Open Collections

UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Structural and functional characterization of the DAXX N-terminal helix bundle and SUMO interaction motifs Escobar-Cabrera, Eric 2010

Your browser doesn't seem to have a PDF viewer, please download the PDF to view this item.

Notice for Google Chrome users:
If you are having trouble viewing or searching the PDF with Google Chrome, please download it here instead.

Item Metadata


24-ubc_spring_2010_escobar-cabrera_eric.PDF [ 7.26MB ]
JSON: 24-1.0060528.json
JSON-LD: 24-1.0060528-ld.json
RDF/XML (Pretty): 24-1.0060528-rdf.xml
RDF/JSON: 24-1.0060528-rdf.json
Turtle: 24-1.0060528-turtle.txt
N-Triples: 24-1.0060528-rdf-ntriples.txt
Original Record: 24-1.0060528-source.json
Full Text

Full Text

STRUCTURAL AND FUNCTIONAL CHARACTERIZATION OF THE DAXX N-TERMINAL HELIX BUNDLE AND SUMO INTERACTION MOTIFS by ERIC ESCOBAR-CABRERA B.Sc., Trent University, 2002 A THESIS SUBMITTED 1N PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY in The Faculty of Graduate Studies (Chemistry) THE UNIVERSITY OF BRITISH COLUMBIA (Vancouver) April 2010 © Eric Escobar-Cabrera, 2010 Abstract The Fas death-domain associated (DAXX) protein was first discovered as an intermediary of a FADD-independent apoptosis signaling pathway. However, subsequent studies have established it as an important player in both transcription and cell cycle regulation. In this thesis, the first structural characterization of DAXX is presented. Sequence alignment and secondary structure prediction algorithms were used to define a number of constructs of DAXX. The C-terminal -l/3 of DAXX was found to be intrinsically disordered, whereas a well- defined folded domain was identified near its N-terminus. NMR spectroscopy was used to solve the three-dimensional structure of this domain, and to characterize its dynamic behavior. The calculated structural ensemble consists of five helices, and hence is named the DAXX Helix Bundle (DHB) domain. This domain has a very different topology to the Sin3 PAH domains, which until now have been used as a model for this region of DAXX. Rassfl C, an important tumor suppressor, was reported recently to interact with DAXX via the DHB domain. This interaction was linked to mitosis progression, and has potential implications in the treatment of cancer. The NMR-derived ensemble of the complex of the DHB domain with an N-terminal fragment of Rassf1 C revealed a short amphipathic a-helix filling the cleft between helices 2 and 5 of the DHB domain. Both hydrophobic and electrostatic interactions mediated complex formation. This structural characterization explains the observed in vivo interaction and provides clues as to how the binding might be regulated. Additionally, two SUMO-interacting motifs at the termini of DAXX, SIM-N and SIM-C, were characterized. Their interactions with SUMO- 1 and SUMO-2 were examined structurally 11 and thermodynamically using NMR spectroscopy. SIM-N was also found to bind intramolecularly to the DHB domain and SIM-C to mediate the interaction of DAXX with the sumoylated Ets- 1 transcription factor. Importantly, the latter did not involve any direct contacts between DAXX and Ets- 1, but rather derived from the non-covalent binding of DAXX SIM-C to SUMO- 1, which in turn was covalently linked in a “beads-on-a-string” fashion to Ets- 1. These results provide new insights into the binding mechanisms and biological roles of DAXX-SUMO interactions. 111 Table of Contents ABSTRACT.ii TABLE OF CONTENTS iv LIST OF TABLES x LIST OF FIGURES xi LIST OF ABBREVIATIONS xiv LIST OF AMINO ACID ABBREVIATIONS xvii ACKNOWLEDGMENTS xviii CHAPTER 1- INTRODUCTION 1 1.1 PREAMBLE 2 1.1.1 The human genome 2 1 .1 .2 Signal transduction 2 1.1.3 Modularproteins 3 1.1.4 Intrinsically disordered proteins 3 1.1.5 DAXX, a multi-domain protein with intrinsically disordered regions 4 1.2 DAXX 5 1.2.1 What is DAXX 5 1.2.2 The mixed roles of DAXX in apoptosis 6 1.2.3 Where is DAXX9 7 DAXX in PML nuclear bodies 10 DAXX in other nuclear compartments 11 Cytoplasmic DAXX? 12 1.2.4 DAXX as a transcription regulator 13 1.2.5 The role of DAXX in mitosis 15 1.3 THE SMALL UBIQUITIN-LIKE MODIFIER PROTEIN SUMO 16 iv 1.3.1 TheSUMOfamily .17 1.3.2 The mechanism of reversible sumoylation 17 1.3.3 Molecular consequences of sumoylation 20 1.3.4 Differences among SUMO paralogues 23 1.3.5 SUMO interacting motifs 24 1.3.6 DAXX and SUMO 25 1.4 THE RAS-ASSOCIATION DOMAIN FAMILY (RAssF) 27 1.4.1 Domain structure of Rassf 27 1 .4.2 Rassfl and cancer 31 1 .4.3 Rassfl and DAXX 32 1.5 THESIS OVERVIEW 35 CHAPTER 2- STRUCTURE OF THE N-TERMINAL DAXX HELIX BUNDLE 37 2.1 DAXX SEQUENCE AND SECONDARY STRUCTURE 38 2.2 CHARACTERIZATION OF DAXX CONSTRUCTS 42 2.3 MAPPING OF THE DAXX N-TERMINAL STRUCTURAL DOMAIN BOUNDARIES 46 2.3.1 Chemical shift assignments 47 2.3.2 Identification of the folded domain 49 DAXX4616°secondary structure 49 ‘5N DAXX4616°relaxation analysis 50 Comparison of various DAXX constructs 50 2.4 DAXX5544 STRUCTURE DETERMINATION 51 2.4.1 Assignments of side-chain signals 51 Assignment of signals from side-chain aliphatic nuclei 54 Assignment of signals from side-chain aromatic nuclei 57 2.4.2 DAXX55144 secondary structure 61 2.4.3 DAXX55144 tertiary structure 62 V 2.4.4 DAXX55144 dynamics .68 2.5 STRUCTURAL COMPARISONS OF THE DHB DOMAIN 74 2.5.1 Comparison with Sin3 74 2.5.2 Comparison with other helix bundles 78 2.6 MATERIALS AND METHODS 81 2.6.1 Cloning 81 2.6.2 Protein expression tests 82 2.6.3 Protein purification 83 2.6.4 NMR spectroscopy 83 NMR relaxation measurements 84 Amide proton-deuterium exchange 84 2.6.5 Structure calculation 84 2.6.6 lHNl5N Residual dipolar coupling measurements 85 CHAPTER 3- STRUCTURE OF THE DAXXJRASSF1C COMPLEX 88 3.1 IDENTIFICATION OF RASSF 1 C AS AN INTERACTION PARTNER OF THE DHB DOMAIN 89 3.2 NMR-M0NITORED TITRATION OF RASSF1C’5°AND DAXX55’44 92 3.3 NMR-M0NITORED TITRATIONS OF TRUNCATION FRAGMENTS OF RASSF1C WITH DAXX55144 94 3.3.1 RassflC838 94 3.3.2 RassflC82’ 95 3.3.3 Rassf1C2338’’ 95 3.4 SCREEN OF CONDITIONS FOR THE FORMATION OF A TIGHTER COMPLEX 103 3.5 DETERMINATION OF THE TERTIARY STRUCTURE OF THE COMPLEX 105 3.5.1 Changes in DAXX55144 secondary structure upon binding 105 3.5.2 Changes in Rassf1C2338vsecondary structure upon binding 106 3.5.3 Tertiary structure of the DAXX55144/RassflC2338”complex 109 3.6 STRUCTURAL IMPLICATIONS 115 vi 3.6.1 Structural verification that the DAXX DHB and Sin PAH domains differ in binding mechanisms 115 3.6.2 Other structural relatives share helix packing patterns 118 3.6.3 Other binding partners of DAXX 118 3.6.4 Possible new DIM-containing interacting partners of DAXX 122 3.7 BIoLoGIcAL IMPLICATIONS OF THE RASSF1C-DAXX INTERACTION 124 3.8 MATERIALS AND METHODS 126 3.8.1 Cloning 126 3.8.2 Sample purification 126 RassflC15° 127 Rassf1C2344Nand RassflC235° 128 3.8.3 NMR-rnonitored titrations 128 3.8.4 Structure of the complex 129 Sample preparation 129 Resonance assignments 132 Structure calculation 133 CHAPTER 4- SUMO-INTERACTING MOTIFS (SIM) IN DAXX 134 4.1 DAXX SUMO-INTERACTING MOTIFS (SIM-N AND SIM-C) 135 4.1.1 Identification of SIM-N and SIM-C 135 4.1.2 Role of DAXX SIMs in its localization 136 4.1.3 Research objectives 136 4.2 NMR-M0NIT0RED TITRATIONS OF DAXX-SIM-N AND SIM-C WITH SUMO-1 AND SUMO-2 138 4.2.1 Preamble. SIM binding by SUMO-1 and SUMO-2 138 4.2.2 NMR characterization of SIM-N, SIM-C, SUMO-1 and SUMO-2 142 4.2.3 NMR-monitored interactions 142 Mapping of the binding interface 143 vii SIM-N binding to SUMO-1 and SUMO-2 .143 SIM-C binding to SUMO-1 and SUMO-2 146 4.3 COMPARISON WITH VALUES FROM THE LITERATURE 149 4.4 Two ORIENTATION BINDING OF DAXX SIM-C TO SUMO-1 151 4.5 DETERMINATION OF THE BINDING MODE OF DAXX SIM-N WITH SUMO-1 155 4.6 INTRAMOLECULAR BINDING OF SIM-N TO THE DHB DOMAIN 160 4.7 CHARACTERIZATION OF THE INTERACTION BETWEEN SUMOYLATED ETs-1 AND DAXX STM-C 166 4.8 SUMMARY AND SIGNIFICANCE 170 4.9 MATERIALS AND METHODS 171 4.9.1 Constructs 171 4.9.2 Protein expression 172 4.9.3 Protein purification 172 4.9.4 NMR spectral assignments 174 4.9.5 NMRtitrations 174 4.9.6 Dissociation constants 176 4.9.7 Spin labeling 176 4.9.8 Paramagnetic relaxation enhancement measurements 176 CHAPTER 5- CONCLUDING REMARKS AND FUTURE EXPERIMENTS 178 5.1 STRUCTURE0FDAXX 179 5.2 SIGNIFICANCE FOR FUTURE TRUNCATION CONSTRUCTS OF DAXX 182 5.3 PROPOSED MECHANISM FOR ACTIVATION OF THE DHB DOMAIN 184 5.4 FUTURE EXPERIMENTS IN DEFINING DHB DOMAIN INTERACTING MOTIFS (DIMS) 189 5.5 DAXX SIMS 190 5.6 REMAINING QUESTIONS ABOUT THE DAXX/RASSF1C INTERACTION 191 5.7 CONCLUDING REMARKS 192 REFERENCES 194 Vlll APPENDICES .214 APPENDIX 1 CIRCULAR DICHROISM (CD) SPECTRA OF DAXX34742° 215 APPENDIX 2 EXPREssIoN TESTS OF DAXX CONSTRUCTS 216 APPENDIX 3 ANNOTATED 1H 15N HSQC SPECTRA 221 APPENDIX 4 CALCULATION OF DISSOCIATION CONSTANTS IN NMR-M0NIT0RED TITRATIONS224 APPENDIX 5 DISSOCIATION CONSTANT PLOTS FROM SIM/SUMO NMR-.M0NITORED TITRATIONS 228 APPENDIX 6 EXCHANGE REGIMES 231 APPENDIX 7 ASSIGNMENT OF THE DHB-D0MAIN 238 APPENDIX 8 ASSIGNMENT OF RASSF1C2338’’IN THE DHB/RASSF1C2338WCOMPLEX 245 APPENDIX 9 ASSIGNMENT OF THE DHB DOMAIN IN THE DHB/RAsSF1C2338’’COMPLEX 247 APPENDIX 10 TRIANGULATION OF THE SPIN LABEL 253 ix List of Tables TABLE 2.1 - NMR RESTRAINTS AND STATISTICS FOR THE ENSEMBLE OF THE TEN LOWEST ENERGY STRUCTURES CALCULATED FOR DAXX55144 65 TABLE 2.2 - SUMMARY OF AMIDE RELAXATION DATA FOR DIFFERENT STRUCTURAL REGIONS OF THE DHB DOMAIN 70 TABLE 2.3 - INTERHELICAL ANGLE (Q)IN SIN3A PAH2 (IG1E.PDB) AND THE DHB DOMAIN CALCULATED WITH PROMOTIF 75 TABLE 3.1 - DISSOCIATION CONSTANTS OF DAXX55’44AND VARIOUS RASSF1C CONSTRUCTS UNDER DIFFERENT CONDITIONS 104 TABLE 3.2 - NMR RESTRAINTS AND STATISTICS FOR THE ENSEMBLE OF THE TEN LOWEST ENERGY STRUCTURES CALCULATED FOR THEDAXX5544/R SSF1C2338vCOMPLEX 110 TABLE 3.3 - PROTEINS PREVIOUSLY REPORTED IN THE LITERATURE TO INTERACT WITH THE N- TERMINUS OF DAXX 120 TABLE 3.4- BLAST SEARCH FOR SEQUENCES SIMILAR TO RASSF1C2338 123 TABLE 4.1 - DISSOCIATION CONSTANTS BETWEEN SIM-N” AND SIM-C” WITH SUMO-1’’ AND SUMO-2’ 149 TABLE 4.2 - SUMMARY OF REPORTED DISSOCIATION CONSTANTS OF SIMs/SUMO 150 TABLE 4.3 - INTERMOLECULAR NOES BETWEEN UNLABELED SIM-N AND 13C/’5NLABELED SUMO-I 156 TABLE 4.4 - SAMPLE CONDITIONS USED FOR TITRATIONS PRESENTED IN CHAPTER 3 175 x List of Figures FIGuRE 1.1 - FuNcTIoN AND LOCALIZATION OF DAXX .8 FIGuRE 1.2 - GENERIC MODELS FOR TRANSCRIPTIONAL REPRESSION IN EUKARYOTIC SYSTEMS 14 FIGURE 1.3 - SUMOYLATION MECHANISM 19 FIGURE 1.4 - MoLECuLAR CONSEQUENCES OF SUMOYLATION 22 FIGURE 1.5 - THE RAS-ASSOCIATION DOMAIN FAMILY (RASSF) AND THEIR DOMAINS 29 FIGURE 1.6- PROPOSED MODELS FOR MECHANISMS CONTROLLING DAXX/RASSF INTERACTIONS 33 FIGURE 2.1 - SUGGESTED “DOMAIN” ORGANIZATION OF DAXX 39 FIGURE 2.2 - ALIGNMENT OF THE DAXX SEQUENCE 41 FIGURE 2.3 - IDENTIFICATION OF STRUCTURED DOMAINS IN DAXX THROUGH DELETION ANALYSES 44 FIGURE 2.4 -1H-’5NHSQC SPECTRA OF SOLUBLE CONSTRUCTS OF DAXX 45 FIGURE 2.5 - MAGNETIZATION TRANSFER PATHWAYS FOR EXPERIMENTS USED FOR BACKBONE ASSIGNMENT 48 FIGuRE 2.6 - IDENTIFICATION OF THE BOUNDARIES OF THE STRUCTURAL DOMAIN WITHIN DAXX4616° 52 FIGURE 2.7 - ANNOTATED ‘H-’5NHSQC SPECTRUM OF DAXX55144 53 FIGURE 2.8 - ExPERIMENTS USED FOR THE ASSIGNMENT OF NMR SIGNALS FROM THE SIDE-CHAIN ALIPHATIC ‘3C AND 1H NUCLEI OF UNIFORMLY‘3C/15N-LABELED DAXX55144 55 FIGURE 2.9 - STEPS INVOLVED IN THE ASSIGNMENT OF SIGNALS FROM THE AROMATIC SIDE-CHAINS OF TYR AND PHE RESIDUES IN UNIFORMLY‘3C/’5N-LABELED DAXX5544 59 FIGURE 2.10 - ASSIGNMENT OF SIGNALS FROM HISTIDINE RESIDUES 60 FIGuRE 2.11 - NMR-DERIvED SECONDARY STRUCTURE OF DAXX55144 63 FIGuRE 2.12 - STRUCTURE OF THE DHB DOMAIN 66 FIGURE 2.13 - THE GLOBAL AND LOCAL BACKBONE DYNAMICS OF THE DHB DOMAIN 72 FIGURE 2.14- COMPARISON OF THE DHB DOMAIN WITH THE SIN3 PAH DOMAINS 76 FIGURE 2.15 - STRUCTURAL COMPARISON OF THE DHB DOMAIN WITH SIMILAR HELIX BUNDLES 80 FIGURE 2.16 - USE OF RDC VALUES IN THE STRUCTURAL CALCULATION OF THE DHB DOMAIN 87 xi FIGuRE 3.1 - RAssF Ic1334, BUT NOT AxIN576598,BOUND THE DHB DOMAIN 91 FIGURE 3.2 - DETERMINATION OF THE REGION OF RAssF1C’5°THAT MEDIATED THE INTERACTION WITH DAXX5544 93 FIGURE 3.3 - IDENTIFICATION OF THE RAssF1C838 RESIDUES THAT INTERACTED WITH DAXX55’44 96 FIGuRE 3.4 - ASSFIC233Sw, BUT NOT RASSF1C82,BOUND DAXX55’44 97 FIGURE 3.5 - ELECTROSTATIC INTERACTIONS CONTRIBUTED TO THE BINDING OF RASSF1C2338’’AND DAXX55144 100 FIGuRE 3.6 - IDENTIFICATION OF THE RAssF1C2338’BINDING INTERFACE ON DAXX5544 102 FIGURE 3.7 - CHANGES IN THE SECONDARY STRUCTURES OF RAssF1C2338”’AND DAXX55144 UPON BINDING 107 FIGURE 3.8 - CHANGES IN THE AMIDE CHEMICAL SHIFTS AND RELAXATION OF RASSF1C2338wUPON BINDING DAXX5544 108 FIGURE 3.9 - TERTIARY STRUCTURE OF THE DAXX5544/RASSFIC2338WCOMPLEX 111 FIGURE 3.10 - RESIDUES MEDIATING THE INTERMOLECULAR INTERACTION BETWEEN DAXX55144 AND RAS5F1C2338w 113 FIGURE 3.11 - DAXX AND SIN3 PAH DOMAINS USE DIFFERENT BINDING MODES. OTHER STRUCTURAL RELATIVES OF THE DHB/RAssF1C COMPLEX 117 FIGURE 3.12 - PREPARATION OF DAXX5544/R SSF1C2338’’COMPLEXES WITH COMPLEMENTARY LABELING FOR STRUCTURAL STUDIES 130 FIGURE 4.1 - D1FFERENCEs BETWEEN SUMO-1 AND SUMO-2 139 FIGuRE 4.2 - MAPPING OF THE BINDING INTERFACE OF SUMO-1 AND SUMO-2 WITH SIM-N AND SIM-C 144 FIGURE 4.3 - NMR-M0NIT0RED TITRATIONS DEMONSTRATED THAT SIM-N’ BOUND BOTH SUMO-I’” AND SUMO2w 147 FIGURE 4.4 - NMR-M0NIT0RED TITRATIONS DEMONSTRATED THAT SIM-C’’ BOUND BOTH SUMO1w AND SUMO-2” 148 FIGURE 4.5 - SIM-C BOUND SUMO-1 IN BOTH PARALLEL AND ANTI-PARALLEL ORIENTATIONS 152 FIGURE 4.6- SIM-N BOUND SUMO-1 IN A PARALLEL ORIENTATION 157 FIGURE 4.7 - DAXX44 SELF-ASSOCIATION WAS IONIC STRENGTH DEPENDENT 161 FIGURE 4.8 - DAXX156 INTERACTED WITH DAXX5544 THROUGH SIM-N 163 FIGuRE 4.9 - SUMOYLATED ETs-l BOUND SIM-C OF MDAXX567739 VIA THE COVALENTLY ATTACHED SUMO 168 xii FIGURE 5.1 - DOMAIN STRUCTURE OF DAXX.180 FIGURE 52 - PROPosED AUTO-REGULATORY MECHANISM OF THE DHB DOMAIN. THE COMPETITIVE/CO-OPERATIVE COUPLED EQUILIBRIA (CCCE) MODEL 186 Xli List of Abbreviations 1 D one-dimensional 2D two-dimensional 3D three-dimensional aa amino acid AR androgen receptor ASKI apoptosis signal-regulating kinase 1 ATM ataxia telagiectasia mutated kinase ATP adenosine 5’-triphosphate ATRX cL-thalassaemia syndrome protein BLAST basic local alignment search tool CBP CREB binding protein CC coiled-coils CCCE competitive/co-operative coupled equilibria CD circular dichroism CK2 casein kinase 2 CREB cAMP-response element-binding protein CSI chemical shift index Da Dalton DAG diacyiglycerol/phorbol ester DAXX Fas death-domain associated protein DHB DAXX-Helix-Bundle (DAXX55144) DHD putative DAXX helical domain DIM DHB-domain interacting motif Dr ‘3C/15N-DAXX55144/unlabelecl RassflC2338” DSSP define secondary structure of proteins DTT dithiothreitol El SUMO/ubiquitin-activating enzyme E2 SUMO/ubiquitin-conj ugating enzyme E3 SUMO/ubiquitin-protein ligase EDTA ethylenediaminetetraacetic acid xiv FADD Fas-associated protein with Death Domain FAT focal adhesion targeting domain GR glucocorticoid receptor GST glutathione-s-transferase HDAC histone deacetylase hDAXX human DAXX HEPES 4-(2-hydroxyethyl)- 1 -piperazineethanesulfonic acid HIPI R huntingtin -interacting protein-i related HIPK-i homeodornain-interacting protein kinase 1 HMBC heteronuclear multiple bond correlation HPLC high performance liquid chromatography HSQC heteronuclear single quantum coherence IDPs intrinsically disordered proteins IPTG isopropyl 13-D- 1 -thiogalactopyranoside ITC isothermal titration calorimetry INK Jun N-terminal kinase KD dissociation constant kDa kilodalton MA LD 1-To F-MS matrix-assisted laser desorption/ionization-Time of Flight-Mass Spectrometry rnDAXX murine DAXX MEF mouse embryonic fibroblast MES 2-(N-morpholino)ethanesulfonic acid MOPS 3-(N-morpholino)propanesulfonic acid MSP58 nucleolar microspherule protein MTSL S-(2,2,5,5-tetramethyl-2,5-dihydro- I H-pyrrol-3 -yl)methyl methanesulfonothioate MW molecular weight MWCO molecular weight cutoff N CB I national center for biotechnology information (http://www.ncbi NDSM negatively-charged amino-acid dependent sumoylation motif NMR nuclear magnetic resonance NOE nuclear Overhauser effect NOESY nuclear Overhauser effect spectroscopy PAGE polyacrylamide gel electrophoresis xv PAH paired amphipathic helices PCR polymerase chain reaction PDB protein data bank ( PDSM phosphorylation-dependent sumoylation motif p1 isoelectric point PIAS protein inhibitors of activated STAT PML-NB s promyelocytic leukemia oncoprotein nuclear bodies ppm parts per million PRE paramagnetic relaxation enhancement RA Ras-association Rassf Ras-association domain family Rd ‘3C/15N-RassflC2338”/unlabeled DAXX55144 sample R[)C residual dipolar coupling RNAI RNA interference rmsd root mean square deviation SARAH Sav/RASSF/HPO domain SDS sodium dodecyl sulphate SENP sentrin specific protease SIM SUMO-interacting motif SIM-C” DAXX C-terminal SUMO-interacting motif. (mDAXX’718739) SlMNw DAXX N-terminal SUMO-interacting motif. (DAXXII9W) SPT Ser-Pro-Thr rich region in DAXX. S S P secondary structure propensity SUMO small ubiquitin-like modifier protein T1 longitudinal relaxation time T2 transverse relaxation time TEMPO 2,2,6,6-tetramethylpiperidine- 1 -oxyl TF transcription factor TFA trifluoroacetic acid TGF-3 transforming growth factor f3 TNF-cL tumor necrosis factor a. TOCSY total correlation spectroscopy xvi List of Amino Acid Abbreviations A Ala alanine C Cys cysteine D Asp aspartic acid F Gly glutamic acid F Phe phenylalanine G Gly glycine H His histidine I lie isoleucine K Lys lysine L Leu leucine M Met methionine N Asn asparagines P Pro proline Q Gin giutamine R Arg arginine S Ser serine T Thr threonine V Val valine W Trp tryptophan Y Tyr tyrosine xvii Acknowledgments My most sincere gratitude goes to Dr. Lawrence P. McIntosh, who throughout the years has been an excellent supervisor and a good friend. He was always available to answer my questions, and was open to my suggestions and ideas. He criticized my work always constructively, and made me grow as a scientist. His “work hard, play hard” policy was inspirational. He motivated me to get results, but also to enjoy the great things that the mountains and the ocean around Vancouver have to offer. I cannot emphasize enough how much I enjoyed working in his laboratory. I am forever grateful to my wife Claire, whom I married at the beginning of my graduate work. The research process often takes you through depressions and frustrations, and without her support, my success would not have been possible. Our son Natiü was born when I was in the middle of the experiments described in Chapters 3 and 4, and I can say for sure that without Claire’s extra support, I would not have been able to complete my graduate studies. I dedicate this thesis to her and our son. Of course, I would not be here without the support of many people. My mother Patricia, and my father Gerardo always motivated me to succeed academically, and this led to my scholarship to the United World College of the Adriatic (Italy) when I was 17. There, my chemistry teacher Anne Brearley inspired me to become a scientist. At my undergraduate studies, Dr. Steven Rafferty and Dr. Mark J. Parnis inspired me to go to graduate school. In Dr. McIntosh’s laboratory, I had the privilege to work with great people: Meena, Cameron, Mario, Matt, Gary, Manuela, Dave, Greg, Hyun-Seo, Martin, Markus, Chris, Shaheen, xviii Mark, Patrick, Genevieve, Jerome, Desmond and Simon. In particular, Genevieve and Hyun-Seo became really good friends and rock climbing partners. I thank them for their friendship through the good and bad times. My special gratitude goes to our collaborator Dr. Alexander Ishov for bringing a much needed biological perspective to my project. I would also like to thank the following people for their help in some of the work presented in this thesis. The sumoylated Ets’’38 studied in section 4.7 was prepared by the summer student Christopher Dart. The intermolecular NOEs between unlabeled SIM-N” and ‘3C/15N labeled SUMO-1 (Table 4.3 and Figure 4.6) were assigned by Dr. Mark Okon. Finally, I want to thank once again Dr. McIntosh, Dr. Gary Yalloway and Genevieve Desjardins-Seguin for reading my thesis and suggesting me how to make it better. xix Chapter 1 Introduction The Fas death-domain associated (DAXX) protein was first discovered over a decade ago as an intermediary of a FADD-independent Fas signaling pathway. However, subsequent studies have established it as an important player in both transcription and apoptosis regulation. Recently, a role of DAXX as a cell cycle regulator is emerging as well. Despite these important biological functions, the structural and dynamic properties of DAXX remained unknown prior to this thesis research. In this chapter, an overview of the biological functions of DAXX is presented, with emphasis on its interacting partners Rassfl C and the small ubiquitin-like modifier SUMO. An explanation of how DAXX mediates numerous essential biological functions requires detailed structural and thermodynamic analyses of DAXX with these protein partners. These experimental goals will be addressed in the following chapters of this thesis. 