Ruben, L. N., and Clothier, R. H. (2000). Phosphatidylserine expression on apoptotic lymphocytes of Xenopus laevis, the South African clawed toad, as a signal for macrophage recognition. Dev. Comp. Immunol. 24, 641–652. Nishiwaki, K. (1999). Mutations aVecting symmetrical migration of distal tip cells in Caenorhabditis elegans. Genetics 152, 985–997. Nolan, K. M., Barrett, K., Lu, Y., Hu, K. Q., Vincent, S., and Settleman, J. (1998). Myoblast city, the Drosophila homolog of DOCK180/CED-5, is required in a Rac signaling pathway utilized for multiple developmental processes. Genes Dev. 12, 3337–3342. Oberhammer, F., Wilson, J. W., Dive, C., Morris, I. D., Hickman, J. A., Wakeling, A. E., Walker, P. R., and Sikorska, M. (1993). Apoptotic death in epithelial cells: Cleavage of DNA to 300 and/or 50 kb fragments prior to or in the absence of internucleosomal fragmentation. EMBO J. 12, 3679–3684. Parrish, J., Li, L., Klotz, K., Ledwich, D., Wang, X., and Xue, D. (2001). Mitochondrial endonuclease G is important for apoptosis in C. elegans. Nature 412, 90–94. Parrish, J. Z., and Xue, D. (2003). Functional genomic analysis of apoptotic DNA degradation in C. elegans. Mol. Cell 11, 987–996. Parrish, J. Z., Yang, C., Shen, B., and Xue, D. (2003). CRN-1, a Caenorhabditis elegans FEN-1 homologue, cooperates with CPS-6/EndoG to promote apoptotic DNA degradation. EMBO J. 22, 3451–3460.


141

Pawson, T., and Scott, J. D. (1997). Signaling through scaVold, anchoring, and adaptor proteins. Science 278, 2075–2080. Pearson, A., Lux, A., and Krieger, M. (1995). Expression cloning of dSR-CI, a class C macrophage-specific scavenger receptor from Drosophila melanogaster. Proc. Natl. Acad. Sci. USA 92, 4056–4060. Platt, N., da Silva, R., and Gordon, S. (1998). Recognizing death: The phagocytosis of apoptotic cells. Trends Cell Biol. 8, 365–372. Reddien, P. W., Cameron, S., and Horvitz, H. R. (2001). Phagocytosis promotes programmed cell death in C. elegans. Nature 412, 198–202. Reddien, P. W., and Horvitz, H. R. (2000). CED-2/CrkII and CED-10/Rac control phagocytosis and cell migration in Caenorhabditis elegans. Nat. Cell Biol. 2, 131–136. Rehorn, K. P., Thelen, H., Michelson, A. M., and Reuter, R. (1996). A molecular aspect of hematopoiesis and endoderm development common to vertebrates and Drosophila. Development 122, 4023–4031. Ridley, A. J., and Hall, A. (1992). The small GTP-binding protein Rho regulates the assembly of focal adhesions and actin stress fibers in response to growth factors. Cell 70, 389–399. Ridley, A. J., Paterson, H. F., Johnston, C. L., Diekmann, D., and Hall, A. (1992). The small GTP-binding protein Rac regulates growth factor-induced membrane ruZing. Cell 70, 401–410. Robertson, A. M. G., and Thomson, J. N. (1982). Morphology of programmed cell death in the ventral nerve cord of Caenorhabditis elegans larvae. J. Embryol. Exp. Morph. 67, 89–100. Rodrigues, V., Cheah, P. Y., Ray, K., and Chia, W. (1995). malvolio, the Drosophila homologue of mouse NRAMP-1 (Bcg), is expressed in macrophages and in the nervous system and is required for normal taste behaviour. EMBO J. 14, 3007–3020. Sakahira, H., Enari, M., and Nagata, S. (1998). Cleavage of CAD inhibitor in CAD activation and DNA degradation during apoptosis. Nature 391, 96–99. Sakahira, H., Iwamatsu, A., and Nagata, S. (2000). Specific chaperone-like activity of inhibitor of caspase-activated DNase for caspase-activated DNase. J. Biol. Chem. 275, 8091–8096. Savill, J., and Fadok, V. (2000). Corpse clearance defines the meaning of cell death. Nature 407, 784–788. Savill, J. S., Fadok, V., Henson, P., and Haslett, C. (1993). Phagocyte recognition of cells undergoing apoptosis. Immunol. Today 14, 131–136. Savill, J., Hogg, N., Ren, Y., and Haslett, C. (1992). Thrombospondin cooperates with CD36 and the vitronectin receptor in macrophage recognition of neutrophils undergoing apoptosis. J. Clin. Invest. 90, 1513–1522. Schlegel, R. A., and Williamson, P. (2001). Phosphatidylserine, a death knell. Cell Death DiVer. 8, 551–563. Sears, H. C., Kennedy, C. J., and Garrity, P. A. (2003). Macrophage-mediated corpse engulfment is required for normal Drosophila CNS morphogenesis. Development 130, 3557–3565. Shaham, S., and Horvitz, H. R. (1996). Developing Caenorhabditis elegans neurons may contain both cell-death protective and killer activities. Genes Dev. 10, 578–591. Smits, E., Van Criekinge, W., Plaetinck, G., and Bogaert, T. (1999). The human homologue of Caenorhabditis elegans CED-6 specifically promotes phagocytosis of apoptotic cells. Curr. Biol. 9, 1351–1354. Song, Z., and Steller, H. (1999). Death by design: Mechanism and control of apoptosis. Trends Cell Biol. 9, M49–52. Songyang, Z., and Cantley, L. C. (1995). Recognition and specificity in protein tyrosine kinasemediated signalling. Trends Biochem. Sci. 20, 470–475.