1 Chapter 1 — Introduction 11 Preamble 1.1.1 The human genome A milestone in molecular biology was reached in 2001 when it was announced that the human genome had finally been sequenced (Lander et a!., 2001; Venter et a!., 2001). Exciting, yes, but intriguing as well, as it raised as many questions as it provided answers. The biggest question perhaps being how this sequence dictates the formation of a human being? It is clear that RNA and proteins, not coding DNA, do the “work” in biological systems, so knowledge of a genome by itself is insufficient to understand what makes living beings. Proof of this is the observation that the latest estimate of the number of protein-encoding genes in the human genorne is around 25,000, a number close to that of C. elegans, a free-living one millimeter-long transparent worm (Pennisi, 2003). Obviously it is not the mere number of genes that defines the complexity of a species, but rather the properties of the encoded proteins, the mechanisms of their regulation, and the complicated networks that arise from protein-protein interactions and post-translational modifications. 1.1.2 Signal transduction Such complicated protein-protein-interaction networks are crucial for cell communication in multi-cellular beings, where cells die, differentiate, divide, or survive for the good of the organism as a whole. In a general sense, cell communication involves a receptor protein at the cell surface, which is in charge of detecting a signal. The receptor protein reacts to the signal by activating an intracellular signaling protein, which initiates a signaling transduction pathway. This involves multiple enzymes and adaptor proteins (scaffolds) that work together to convey the message to the transcription machinery in the nucleus, where genes for each cellular process are expressed or repressed. 2 Chapter 1 Introduction 1.1.3 Modular proteins Interactions between signaling proteins are mediated by modular binding domains, which are typically defined as “structural, functional and evolutionary components of proteins, which can often be expressed as a single unit” (Han et al., 2007). It is estimated that out of 34,000 proteins encoded by the human genome, between 65% and 80% of proteins consist of two or more domains (Han et al., 2007). The modular nature of proteins facilitates their study by a “divide-and-conquer approach, in which the structure and function of isolated domains are studied, and their contribution to the overall function of the full-length protein is subsequently derived. Such an approach has proved useful in elucidating the functions of a number of proteins (Vogel et al., 2004). However, in order to understand the integrated roles of protein modules fully, it is critical to complement such reductionist approach with a “re-constructionist” approach, and study the wild type protein in its native cellular context at a system level. 1.1.4 Intrinsically disordered proteins Captivatingly, about 25% of the predicted protein sequences from eukaryotic genomes cannot be attributed to any known domains (Moore et al., 2008). Some of these cases certainly reflect inadequacies in sequence-based “homology searching” as well as yet undiscovered “novel” domains. However, it is clear that a significant amount of these sequences form natively or intrinsically disordered proteins (IDPs) or segments of proteins, which at least in isolation lack any predominant secondary or tertiary structural elements. There is a growing recognition that IDP sequences play crucial roles ranging from flexible linkers between structured domains in modular proteins (“beads-on-strings”) to linear recognition epitopes (motifs) for forming protein protein complexes. IDPs, as well as misfolded or aggregated proteins, are also directly linked to 3 Chapter 1 — Introduction a number of human diseases, including cancer, cardiovascular diseases, neurodegenerative diseases and diabetes (Uversky, 2009). The intrinsically disordered regions of proteins represent a challenge to the above structure- based definition of a domain. That is, they are often evolutionally and functionally conserved, but they lack a well-defined three-dimensional structure. However, as their presence is crucial to the function of the protein, they need to be studied as well. Proteins often contain both: well- defined structural domains juxtaposed to intrinsically disordered regions that often play a role in regulating the function of the full-length protein. 1.1.5 DAXX, a multi-domain protein with intrinsically disordered regions The subject of this study is DAXX, an intracellular signaling adaptor protein with multiple structural domains and intrinsically disordered regions. Since its discovery in 1997 as a “Death Associated” factor (Yang et al., 1997), DAXX has been found to be an essential player in multiple cellular mechanisms such as apoptosis (programmed cell death) and transcription regulation. This brings special meaning to its structural characterization, which is presented in this thesis. In other words, by learning where its domains and intrinsically disordered regions are, one can better understand how it comes to play such an important role in so many mechanisms. The role of DAXX in various cellular processes is described in more than 250 publications. These roles are both widely varied and at times seemingly controversial. Thus, these publications are often confusing in the big-picture perspective. Fortunately, there are five reviews on DAXX (Lindsay et al., 2008; Michaelson, 2000; Salomoni and Khelifi, 2006; Salomoni and Michod, 2009; Shih et a!., 2007), to which the reader is urged to go for a more extensive summary of the numerous facets of this fascinating protein. 4 Chapter 1 — Introduction In this introductory chapter, a shorter review of the functions and localization of DAXX is first presented (section 1.2). This is followed by an introduction to SUMO (section 1.3) and Rassfl C (section 1.4), two important partners of DAXX, which were part of this study. Finally, at the end of this chapter, the main goals and organization of this thesis are described (section 1.5). 1.2 DAXX 1.2.1 What is DAXX? DAXX, a 740 aa protein, was first discovered and characterized in 1997 by Yang et al. (Yang et a!., 1997), who found that DAXX expressed in all mouse organs ubiquitously, and associated with the Fas death-domain (DAXX stands for Fas death-domain ssociated protein). This association was thought to initiate the activation of the Jun N-terminal kinase (INK) pathway, which ultimately led to cell death (apoptosis) (Figure 1. 1A). At the time, this appeared as a breakthrough in the field of death receptors, providing an alternative pathway for Fas-controlled death that is independent of the adaptor FADD (Fas-associated protein with Death Domain). The same group identified a year later the poptosis igna1 regulating inase 1 (ASK1) as the link between DAXX and the JNK-pathway, reinforcing their original model (Figure 1. 1A) (Chang et al., 1998). However, conflicting reports on the role of DAXX started appearing shortly after its discovery. First, mouse embryo fibroblasts deficient in FADD or caspase-8 were unable to produce apoptosis in response to Fas activation, suggesting that DAXX did not dictate a parallel signaling pathway (Juo et al., 1998; Zhang et al., 1998). Second, Toni et a!. failed to reproduce the DAXX-Fas direct interaction and found that DAXX was primarily located in the nucleus, 5 Chapter 1 — Introduction particularly in PML-nuclear bodies, and not in the cytoplasm (section Additionally, they identified DAXX as a transcription repressor and speculated that it modulated the genes activated by Fas-induction from the nucleus (Toni et al., 1999). Overall, these initial reports highlighted the two primary functions attributed to DAXX: a mediator in apoptosis, and a transcription regulator. 1.2.2 The mixed roles of DAXX in apoptosis In addition to the initial studies which implicated DAXX in Fas-induced apoptosis, another group identified DAXX as a binding partner of the transforming growth factor-13 (TGF-f3) and as a regulator of TGF-f3-induced apoptosis again via JNK activation (Perlman et al., 2001). This study was important as it established for the first time the mechanism by which TGF-f3 controlled apoptosis (Figure l.1A). Based on the reports implicating DAXX as a pro-apoptotic protein, one would predict that a knock-out mouse for its gene would display proliferation abnormalities. Such abnormalities were observed with p53 tumor suppressor knock-out mice, which were able to live but developed some type of cancer within 6 months of birth (Donehower et al., 1992). Surprisingly, the opposite was observed with DAXX. That is, DAXX was an essential protein for embryo development, as embryos with a truncated form of DAXX or with a complete knock-out of the DAXX gene displayed extensive apoptosis and lethality at embryonic day 9.5-10.5 (Ishov et al., 2004; Michaelson et al., 1999). These studies therefore suggested that DAXX plays an essential anti-apoptotic role, at least in embryonic development. Further evidence of an anti-apoptotic role for DAXX was shown by two independent studies in which DAXX expression was silenced via RNA interference (RNAi) experiments. Both 6 Chapter 1 — Introduction studies found that DAXX silencing resulted in sensitization of the cell to apoptosis induced by Fas, tumor necrosis factor a. (TNF-ct), or even ultraviolet light irradiation (Chen and Chen, 2003; Michaelson and Leder, 2003). It is then conceivable that DAXX has both pro- or anti-apoptotic roles in different situations. It has been suggested that this duality may arise from the distribution of the protein in cell compartments (Michaelson, 2000). 1.2.3 Where is DAXX? The function of DAXX might be dependent upon its cellular location. Obviously, to interact with the Fas-death domain, it needed to be present in the cytoplasm. It was then surprising that the first study reporting the cloning and sequencing of human DAXX predicted that it would be localized in the nucleus due to the presence of nuclear localization signals (Kiriakidou et aL, 1997). Soon thereafter, multiple reports confirmed that DAXX was indeed predominantly a nuclear protein (Everett et a!., 1999; Ishov et al., 1999; Toni et a!., 1999). The cell nucleus is a complicated organelle where many essential processes take place. This is where gene expression, as well as DNA replication, repair and recombination occur. It is organized in a number of dynamic compartments that, in contrast to their cytoplasmic counterparts, do not have delineating membranes. However, they can be distinguished by light and electron microscopy and each contains defining subsets of resident proteins (Dundr and Misteli, 2001; Shiels et a!., 2007). Nuclear compartments include the nucleolus, the splicing factor compartments, the Cajal bodies, and the prornyelocytic leukemia oncoprotein (PML) nuclear bodies (Figure 1.1 B). The latter are where DAXX has often been reported to reside. 7 Chapter 1 — Introduction Figure 1.1 - Function and localization of DAXX. (A) Attributed pro-apoptotic roles of DAXX. DAXX was discovered as a binding partner of the death domain of the Fas receptor (DD), which activated the INK-signaling pathway via ASK1 and ultimately led to cell apoptosis (Chang et al., 1998; Yang et al., 1997). This pathway was the first proposal of an alternative to the well- established FADD-caspase signaling pathway activated by Fas (Strasser et al., 2009). The cytoplasmic TRII domain of the Transforming Growth Factor-13 was also shown to interact with DAXX and activate the same signaling pathway (Perlman et a!., 2001). (B) Schematic representation of the compartmental organization within a cell nucleus, showing chromosome territories (green), cajal bodies (purple), nucleoli (violet), splicing-factor compartments (red), PML-NBs (light green) and nucleoplasm (white). Within chromosome territories, dots symbolize the location of euchromatin where transcription occurs, and stars symbolize the location of heterochromatic regions. This figure was adapted from (Dundr and Misteli, 2001; Schmitz and Herrmann, 2008; Shiels et a!., 2007). (C) Compartmental localization of DAXX is believed to control its function. When DAXX is recruited into PML-NBs by the PML protein during interphase, it cannot act as a transcription repressor. However, different stimuli can result in its release from PML-NBs into the nucleoplasm, where it can repress sumoylated transcription factors (TF) via its SUMO interaction motif (SIM) in the Ser-Pro-Thr rich region (SPT), or it can localize into the nucleolus, centromeres or heterochromatic regions. The function of DAXX within these compartments has not been fully characterized, although it has been proposed that nucleolus sequestration also inactivates DAXX, and that interaction with the ATP-dependent chromatin remodeling protein ATRX might play a role in its transcription regulation role. See text for details. Figure adapted from (Shih et al., 2007). 8 Chapter 1 — Introduction Extra-cellular matrix Cytoplasm Fas receptor \/ ASK1 Caspase-8 Apoptosis Nucleoplasm \ A B C Splicing-Factor Compartments JNK Apoptosis NucleoliPML-NBs _y7_______ Chromosome tories Transcription Cajalsites I-Ieterochromatin bodies ATRXAXX Nucleoplasm Or Ets-1 p53 Smad4, etc. 9 Chapter 1 — Introduction DAXX in PML nuclear bodies PML nuclear bodies (PML-NBs; also known as nuclear domains-lO (ND-lU), PML oncogenic domains (PODs), or Kremer bodies) are present in 1-30 copies in most mammalian cell nuclei and typically measure 0.2-1.0 p.m (Bernardi and Pandolfi, 2007). They regulate a number of different cellular functions, including gene expression, DNA repair, apoptosis and viral infection control (Bernardi et al., 2008; Dellaire and Bazett-Jones, 2004; Tavalai and Stamminger, 2008). PML-NBs are composed of numerous proteins, with nucleic acids at the periphery of the bodies. Although variable in exact composition, key to PML-NB formation is the PML protein. Knock-out mice without the PML protein failed to form PML-NBs, and consequently were more resistant to apoptosis after DNA damage and prone to carcinogenesis (Wang et a!., 1998). The first groups that detected DAXX in the nucleus also reported that it was located in PML NBs, and identified DAXX as a binding partner of PML itself (Ishov et a!., 1999; Li et al., 2000a). In these studies, sumoylation of PML was proposed to be necessary for the recruitment of DAXX. As will be discussed shortly, the ability of PML to be sumoylated and to bind other sumoylated proteins is also essential for the formation of PML-NBs (Shen et a!., 2006). An extensive recent study confirmed that a UMO-jnteracting motif (SIM) within DAXX was needed for its recruitment to PML-NB, thus explaining the above proposition (Figure 1.1 C) (Lin et al., 2006). DAXX recruitment into PML-NBs has been proposed to serve as a regulator of its function (Ishov et al., 2004; Li et a!., 2000a; Shih et al., 2007). In this model, PML-NBs serve as “storage depots”, sequestering DAXX in an inactive state. Indeed, studies have reported that overexpression of the PML protein reversed DAXX-controlled repression by recruiting it from transcription sites into PML-NBs (Lehembre et al., 2001; Li et a!., 2000a). Overall these studies 10 Chapter 1 — Introduction highlighted the functional importance of the localization of DAXX, and the role that sumoylation and SIMs play in mediating its localization to PML-NBs (section 1.3). DAXX in other nuclear compartments In a study by Ishov et al., DAXX mice did not survive beyond embryonic day 10.5, but PML’ mice, which lacked PML-NBs, were viable (Ishov et a!., 1999). Therefore, PML-NB recruitment of DAXX is not essential for its biological function. Interestingly, in PML’ cells, DAXX was mostly found in areas of condensed heterochromatin. The association of DAXX with heterochromatin was also indirectly implied by Hollenbach et al., who used co immunoprecipitation and gel filtration co-fractionation to demonstrate that DAXX was part of a large complex that included core histones, chromatin associated protein Dek and histone deacetylase 2 (HDAC-2) (Hollenbach et al., 2002). Numerous groups have subsequently reported a direct or indirect association of DAXX with histone deacetylases (Ecsedy et a!., 2003; Greger et al., 2005; Huang et al., 2008; Kuo et a!., 2005; Morozov et a!., 2008; Poleshko et a!., 2008). Since HDACs play a role in remodeling chromatin to an inactive state, their association with DAXX is believed to be key to the function of DAXX as a transcription repressor (discussed in section 1.3.3). However, many details of this model remain unknown. For example, what are the mechanisms that make DAXX move from PML-NBs to chromatin regions? This is unclear at the moment, although interaction between DAXX and the homeodomain-interacting protein kinase 1 (HIPK- 1), and the ct-thalassaemia syndrome protein (ATRX) have been shown to play a role in this process. In a study by Ecsedy et a!., HIPK-1 physically interacted with DAXX, re-localizing it from PML-NBs to chromatin via a direct interaction (Ecsedy et a!., 2003). The kinase also 11 Chapter 1 — Introduction phosphorylated DAXX at Ser669, increasing its repression ability by an unknown mechanism. ATRX, an ATP-dependent chromatin remodeling protein, and DAXX have also been shown by independent groups to form a complex (Ishov et a!., 2004; Tang et al., 2004; Xue et a!., 2003). This complex was relevant in inactivating ATRX via recruitment into PML-NBs in a cell-cycle dependent manner (Ishov et a!., 2004). Apart from heterochromatic regions, DAXX has been shown to interact with the centromere protein CENP-C and thereby it associated with centromeres (Pluta et al., 1998). DAXX also bound the nucleolar microspherule protein MSP58 and accumulated at the nucleolus (Lin and Shih, 2002). The role of DAXX in centromeres is unknown, whereas the latter report suggested that DAXX was recruited in the nucleolus with a functional consequence similar to PML-NB sequestration. That is, nucleolar DAXX was inactive and unable to mediate transcription. Cytoplasmic DAXX? Although the overwhelming evidence points to a nuclear localization of DAXX, a number of studies have reported DAXX trafficking from the cell nucleus to the cytoplasm under stress conditions (Akterin et al., 2006; Charette and Landry, 2000; Jung et al., 2008; Jung et al., 2007; Junn et al., 2005; Karunakaran et al., 2007; Song and Lee, 2004). However, most of these studies relied on immunofluorescence staining of transiently over-expressed (and hence unnaturally abundant) proteins. Further doubt to these findings has arisen from a very recent report. Lindsay et al. applied multiple approaches, including subcellular fractionation by various methods, and time lapse visualization by immune-fluorescent microscopy, to unequivocally show that DAXX does not translocate to the cytoplasm under previously reported conditions of stress or cell line type (Lindsay et a!., 2009a). 12 Chapter 1 — Introduction Therefore, the cytosolic association of DAXX with Fas, Ask-i or other cytoplasmic proteins remains in question. The authors suggested that these associations may be cell-cycle dependent (Lindsay et a!., 2007). Indeed, the same group has shown recently that DAXX interacts with RassflC, a cytoplasmic protein, once the nuclear envelope has broken down during mitosis ((Lindsay et a!., 2009b), discussed in section 1.2.5). L24 DAXX as a transcription regulator A role of DAXX as a transcription co-repressor was first proposed by Hollenbach et al., who observed transcriptional repression of Pax3 when cells were co-transfected with DAXX (Hollenbach et al., 1999). Ten years later, the list of transcription factors regulated by DAXX includes more than 25 examples (summarized by Lindsay et al., 2008). Although some of these transcription factors appeared to be activated by DAXX (Emelyanov et a!., 2002; Huang and Shih, 2009), in most cases DAXX played a role in transcriptional repression. DAXX associates with HDACs, and this association is believed to be the main route to repression of gene expression (Figure 1 .2E and 1 .4B). Hollenbach et a!. proposed that DAXX functions in a similar way as the Sin3 co-repressor. Sin3 is a large multi-domain protein, which serves as a scaffold, linking HDACs with other transcription regulators and transcription factors (Grzenda et al., 2009; Silverstein and Ekwall, 2005). Key to the formation of these large complexes is the presence of four domains within Sin3 named “paired amphipathic helices” (PAH). 13 —Activator -RNA Pol H Holoenzynw TFHD Repressor C j ____ E Nucleosome Deacetylated histone tails Gene t Gene Gene Chapter 1 — Introduction B4 __ Ix. i D rii iw1’ Gene Remodeled Chromatin remodeling Nucleosome complex F • JIVj 1t Figure 1.2 - Generic models for transcriptional repression in eukaryotic systems. (A) Transcription activation requires the recruitment of the RNA polymerase II holoenzyme complex, which contains a large number of proteins that are omitted for clarity (Roeder, 2005; Sikorski and Buratowski, 2009; Thomas and Chiang, 2006). Recruitment is normally achieved by activators, which are multi-domain proteins with a DNA binding module and one or several protein-protein interaction domains. Activators not only recruit RNA polymerase II, but they also recruit co-activators such as histone acetylases for chromatin remodeling. Activators can be modulated by post-translational modifications, such as phosphorylation and sumoylation (not shown). Transcriptional repression can be achieved by many different pathways, but always involve one or more repressors, which can interfere with the recruitment of RNA polymerase II holoenzyme (B), neutralize activators (C), competitively bind activator DNA-binding regions (D) or recruit other complexes such as histone deacetylases (E) or chromatin remodeling complexes (F). Figure adapted from (Alberts, 2002). A 14 Chapter 1 — Introduction Sin3 PAH domains mediate interactions with transcription factors (TF5) and other co repressors. The cell tightly regulates these interactions (and thereby controls gene expression) via post-translational modifications to the TFs and competitive binding of activators to the same interaction interface on the TF. Hollenbach et a!. proposed the existence of two PAR domains in DAXX that could control its function in a similar way as the Sin3 PAR domains (this model is revisited in Chapters 2 and 3, as the N-terminal domain of DAXX is compared with the known structures of Sin3 PAR domains). Finally, in contrast to Sin3, a common theme that has appeared in recent literature is that DAXX-mediated repression depends on its ability to interact with sumoylated transcription factors. This is discussed in section 1.3. 1.2.5 The role of DAXX in mitosis As discussed so far, previous studies have demonstrated that DAXX functions as an apoptosis and transcription factor mediator. Rowever, a new and exciting role of DAXX as a mitosis regulator has lately emerged. The evidence that DAXX is involved in cell cycle controlling processes is manifold. First, Lindsay et al. noticed an abnormal number of chromosomes (aneuploidy) in DAXX ‘ mouse embryonic fibroblasts (MEF5), suggesting an important role of DAXX in cell division (Lindsay, 2009). Second, DAXX cells showed different cell-cycle progression with respect to control cells. Third, cells depleted of DAXX via siRNA underwent cell division at a different rate with respect to control cells (Lindsay et al., 2009b). Fourth, Lindsay et al. noticed that DAXX localized at the mitotic spindle during mitosis, specifically during the prometaphase stage (discussed in section 1.4.3 and 3.7). 15 Chapter 1 — Introduction To understand the role of DAXX in cell-cycle control, Lindsay et a!. set up a yeast-two hybrid screen for interactors of DAXX that could be directly involved in cell-cycle controlling processes. In this way, they identified Rassfl C, a member of the tumor suppressor Ras associated domain family, as a DAXX partner (section 1.4.3). Further studies revealed that the DAXX/Rassfl C interaction was indeed important in mediating mitosis, and more importantly, in determining how cancer cells respond to a common chemotherapy agent, taxol. The details of their findings are discussed in Chapter 3, as the structure of the DAXX/Rassfl C complex is presented. t3 The small ubiquitin-like modifier protein SUMO The small ubiquitin-like modifier protein (SUMO) derives its name from its structural resemblance to ubiquitin (rmsd of backbone C 3 A), despite having only 18% sequence identity with it. It was discovered in 1996 by three independent groups (Boddy et al., 1996; Okura et al., 1996; Shen et al., 1996). Since then, the research field of SUMO has grown immensely, with well over 1500 publications and with a growing number of publications per year (in 2008 alone there were 265). This is not surprising, as SUMO has been implicated in multiple fundamental biological processes such as transcription control, DNA repair, nucleocytoplasmic trafficking, mitochondrial fragmentation, and viral infection. There are over 100 reviews on various aspects of SUMO functions. Excellent recent reviews on SUMO include: (Bergink and Jentsch, 2009; Garcia-Dominguez and Reyes, 2009; Geiss-Friedlander and Melchior, 2007; Geoffroy and Hay, 2009; Hay, 2005; Johnson, 2004; Kerscher, 2007). Here, an overview of the general principles and mechanisms of sumoylation is provided, with an emphasis on those that are relevant to DAXX localization and function. 16 Chapter 1 — Introduction L3.1 The SUMO family Despite limited sequence similarity, SUMO family members have three-dimensional structures very similar to ubiquitin. However, SUMO proteins contain an N-terminal unstructured 10-25 amino-acid segment that is not found in ubiquitin. Until now, the only function known of this disordered region is formation of SUMO chains. The human genome encodes four different SUMO paralogues (SUMO-1 to SUMO-4). SUMO-2 and SUMO-3 are virtually identical, with 97% sequence identity, and are commonly referred as SUMO-2/3. SUMO- 1 and SUMO-2/3 are expressed in all cells, whereas SUMO-4 seems to be mainly expressed in lymph nodes and the spleen (Guo et al., 2004). SUMO-2/3 have 50% sequence identity with SUMO-1, and 87% sequence identity with SUMO-4. All SUMO paralogues are expressed in inactive forms, with an extended C-terminus that needs to be removed proteolytically so that mature SUMO can be covalently attached to its target via a Gly-Gly motif (Figure 1 .3A). Whether maturation by SUMO-specific proteases is regulated or constitutive is unknown at the moment. Once matured, SUMO paralogues can be covalently attached to substrates via a reversible mechanism. 1.3.2 The mechanism of reversible sumoylation The process of sumoylation is very similar to ubiquitylation. It involves an El activating enzyme (Figure 1 .3B), an E2 conjugating enzyme (Figure 1.3 C), and in some cases, an E3 ligating enzyme (Figure 1 .3D). The latter ultimately attaches the C-terminal Gly carboxyl of the SUMO moiety to the amine of a lysine residue in the substrate via an isopeptide linkage. This reaction is reversible, as a number of sentrin-specific proteases (SENP) can specifically remove SUMO from substrates (Kim and Baek, 2009; Yeh, 2009) (Figure 1 .3E). 17 Chapter 1 — Introduction Clear differences to ubiquitylation exist, however. For example, a consensus SUMO-acceptor site has been recognized as ‘PKxE (where P is a branched aliphatic amino acid and x is any amino acid) (Rodriguez et a!., 2001). Recently this acceptor site has been expanded to a phosphorylation-dependent sumoylation motif (PDSM), or a negatively-charged amino-acid dependent sumoylation motif (NDSM), both of which depend on negative charges that are either added by phosphorylation (PDSM) or are constitutive of the residues downstream of the sumoylation site (Hietakangas et al., 2006; Yang et a!., 2006). Ubiquitylation-consensus sites, on the other hand, have not been identified. Furthermore, sumoylation involves only one E2, and not always an E3, whereas ubiquitylation involves more than 20 different E2s and hundreds of E3s (Geiss-Friedlander and Meichior, 2007). Sumoylation of substrates occurs very rapidly. Golebiowski et al. reported maximal sumoylation of proteins only 5 minutes after heat shock (Golebiowski et a!., 2009). Interestingly, while most heat-shocked-sumoylated proteins were de-sumoylated 2 hours after heat shock, heat shock-desumoylated proteins were not re-sumoylated over the same period of time. This hints to a reversible mechanism that is tightly and quickly regulated. 18 Chapter 1 — Introduction SUMO c=o rf Figure 1.3 - Sumoylation mechanism. Before SUMO can be conjugated, it must be matured by sentrin specific proteases (SENPs; A). This leaves a C-terminal Gly-Gly that can be covalently attached to Cys 173 in the El activating enzyme (AOS 1 /UBA2 in humans) by an ATP-dependent reaction (B). Cys92 of the E2 conjugating enzyme (UBC9) receives the activated SUMO from El (C). Finally, a lysine side-chain of a substrate (a transcription factor (TF) in the figure) is covalently modified to form an isopeptide linkage (D). This step is usually aided by an E3 ligating enzyme. The mechanism is reversible as SENPs can remove SUMO from the substrate (E). Structures used for this figure are: SUMO-1 (la5r.pdb); Senp2 (ltgz.pdb); Aosl/Uba2 (ly8r.pdb); Ubc9 (lu9a.pdb); Ets-1 TF (2jv3.pdb). .. ..-.. SF\P + SUM() I II SUMO . --.. O) a. + I SUMO GG-COO + E3? SUMO ATP j qt 19 Chapter 1 — Introduction Different mechanisms for regulating sumoylation have been identified. Various research groups have noticed that the expression levels of the sumoylating enzymes (i.e. El, E2 and E3s) changed under different conditions, such as in hypoxia, calcium signaling, or progesterone signaling (Comerford et a!., 2003; Deyrieux et al., 2007; Jones et a!., 2006; Liu and Shuai, 2008). Apart from expression levels, sumoylating enzymes have been reported to be regulated by inactivation under certain conditions or by translocation to different cellular compartments. For example, reactive oxygen species caused the formation of a disulfide bond between the catalytic cysteine residues of El and E2 in HeLa cells (Bossis and Melchior, 2006). Similarly, pro- inflammatory stimuli such as TNF-a induced phosphorylation of PIAS 1 (an E3 enzyme) and controlled its translocation to the promoters of necrosis factor-kB (Liu et al., 2007). 1.3.3 Molecular consequences of sumoylation Sumoylation can affect substrates in a number of ways. In a general sense, sumoylation can protect a lysine side-chain from other post-translational modifications such as ubiquitylation or acetylation. Most often, however, it alters the intra- or inter-molecular interactions of the substrate. This can either alter the activity of the substrate (if it is an enzyme), or it can promote or inhibit the formation of large complexes that carry out a process that is important for the substrate (Figure 1 .4A). In principle, SUMO can either add or block a binding interface from the substrate, or it can induce a conformational change in the substrate that affects its activity. Examples of the first are RanGap 1 (which interacts with RanBP2 when sumoylated), and the transcription repressor ZNF76 (which binds to a TATA-binding protein only when it is SUMO-free) (Mahajan et a!., 1997; Zheng and Yang, 2004). Consistent with the model of SUMO serving as a structurally independent tag on a protein, Macauley et a!. showed that SUMO-RanGAP behave as two 20 Chapter 1 — Introduction independent domains linked by a flexible isopeptide bond (Macauley et al., 2004; Macauley et al., 2006). In contrast, only one example of a protein whose conformation changes significantly upon sumoylation has been reported: thymine DNA glycosylase. This DNA-repair enzyme requires sumoylation to undergo a structural change that allows it to release its substrate once repaired (Baba et a!., 2005; Hardeland et al., 2002). Ultimately, it is not easy to predict the effect that sumoylation will have on a particular substrate. However, the observation that the vast majority of sumoylated proteins are transcription factors suggests that some common themes exist. These include the recruitment of chromatin remodeling complexes and histone-modifying factors such as histone deacetylases (Garcia-Dominguez and Reyes, 2009). It is worth noting that such chromatin remodeling is long-lasting. That is, a small amount of sumoylated transcription factor can have a big effect by leaving permanent repressive changes once the transcription factor or its repressive complexes are removed by de-sumoylation. This is important as only 5% of the transcription factors are sumoylated in vivo at the “steady-state” (Geiss-Friedlander and Meichior, 2007; Hay, 2005). Recruitment of chromatin remodeling complexes does not occur via a direct interaction with transcription factors, but rather it is thought to be mediated by scaffold proteins such as DAXX (Figure 1 .4B, section 1.3.6). 21 Chapter 1 — Introduction A Surnoviation rSubst Lstra ( I Desurnoylation tratZ II1LLraLIor A • k - (SIts) :Subs Ac Ac ____ £TF) Nucleosome ‘ Figure 1.4 - Molecular consequences of sumoylation. (A) Before sumoylation, a substrate can either be free (and unable to interact with interactor B) or interact with another protein (interactor A). Sumoylation can have a number of consequences, including: (1) disrupt the complex of the substrate with interactor A; (2) induce a conformational change on the substrate; (3) supply a new surface for recruiting a new protein (interactor B). In principle, this could occur either independently from the substrate or with cooperative binding from both the substrate and SUMO. (B) Proposed mechanism for DAXX-mediated transcription repression. DAXX is unable to bind a transcription factor (TF) when it is not sumoylated (in this example the TF is present in active chromatin with acetylated histone tails). Once the TF is sumoylated, it recruits a large repression complex via the binding of DAXX to a SUMO. The repression complex may contain histone deacetylases (HDACs) that deacetylate the histone tails and remodel chromatin to prevent gene transcription. B Ac Ac Sumoykition 22 Chapter 1 — Introduction 1.3.4 Differences among SUMO paralogues Since SUMO-4 was discovered more recently, its function is still unclear. There is also a controversy as to whether or not SUMO-4 can even be covalently matured and attached to substrates. Thus, its biological function appears to differ from that of the other SUMO paralogues (Owerbach et al., 2005; Wei et al., 2008). Most studies to date have focused on SUMO-1 and SUMO-2/3 differences. Saitoh et al. were the first to address some of the functional differences between these SUMO paralogues (Saitoh and Hinchey, 2000). In their study, they observed the following differences: (1) SUMO-2/3 was about 4 times more abundant than SUMO- 1. (2) SUMO-1 was mostly in a conjugated form, whereas there was a large pooi of free SUMO-2/3 in the cell. Overall, there was about fifty times more free SUMO 2/3 than SUMO-l. (3) Some substrates were modified by specific SUMO paralogues. For example, RanGapi was modified by SUMO-1 and not by SUMO-2/3. Indeed, a recent proteomic study found SUMO paralogue specificity is quite common among SUMO substrates. Out of 53 identified sumoylated proteins in HeLa cells, only 9 could be modified by either SUMO-1 or SUMO-2/3, whereas 25 were preferentially modified by SUMO-l, and 19 by SUMO-2/3 (Vertegaal et al., 2006). (4) Stress signals such as heat shock depleted the pool of free SUMO-2/3. This suggested that a number of proteins are sumoylated in response to these stimuli. Indeed, a very recent study identified 766 SUMO-2/3 substrates, many of which underwent sumoylation changes upon treatment with heat shock (Golebiowski et al., 2009). 23 Chapter 1 — Introduction Additionally, SUMO paralogues have been shown to have different localization and mobility in vivo by Ayaydin and Dasso (Ayaydin and Dasso, 2004). In their experiments, they demonstrated that only SUMO- 1 is located at the nuclear envelope and in the nucleolus, whereas all paralogues were found throughout the nucleoplasm. Furthermore, photobleaching experiments revealed that SUMO- 1 mobility was much slower than that of SUMO-2/3. Other differences among SUMO paralogues include their ability to form polymers and their specificity towards interactors via non-covalent bonds. For example, SUMO-2/3 can form chains through its flexible N-terminal segment, whereas SUMO-1 cannot. These chains have been shown to be important in signaling and promotion of macromolecular complexes (Ulrich, 2008). Finally, differences in the specificity towards binding SUMO interacting motifs from different paralogues have been discovered recently (Hecker et al., 2006; Meulmeester et al., 2008). These differences are discussed in detail in Chapter 4. 1.3.5 SUMO interacting motifs The first group that reported a specific SUMO-interacting motif (SIM; also known as SUMO-binding motif, SBM) noticed that proteins interacting with sumoylated p’73, a member of the p53 family, contained a common Ser-X-Ser sequence juxtaposed with hydrophobic and acidic residues (Minty et al., 2000). Four years later, Song et al. identified an entirely different SIM by NMR spectroscopy, and isothermal titration calorimetry (ITC). This SIM consisted of Val/Ile-X-Val/Ile-Val-Ile (Song et al., 2004). A year later, a crystal structure of SUMO 1/RanGapl-UBC9-Nup358 complex revealed that SUMO-l and RanGapi indeed interacted through this motif and that the hydrophobic core is fundamental for the interaction (Reverter and Lima, 2005). This example highlights the importance of a structural characterization to unequivocally determine the basis for the interaction between molecules. 24 Chapter 1 — Introduction A number of SIM-containing proteins have been documented since then (reviewed in (Kerscher, 2007)). SIMs contain a hydrophobic core juxtaposed with a cluster of charges. Common characteristics among SIMs are: (1) The hydrophobic core is essential for the interaction, and a single amino-acid change can significantly reduce its affinity for SUMO. (2) A cluster of charges juxtaposed to the hydrophobic core is almost always present, and helps to dictate the binding orientation of the motif (Hecker et al., 2006; Song et al., 2005). (3) Charges within the hydrophobic core and the juxtaposing residues can also influence SUMO-paralogue binding specificity (Hecker et al., 2006; Meulmeester et al., 2008). (4) SIMs have dissociation constants in the range of 2-3 jtM, whereas ubiquitin binding motifs are more in the range of 10-500 tM (Hicke et al., 2005). (5) Some proteins use multiple SIMs to increase their affinity towards SUMO-chains. An example of this is RNF4, an E3 ubiquitin ligase that bound weakly to monomeric SUMO-2/3, but strongly to SUMO-2/3 chains. Consistently, RNF4 only ubiquitinated PML that had been modified by SUMO-2/3 chains in vitro (Tatham et al., 2008). 