142

Zhou et al.

Sonnenfeld, M. J., and Jacobs, J. R. (1995). Macrophages and glia participate in the removal of apoptotic neurons from the Drosophila embryonic nervous system. J. Comp. Neurol. 359, 644–652. Soto, M. C., Qadota, H., Kasuya, K., Inoue, M., Tsuboi, D., Mello, C. C., and Kaibuchi, K. (2002). The GEX-2 and GEX-3 proteins are required for tissue morphogenesis and cell migrations in C. elegans. Genes Dev. 16, 620–632. Staley, K., Blaschke, A., and Chun, J. (1997). Apoptotic DNA fragmentation is detected by a semi-quantitative ligation-mediated PCR of blunt DNA ends. Cell Death DiVer. 4, 66–75. Stanfield, G. M., and Horvitz, H. R. (2000). The ced-8 gene controls the timing of programmed cell deaths in C. elegans. Mol. Cell 5, 423–433. Su, H. P., Brugnera, E., Van Criekinge, W., Smits, E., Hengartner, M., Bogaert, T., and Ravichandran, K. S. (2000). Identification and characterization of a dimerization domain in CED-6, an adapter protein involved in engulfment of apoptotic cells. J. Biol. Chem. 275, 9542–9549. Su, H. P., Nakada-Tsukui, K., Tosello-Trampont, A. C., Li, Y., Bu, G., Henson, P. M., and Ravichandran, K. S. (2002). Interaction of CED-6/GULP, an adapter protein involved in engulfment of apoptotic cells with CED-1 and CD91/low density lipoprotein receptor-related protein (LRP). J. Biol. Chem. 277, 11772–11779. Sulston, J. E. (1976). Post-embryonic development in the ventral cord of Caenorhabditis elegans. Philos. Trans. R. Soc. Lond. B Biol. Sci. 275, 287–297. Sulston, J. E., Albertson, D. G., and Thomson, J. N. (1980). The Caenorhabitis elegans male: Postembryonic development of nongonadal structures. Dev. Biol. 78, 542–576. Sulston, J. E., and Horvitz, H. R. (1977). Post-embryonic cell lineages of the nematode, Caenorhabditis elegans. Dev. Biol. 56, 110–156. Sulston, J. E., Schierenberg, E., White, J. G., and Thomson, N. (1983). The embryonic cell lineage of the nematode Caenorhabditis elegans. Dev. Biol. 100, 64–119. Susin, S. A., Lorenzo, H. K., Zamzami, N., Marzo, I., Snow, B. E., Brothers, G. M., Mangion, J., Jacotot, E., Costantini, P., LoeZer, M., et al. (1999). Molecular characterization of mitochondrial apoptosis-inducing factor. Nature 397, 441–446. Tepass, U., Fessler, L. I., Aziz, A., and Hartenstein, V. (1994). Embryonic origin of hemocytes and their relationship to cell death in Drosophila. Development 120, 1829–1837. Thellmann, M., Hatzold, J., and Conradt, B. (2003). The Snail-like CES-1 protein of C. elegans can block the expression of the BH3-only cell-death activator gene egl-1 by antagonizing the function of bHLH proteins. Development 130, 4057–4071. Van Aelst, L., and D’Souza-Schorey, C. (1997). Rho GTPases and signaling networks. Genes Dev. 11, 2295–2322. van den Eijnde, S. M., Boshart, L., Baehrecke, E. H., De Zeeuw, C. I., Reutelingsperger, C. P. M., and Vermeij-Keers, C. (1998). Cell surface exposure of phosphatidylserine during apoptosis is phylogenetically conserved. Apoptosis 3, 9–16. Wang, X., Yang, C., Chai, J., Shi, Y., and Xue, D. (2002). Mechanisms of AIF-mediated apoptotic DNA degradation in Caenorhabditis elegans. Science 298, 1587–1592. White, K., Grether, M. E., Abrams, J. M., Young, L., Farrell, K., and Steller, H. (1994). Genetic control of programmed cell death in Drosophila. Science 264, 677–683. WolV, T., and Ready, D. F. (1991). Cell death in normal and rough eye mutants of Drosophila. Development 113, 825–839. Wu, Y. C., Cheng, T. W., Lee, M. C., and Weng, N. Y. (2002). Distinct Rac activation pathways control Caenorhabditis elegans cell migration and axon outgrowth. Dev. Biol. 250, 145–155. Wu, Y., and Horvitz, H. R. (1998a). The C. elegans cell corpse engulfment gene ced-7 encodes a protein similar to ABC transporters. Cell 93, 951–960.