1.3.6 DAXX and SUMO DAXX and SUMO interact in two possible ways: DAXX itself can be sumoylated, or DAXX and SUMO can interact non-covalently via SIMs within DAXX. In the first case, DAXX sumoylation can occur in a number of sites. DAXX has 15 lysine amino acids, of which only one has the PKxE sumoylation motif: Lys6O (YKLE). However, Jang et a!. failed to detect any sumoylation effect on K6OA or K385A DAXX mutants. Instead, they identified Lys630 or 25 Chapter 1 — Introduction Lys63 1 as possible non-consensus sumoylation sites in BOSC23 cells (Jang et al., 2002). These sumoylation sites were corroborated by a later study, which also identified Lys6O as a sumoylation site. However, Lys6O was sumoylated only if expressed as a peptide comprising residues 46-75, and not if expressed as 1-282 (Chen et al., 2006). Thus, it is unclear if this site is functional in the wild type protein. The sumoylation sites within DAXX have been suggested to be cell-type dependent. For example, Lin et al. reported detectable sumoylation at all lysine residues in HeLa cells (Lin et al., 2006). More interestingly, Lin et al. also reported that DAXX sumoylation was dependent on one of its SIMs, as mutations of the SIM-binding cleft of SUMO, or of the DAXX C-terminal SIM, significantly reduced DAXX sumoylation. A simple model for this effect is that DAXX interacted non-covalently with E2-SUMO via its SIM-C, thereby increasing the effective concentration of its acceptor lysine(s) towards the conjugating enzyme complex. The biological significance of DAXX sumoylation remains to be established. A mutant lacking all Lys residues showed no functional difference to the wild type, localizing in PML NBs and retaining its transcription repression ability (Lin et al., 2006). However, recent reports suggested that sumoylation of DAXX was up-regulated in viral infections or in heat shock (Golebiowski et al., 2009; Hwang and Kalejta, 2009). Whether this is biologically relevant remains to be tested. On the other hand, non-covalent SUMO binding by DAXX has been demonstrated to be essential for its recruitment into PML-NBs and/or transcription regulation. Two SIMs within DAXX that mediated non-covalent SUMO-binding have been reported in the literature, one at its N-terminus (SIM-N), and one at its C-terminus (SIM-C) (Lin et al., 2006; Santiago et al., 2009). 26 Chapter 1 — Introduction In Chapter 4, a detailed analysis of how these SIMs interact with SUMO- 1 and SUMO-2 is presented. In Chapter 5, some other interesting models for DAXX SIMs that have arisen from the structural characterization of DAXX are discussed. 1.4 The Ras-association domain family (Rassf) Rassfl C is a member of the tumor suppressor Ras-association domain family (Rassf). Its founding member, Nore 1 (now known as Rassf5), was first characterized a decade ago, almost at the same time as SUMO and DAXX were first identified (Vavvas et al., 1998). Ten years later, this family is at the center stage for cancer research. It is now known that many members of the Ras-association domain family play important roles as tumor suppressors and their inactivation is a common characteristic of various types of tumors. The Rassf family has been the subject of a number of recent excellent reviews (Avruch et al., 2009; Hesson et a!., 2007a; Richter et a!., 2009; van der Weyden and Adams, 2007). In the next sections, a brief overview of the characteristics of this family and its association with DAXX is presented. 1.4.1 Domain structure of Rassf The defining characteristic of all Rassf members is the presence of a Ras-association (RA) domain (Figure 1.5). Ras GTPases are a superfamily of monomeric GTPases that broadcast signals from the cell surface to the nucleus. When activated indirectly via cell surface receptors, they interact with different effectors that in turn initiate signaling transduction pathways to ultimately regulate cell proliferation, differentiation, motility and apoptosis (Alberts, 2002). So far, only Rassf5 has been established as a true Ras-effector, and the remaining members of the family are thought to either exert functions independent of Ras, or mediate Ras-activated processes via heterodimerization with Rassf5 (Avruch et a!., 2009). 27 Chapter 1 — Introduction Most members of this family also contain a Sav/Rassf/Hpo (SARAH) domain, a 50 amino- acid stretch shown to form heterodimers with proteins involved in cell cycle control and apoptosis development. Some members contain a Cl-type zinc finger domain, also involved in protein-protein interactions controlling apoptosis. Finally, Rassfl A and Rassfl C have an ataxia telangiectasia mutated kinase (ATM) recognition motif, which is phosphorylated by the ATM, an enzyme of central importance in DNA-repair and apoptosis (Figure 1 .5A) (Richter et a!., 2009). At the structural level, the domains in the Rassf members are just beginning to be characterized. The structure of the Cl domain of Rassf5 was solved by NMR methods recently, and was found to bind intramolecularly with its RA domain (Harjes et a!., 2006). Interestingly, this intramolecular interaction was disrupted by Ras binding, freeing the Cl domain to presumably interact with other partners. The SARAH domain structure of Mstl was solved by NMR methods a year later, and it was found to mediate homo- and heterodimerzation through two short amphipathic helices (Hwang et al., 2007). Finally, the crystal structure of the RA domain of Rassf5 bound to Ras was reported last year (Stieglitz et a!., 2008), revealing a novel mode of Ras binding with very tight association (KD between 0.02-0.3 tiM, depending on the method used) (Wohlgemuth et al., 2005) (Figure 1 .5B). To our knowledge, the work presented in Chapter 3 is the first to structurally characterize the unique N-terminus of Rassfl C. 28 Chapter 1 — Introduction Figure 1.5 - The Ras-Association Domain Family (RASSF) and their domains. (A) Schematic of the most prevalent members of the family. The defining characteristic of this family is the presence of a Ras-Association (RA; red) domain. Most of the members have this domain preceding the Sav/RASSF/HPO (SARAH) domain (green) at their C-terminus. A newly discovered sub-family (Rassf7-RassflO) has the RA domain at its N-terminus. RassflA and Rassf5A-B have a cysteine-rich diacyiglycerol/phorbol ester (DAG)-binding domain, also called protein kinase C conserved region 1 (blue; Cl). RassflA, C have an ataxia telagiectasia mutated (ATM) kinase phosphorylation motif (magenta). The NCBI accession number of each member is shown as reference. (B) Cartoon representation of the structural domains of Rassf5A. The structure of the apo C-i domain (blue, 1 rth.pdb) with Zn2 atoms ; the RA domain (red; 3ddc.pdb) bound to RAS; and the SARAH domain as a homodimer (green/gray; 2jo8.pdb). The structures are not to scale. 29 AChapter 1 — Introduction 230 RASSF1A NP 009113 I . — 341) RASSF1C NP 733831 LZZZZZEZZZ — 270 RASSF2 NP 739580 TZZ._ . — 326 RASSF3 NP 835463 RASSF4 NP 114412 • RASSF5A NP 872604 -• •1 410 RASSF5B NP 872605 336 ASSF5C NP 872606 265 RASSF6A NP 803876 337 RASSF6B NP 958834 1 - - ZZZZZZ IZ 369 RASSF7 NP 003466 ZZZZZ ZZZZZI — 373 RASSF8 NP 009142 ZZ — 392 RASSF9 NP 005438 -- 435 RASSF1O NP 001073990 --- ------------—-- 615 349 ‘4O3 RASSF5A 366 413 ------- -- _s-__ 122 170 274 364 418 B 166 RAS 357 30 Chapter 1 — Introduction 1.4.2 Rassfl and cancer Rassfl A and Rassfl C are identical at their C-termini, as they are different splice forms of the same gene, located on chromosome 3p2l.3. Their transcription uses two different promoters (3.5kb apart), resulting in a unique 120 and 50 aa N-terminus for RassflA, and RassflC, respectively. Very interestingly, deletions of the 3p2l .3 region are a common and early event in the formation of lung, breast, kidney and other cancers. Rassfl A inactivation is one of the most frequent marks so far described in human cancer (Hesson et al., 2007a). Mutations in the RassflA gene are rare in cancer cells, but rather the whole gene is often silenced by DNA methylation of its promoter, occurring in up to 100% of liver cancer cells, -60% of breast cancer cells, 80% of testicular cancer cells, -50% of ovarian cancer cells, and so on (reviewed in (Agathanggelou et al., 2005)). Since RassflC uses a different promoter, its inactivation by DNA methylation is not as frequent as for the Rassfl A gene, although loss of expression of RassflC does occur in some cancer cells (Li et al., 2004; van der Weyden and Adams, 2007). These observations have led a number of groups to use the Rassfl A as a tumor biomarker, prognostic marker, and even as a candidate for treatment with inhibitors of DNA methyltransferases (Hesson et al., 2007b). Based on this evidence, it is obvious that Rassfl genes must play an important role in normal cell function. So far, members of this family, particularly RassflA, have been implicated in the regulation of Ras signaling, cell cycle control, microtubule stability, and apoptosis regulation (Richter et a!., 2009). It is evident that they perform this by regulating protein-protein interactions with various players, and therefore, like DAXX, appear to function mainly as scaffolding proteins. 31 Chapter 1 — Introduction 1.4.3 Rassfl and DAXX Apart from Lindsay’s work (discussed in section 1.2.5), two papers have already implicated a direct interaction of DAXX with Rassfl proteins. Kitagawa et al. reported that DAXX and Rassfl C interacted directly at the PML-NBs during interphase. Interestingly, stress stimuli that produced DNA damage also induced degradation of DAXX, and released Rassfl C from the PML-NBs. Rassfl C then translocated to the cytoplasm and activated the INK signaling pathway, thus ultimately activating apoptotic pathways (Kitagawa et al., 2006) (Figure 1 .6A). This suggested mechanism is simple and appealing. However, Lindsay et al. failed to reproduce these results, and found that Rassfl C and DAXX are totally separated during interphase, with DAXX being recruited in PML-NBs while RassflC remained in the cytoplasm. Once in mitosis, they then did interact and presumably mediated cell- cycle checkpoints (Figure 1.6C, discussed in Chapter 3) (Lindsay et al., 2009b). It is unclear at the moment what conditions determined these contradictory results, but nevertheless the DAXX/Rassfl C interaction has (an) important biological function(s). Interestingly, both studies identified roughly the same interacting regions of both proteins: the N-terminus of DAXX and the N-terminus of Rassfl C. More recently, Song et al. reported a direct interaction between RassflA and DAXX (Song et al., 2008). This depended on the translocation of RassflA from the cytoplasm to the nucleus, which they speculated occurred upon DNA damage. They found that RassflA disrupted the interaction of DAXX with Mdm2 and Hausp, two regulators of p53. Such disruption resulted in the activation of p53 genes, which mediate apoptotic pathways (Figure 1.6B). 32 Chapter 1 — Introduction Figure 1.6 - Proposed models for mechanisms controlling DAXX/RASSF interactions. (A) Kitagawa et a!. reported that DAXX and RASSF1C interact at PML-NBs during interphase, and induced DNA damage resulted in DAXX degradation (presumably by ubiquitylation) and freed Rassfl C from PML-NBs. Subsequently, Rassfl C translocated to the cytoplasm and mediated activation of the INK-signaling pathway (Kitagawa et a!., 2006). (B) Song et al. reported direct interaction between RassflA and DAXX. This disrupted the DAXX-Mdm2-Hausp complex (Tang et a!., 2006), which regulated the ubiquitylation of the transcription factor p53. Rassfl A promoted Mdm2 auto-ubiquitylation and the increased level of p53 resulted in transcription activation (Song et a!., 2008). (C) Lindsay et al. reported different locations between RassflA/C (microtubules at the cytoplasm), which formed a heterodimer, and DAXX (PML-NB5 at the nucleus) during interphase. Once in mitosis, DAXX translocated to the microtubules via Rassf1 C, and they mediated a mitotic checkpoint important for the progression of cell division (Lindsay et a!., 2009b). 33 Chapter 1 — Introduction A PML bodies (Rascflc DAXX DNA damage Ub DAXX Proteasome degradation - &19 Cytoplasm , translocation iNK pathway i Apoptosis DAXX) MDM2 _—.-- DNA damage p53 _________ Pr:osorne degradation Proteosome degradation bX MDM2 (ssf:9 4 B ssf1A) DAXX - : () C Interphase Mitosis (prornetaphase) 34 Chapter 1 — Introduction Importantly, Song et al. found that the N-terminus of DAXX once again mediated the interaction, but identified the C-terminus of Rassf1 A (common to Rassfl C) as the region regulating the interaction. On the other hand, Lindsay et a!. identified the unique 50 aa N- terminus of Rassfl C as the DAXX interacting region. In Chapter 3, it is confirmed that indeed this region of Rassfl C contains a DAXX interacting motif (DIM). It is possible that a second DIM is present at the C-terminus of RassflAIC, which would explain Song’s results. Further experiments will help to clarify this. t5 Thesis overview It is well established that DAXX performs a number of important functions in the cell. Most of the knowledge on how DAXX functions as a scaffold protein has arisen from the ability of DAXX truncation constructs to bind the many effectors that DAXX regulates. However, these constructs have been designed based solely on a very poor knowledge of the structural domains of DAXX, derived only from sequence alignments with homologues and extrapolation of their function. The overall goal of this thesis is to provide a clear structural characterization of the domain boundaries of DAXX and the roles of the domains in mediating protein-protein interactions. This will aid future biological experiments that will improve our understanding of the detailed mechanisms that mediate the function of this important protein. Chapter 2 presents the first study of the domain composition of DAXX. Different constructs of DAXX that contained predicted domains based on a consensus of secondary structure and disordered region prediction algorithms were designed. This less biased approach led to the discovery of a well folded eighty-residue domain that is predicted to be separated from other structured domains by long intrinsically disordered regions. Using NMR spectroscopy, its three 35 Chapter 1 — Introduction dimensional structure was solved. Although this domain did form a helix bundle, its fold was clearly different to that of the reported Sin3 PAR domains. Thus it was named as the DAXX Helix-Bundle (DHB) domain. The fold of the DHB domain suggested that it would also mediate protein-protein interactions in a different mode to the one reported for Sin3 PAH domains. Chapter 3 presents the investigation of the interaction between the DHB domain and the unique 50 aa N-terminus of Rassfl C. Indeed, an interaction by NMR-monitored titrations was corroborated. After a long iterative process, the optimal conditions for the formation of a tight and soluble complex were derived, and thereby its three dimensional structure was solved by NMR methods. This structure will be valuable in understanding how DAXX mediates protein-protein interactions. Excitingly, it explains previously reported interactions and it might even predict future binding partners of DAXX. Finally, Chapter 4 presents the characterization of the SUMO interacting motifs (SIMs) within DAXX. These are localized in conserved disordered regions at both ends of the protein. The strength of the SUMO/DAXX-SIMs interactions under various conditions was measured and their binding mode was determined by NMR-methods. Moreover, how a sumoylated transcription factor, Ets- 1, is able to interact with DAXX was studied. Interestingly, this interaction is found to be completely dependent of the presence of SUMO, as unsumoylated Ets 1 cannot bind DAXX. Additionally, no cooperativity in the binding is observed, suggesting that the ability of DAXX to bind sumoylated transcription factors depends solely on SUMO, and not on the transcription factor. Finally, an exciting finding of an intramolecular binding of SIM-N with the DHB domain is shown. This could open the door to new mechanisms at how DAXX is regulated in the cell, discussed in Chapter 5. 36 Chapter 2 Structure of the N-terminal DAXX Helix Bundle The first structural characterization of DAXX is presented in this chapter. Sequence alignment and secondary structure prediction algorithms were used to define a number of constructs of DAXX. After cloning and expression, the C-terminal 1/3 of DAXX was found to be intrinsically disordered. In contrast, a well-defined folded domain was indentified near its N terminus. NMR spectroscopy was used to define the boundaries of this domain, to solve its three dimensional structure, and to determine its dynamic behavior. The calculated structure consists of five helices, and hence it is named the DAXX Helix Bundle (DHB) domain. This domain has a very different topology to the Sin3 PAH domains, which until now had been used as models for this region of DAXX. The differences in the structure between the DHB domain and Sin3 also suggest that each utilizes a very different mode of binding. In Chapter 3, this is confirmed by the structural analysis of a Rassfl Cl peptide with the DHB domain. This work provides the first clear domain boundaries in DAXX, which in turn will provide clues as to how this key regulatory protein operates at the molecular level in a number of diverse, important cellular processes. 37 Chapter 2 — Structure of the DHB domain 2.1 DAXX sequence and secondary structure Due to the lack of experimentally-determined structural information, various researchers used sequence comparisons to provide “best guesses” regarding the structure-function properties of DAXX (Figure 2.1). The first group to do so suggested the existence of two coiled-coils formed by residues 180-212 and 356-388 (Pluta et al., 1998). Coiled-coils are known to regulate oligomerization or protein-protein interactions (Grigoryan and Keating, 2008). A second group shortly thereafter predicted a “Pair of Amphipathic Helices” (PAH domains) between residues 64-108 and residues 192-240. This was based on weak sequence similarities with the transcription repressor Sin3 (Hollenbach et al., 1999). Researchers have also defined “domains” in the C-terminal portion of DAXX based on unusual amino acid content (Figure 2.1, top). The first one, between amino acids 434-485, contains 80% of GluJAsp and hence was denoted “acidic”. The second one is -48% Ser-Pro-Glu residues (SPE: residues 495-596). The third one (SPT: residues 665-740), which is 46% Ser-Pro Thr residues, was reported to bind the transcription factor Ets-1 (see section 4.6). These sequence analyses reflect very low resolution views of DAXX, and thus a more exhaustive bioinformatic approach was undertaken to better identify possible structural domains within this protein. It is clear that three-dimensional structural information is needed for a detailed understanding of the function of DAXX. However, unstructured and disordered regions often prevent successful crystallization of a protein, or make the analysis of its NMR spectra much more difficult. This is particularly true for large, multi-domain proteins, which often behave as “beads on-a-string”. Therefore, following a reductionist’s approach, it is often necessary to define minimal structural domains for bacterial expression and in vitro characterization. 38 Chapter 2 — Structure of the DHB domain < >s > <-> CL CC CC 55 DHB 144 180212 35__fi DAXX []] I’AmI idi4 SPE I SPT[] 1 17 64 108 192 240 434 485 495 5% 665 723 740 PHD HNN SOPMA NNPREDICT JNET SCRATCH PORTER PSIPRED 0 50 100 150 200 250 300 350 400 450 500 550 600 650 700 750 Residue Figure 2.1 - Suggested “domain” organization of DAXX. (Top) Two “Pair of Amphipathic Helices” domains (PAH1 and PAH2) have been proposed to be located in the N-terminal portion of the protein (Hollenbach et a!., 1999). A region between amino acids 434-485 contains 80% Glu/Asp residues and hence termed “Acidic”. The C-terminus contains regions rich in Ser/Pro/Glu residues (SPE; residues 495-596), and in Ser/Pro/Thr residues (SPT; residues 665- 740). Two segments were suggested to form coiled-coils (CC) (Pluta et a!., 1998), the first one within the putative PAH2 domain (residues 180-212), and the second one between PAH2 and the acidic region (residues 356-388). Based on the analysis of Figure 2.2 using more extensive database, DAXX has three regions of conserved sequence, identified as CL (green, 60-145), 3 (orange, 1 85-400) and y (blue, —430-470), as well as two SIMs (red) and the DHB domain characterized in this chapter (magenta). (Bottom) Secondary structure prediction of DAXX using the following sequence-based algorithms: PHD (Rost, 1996), Hierarchical Neural Network (HNN; Guermeur et a!., 1999), the improved self-optimized prediction method (SOPMA; Geourjon and Deleage, 1995), Neural Network Predict (NNPredict; (Kneller et a!., 1990), JNET (Cuff and Barton, 2000), SCRATCH (Cheng et al., 2005), PORTER (Pollastri and McLysaght, 2005), and PSIPRED (McGuffin et a!., 2000). Predicted cL-hehces are shown in black and predicted 13-strands in red. 39 Chapter 2— Structure of the DHB domain Comparison of the sequences of DAXX from various vertebrate organisms shows proteins of similar size with a number of regions that are highly conserved (Figure 2.2). Remarkably, the N- terminal and C-terminal segments of the protein each contain an essentially invariant SUMO Interacting Motif (SIM), which forms the basis of Chapter 4. Two additional regions that show a high degree of conservation are located between residues 60-145 (termed region c) and residues 182-397 (region 13). A more variable-length conserved region rich in Asp and Glu amino acids (residues 434-485, region y) is also present in each organism. These conserved regions suggest the existence of functional domains in the protein. As a first step to define the boundaries of possible structural domains, the presence of secondary structure elements in the protein was searched, particularly in the conserved regions. Recent advances in bioinformatics have provided a number of sequence-based algorithms that predict the secondary structure of proteins with relatively high reliability (Rost, 2001). Since each algorithm has strengths and weaknesses, a consensus of the following web-based servers to predict the secondary structure of DAXX was used: HNN (Guermeur et al., 1999); NNPredict (Kneller et a!., 1990); INET (Cuff and Barton, 2000); SCRATCH (Cheng et al., 2005); PORTER (Pollastri and McLysaght, 2005); PSiPred (McGuffin et al., 2000); PHD (Rost, 1996); and SOPMA (Geourjon and Deleage, 1995). 40 000 ______________ REDS 6CR - ATTLE MONKEY _____ 3ICU.CE .IALMCCI DCCC C. LCAr.A CR31 LE 7. lIKES 5CC-IAN -VINE DOG ICEWAC ClAD -. 6150655. STYLE V .6V7NEY ClAD 55514(5 DoG CCRWAY RAT ___ V. 615351CR CATTLE V MCNCCSY HOC-IAN MOOSE SALMON 05CC NC 54155 HAl -. MEORCA CATTLE C. MONKS? HAS ODE .IELLVIEV’. FE:lCA:I’c: F DIV I F5.CSE5COCDTVl.Z .. -C 5Y67.’V.. 11l7:sGf P . C5DD 233 .CNCEPNFCCCCIC 0 556 - .LLEP3CDL?\ ‘_AELEC-4153P5’ - ECI ‘hyC’sED.:DCIJDL 5’ árWCZADlL C 0.3,, LLGGCYIKEC I4A.,VJ .0 553 SON: s-s4:oo H\’SADHISTsC’-D5YAsp .:-.DLO-OS:vzINT410 DOT :.,E1jRSDC DrrE3TVAD C KECOK: YiVL2s0: :020.1(3 041 INFD’,COSPJ ‘LFEI.ERAL’1FPEFFDI ‘.11150 EERL”TVCF 010 .‘CCC :Ts, -il B AEPD0N-YUEP.C ‘ASs Cl 1- C— L I El- F A C 51 Cl IL IF F L 1EE LE k TF 1 T/ 1 F 71111 <—> SIMC 74 577 DNDA——-.61 1l- CCI 743 6115 sppsFLC’-:DLIVA:cAP::KrpCIs:FHDFGTI 737 roSSPLA’.:,:_.. 1’ J L’C SC.I_-JCI’. ‘355 6011 078 - ‘D’ A .,‘53H COD F DIII’ 576 SPFSP:?I- OD4 ‘41 IT 1 :C4:51T60:OSSOPGTI 74E 664 32605 1, 5- ‘•- 3. CDA4L_SC110_’, 16656531 766 Figure 2.2 - Alignment of the DAXX sequence. The following organisms were used for the alignment: (common name, accession numbers): Homo sapiens (Human, CAG33366), Mus musculus (Mouse, AAC53284), Salmo salar (Salmo salar ABQ59664), Canis familiaris (dog, CAl 11447), Rattus norvegicus (Norway rat, NP_543 167), Orzias latipes (Japanese Medaka, BAD93268), Bos taurus (Cattle, AA134603), and Cercopithecus aethiops (Vervet monkey, AAB66586). Positions with 90% conservation identity are highlighted in black, residues with 90% conservation similarity are highlighted in gray. There are two SIMs in all the forms of DAXX, one at its N-terminus (SIM-N) and one at its C-terminus (SIM-C), shown with red double-headed arrows. Three regions with high degree of conservation and predicted to have secondary structure elements are also identified with arrows and named in order as (green), f3 (orange) and ‘y (blue). The structured residues of the DAXX Helix Bundle (DHB) domain characterized in this chapter are marked (magenta box). Chapter 2— Structure of the DHB domain E—> SIM-N - I LE1E; Ljj !! 116 Dli —is —ii 1031,2 ‘1 SCLl.CEKRiD5E3I /C4V.CI.SL LHECs1AC.4 CF 0. — _Jii! I F S H ER HzI I; V Y ICC —. -i-IF F F EEE1T 11 MI5NDOFL’ IDEFEHAA .C’T)C!NS?N —. ,OICCC’KlCC FL 1105.0 1.C,sClCC’p:o’p,CI LSCCFPLAPIC.It ‘“6 456 EECE EENT 5L3ENG SEGNE— ———GCIESFD PVFFH2R—NSE?HEN’LFT?EC54OiCLTEP55Po.ADLDPIVDL LI 467 - ..- -I 1-CA: - - -5503-Es:-. C /EI1FEI .IV’1E-O’C EEDENAEVOIN-—:lA1SGN3PI:QSYT. —:.:NLvDE5510:E 448 N C cE:T; I SI-LIE.Lo5: :—co:s:ssto ‘:61 Fpr,ISTE1:l;LEFSC:3l3R:MSIV.:. FTVF5-—:ECI LAFC::0 Cl 445 5 6 - KEs’I- ISO-CHILI 1 s1CE 1’6 -. C CDI ,‘l’IIF 211141 ‘Cll(RCCDGl LENL.CLDS1ASiLN-C.0 GLFVL 415 --—-16” IA: rQs:-. C C - -366551 ICEs]- -—0 .. C:CIFA-EE——Es:Tt,C,-po,:IE000, :-[-CASIFAHSE’;C -0 445 LEEI6LE50LEEE_l - 1 .1-1655) C 1_E5RE___CDC C, ‘CC OIIiTARICLLFECEEC. Es-’.&.,DJFCCLSEEIFAFS .411. 4-CF . r - FT ‘‘SLFDCFD 5FCsHp]IA’,C’0 5 QT1TEBFiLFlKo7H ‘VIE. .‘:——--.p C C,:lAEPLCP.I.CIF - ‘C HOC-CAD ICE 053.-lOll 304 IV MEOAFS. V. 14115110 81465 -V-C 56 15515011 DOS N 41WAI i-IL V. ME-ASIA CADDLE 7. IKNREY 41 Chapter 2— Structure of the DHB domain As summarized in Figure 2.1, the N-terminal 2/3 of DAXX is predicted to contain many c helices that cluster roughly into the three regions (cL, i, and y) of highest sequence conservation (Figure 2.2). In contrast, the C-terminal —l/3 of DAXX is predicted to be predominantly unstructured (“intrinsically disordered”), as expected from the abundance of polar Ser/Thr/Pro/Glu residues. Thus, secondary structure prediction algorithms and sequence conservation suggested the presence of clear domain boundaries. 2.2 Characterization of DAXX constructs With the aim of using NMR spectroscopy to characterize potential structural domains of DAXX, a set of constructs encoding various truncation fragments corresponding to the two most conserved regions, c. and f3, were cloned (Figure 2.3). The region ‘ was not investigated, as it contains the acidic “domain” with 31 Glu and 12 Asp residues and thus would not be expected to be well structured. The constructs were tested first for protein expression in E. coil BL21 (2DE3) cells. In the cases where protein of the expected size was indeed produced, cells were fractionated to differentiate soluble DAXX fragments from those forming insoluble inclusion bodies (see section 2.6.2 for details on the procedure). Of twenty three constructs tested, four expressed soluble proteins (Figure 2.3, blue boxes, Appendix 2): human DAXX46’60,DAXX’61243, DAXX347420,and murine mDAXX566739.The first two contain the putative PAH1 and coiled coil/PAH2 domains respectively. The third construct covered another predicted coiled-coil (Figure 2.1). The fourth construct contained the C-terminal SPT region (including SIM-C). In the absence of any functional assays, the four soluble DAXX fragments were examined for structure by NMR spectroscopy. By far the best construct was DAXX46’60,yielding a 42 Chapter 2— Structure of the DHB domain spectrum that contained over one hundred dispersed, sharp signals indicative of a likely monomeric, well structured protein (Figure 2.4). This region contains the putative PAH1 domain, and thus it was decided to solve its three-dimensional structure as a route towards understanding its function. In contrast, DAXX’61-243 precipitated when dialyzed into NMR buffer (see section 2.6), and was found to be stable only in Ni2 elution buffer which is high in ionic strength (500 mM NaC1) and has a pH of 7.4. The spectrum of this construct under the latter conditions still showed only a limited number of broad signals, indicative of aggregation. Furthermore, mass spectrometry revealed that the sample was also degraded. This result was disappointing as DAXX’61243 contains the putative PAH2 domain. Further deletion or addition studies may be necessary to obtain fragments of this region that remain stable and soluble, as required for any structural analysis. DAXX34742°was stable in NMR sample buffer and gave a good ‘H-’5N HSQC spectrum, albeit with much less dispersion than exhibited by DAXX46’6°(Figure 2.4). Although such limited 1H’ dispersion is often indicative of a predominantly unstructured polypeptide, coiled coils often give poor dispersion despite having secondary structure (Kovacs et al., 2002). Indeed, the circular dichroism (CD) spectrum of this construct did show features diagnostic of cL-helix content (Appendix 1). It is interesting to note that this construct contains the region originally predicted to contain a coiled coil (Figure 2.2) (Pluta et al., 1998). However, an improved prediction algorithm for coiled coils gave a probability score of only l6% for residues 359-399 (McDonnell et al., 2006), so further structural characterization is needed to determine if this region is indeed involved in coiled coil formation. 43 Chapter 2— Structure of the DHB domain [J I’I I 17 64 100 Figure 2.3 - Identification of structured domains in DAXX through deletion analyses. All constructs were His6-tagged except for two that were GST-tagged (*). Expression tests were performed in BL21 (2DE3) cells grown at 30 °C for 5 hours. Blue frames highlight constructs that expressed soluble proteins. The last construct (566-739) was made using murine DAXX (mDAXX) instead of human DAXX. For reference, a schematic drawing of the proposed domain organization (described in Figure 2.1) of DAXX is shown on the top. <>4— cc 100 212 cc 356 360 PAHZ [idj SPE SPT Expression Solubility 192 240 434 480 495 596 665 724 730 1-160 No - 46-160 Yes Yes 46-420 < Yes No 55-243 Yes Partial 161-243 Yes Yes 161-342 Yes No 161-420 Yes No 181-335 Yes No 181-365 Yes No 181-385 Yes No 181-407 Yes No 181-365 Yes No *181407 Yes No 206-335 Yes No 206-365 Yes Partial 206-385 Yes Partial 206-407 Yes I’artial 246-342 —> Yes No 261-420 ——.——> Yes No 261-342 No - 279-342 No - [420 Yes rn566-739 Yes Yes 47-560 1 Yes No 44 10 9 8 Figure 2.4 - 1H-’5N HSQC spectra of soluble constructs of DAXX. All spectra were recorded 161-243 .2+in NMR sample buffer at 25 C except for DAXX , which was recorded in Ni binding buffer (50 mM K2HPO4buffer, 500 mM NaC1, pH 7.5) to minimize precipitation at lower ionic strength. The dispersion of the amide resonances in the DAXX46’6 spectrum is indicative of stable secondary and tertiari structure elements. DAXX161 243 gave a poor spectrum, probably due to aggregation. DAXX 47-420 showed slight IHN dispersion, which along with CD studies (Appendix 1), suggested that it may be partially structured. mDAXX566739 had random coil IHN shifts, indicating that it was predominantly unstructured. Chapter 2 — Structure of the DHB domain DAXX’61243 •0, D.A)cX466° 0 e • rO o 0 C 0 0 0 C C 110 115 120 125 110 115 120 125 DAXX742° C 0 000 • 0 C ‘H (ppm) 7 ‘H (ppm) 45 Chapter 2— Structure of the DHB domain mDAXX566739 gave a typical spectrum of an intrinsically unstructured (or natively disordered) random coil polypeptide, with relatively sharp amide 11N resonances clustered in the narrow 7.5 - 8.5 ppm range (Figure 2.4). Interestingly, this region has been reported to interact with a number of proteins (discussed in Chapter 5, Figure 5.1C). It is hypothesized that these interactions may occur via “linear peptide recognition sequences,” such as the conserved SIM-C (Figure 2.1). Indeed, SIM-C was able to bind independently to a sumoylated transcription factor (section 4.6). It is very likely that the partially conserved region 675-699, which is very rich in Ser/Thr residues, has a yet uncharacterized biological role as well. 2.3 Mapping of the DAXX N-terminal structural domain boundaries Unstructured polymer segments typically present a challenge for the elucidation of the three dimensional structures of the ordered regions of biomolecules, including proteins, nucleic acids, and carbohydrates. In the case of X-ray crystallography, they both interfere with the crystallization of the molecule and fail to give detectable electron density in any resulting crystal. In the case of NMR spectroscopy, they often yield strong, poorly dispersed signals that overlap other resonances co-incidentally in the same spectral region. Furthermore, they are often susceptible to aggregation, chemical degradation or proteolytic cleavage, thus producing a heterogeneous sample that is difficult to analyze. Also, disordered regions typically do not yield useful long range NOEs that might otherwise provide possible restraints on their conformations. Thus, unless the unstructured region has a known biological function, it is usually prudent to first work with minimal-sized constructs containing only the folded domain of interest. Once structurally characterized via this reductionist approach, one can then undertake a constructionist approach to investigate the domain within its native context. This is particularly true given the more recent recognition that disordered segments of proteins indeed have numerous important 46 Chapter 2 — Structure of the DHB domain functions. With this goal, the main-chain ‘H, ‘3C, and ‘5N resonances of DAXX46’6°were assigned, to enable a better definition of the boundaries of the structural domain within this truncation fragment via chemical shift analyses and ‘5N relaxation measurements. 2.3.1 Chemical shift assignments NMR spectroscopists have a number of powerful multi-dimensional heteronuclear experiments at their disposal for assigning signals from the main-chain or backbone nuclei of proteins (a description and pulse sequences of these experiments are reviewed in (Sattler et al., 1999)). Two of these, namely the HNCACB and CBCA(CO)NH, were used to assign the backbone nuclei of DAXX46’60. The HNCACB pulse sequence transfers magnetization “out-and-back” from the amide proton to the directly bonded nitrogen ‘5N nucleus, then to both the ‘3C° and 13C of the same residue (i, intra-residue) and of the preceding residue (i-i, inter-residue) (for frequency labeling in ti), then back to the nitrogen (t2) and finally back to the amide proton for detection (t3) (Figure 2.5A). This means that in the final three dimensional spectrum, at each amide 1H” (wa) and ‘5N (02), there will be four cross-peaks: two from the intra-residue 13Cx and ‘3C nuclei, and two from the preceding ‘3C’ and ‘3C nuclei (Figure 2.5B). Conveniently, the 13Cu and ‘3C signals have opposite signs due to a constant time delay used for the pulse sequence. The CBCA(CO)NH pulse sequence transfers magnetization from the ‘H”/’H to the 13C/’3 (t,), then via the carbonyl to the nitrogen of the next amide (t2) and finally to the amide 1HN for detection (t3) (Figure 2.5A). The result is that each residue gives only two cross-peaks in the three dimensional spectrum: one for the 13C nucleus and one for the ‘3C nucleus of the residue (i-i, w,) preceding the amide (i; 02,03) in question (Figure 2.5B). 47 Chapter 2 — Structure of the DHB domain 111 (ppm) Figure 2.5 - Magnetization transfer pathways for experiments used for backbone assignment. (A) Magnetization transfer pathway for the I-fl\TCACB (1) and CBCA(CO)NH (2) experiments. Arrows indicate the direction of the magnetization transfer, and circles define the atoms whose chemical shifts are recorded in the experiment. (B) Example of how the HNCACB and CBCA(CO)NH experiments were used to assign uniformly‘3C/’5N-labeled DAXX46’60. Both experiments give cross peaks for 13C and ‘3C resonances of the residue before (i-i), but HNCACB also gives cross peaks for the intra-residue (1) ‘3C and ‘3C. Additionally, in I-LNCACB spectra, positive peaks arise from ‘3C nuclei (black), and negative from ‘3Cr nuclei (red). All this information can be used to connect sequential residues (dotted line). Shown are o (vertical; 13c) - (03 (horizontal; 1H’) strips taken at the ‘5N chemical shift (o2) of the indicated residues. HNCACB CBCA(CO)NHA B H—,.©—H H—©çH çII) — N — C —C&•—©— C I I II II H H 0 0 r H—©—H H—Co—H c1I 0. cr - Cr -•+ C - Cr oS c;1 AIO3NH SIO2NH II II It A1OINH LIOONH 48 Chapter 2— Structure of the DHB domain By using these two experiments, one “walks” along the backbone of a protein, linking crosspeaks from the intra-residue (i), as well as the inter-residue (i-i) l3Ca and ‘3C correlations with the IHN and 15N signals of each amide (i). Ala, Gly, Ser and Thr have very diagnostic l3Cx and 13C chemical shifts, and thus are easily identified as starting points for this process. In the ideal case it is possible to unambiguously assign most, if not all, of signals from the backbone nuclei of a protein using this approach. In cases of ambiguity, due for example to extensive spectral overlap, complementary pulse sequences such as those that detect the carbonyl carbon can also be used to resolve problem areas. However, for DAXX46’60,the H1TCACB and CBCA(CO)NH experiments were sufficient to assign most of the resonances from the backbone nuclei in this fragment. 2.3.2 Identification of the folded domain With spectral assignments in hand, the boundaries of the folded domain within DAXX46’6° were easily refined. This was done by determining the secondary structure of each residue (section, by studying the relaxation dynamics of its backbone ‘5N (section, and by making a number of constructs differing slightly in length to determine the boundaries of the minimum folded domain within the DAXX46’6°fragment (section DAXX4616°secondary structure The chemical shift of a ‘3C’ nucleus (as well as 13Ca, ‘3C, ‘He) is extremely sensitive to its secondary structure. In particular, the 3C of an amino acid shifts downfield with respect to the corresponding random coil value in x-helices and upfield in 13-sheets (Wishart and Sykes, 1994). Therefore a plot of “secondary chemical shifts” (l3Caobserved — 13Crandom coil) is diagnostic of the secondary structure of a protein, with contiguous stretches of residues having positive 49 Chapter 2— Structure of the DHB domain 13Cu secondary chemical shift values readily identifiable as being in helices, and negative values indicating the presence of a f3-strand. Figure 2.6A shows that amino acids 60-136 in DAXX46’6° contain a number of c-helices, and no n-strands. Remarkably, these boundaries agreed well with the secondary structure prediction algorithms (Figure 2.2) and with the conserved region a. in the alignment of DAXX sequences from various organisms (Figure 2.1). 15N DAXX4616°relaxation analysis A complementary way to determine the boundaries of the folded domain is to examine the NMR relaxation behavior of its backbone. ‘5N relaxation data reflect the local mobility of the amide bonds in the polypeptide backbone on a residue-specific basis. In general, the heteronuclear NOE decreases from approximately +0.8 to -3.6 with increasing fast timescale mobility of the ‘5N-’H bond (Kay et al., 1989). Decreasing ‘5N {‘H}-NOE values indicate enhanced flexibility on a nsec-psec timescale. Folded regions tend to have elevated ‘5N {‘H} NOE values (i.e. > 0.6) due to the lack of fast timescale flexibility in the backbone, whereas unfolded regions have much smaller or even negative ‘5N{1H}-NOE values. Figure 2.6B shows that amino acids 60-140 of DAXX46’6°were well ordered, as expected based upon their predicted cL-helical secondary structure, whereas the flanking residues were significantly more mobile and thus conformationally disordered. Comparison of various DAXX constructs To refine the boundaries of the DAXX fragment containing a structural domain for minimal size and optimal expression and spectral quality, four additional constructs based on chemical shifts and relaxation analyses were made. DAXX59’4was first examined, which contained the minimum folded domain and does not have an oxidizable N-terminal Cys58. However, this 50 Chapter 2 — Structure of the DHB domain fragment was almost completely insoluble, and thus presumably too short to fold stably (Figure 2.6C). The N-terminus was then extended and Cys58 was replaced by Gly. DAXX5544C8G proved to be more soluble, but still with a significant amount of protein in the insoluble fraction of an E. coil extract. In contrast, DAXX55’4retained the wild-type Cys residue, and exhibited improved solubility. Finally, the C-terminus was extended slightly to produce DAXX55’44.This construct was completely soluble when expressed in E. coil (Figure 2.6C). A superimposition of the ‘H-15N HSQC spectra of DAXX55’44 and DAXX46’6°showed that the optimized fragment indeed retained the structured elements from DAXX46’60.The only significant spectral changes were the absence of signals with random coil 1HN chemical shifts. HNCACB and CBCA(CO)NH spectra of DAXX55’44 were recorded to confirm previous assignments and assign signals from residues close to the new N- and C-termini, which expectedly differed slightly from DAXX46’60.Figure 2.7 shows the fully annotated ‘H-’5N HSQC spectrum of the optimized construct DAXX55’44. 