143

Wu, Y., and Horvitz, H. R. (1998b). C. elegans phagocytosis and cell-migration protein CED-5 is similar to human DOCK180. Nature 392, 501–504. Wu, Y. C., Stanfield, G. M., and Horvitz, H. R. (2000). NUC-1, a Caenorhabditis elegans DNase II homolog, functions in an intermediate step of DNA degradation during apoptosis. Genes Dev. 14, 536–548. Wu, Y. C., Tsai, M. C., Cheng, L. C., Chou, C. J., and Weng, N. Y. (2001). C. elegans CED-12 acts in the conserved CrkII/DOCK180/Rac pathway to control cell migration and cell corpse engulfment. Dev. Cell 1, 491–502. Wyllie, A. H. (1980). Glucocorticoid-induced thymocyte apoptosis is associated with endogenous endonuclease activation. Nature 284, 555–556. Wyllie, A. H., Kerr, J. F., and Currie, A. R. (1980). Cell death: The significance of apoptosis. Int. Rev. Cytol. 68, 251–306. Yokoyama, H., Mukae, N., Sakahira, H., Okawa, K., Iwamatsu, A., and Nagata, S. (2000). A novel activation mechanism of caspase-activated DNase from Drosophila melanogaster. J. Biol. Chem. 275, 12978–12986. Yuan, J. Y., and Horvitz, H. R. (1990). The Caenorhabditis elegans genes ced-3 and ced-4 act cell autonomously to cause programmed cell death. Dev. Biol. 138, 33–41. Zhou, Z., Caron, E., Hartwieg, E., Hall, A., and Horvitz, H. R. (2001a). The C. elegans PH domain protein CED-12 regulates cytoskeletal reorganization via a Rho/Rac GTPase signaling pathway. Dev. Cell 1, 477–489. Zhou, Z., Hartwieg, E., and Horvitz, H. R. (2001b). CED-1 is a transmembrane receptor that mediates cell corpse engulfment in C. elegans. Cell 104, 43–56.