2.4 DAXX55-144 structure determination 2.4.1 Assignments of side-chain signals The next step towards determining the three dimensional structure of DAXX55’44 was to assign the signals from its side-chain ‘H, ‘3C, and ‘5N nuclei. The successful calculation of a protein structure by NMR methods generally requires the assignment of at least 90% of all its side-chain 1H signals (Habeck et al., 2004). Therefore, a suite of three dimensional heteronuclear experiments was employed to assign the chemical shifts of nuclei in the aliphatic and aromatic side chains of DAXX55’44(Figures 2.8-2.10). 51 C --:a... •s •‘— -*1’ SW— -e—— e= _____ . 55-141 55-141C’50C 59-141 55-144 0,0 7,5 7.5 ‘El (ppm) Figure 2.6 - Identification of the boundaries of the structural domain within DAXX46’60. (A) The difference between the observed and corresponding random coil l3Ca chemical shifts. A stretch of residues whose ‘3C values shifted more than ‘—1 ppm downfield (dotted line) indicate the presence of an x-helix, whereas a stretch of residues with ‘3Cc values that are ‘—1 ppm or more upfield shifted (dotted line) indicate the presence of a 13-strand. According to this criterion, residues 60-136 contained at least 4 x-he1ices (double-headed arrow). (B) Heteronuclear 15N {‘H}-NOE values of DAXX46’6°reveal that the fast timescale dynamics of residues 60-140 correspond to those of a well ordered domain. Missing points correspond to prolines or residues with overlapping or anomalously weak signals. (C) Expression tests of DAXX55’41,DAXX55 141C58G DAXX59’41,and DAXX55’44.Lanes identified as M: molecular weight marker (see Appendix 2 for size marks); un: cell lysate of cells prior to induction; 5h: the total cell lysate 5 hours after induction; sol: soluble proteins in the cell lysate 5 hours after induction; ins: extracted proteins from the insoluble pellet 5 hours after induction. DAXX55’44 showed the greatest solubility of all the constructs. (D) Superimposition of the ‘H-’5N HSQC spectra of DAXX46’6° (red) and DAXX55’44 (black). The excellent overlap between the two spectra indicated that the DAXX55144 construct had the same secondary and tertiary structures as in DAXX46160. The spectra differed by the absence of signals within the random coil region. Chapter 2—Structure of the DHB domain 8 B I -2 -4 C z -0.5 z 110 Is 120 125 I D Q e . eQ 0 0 O •- 0 • 0 e 50 60 70 80 90 100 110 120 130 140 150 Residue 9,0 0,5 160 52 Chapter 2 — Structure of the DHB domain 110 115 I 120 125 9 8 7 ‘H (ppm) Figure 2.7 - Annotated 1H-5N HSQC spectrum of DAXX55’44.(1 mM, 10 mM phosphate, 100 mM KC1, pH 6.5, 25 °C). Signals from main-chain and AsnlGln side-chain amides are identified. The peaks corresponding to the side chain‘5N2H of Asn63 were aliased from an unusual chemical shift of 103.4 ppm in the ‘5N dimension, due to proximity to the aromatic ring of PhelO5. T78 G55 fJQ77NC S114 S118 N107NS )110 4 Yi24 ‘1R117 -Ii 0S138 ‘Q93N Q3 Q77 8137 S102 4C74 ‘C131 UK122 N90 )K91 E104 L61 N107 -H81 P132 NJ. N63Ø Y89 iDJ24fl128 870E83N—HN H96. 4 F.67t rI 3F9911271 A121—.1tKG 1 865 A136L123 L109 0 71116 KV84 L73114 ‘:)C106F/O.. 4 1 I 1k 3JALO1 4L8s 1108 v12S ?L71 11)E62 - yKi35 )RliJ P126 E129 - KS ‘.1. . 4V1 33 K75 .741, ‘L98 1,97 8147 7E64 O81444cs8 ,,.7119 L143’O N636 03 .3A139 53 Chapter 2— Structure of the DHB domain Assignment of signals from side-chain aliphatic nuclei The side-chain assignment strategy naturally uses the assigned main-chain or backbone nuclei as starting points. Figure 2.8 shows how this is achieved for A1a95 and Arg94. In the first experiment, the (H)CC(CO)NH-TOCSY, signals from all of the ‘3C nuclei in residue (i-i) preceding the amide group of residue i are detected. It is often possible to identify these signals as arising from 13Ca, ‘3C, ‘3Cr, l3Co or 13Cc based on expected chemical shift ranges. in case of ambiguity due to overlap or to anomalous shifts (i.e. resulting from tertiary structural effects such as the close proximity to an aromatic ring), additional experiments that detect only specific signals can be used. For example, the previously introduced CBCA(CO)NH only detects ‘3C and The assignment of signals from aliphatic hydrogen nuclei follows a similar strategy. The H(CC)(CO)NH-TOCSY experiment detects the side-chain ‘H of the residue (i-i) preceding a specific amide (i). The possible degeneracy of ‘Hp, ‘H and ‘H chemical shifts in some residues, such as lysine, is worse than with their carbon counter-parts, so the HBHA(CO)NH is very useful, since it displays only the ‘H and ‘H chemical shifts. A complementary solution is to use the HCCH-TOCSY experiment, which both correlates signals from all the aliphatic carbon and hydrogen nuclei in a residue and identifies the directly bonded‘3C-’H pairs. One problem with the amide-directed experiments is that they fail to provide assignments for residues that are followed by a proline. For those residues, a‘5N-TOCSY-HSQC experiment was used, which correlates signals from intra-residue side-chain protons with that of the main-chain 1HN 54 Chapter 2— Structure of the DHB domain Figure 2.8 - Experiments used for the assignment of NMR signals from the side-chain aliphatic 13C and ‘H nuclei of uniformly‘3C/’5N-labeled DAXX55’44.(Top) Magnetization transfer pathways for (H)CC(CO)NH-TOCSY (1) and CBCA(CO)NH (2), H(CC)(CO)NH TOCSY (3), HBHA(CO)NH (4), HCCH-TOCSY (5), and‘N-TOCSY-HSQC (6) experiments. Arrows indicate the direction of the magnetization transfer (open arrows for TOCSY steps), and circles define the atoms whose chemical shifts are recorded in the experiment. (Bottom) Representative example of the assignment of the side-chain atoms of A1a95 and Arg94 using the above experiments, using o (vertical; ‘3C or ‘H) - o (horizontal; 1HN) strips taken at the ‘5N chemical shift (02) of the indicated residues. 55 Chapter 2 — Structure of the DHB domain H-H —N——C—() (H)CC(CO)NH-TOCSY cD —N—C I I’) II I.) HBHA(CO)NH R94-C’ o. 1I ;) i CBCA(CO)NH © N I I’) II Fl © 0 H HCCH-TOCSY —N— I I) N 1) H®O® H(CC)(CO)NH-TOCSY © -®-C—C— N\CJW TOCSY-HSQC 22 20 © A95-I-1 R94-H R94-H’ R94-H -3 R94-H 20 30 2 p40. 50 60 70 0 O R94-C 40 ‘0 R94-C’ 50 O R94-C’ 60 70 A95-NH 2 ) R94-C ê4 R94CA A95-NH A95—NH -3 3 3 RM1T 0 4 4 A95-H” A95-NH A95-NH 56 Chapter 2— Structure of the DHB domain Stereospecific assignments Va! and Leu methyl groups were stereospecifically assigned using the method described by Senn et a!. (Senn et al., 1989). In short, DAXX55’44was expressed in M9 media containing 10% ‘3C6-giucose and 90%‘2C6-giucose. Stereospecific biosynthesis makes the pro-S‘3C-labeied methyl groups (Val C2 or Leu C62) have a‘2C-labeied neighbor, whereas the pro-R‘3C-labeied methyl groups (Val C’ or Leu C61) have a‘3C-labeled neighbor. These can be distinguished by 13C-’3 couplings in a high resolution ‘H-’3C HSQC spectrum or by the sign in a ‘H-13C CT HSQC spectrum. Stereospecific assignments of the side-chain amides of Gin and Asn were carried out using the EZ-HMQC experiment (McIntosh et ai., 1997). This approach exploits the difference in the 3JC-H coupling between ‘3C (y for Gin, 13 for Asn) and its Z- and E- amide proton (7 Hz for the former, negligible for the latter). Assignment of signals from side-chain aromatic nuclei The assignment of aromatic side-chain signals follows a different strategy, since Trp, His, and Phe/Tyr all have distinctly different spin systems (and ring nuclei do not show correlations in standard main-chain-detected experiments). Figure 2.9 shows the procedure used for assigning signals from the aromatic rings of Tyr89 and Phe67. The first step is to link the previously assigned ‘3C shift with the ‘H6 shifts of aromatic resides using the (HB)CB(CGCD)HD experiment. The corresponding ‘3C6 shift can then be identified in a ‘H-’3C CT-HSQC spectrum. The IHe shift can be obtained either from the HCCH-TOCSY spectrum using these ‘3C6 and ‘H6 shifts, or from the (HB)CB(CGCDCE)HE experiment. Finally, for Phe residues, the ‘3CC and ‘H shift can be obtained from the HCCH-TOCSY spectrum. 57 Chapter 2— Structure of the DHB domain Figure 2.10 shows the steps in assigning signals from histidine imidazole rings using His96 as example. The first steps are as above, assigning the ‘3C62 and 11162 using the (HB)CB(CGCD)HD and CT-HSQC experiments. Then, the ‘H-’5N HMBC experiment was used, which exploits the small 2JNH and 3JNH couplings to assign the 15NE2, ‘W’, ‘5N6’ and 11162 shifts. Note that the labile nitrogen-bonded protons of histidines are usually not observed due to rapid exchange with water, and that the chemical shifts of an imidazole ring are highly diagnostic of its charge state and, if neutral, its tautomeric form (Pelton et al., 1993). His96, and His53 were found to be neutral with the proton on the HE2 position, and His8l, and His137 were found to be positively-charged. Although DAXX55’44 does not contain any Trp residues, the RassflC2338”peptide discussed in Chapter 3 contains one. The steps in assigning Trp residues are similar, using the (HB)CB(CGCD)HD experiment to obtain the ‘H6’ shift as starting point, and the 15NE/H shifts are easily derived from a‘5N-TOCSY-HSQC and5N-NOESY-HSQC spectra. The rest of the aromatic ring can be assigned using HCCH-TOCSY and‘3C-NOESY-HSQC experiments. Combining all of these strategies, it was possible to assign signals from 98% of all non-labile ‘H, ‘3C, and ‘5N nuclei in DAXX55’44(Appendix 7). 58 Chapter 2— Structure of the DHB domain V H% / C\- :: t I I II I I II I I II I I II H H C) H H 0 H H 0 H H 0 (HB)CB(CGCL))HD (HB)CB(CGCDCE)HE CT-HSQC HCCH-TOCSV (HB)CB(CGCD)HD (HB)CB(CGCDCE)HE38 - .j 38 \9Y89-C1_H () 1 Y89-C-H I F67-CH 40 F7_C_Hb 40 ______ CT-HSQC _________ I. — _ :F679 F67Sc4 :: — 133.0 V 131.0 ‘\\ \%/ 118.5 1335 Y89-C-H’ 1315 1190 - HCCH.TOCSY 6.8 6.6 7.1 q.N 7.1 F67_C_HiHh I - . F67-C -H —H72 7,1 72 7.3 .. 7.3 .. - 7A Y89-C H -H Y89-c6-H-H F67-C -H -H 7.4 7.2 7.0 6.8 6.6 7.4 7.2 7.0 6.8 6.6 ‘H (ppm) ‘H (ppm) Figure 2.9 - Steps involved in the assignment of signals from the aromatic side-chains of Tyr and Phe residues in uniformly‘3C/’5N-labeled DAXX55’44.Magnetization transfer pathway for a Phe residue in the (HB)CB(CGCD)HD, (HB)CBCGCDCE)HE, CT-HSQC and HCCH-TOCSY experiments. Arrows indicate the direction of the magnetization transfer (open arrows for TOCSY steps), and circles define the atoms whose chemical shifts are recorded in the experiment (see text for details). 59 Chapter 2— Structure of the DHB domain Figure 2.10 - Assignment of signals from histidine residues. The aromatic ring of His96 in uniformly‘3C/’5N-labeled DAXX55’44was assigned using (HB)CB(CGCD)HD, CT-HSQC, and HMBC experiment. The sign of the peaks in the CT-HSQC experiment depends on the number of neighboring ‘3C nuclei. Positive peaks (black) have either zero or two ‘3C nuclei directly connected, whereas negative peaks (red) have one. The circles indicate the atoms detected in each particular spectrum. (HB)CB(CGCD)HD H96Ct-H CT.-HSQC H96C II 1T96 N II 01196 N II 1-IMBC Ic //“.\ Ic /(,.\ II— —II (I) u— /I\ CT-HSQC 31.5 - 32.0 9 32.5 1175 - 118.0 9 118.5 2 P168 € :172 138.5 - 139.0 9 139.5 2 9 F I I96-C”-1-1 Q H96-N-H SAl H96.N’-H’ 7.5 7.0 H (ppm) HMBC L244 -248 6.5 6.0 60 Chapter 2 — Structure of the DHB domain 2.4.2 DAXX55144 secondary structure As discussed above in section, the “secondary chemical shifts” (sobs - random coji) of main-chain ‘He, 13Ca, ‘3C, and ‘3C’ nuclei of a given residue are dependent upon secondary structure within a protein. Wishart et al. were the first to formulate a simple algorithm, termed the “Chemical Shift Index” (CSI), to predict the secondary structure of a protein from these main-chain chemical shifts (Wishart and Sykes, 1994; Wishart et al., 1992). In this algorithm, residues are given scores of + 1, 0, or -1 depending upon the “secondary chemical shifts” of each of their main chain nuclei, followed by averaging over all nuclei. A more sophisticated “Secondary Structure Propensity” (SSP) algorithm includes additional weighing and scaling terms to yield a score ranging from -1(100% 3-strand) to +1(100% x-helix) that indicates the propensity of a residue to adopt a given secondary structure (Marsh et al., 2006). Alternatively, the program TALOS uses a database of known high resolution structures and chemical shifts to estimate the backbone dihedral p and w angles for each residue, which in turn reflect secondary structure (Cornilescu et al., 1999). Complementing chemical shift analyses, the secondary structure of a protein can also be determined from patterns of near-neighbor ‘H-’H NOE interactions. For example, c-helices typically show strong lHNlHN+1NOEs (historically denoted as a short dNN(i,i+]) distance), as well as the presence of NOE reflecting intermediate daN(I, 1+3), d(1, i+3) and dc,N(i, 1+4) distances. Finally, secondary structure elements can be confirmed by determining which amide groups are protected from rapid hydrogen exchange (HX) when the sample is transferred to a D20 solvent. This protection is usually interpreted to result from inter-residue amide-carbonyl 61 Chapter 2 — Structure of the DHB domain hydrogen bonding, as present between residues (I, i+4) along a helix or between paired 13-strands of a 13-sheet. As summarized in Figure 2.11, a combination of the above methods clearly demonstrates five ct—helices in DAXX55’44. 2.4.3 DAXX55144 tertiary structure The three dimensional structure of DAXX55’44 was determined using distance restraints derived from inter-proton NOEs, dihedral angle restraints from chemical shifts, and orientational restraints from residual dipolar couplings. The NOEs were measured using simultaneous aliphatic ‘3C- and‘5N-NOESY-HSQC-, aromatic‘3C-NOESY-HSQC, and CT-methyl-methyl and amide-methyl-NOESY spectra. The tertiary structure of DAXX55’44was calculated from these NMR-derived restraints using ARIA 1.2 (Linge et al., 2001). Manual assignments of unambiguous peaks were incorporated in the initial calculation, although most NOEs were automatically assigned by ARIA, followed by a manual correction of violated restraints (details in the Materials and Methods section). A final ensemble often water-refined structures was obtained with an RMSD value of 0.54 A for backbone heavy atoms in structured regions of the protein (Table 2.1). Based on the algorithms DSSP, PROMOTIF and PROCHECK, the boundaries for each helix are: x1 (residues 60-77); a2 (85-93); c3 (97-100); ct4 (103-118); and c5 (123-136) (Figure 2.12A) (Hutchinson and Thornton, 1996; Kabsch and Sander, 1983; Laskowski et al., 1996). This is in good agreement with the secondary structure analysis described in the previous section. 62 Chapter 2 — Structure of the DHB domain Figure 2.11 - NMR-derived secondary structure of DAXX55144.A combination of HX, CSI, SSP, TALOS and NOE-derived parameters defined the helical regions of DAXX55’44.For the HX experiment, amide protons that were detectable 45 minutes after transferring the sample to D20 are indicated with a . For the CSI analysis, only residues predicted to be c-helica1 using a combination of ‘3C’ and ‘H chemical shifts are shown as ‘H’. For the SSP, only residues with a helical propensity score greater than 0.6 are shown as ‘H’. Similarly, residues with TALOS-predicted p and w angles of -57 ± 20° and 47 ± 20°, respectively, in nine out of the ten closest database entries are labeled as H. Bars connect atoms for which the indicated NOE crosspeak (denoted as a short distance between two nuclei) has been identified. The thickness of the bars indicates the intensity of the NOE. The presence of strong dNN(i, 1+1), as well as intermediate daN(I, i+3), d(i, 1+3) and dcN(1, 1+4) distances, is indicative of a-helices (Wüthrich, 1986). For comparison, the cylinders above the sequence show the secondary structure elements identified by PROCHECK and DSSP using the three dimensional structural ensemble of DAXX55’44derived in section 2.4.4. 63 I- (1 z z t t I:I4 x x I,, IH X Z : > i::i x Cl) X . Qz P1 r P1 U -K (l) . P 1 x U-K I1 i::i x x z Z x x X X C/) x I IeI I’-4 U I i1 ‘ - — C) P1 Cl) II (Y C!) P1 U — — c - c c 1- — — r - r t ’ t — — - o - - ’ - - - - - - - - o - - - - - - - - H - z 1 ) U ) z o Chapter 2— Structure of the DHB domain Table 2.1 - NMR restraints and statistics for the ensemble of the ten lowest energy structures calculated for DAXX55144. Summary of restraints. NOESa Intra-residue 514 (69) Sequential 468 (151) Medium range (i-j 2-4) 463 (330) Long range (i-j 5) 369 (278) Total 1814 (828) Dihedral angles N’ 70 4) 70 ‘H-’N Residual Dipolar Coupling (RDC) 87 Deviation from restraints NOE(A) 0.05 ± 0.02 Dihedral restraints (degrees) 1.8 ± 0.8 RDC(Hz) 1.1±0.1 Deviation from idealized geometry Bonds (A) 0.005 ± 0.000 Angles (degrees) 0.71 ± 0.07 Improper angles (degrees) 1.95 ± 0.25 Residues in generously allowed regions of the Ramachandran plot (%) 97.5 Mean energies” (kcal moF’) Evdw 205±61 Eb0d 36±8 Eangies 210 ± 40 Eimproper 170 ± 60 ENQE 190±90 EdIh 30± 30 Esani 110±20 Residual dipolar couplings Qvalue 0.047 RMSD from average structure’ (A) Residues in c-helices Backbone atoms 0.54 ± 0.07 Heavy atoms 1.28 ± 0.05 All residues Backbone atoms 0.73 ± 0.15 Heavy atoms 1.41 ± 0.11 a number of unambiguous restraints, with ambiguous restraints in parentheses. b final ARIA/CNS energies for van der Waals (vdw), bonds, angles, NOE restraints (NOE), dihedral restraints (cidh) and residual dipolar coupling restraints (sani). C Residues in a-helices: 60-77, 85-93, 97-100, 103-118, 123-136. All residues: 60-136. 65 Chapter 2— Structure of the DHB domain Figure 2.12 - Structure of the DHB domain. (A) Ribbon diagram of one low-energy NMR derived structure of the DHB domain contained in DAXX55’44with c-helices shown in red. The N- and C-termini are labeled for reference. (B) Superimposed backbone atoms from the structural ensemble of the DHB domain. (C) Surface of one low-energy NMR-derived structure of the DHB domain colored according to an electrostatic potential (blue positive; red negative; white neutral; calculated using MOLMOL with the “simplecharge” method and 100 mM salt). Selected residues are labeled for reference. A cluster of positively-charged residues corresponding to Arg9l, Arg94, Lys135, His137, Lysl4O, Lysl4l and Lys142 stands out on one side of the molecule, whereas a cluster of negatively-charged residues corresponding to G1u62, G1u64, G1u68, G1u69 and G1u72 is present on the opposite side of the molecule. (D) Exposed hydrophobic residues (green) on the surface of the DHB domain (polar residues in gray). A cluster of hydrophobic residues comprising Pro86, Phe87, Va184, Tyr124, Va1125, 11e127 dominates one side of the molecule. 