5

____________________________________________________________________________

Beginning and Ending an Actin Filament: Control at the Barbed End Sally H. Zigmond Biology Department University of Pennsylvania Philadelphia, Pennsylvania 19104

I. Introduction A. Actin Dynamics Depends on Regulated Availability of the Barbed End of Actin Filaments II. Locally Create a Free Barbed End A. De Novo Nucleation B. Cutting: ADF/Cofilin, Formins C. Uncapping III. Enhance Barbed-End Elongation IV. Capping Barbed Ends A. Barbed-End Capping Proteins B. Regulation of Barbed-End Capping V. Replenish G-Actin Pool by Pointed-End Depolymerization of F-Actin VI. New Directions References

Dynamic actin filaments contribute to cell migration, organelle movements, memory, and gene regulation. These dynamic processes are often regulated by extracellular and/or cell cycle signals. Regulation targets, not actin itself, but the factors that determine it’s dynamic properties. Thus, filament nucleation, rate and duration of elongation, and depolymerization are each controlled with regard to time and/or space. Two mechanisms exist for nucleating filaments de novo, the Arp23 complex and the formins; multiple pathways regulate each. A new filament elongates rapidly but transiently before its barbed end is capped. Rapid capping allows the cell to maintain fine temporal and spatial control over F-actin distribution. Modulation of capping protein activity and its access to barbed ends is emerging as a site of local regulation. Finally, to maintain a steady state filaments must depolymerize. Depolymerization can limit the rate of new filament nucleation and elongation. The activity of ADF/cofilin, which facilitates depolymerization, is also regulated by multiple inputs. This chapter describes (1) mechanism and regulation of new filament formation, (2) mechanism of enhancing elongation at barbed ends, (3) capping proteins Current Topics in Developmental Biology, Vol. 63 Copyright 2004, Elsevier Inc. All rights reserved. 0070-2153/04 $35.00

145

146

Sally H. Zigmond

and their regulators, and (4) recycling of actin monomers from filamentous actin (F-actin) back to globular actin (G-actin). C 2004, Elsevier Inc.

I. Introduction A. Actin Dynamics Depends on Regulated Availability of the Barbed Ends of Actin Filaments Dynamic actin filaments contribute to cell migration, cytokinesis, subcellular movements including endocytosis and secretion, memory, and gene regulation. Often the actin filaments are turning over rapidly: filaments in the leading lamella have a half-life of 30 s (Cassimeris et al., 1990), those in stress fibers and in contractile ring, a half life of 5 min (Turnacioglu et al., 1998). Temporal and spatial control of this turnover focuses on two key points. Regulation and nucleation of new filaments determines the temporal and spatial distribution of filamentous actin. Regulation of depolymerization controls steady state dynamics. 1. In Vivo, Actin Filament Elongation Occurs at the Barbed End Many accessory proteins involved in the regulation modulate properties that are intrinsic to the actin itself. Actin filaments are polar structures because each monomer that assembles in a helix to form the filament is oriented in the same direction along the axis of the filament. Actin domains 1 and 3 face one end (the plus or ‘‘barbed’’ end) and domains 2 and 4 face the other (the minus or pointed end). The resultant filament polarity determines the direction of myosin-mediated movement: most myosins move toward the barbed end, the end anchored in the Z-line of muscle, while myosin type VI moves (slowly) toward the pointed end. The two ends of the filament, being diVerent, also exhibit diVerent kinetics of binding an actin monomer (G-actin) (Pantaloni et al., 2001). The monomer binds to the barbed end (the on rate) about 10 times faster than to the pointed end. This diVerence, when ATP is present, results in a situation where the aYnity of the barbed end for actin monomer (the barbed end critical concentration) is about 10 times higher than for the pointed end, Kd-barbed 0.5 M; Kd-pointed 5 M (for a clear description of these issues see Pollard and Borisy 2003; Weber 1999). Since both ends of a given filament are exposed to the same pool of G-actin, this diVerence in aYnity causes the actin filament, at steady state in ATP, to exhibit treadmilling: a given filament undergoes net elongation at its barbed end and net depolymerization at its pointed end.