66 Chapter 2— Structure of the DHB domain BA C 055 900 H-’ Ri 15— 180° K141 R115 11 V. K56 1 -i K57 K142 D 180° 67 Chapter 2— Structure of the DHB domain The ioop regions connecting cd—a2, cL2—x3, c3—a4, and c4— cL5 are also well defined in the structural ensemble. This is consistent with their low mobility by ‘5N relaxation measurements (next section) and likely reflects a number of potential hydrogen bonds. DSSP and PROMOTIF (but not PROCHECK) identified residues 120-122 (linker a4— c5) as a 3io helix. Due to its helical content, this structure will be named the DAXX N-terminal Helix Bundle (DHB) domain. An examination of the surface the DAXX55’44 ensemble reveals a number of physicochemical features that might be relevant to the function of the DHB domain. First of all, the molecule is predominantly positively-charged, in agreement with its theoretical p1 value of 9.45 (Wilkins et al., 1999). More so, a cluster of positively-charged residues (Arg9l, Arg94, Lysl35, Hisl37, Lysl4O, Lysl4l and Lys142) stands out on one side of the molecule. In contrast, several negatively-charged residues (Glu62, Glu64, G1u68, Glu69 and G1u72) are located on the opposite side of the molecule (Figure 2.12C), thus giving the DHB a very asymmetric electrostatic surface. Exposed hydrophobic residues (Figure 2.1 2D, green) on the surface of the DHB domain also show a clear clustering. In particular, Pro86, Phe87, Val84, Tyr124, Val125, and 1le127 are all located on one side of the molecule, surrounding the positively-charged residues. In the next chapter, it will be shown that this feature is key to the function of the DHB domain as a protein protein interaction module. 2.4.4 DAXX55144 dynamics The dynamic properties of the DHB domain in DAXX55’44 were investigated using ‘5N relaxation methods (Figure 2.13A-C). From the measured amide ‘5N T1 and T2 values, the effective correlation time for the global isotropic tumbling of DAXX55’44was determined to be 68 Chapter 2— Structure of the DHB domain 7.5 ns (Tensor2 (Dosset et al., 2000)). This value is within the range expected for a monomeric species having the molecular weight of the DHB domain (l1 kDa) (Daragan and Mayo, 1997). In addition to providing a measure of the global rotational diffusion of a protein, ‘5N relaxation data reflect the local mobility of its polypeptide backbone on a residue-specific basis. In general, reduced 15N { ‘H } -NOE values and elevated T2 lifetimes indicate enhanced flexibility on a nsec-psec timescale, whereas short 12 lifetimes arise from conformational exchange broadening on a msec-jisec timescale. The ‘5N {‘H}-NOE, T,, and T2 values of the DHB domain, summarized in Table 2.2, are typical of a well ordered domain, with an average value of 0.75 ±0.07 (NOE), 0.48 ± 0.02 s (Ti), and 0.10 ± 0.01 s (T2) for residues 60-136, which include the loops. Often ioops tend to be more flexible than helical or strand regions. However, in the case of the DHB domain, there is very little difference in the relaxation parameters of the loops versus a-helical regions. This can be explained by the likely presence of a number of hydrogen bonds in the short turns of the loops. For example loop a4-a5 adopts a helical conformation, due to the inferred hydrogen bonding between the carbonyls of residues Arg 119, Pro 120, and Lys 122 and the amides of Lys122, Leu 123, and Val125, respectively. This inferred hydrogen bonding network likely contributes to the low RMSD value of the ioops in the NMR structure ensemble (Figure 2.12B). In contrast to the core residues 60-136, the N- and C-terminal residues of DAXX55’44 are clearly conformationally disordered, based on both their ‘5N relaxation properties and high RMSD deviations in the structural ensemble. 69 Chapter 2— Structure of the DHB domain Table 2.2 - Summary of amide 15N relaxation data for different structural regions of the DHB domain. The residues comprising each region, as determined by PROMOTIF, PROCHECK and DSSP, are indicated in parenthesis. Standard deviations are also shown. Region 15N{1H}-NOE T1 (s) T2 (s) Helixi (60-77) 0.75 ± 0.06 0.49 ± 0.02 0.10 ± 0.01 Helix2 (85-93) 0.77 ± 0.05 0.48 ± 0.02 0.09 ± 0.00 HeIix3 (97-100) 0.68 ± 0.04 0.46 ± 0.03 0.10 ± 0.01 Helix4 (103-118) 0.79 ± 0.04 0.48 ± 0.01 0.10 ± 0.00 Helix5 (123-136) 0.76 ± 0.06 0.48 ± 0.01 0.10 ± 0.00 Loop cd—c2 (78—84) 0.69 ± 0.05 0.49 ± 0.02 0.08 ± 0.01 Loop a2—a3 (94—96) 0.77 ± 0.06 0.48 ± 0.02 0.10 ± 0.01 Loop ct3—a4 (101—102) 0.79 ± 0.01 0.48 ± 0.02 0.10 ± 0.00 Loop a4—cL5 (119—122) 0.77 ± 0.04 0.48 ± 0.02 0.10 ± 0.01 All residues 60-136 0.75 ± 0.07 0.48 ± 0.02 0.10 ± 0.01 Amide proton-deuterium exchange (HX) rates were also measured to probe the local stability and dynamics of the DHB domain (Figure 2.1 3D). In addition to the known effects of sequence, the exchange rate of specific amide hydrogen atoms in a protein is dependent upon its structural environment, including hydrogen bonding and solvent accessibility, as well as local and global fluctuations leading to water contact (Hoofnagle et al., 2003). Under EX2 conditions, protection factors can be interpreted as the inverse of an equilibrium constant for fluctuations between a closed non-exchangeable state and a transiently open exchange-competent state. Thus protection factors provide a measure of residue-specific free energy changes, AG1 RTln(krc/kex), 70 Chapter 2— Structure of the DHB domain governing local and global conformation dynamics allowing exchange. Based on the largest protection factor (Leu7 1, 1.1 x 1 Os), the minimum estimate of the free energy changes for global unfolding of the DHB domain is 6.8 kcal/mol. The measured exchange lifetimes (1/kex) of the DHB domain reveal a striking difference in the protection of each helix to exchange with the buffer. Residues in helix 1 and helix 4 had the largest protection factors, residues in helix 3 and helix 5 had intermediate protection, and residues in helix 2 underwent facile HX (Figure 2.13D). It is interesting to note that functionally important residues are often more dynamic, which enables them to accommodate conformational changes needed to accommodate ligands, substrates or other proteins (Elcock, 2001). Indeed helices 2 and 5, the least stable of the bundle, are involved in substrate binding (Chapter 3). 71 Chapter 2— Structure of the DHB domain Figure 2.13 - The global and local backbone dynamics of the DHB domain. (A) Heteronuclear ‘5N {1H}-NOE. (B-C) ‘5N T1 (B) and ‘5N T2 (C) relaxation data measured at 25 °C, pH 6.5 using a 500 MHz spectrometer. Missing data points corresponded to prolines or residues with overlapping signals. (D) HX lifetimes for the DHB domain at pH 6.5 and 25 °C. Missing data were for prolines, residues with no detectable signal in the first spectra recorded (i.e. helix 2), or residues with overlapping signals. Overall, with the exception of the terminal residues, the DHB domain was well ordered on the nsec-psec timescale. 72 Chapter 2— Structure of the DHB domain 10cI_j5 10 i0 i03 102 60 70 80 90 A 1.0• 0.5 0- iizi IJ. C z z I•1 I I I I I I I I I I I I I I I I -1.0J,- 1.0- 0.8 0.6 0.4 0.2 I ——————————————————0 0.5 —— I I iiibi1 Hii 100 110 120 130 140 Residue 73 Chapter 2 — Structure of the DHB domain 2.5 Structural comparisons of the DHB domain 2.5.1 Comparison with Sin3 Previous to this study, the region encompassing the DHB domain was predicted to contain a ‘Pair of Amphipathic Helices” (PAH) domain (Hollenbach et al., 1999). Historically, the name PAH domain comes from an early sequence analysis of the yeast Sin3 gene (Wang et al., 1990). The encoded Sin3 transcriptional repressor was simply predicted to contain two adjacent amphipathic helices. The first structural analysis of a Sin3 fragment demonstrated that the so-called PAH domain is in fact a four helix bundle that binds partners (such as a short helical segment of Mad 1) using a hydrophobic patch between all helices (Figure 2.14A, see also Figure 3.11C) (Brubaker et a!., 2000; Spronk et a!., 2000). Sponk et a!. suggested the name of “wedged helical bundle” for the domain, which may be more appropriate in a descriptive sense. However, the somewhat erroneous PAH nomenclature has prevailed in the literature of Sin3. Since these original studies, five additional structures of Sin3-fami!y PAH domains have been reported (Nomura et al., 2005; Sahu et al., 2008; Swanson et a!., 2004; van Ingen et al., 2006). Figure 2.14A summarizes the structure of a!! the PAH domains known to date. It is clear that all of these structures are very similar, including the utilization of the same cleft between helix 1 and helix 2 to bind their targets. At first sight the DHB domain indeed looks like a four-helix bundle similar to the PAH domains. However, closer inspection reveals a number of very significant differences. First of all, the DHB domain contains an additional helix (helix 3 in Figure 2.14). Although short, this helix is well defined and traverses diagonally from helix 2 to helix 4. As a result of this cross 74 Chapter 2 — Structure of the DHB domain over, the Sin3 PAH domains and the DHB domain have a different topological arrangement of their core 4 helix bundles (see next section). Second, helix 3 of the DHB blocks the cleft used by Sin3 PAR domains to bind their targets. This immediately suggests that the binding mode of the DHB domain is different to the one used by Sin3 PAH domains. Third, the interhelical angles (Q) between helices are quite different (Table 2.3). In all of the structures of Sin3 PAR domains helices 1 and 2 have an angle approximately of 1400 to accommodate substrate binding (Figure 2.1 4A), whereas in the DHB domain that angle is only l 65°, providing much less “room” to accommodate a binding partner. Fourth, the isolated DHB domain is very well structured in isolation, whereas the two available structures of ligand-free Sin3 PAR domains show a substantial increase in the RMSD of helices 1 and 4 (Figure 2.14A), suggesting that the unbound domain is relatively dynamic. Table 2.3 - Interhelical angle (Q)a in Sin3A PAH2 (1G1E.pdb) and the DHB domain calculated with PROMOTIF. DHB Sin3B 1 2 3 (4)b 4(5)b 1 -164 27 -140 2 137 -169 24 3 (4)b 35 172 167 4(5)b -159 -23 165 a Angles are in degrees. The sign convention is explained in (Drohat et al., 1996). In short, to obtain the interhelical angle, it is necessary to (1) orient the two helices of interest (if) such that they are in a plane parallel to the screen; (2) align the first helix (i) vertically (0°) with its N —* C vector pointing upwards. (3) transpose the vector of the second helix (I) such that its tail is at the same location as the tail for the vector of the first helix (i) (4) rotate an imaginary vector aligned vertically (0°) with its tail at the N-terminus of both helices until it aligns with the N —k C vector of the second helix (f). A clockwise rotation gives a positive interhelical angle, and a counterclockwise rotation gives a negative value. b Helix numbering in Sin3 is used as reference, and numbering in DAXX is shown in parenthesis for comparison. 75 Chapter 2— Structure of the DHB domain Figure 2.14 - Comparison of the DHB domain with the Sin3 PAH domains. (A) Structure comparisons of the PAR domain with the DHB domain. There are seven available NMR-derived structures of Sin3 PAR domains: Sin3B PAH2 with a target peptide from Madi (lE9lpdb); Sin3A PAH2 with Madi (1G1E.pdb); Sin3B PAH2 with HBP1 (1S5Q.pdb); Sin3B PAH1 with NSRF/REST (2CZY.pdb); apo-Sin3B PAH2 (2F05.pdb); apo-Sin3A PAHI (2RMR.pdb); and Sin3A PAH1 with SAP25 (2RMS.pdb). Helices 1, 2, 3 and 4 of the four-helix bundle are shown in red, orange, blue and purple, respectively. The DHB domain contains a different fold in which a fifth helix blocks the position where the Sin3 PAH domains bind their targets (green). This suggests a distinct mechanism of ligand binding for the DHB domain. (B) Topographical representation of the left-handed DHB and Sin3B PAH2 and, as an example, right-handed cytochrome b562 helical bundles. To determine the handedness of the bundle, it is necessary to place the N-terminus of helix 1 at the bottom of the page and its C-terminus to the top, then place a hand with the thumb aligned with helix 1. The handedness is determined by which hand follows helices 2 and 3 with its fingers. Although differing in handedness, Sin3B PAH2 and cytochrome b562 both adopt the more common all anti-parallel configuration. In contrast, the DHB domain has helices 1 and 4 and helices 2 and 5 parallel to each other. In addition, cytochrome b562 and Sin3B PAH2 have their N- and C-termini next to each other, whereas the DHB domain has them on opposite sides of (i.e. diagonally across) the molecule. 76 Chapter 2 — Structure of the DHB domain Lefl hand SIN3B PAH2 HBP 1 SIN3A PAll SAP25 Right hand SIN3B PAH1 NRSF/REST A B Mad 1 SIN3B PAH2 SIN3A PAH2 Mad 1 SIN3B PAH2 SIN3APAH1 C N Left hand DHB b562DHB 77 Chapter 2 — Structure of the DHB domain 2.5.2 Comparison with other helix bundles The DHB domain can be described as having a core four-helix bundle (with an additional short helix between the 2’ and 3’ helices of the bundle). Four-helix bundles, which are found frequently both as independent structural units and as parts of larger domains, can be categorized by a number of descriptors, including (i) connectivity, (ii) handedness, and (iii) helical orientation (Kamtekar and Hecht, 1995). The polypeptide connectivity of the core helical bundle (i.e. helices 1, 2, 4, and 5 in Figure 2.14B) of the DHB domain can be described as up-down-up-down, which is similar to that of Sin3 PAH (Figure 2.14B). Furthermore, the DHB domain and the Sin3 PAH domains are also both left-handed (Figure 2. 14B), although this is by no means uncommon (Kamtekar and Hecht, 1995). Importantly, however, the unit direction vectors of the individual helices in the DHB domain are not all anti-parallel to each other, as seen in most helical bundles including Sin3 (Kamtekar and Hecht, 1995; Presnell and Cohen, 1989). This is due to the crossover of helix 3, making helices 1 and 4 and helices 2 and 5 parallel to each other. Based on these factors, the DHB domain can be classified in the same topological family as that of the domain one of the transcription elongation factor TFIIS (Figure 2.1 5A). However, DHB and TFIIS only share 12% sequence identity and 26% sequence similarity (Figure 2.15B) (Booth et al., 2000), and the C backbone atoms of their helices align with an rms deviation of 4.7 A. According to the “hierarchic classification of protein domain structures” database CATH v3.2.0 (Orengo et al., 1997), the ubiquitin-protein ligase CBL (CATH code 1.20.930.20) (Gay et al., 2008; Zheng et al., 2000), also contains a helical bundle within its phosphotyrosine-binding domain with a similar topology to that of TFIIS and the DHB domain. Despite the low sequence 78 Chapter 2— Structure of the DHB domain identity (18%) and similarity (27%) with the DHB domain, the backbone Ca atoms of the CBL helical bundle align with the DHB domain at a modest level (2.9 A rmsd) (Figure 2.15B- C). Indeed, the DALI server, which searches for proteins sharing similar tertiary structures, returned the CBL helical bundle domain as being most similar to the DHB domains (a high Z score of 6.2) (Holm and Sander, 1995) (Figure 2.15B-C). As noted above, the presence of the cross-over helix 3 almost certainly prevents the DHB domain from binding any partner proteins via the “wedged helix” mechanism exhibited by Sin3 PAH domains. Although sharing similar topological features, both TFIIS (an RNA polymerase II elongation factor) and CBL (a ubiquitin ligase) have low sequence similarity to the DHB domain and thus provide no functional clues as to its role with DAXX. In the next chapter, it will be examined how the DHB recognizes a motif from Rassfl C, which provides insight on the function of this important domain. 79 ATFIIS Chapter 2— Structure of the DHB domain B I -- 2 j----------3 TFII S 1 DHB 55 2---tZJ-- 3 I 1 4 TFIIS 50 VNFK3TN- OK VKKIS— PA 77 0KB 115 PAS?AKLY I N CTVKAF KFI 144 I_--[—1-- . 2 ---- CBL 47 P ‘VDKKMVEKCW KV’ NPKLALKNSPPY DT LRTI SRYEGKME mip 55 -- KC--YKLENE FEFLE MQTADHP-----E YNR -All LA -ZZZZ.LJZZ R---- IZI-- - —--[Z] 4 - CBL 108 TLGENEYFR E KTKQTI LFKEGKERMYEENSQPRRN TKLSLIFSH ELKGI DK3 102 SA NI SVLSRA P AK YVYINELCT KAHS- -Ej- C Figure 2.15 - Structural comparison of the DHB domain with similar helix bundles. (A) The establishing member of the family, TFIIS (1EOO.pdb), CBL (black) in complex with UBCH7 (gray) and a ZAP7O peptide (red circle) (1FBV.pdb) and the DHB domain have the same fold. Helices 1 (red), 2 (orange), 3 (blue) and 4 (purple) and short helices in the loops (green) of the four-helix bundle are highlighted. (B) Sequence comparisons based on structural alignments between CBL and the DHB domain, as well as between TFIIS and the DHB domain, show little sequence identity (black, 12-18%) or sequence similarity (gray, 27%). The numbering of the helices is based on an ideal four-helix bundle. (C) Structural alignment between CBL and the DHB domain show that helices in CBL are much longer, although the orientation of the four principal helices is remarkably similar, with an RMSD of 2.9A. CBL DHB 80 Chapter 2 Structure of the DHB domain 2.6 Materials and Methods 2.6.1 Cloning Figures 2.3 and 2.6C outline most of the constructs utilized in this chapter. All of the genes encoding His6-tagged DAXX constructs were cloned from a human cDNA library (accession CAG3 3366) from Invitrogen, kindly provided by Dr. Graves (University of Utah) (except for DAXX566739,which was originally cloned from a mouse cDNA library by Cameron Mackareth). Briefly, the desired DNA was amplified by PCR methods, purified using a PCR purification kit (Qiagen), and subjected to restriction enzyme digestion (NdeI/XhoI). After incubation overnight at 37 °C, the DNA was re-purified and used for ligation mixtures with previously digested and dephosphorylated pET28a (Novagen) vector. Ligation reactions (20 jiL) were carried out at room temperature for a minimum of 8 hours. The salt in the ligation buffer was removed by dialysis (10 LL on a “V” series Millipore membrane in Petri dish with 10% glycerol for at least 1 hour) prior to transformation by electroporation into E. coil BL21QDE3). Colonies were grown for less than 16 hours after transformation, and then picked for overnight growth in 2 mL of LB media at 37 °C. The cultures were transferred to 25 mL of fresh LB and grown at 37 °C until the 0D600 was 0.6-0.8. Then a glycerol stock (1 mL, 15% glycerol, frozen at -70 °C) was created for each colony, and the rest of the mixture was induced (1 mM IPTG) and screened for expression (next section). Glycerol stocks from colonies expressing protein of the correct size on SDS PAGE were used to extract the cloned plasmid (Qiagen kit) which was sent for DNA sequencing (NAPS, UBC). GST-tagged DAXX (constructs 181-365 and 181-407) were cloned using the same procedure into pGEX-2T expression vector (GE Healthcare) using BamHI and EcoRI restriction enzyme sites. 81 Chapter 2 — Structure of the DHB domain 2.6.2 Protein expression tests To verify the viability of a construct for expression (summarized in Figure 2.3), 1 mL solution from the 25 mL culture (described in previous section) was extracted prior to induction (uninduced sample), then induced bacteria were grown at 30 °C (or lower temperature for some constructs, see Appendix 2) for 5 hours, their 0D600 was recorded, and 2 mL solution were extracted: 1 mL for total expression analysis, and 1 mL for a solubility test (see below). The extracted 1 mL solutions were immediately centrifuged at 6000 rpm for 1 minute, then the supernatant media was discarded, and the pellet was re-suspended proportionally to the 0D600 (Reference volume: for 0D600 = 0.6 use 100 pL of cracking buffer [0.125 M Tris, p1-I 6.8, 2% SDS, 20% glycerol, 0.002% BPB]). To test the solubility of the construct, 1 mL was extracted from a culture that had been induced for 5 hours. It was then centrifuged, the supernatant discarded, and the pellet re suspended in Bugbuster (Novagen, with 500 mM NaC1) instead of cracking buffer. The volume used for this was half of the volume used with cracking buffer. It was then incubated under constant mixing at 4 °C for at least 30 minutes. After that, sample was centrifuged at 14,000 rpm for 5 minutes. The supernatant (which contained the soluble fraction of proteins) was extracted and diluted 1:1 with cracking buffer, whereas the pellet was re-suspended in 1 mL of water (washing), re-centrifuged, and then finally re-suspended with the same volume of cracking buffer as the total expression sample. All samples were placed in boiling water immediately for 5 minutes and kept at -20 °C until the next step. 82 Chapter 2— Structure of the DHB domain SDS-PAGE was used to compare the intensity of the bands of the expressed protein in the uninduced sample, total expression sample, soluble fraction sample, and insoluble fraction sample (Figure 2.6C, Appendix 2). 