5. Control at the Barbed End

147

Treadmilling of an actin filament on its own is very slow, but factors in cytoplasm can dramatically increase the rate. Thus, actin bound to profilin, which is present at much higher concentrations than free G-actin, can be added to the barbed end (but not the pointed end) about as fast as free G-actin. The actin recycling from the filament back to G-actin is enhanced greatly by the presence of ADF/cofilin (Carlier et al., 1997). Because the surface of actin at the two ends of the filament is diVerent, proteins can bind selectively to one or the other end. A number of barbed-end binding proteins (Table V) and two pointed-end cappers, tropomodulin and the Arp2/3 complex, are known. When a protein binds to the barbed end and prevents addition of monomeric actin, it stops elongation. If monomers cannot be added to the high-aYnity barbed end, the concentration of free G-actin increases, rising from that of the critical concentration of the whole filament toward that of the pointed end. Although small, this diVerence between the critical concentration of the barbed and the pointed ends, when amplified by the various accessory proteins, controls the extent and location of polymerization. In many cells, there is also a reservoir of G-actin bound to a protein such as thymosin 4 (T4) that allows a change in the availability of barbed ends to cause a large change in F-actin. T4 binds ATP-G-actin with a Kd of 0.6 M (Weber, 1999). Thus, when barbed ends are capped and the free G-actin is close to 0.5 M, about half of the T4 will be bound with a G-actin. Yet when barbed ends are free, they consume G-actin, causing the concentration of free G-actin to decrease toward its critical concentration (0.05 M). As the free G-actin decreases, it is replenished by G-actin dissociating from T4. Thus, T4 is not regulated and T4-G-actin serves as a simple reservoir that buVers the free G-actin. The filaments still determine the steady state critical concentration of free G-actin because when the G-actin is above this concentration, filaments will polymerize until the free G-actin reaches the critical concentration. Similarly, if the free G-actin is below the critical concentration, filaments will depolymerize until the critical concentration is achieved. Many qualitative and quantitative features of actin dynamics in a living cell can be observed in neutrophilic leukocytes. After a sudden increase in chemoattractant, the leukocyte F-actin rapidly doubles. This increase is blocked by addition of the drug cytochalasin that binds, or ‘‘caps,’’ barbed ends (Cassimeris et al., 1990). Addition of cytochalasin returns the F-actin to its basal level. This suggests that basal level is determined by the concentration of free G-actin being at the critical concentration of the pointed end, and polymerization depends on free barbed ends. Consistent with this, addition of cytochalasin to unstimulated neutrophils does not change the F-actin level. Yet a drug that sequesters G-actin (latrunculin) can decrease the resting F-actin level, demonstrating that the filaments are able to depolymerize. (Pring et al., 2002). During neutrophil migration, actin is cycling rapidly between filaments and G-actin, with the half-life of a filament in the leading edge being only about 3 s

148

Sally H. Zigmond

(Cassimeris et al., 1990). Although barbed ends are free, the F-actin level is close to that of a resting cell. Presumably, the enhanced rate of depolymerization maintains the critical concentration of the cytoplasm close to that of the pointed end. Indeed, keeping the free G-actin concentration close to that of the critical concentration of the pointed end keeps the pools of profilin-actin high so they can support the rapid rate of elongation required to support neutrophil migration that can occur at 30 m/min. 2. Production of New Filaments Increases the Rate of Actin Turnover The rate of F-actin turnover in most cellular structures is greater than can be explained by treadmilling. Turnover detected by labeled actin incorporation or by photobleaching results from two diVerent processes: treadmilling within an individual filament and turnover of filaments, i.e., creation of new filaments and total loss of existing filaments. The doubling of neutrophil F-actin upon stimulation also involves a doubling in the number of filaments present (Cano et al., 1991). Even in quite stable actin networks, e.g., stress fibers and the contractile ring, the subunits of actin filaments are turning over rapidly (Noguchi and Mabuchi, 2001; Pelham and Chang, 2002). In S. pombe, the formin Cdc12 remains essential for ring maintenance and actin incorporation even after the contractile ring has formed (Pelham and Chang, 2002). Because Cdc12 can create new filaments, this suggests that continual formation of new filaments is occurring. New filaments can be produced by de novo nucleation, i.e., from G-actin, when special nucleating factors such as Arp2/3 or formin proteins are present and active. New filaments are also produced through cutting of existing filaments. If a barbed end were created and allowed to elongate continuously, the cell would quickly lose temporal and spatial control over polymerization. To prevent this, a barbed end remains free only transiently (

Current Topics in Developmental Biology, Volume 63 (Current Topics in Developmental Biology)

Current Topics Developmental Biology (Current Topics in Developmental Biology)

Current Topics Developmental Biology (Current Topics in Developmental Biology)