2.6.3 Protein purification The overall procedure for the expression and purification of DAXX46’6°is the same as will be described in sections 4.7.2 and 4.7.3. The protocol for the expression of DAXX55’44 had the following modifications: (1) The sample bound to the Ni2 column was washed with 125 mM imidazole and eluted with 250 mM imidazole. (2) DTT was added to all buffers except for the thrombin cleavage buffer. (3) After thrombin cleavage, the sample was diluted 1 Ox into 50 mM phosphate buffer, ph 6.7, 10 mM DTT, and applied to an ion exchange column (SP Sepharose, GE Healthcare) and eluted with a salt gradient. DAXX55’44 eluted at approximately 350 mM NaCl. 2.6.4 NMR spectroscopy Unless stated otherwise, all proteins were in NMR sample buffer (5% D20, 10 mM K2HPO4,0.1 mM EDTA, 10 mM DTT, pH 6.5 buffer with 100 mM KC1) with 5% DO added for the lock signal. DAXX46’6°was concentrated to 2OO iM in ‘—280 1iL for assignment of the backbone signals. Due to the low concentration and partial degradation of the sample, long HNCACB and CBCA(CO)NH experiments were recorded for 1 day each. DAXX55’44 used for structural calculation was concentrated to 1.2 mM in ‘—280 1iL. Non-randomly fractional 10% ‘3C DAXX55’44used for stereospecific assignments was 0.8 mM. All spectra were recorded at 25 °C using a 500 MHz Varian Unity and 600 MHz Varian Inova NMR spectrometers. The latter is equipped with a triple resonance cryoprobe. All pulse 83 Chapter 2— Structure of the DHB domain sequences were provided by Dr. Lewis Kay (University of Toronto). NMR data were processed using NMRpipe (Delaglio et al., 1995) and analyzed by SPARKY (Goddard and Kneller, 2004). NMR relaxation measurements Amide ‘5N relaxation parameters were acquired at a proton frequency of 500 MHz as described by Farrow et al. (Farrow et al., 1994). Steady-state heteronuclear ‘5N{1H}-NOE spectra were recorded with and without 3 seconds of ‘H saturation and a total recycle delay of 5.017 seconds. Relaxation rates and correlation times were calculated using Sparky and Tensor2 (Dosset et al., 2000), respectively. Amide proton-deuterium exchange After recording reference ‘H-’5N HSQC spectra, ‘5N-labeled DAXX55’44 in 500 iiL protonated NMR buffer was lyophilized and re-dissolved in an equivalent volume of D20. Amide exchange rates were measured by recording a series of sensitivity-enhanced ‘H-’5N HSQC (Kay et al., 1992) spectra at 25 °C, starting 5 mm after the addition of D20. Initial spectra were obtained with a small number of scans to detect the exchange of the least protected residues, while later spectra were recorded with a greater number of scans for improved signal- to-noise. Spectra were scaled according to the acquisition time and residue-specific exchange rates, kex, determined by fitting peak heights, I(t), to the equation ‘(t) = Ioet + I, where I is the baseline value, close to zero, due to residual protonated water. 2.6.5 Structure calculation The program ARIA vl.2 (Linge et al., 2001; Nilges, 1995; Nilges and O’Donoghue, 1998) was used for structural calculations. A collection of manually picked and partially assigned peaks from 3D simultaneous ‘3C- and‘5N-NOESY-HSQC, simultaneous constant time methyl-methyl 84 Chapter 2 — Structure of the DHB domain and amide-methyl NOESY, and aromatic-NOESY spectra were provided for the generation of distance restraints (all with a 100 ms mixing time). Dihedral angle restraints, obtained from TALOS for residues with agreement with at least nine database entries, were also used for the calculation. After a complete ARIA run, violations were manually corrected by consideration of the resulting structure, and a new run initiated. The ensemble in Figure 2.1 6C, left was obtained after convergence with no violations greater than 0.5 A. This structure was then used as a seeding structure for a final structure calculation that incorporated RDC data (next section). The quality of the ensemble of structures is summarized within the text (Table 2.1). 2.6.6 1HN..15N Residual dipolar coupling measurements DAXX55’44 was partially aligned using acrylamide gel stretching, as the more common method using filamentous phage led to aggregation (Chou et a!., 2001). In short, a control IPAP HSQC on the DAXX55144 was first recorded (Figure 2.16A) (Ottiger et a!., 1998). Then, a 5% (29:1 acrylamide:bis-acrylamide) polyacrylamide pellet was prepared, and dialyzed into NMR buffer. The protein (300 jtL, 0.8 mM) was dialyzed into the polyacrylamide pellet (2 cm, 4.2 mm diameter) in an Eppendorf tube for 24 hours. The gel was stretched by using a Teflon funnel to inject it into an open-bottom NMR tube (New Era #NE-370). The IPAP-HSQC of the aligned protein was then recorded, and compared with the control experiment (Figure 2.1 6A). The D20 splitting produced immediately after loading the gel was 6.7 Hz, which decreased to 5.4 Hz over the course of five days. The resulting average RDC in the amides was 20 Hz (Figure 2.1 6A). The program RDCA (Lewis Kay, University of Toronto) was used to estimate the alignment tensor Da and R values (0.53 and -13.8 Hz, respectively) required for the SANI module in ARIA vi .2. Figure 2.1 6C shows the DHB domain structural ensemble calculated with and without lHNl5N RDCs. Without RDC data, the DHB domain showed an average RMSD 85 Chapter 2— Structure of the DHB domain value of 0.61 ± 0.13 A and 1.21 ± 0.23 A for backbone and heavy chains, respectively (residues 60-136) of a? member ensemble. In comparison, an ensemble of the 7 lowest energy structures calculated using RDC data had an improved RMSD value of 0.49 ± 0.11 A and 1.14 ± 0.21 A for backbone and heavy chains, respectively. One notable difference is that RDC data allowed the ARIA program to orient the loops between helices 1 and 2, and between helices 4 and 5, parallel with the helices, in contrast to the almost perpendicular orientation that they had in the structure ensemble without RDCs. 86 Chapter 2 — Structure of the DHB domain I ‘ I 1i EI I ‘‘ I I “ I N Y89 Y89 1196 1(96 —-1 942Hz 74.1Hz 93.7Hz 117.4Hz ((.05 11.110 ((.75 0.70 8.55 8.80 8.75 0.7(1 ‘H(ppm) 60 70 80 90 100 110 120 130 140 Residue A 30 1l0.0. 20 110.5. 10 E 0 ((9.0. z - ll9.5 —20 120.0. D 25 0 0 0 N 20 0 >0 15- 01 0 c 11: - o 0 0 0 Co o 5 0 10 .2-15 0 CO > Q-factor=O.402 5(7 . 25 20 0) 15- - -10 C) ( -15 0 Q-Iusctor=O.047 -20 -10 0 10 20 30 Expt. dipolar coupling (Hz) Figure 2.16 - Use of RDC values in the structural calculation of the DuB domain. (A) Measurement of the RDC values was carried out by comparing the nitrogen splitting in the IPAP-HSQC spectrum before (left) and after (right) aligning the molecule in a stretched 5% polyacrylamide gel. (B) Histogram showing the RDC values for each amide group in the molecule. Remarkably, residues in helices 1, 2, 4 and 5 showed very similar RDC values, consistent with the parallel alignment that they have in the three dimensional structure. (C) Comparison of the 7 lowest energy structures calculated before (left) and after (right) RDC data was incorporated in the calculation. (D) Plot of calculated vs. experimental dipolar coupling shows a poor correlation for a structure calculated without RDC data, and a remarkable improvement in the correlation for a structure calculated using RDC data. The Q factor(rms(RDC0b5_RDC1D)/rms(RDC0T)) was determined with RDCA 87 Chapter 3 Structure of the DAXX/RassflC complex In this chapter the first structural characterization of a DAXX complex is presented. Through an extensive screening of truncation mutants, the residues 23-3 8 of RassflC were identified as responsible for mediating the interaction of this tumor suppressor with the DHB domain. By systematically screening a selection of buffers, conditions were found for the formation of a tight complex suitable for structural calculations (KD 20 jiM, 10 mM MOPS buffer, 0.1 mM EDTA, pH 6.5). The NMR-derived ensemble of the complex reveals that RassflC2338 formed a short amphipathic a-helix with hydrophobic residues filling the cleft between helices 2 and 5 of the DHB domain. A negatively-charged surface of the Rassfl C peptide also faced the positively charged surface of the DHB domain. Thus both hydrophobic and electrostatic interactions mediated complex formation. Consistent with this structural observation, the affinity of RassflC2338 for the DHB domain significantly decreased with increasing ionic strength. This work provides important information on the function of the DHB domain, and could prove key in elucidating the mechanism by which DAXX is regulated and controls cellular functions through protein-protein interactions. 88 Chapter 3 — Structure of the DAXX!RASSF1C complex 3.1 Identification of RassflC as an interaction partner of the DHB domain With the goal of identifying the function of the DHB domain, proteins in the literature reported to interact with DAXX were examined. Many of these studies used truncation constructs in an attempt to identify the regions of DAXX and its partners responsible for the interactions. Given that there is little structural information on DAXX and most of its partners, these mapping studies at best provided very coarse “guesses” of domain boundaries. Disappointingly, a possible common motif among all of these partners could not be identified through sequence alignments (not shown). Notable among these studies, Li et al. investigated a large number of truncation constructs of Axin, a protein that interacts with DAXX’’97(Li et al., 2007). As documented in Chapter 2, this fragment includes the DHB domain. All of the Axin constructs that showed a positive interaction with DAXX’’97 by co-immunoprecipitation shared residues 576-598 (Figure 3.1). To test whether this region was indeed responsible for mediating the interaction, NMR was used to monitor the titration of purified unlabeled GST-Axin576598 with‘5N-DAXX5’44(Figure 3.1 B D). Disappointingly, the lack of any spectral perturbations indicated that these two constructs do not interact in vitro. Indeed, it is a common observation that reported protein-protein interactions, deduced from co-immunoprecipitation or other such pull-down assays using “crude” cell extracts or cell-free transcriptionltranslation systems, do not stand the test of in vitro biophysical analyses using purified proteins. In the summer of 2008, a fruitful collaboration was started with Dr. Ishov, whose laboratory specializes on studying the function of DAXX. As mentioned in section 1.4.3, Dr Ishov’s group recently identified Rassfl C as a binding partner of DAXX, and found that the interaction mediated a mitotic checkpoint (Lindsay et a!., 2009b). Kitagawa et al. also recognized this 89 Chapter 3 — Structure of the DAXX/RASSF1C complex interaction (Kitagawa et al., 2006), albeit they found the interaction to mediate activation of the INK-signaling pathway (see sections 1.4.3 and 3.7). Nevertheless, it is clear that the DAXX/Rassfl C interaction is biologically important. Importantly, Dr. Ishov’s group mapped the region of each protein that was responsible for the interaction to residues 1-142 of DAXX and 1-50 of Rassfl C. Since an intramolecular interaction between DAXX’2°and DAXX55’44had already been identified (section 4.5), a similar motif in RassflC’5°was searched. Sequence alignment revealed residues 13-34 of RassflC as a likely DAXX-binding sequence (Figure 3.1A). Similar to the SIM-N sequence, this region contains a number of negatively-charged residues juxtaposed by hydrophobic residues. To test if residues 13-34 were the DAXX binding sequence, unlabeled GST-RassflC’334was expressed and purified for titration studies with‘5N-His6-DAXX5’44(Figure 3.1E-G). In contrast to GST-Axin576598,unlabeled GST-RassflC’334 produced profound changes in the ‘H ‘5N HSQC spectrum of‘5N-DAXX5’44 (Figure 3.1F). That is, most peaks from the DHB domain were broadened beyond detection due to the large size of the resulting complex, which also included the GST-tag (27kDa). Such broadening could of course result from an interaction between the GST-tag and the DHB domain. However, this is unlikely as such effects with GST Axin576598 were not observed. Furthermore peak broadening was alleviated by incubating the sample with thrombin to remove the GST-tag (Figure 3.1G). 90 Chapter 3 — Structure of the DAXX/RASSF1C complex A Axin576598 576 KPRSYSENAGTTLSAGDLAFGGKTS 598 DAXX12° 1 MAN2IIVLDDDAAA 20 RASSF1C’° 1 [‘4GEAEAPSFEMTWSSITSGYCSQSSLEQYFTARTSLARRPRRDQD 50 0 0 0 ,$ ‘0 0 00 00 00 0 0 0 00 0 o 0 :0 q 0 00 000 o 00 00 0 0 0 0 000 0 0 0 0 0 .o’° 0000 0 0 0 000 • 0 0 • 0o o 0 e • ) j°o0 00 0 • 0 O 0 0 0 0 0 12S o 0 9.5 9.0 8.5 8.0 7.5 7.0 ° 0 0 9.5 9.0 8.5 8.0 7.5 7.0 Figure 3.1 - RassflC1334,but not Axin576598,bound the DHB domain. (A) Residues 13-34 of RassflC align well with DAXX’20, suggesting that, like this N-terminal segment of DAXX (section 4.5), this portion of Rassfl C will also bind the DHB domain. In contrast, Axin576598 does not share any sequence similarity. (B-G) The ‘H-’5N HSQC spectrum of‘5N-labeled His6- DAXX55’44(177 iM B, 155 .iM E) was not affected when unlabeled GST-Axin576598 was added in excess to a 3:1 molar ratio (C), but was profoundly perturbed when unlabeled GST-Rassfl C’3 was added (4:1; F). The loss of signal was due to formation of a high molecular mass GST RassflC’334 /His6-DAXX55”’complex. Removal of the GST-tag (and His6-tag) by addition of thrombin dramatically sharpened the signals of the resulting smaller DAXX55’44 / Rassfl C’ complex (G). In contrast, the spectrum of labeled DAXX55’44 in the presence of Axin576598 was not significantly altered (D). C DAXX’4 B 0 0 0 0 0 — .: 00 0 00 00 0 0 S •o0o 0 • 0 , 0 0 0 *°0 DAXX55’44+ GST-bait+thrombin 0 DAXX”44+ GST-bait C 00°o * 0 C 0 00 0 OOQ • 0 00 ‘Of 00 0 ° C o%o • •00 • 0 00 0 ° 000°0. 00 0 0 0 0 00 110 115 120 125 110 115 120 E C) ci, H > C) (/2 H ci) ci) CD F 0 0° 00 0. JG 0 0 0000 oo0* 0. Oe 0 0° 0 00 0 00 00 0 : 0 0 00 0eo 0 9.5 9.0 8.5 8.0 7.5 7.0 0H (ppm) 91 Chapter 3 — Structure of the DAXX/RASSF1C complex Importantly, this study confirmed that DAXX and RassflC interact in vitro. However, it was still possible that other additional residues in RassflC’5°flanking 13-34 were also involved. Therefore RassflC’5°was purified and NMR methods to determine the exact region that mediates its interaction with DAXX55’44were employed. 3.2 NMR-monitored titration of RassflC15°and DAXX55-144 Expression and purification of Rassfl C’5° was challenging, as it contained an internal pseudo-thrombin cleavage site at Arg38. The sample was also insoluble when expressed with a His6-tag. After optimizing the conditions for its purification, homogenous‘5N-labeled and a partially proteolyzed (i.e. heterogeneous, with a significant amount of RassflC’38)‘3C/’5N- labeled samples were obtained (see section for details). By combining the information from‘5N-TOCSY-HSQC and‘5N-NOESY-HSQC spectra recorded with the‘5N-labeled peptide, along with HNCACB and (HB)CBCA(CO)NH spectra recorded with the‘3C/’5N-labeled sample, the signals from all of the amide residues in RassflC’5°were assigned (Figure 3.2A). Since its lHa and 3C chemical shift values are typical of random coils (not shown), it was concluded that Rassfl C’5° is predominantly unstructured in isolation. With assignments in hand, ‘H-’5N HSQC experiments were used to monitor the titration of 15N-labeled RassflC’5°with unlabeled DAXX55’44.Suspecting an electrostatic contribution to the interaction of these two species, the titration was performed using a low ionic strength buffer. Profound peak broadening and chemical shift changes in the ‘H-’5N HSQC spectrum of RassflC150, as well as the formation of precipitate, were observed upon addition of DAXX55’44 (Figure 3.2B). At the stoichiometric point, a number of RassflC’5° peaks had 92 Chapter 3 — Structure of the DAXX/RASSF1C complex 2 115 2 0.12 Z 120 0.10 E 0.08 < 0.06 125 0.04 0.02 8.4 8:2 8.0 7:8 ‘H (ppm) Figure 3.2 - Determination of the region of RassflC15°that mediated the interaction with DAXX55’44.(A) Fully annotated 1H5N HSQC spectrum of RassflC50. (B) Superimposed spectra of‘5N-Rassfl C’ -50 (red, 115 tM) upon titration with 0.5 (cyan) and 1 (black) equivalents of unlabeled DAXX55’44 (initially, 1.9 mM) in 10 mM phosphate buffer, pH 6.5. Full/open arrows show peaks that shifted and were detectable/undetectable at the stoichiometric ratio. Most peaks were significantly broadened during the titration, but still observable at a DAXX55 ‘44:RassflC150 ratio of 0.5:1 (cyan). (C) Chemical shift perturbations resulting from addition of DAXX55’44 to ‘5N-labeled RassflC’50 at a DAXX:RassflC ratio of 0.5, calculated as (H)2 +(0.2o,N) , reveal that residues 28-38, and to a lesser extent, 9-18, are perturbed upon binding. A T39 J136 S8 ‘ S27 29 S19 — r-S4O SN s15 Sis S23 R38 Q33— —Q49 -F5 .-R43 —F35 D2 .‘E32 —Y21 E25 - —R46 - MU —L31 [)25 r11o) -- —E3O -R47 Q24 C22 R44 ...WI3A42 - TAI -- - -A3 A37 .D5O ‘H (ppm) 10 20 30 40 50 RassflC residue number 93 Chapter 3 — Structure of the DAXX/RASSF1C complex already broadened beyond detection (Figure 3.2B, open arrows). Fortunately, when the unlabeled DAXX55’44 concentration was at half the stoichiometric point, most signals from RassflC’5° were still visible (Figure 3 .2B, cyan), thus allowing the delineation of the region of Rassfl C that interacts with the DHB domain. Significant chemical shift changes were observed for residues 28-3 8, and to a lesser extent, 9-18 (Figure 3.2C). This showed that the originally tested GST-RassflC’334 construct was indeed too short. With this information at hand, a number of intermediate-sized constructs were designed to determine the optimum Rassfl C peptide for structural studies. 3.3 NMR-monitored titrations of truncation fragments of RassflC with DAXX55-144 3.3.1 RassflC838 The first investigated construct was RassflC838,which contained all the residues perturbed in the RassflC’5°titration. When unlabeled RassflC838 was titrated into a sample of His6-’5N- DAXX55’44,substantial irreversible precipitation occurred if the NMR buffer did not contain any added KC1. However, in the presence of 100 mM KC1, precipitation was completely absent (not shown). Under the latter conditions, the 1H” and ‘5N chemical shifts of‘5N-labeled RassflC changed in the fast exchange regime upon addition of unlabeled DAXX55’44 (Figure 3.3A). This titration confirmed that residues 3 1-36 of RassflC were most affected upon binding (Figure 3.3C). The KD for this interaction was found to be 210 ± 30 iM based on least-squares fitting of chemical shift changes for 6 RassflC amides as a function of total ligand (DAXX55’44) concentration (Figure 3.3B). 94 Chapter 3 — Structure of the DAXX/RASSF1C complex 3.3.2 RassflC8-21 The previous experiments with RassflC’5° showed modest chemical shift changes for residues 9-18 upon binding DAXX55’44 (Figure 3.2C). To examine further the role of these residues, an unlabeled construct, RassflC821 was expressed and purified. The ‘H-’5N HSQC spectrum ofHis6-’5N-DAXX5’44in NMR sample buffer lacking KC1 remained unperturbed upon addition of a stoichiometric amount of this peptide, indicating no measureable binding (Figure 3.4A). Note however, since spectral changes can only be observed when the KD values do not exceed approximately the protein concentration times five (i.e. KD < 5 [P1), the lower limit for the KD value for any interaction between this isolated region and DAXX55’44must be greater than 500 ElM. 3.3.3 Rassf1C23-38’’ A complementary construct, Rassf1C2338v, was produced. This construct includes those residues in RassflC’5°and RassflC838 displaying the greatest chemical shift perturbations upon addition of DAXX55’. In contrast to the case of RassflC821, stoichiometric amounts of RassflC2338”produced significant changes of selective amide signals in the ‘H-’5N-HSQC spectrum ofHis6-DAXX55””(Figure 3.4B). These changes occurred in the fast exchange regime (Figure 3.5A), allowing the determination of the dissociation constant for the binding of RassflC2338”by DAXX55’44 (KD = 65 ± 16 1iM, Figure 3.5E). The reverse titration, adding unlabeled DAXX55” to‘5N-labeled RassflC2338”, produced much larger chemical shift () Constructs identified with a ‘w’ contain non-native C-terminal Trp residue, which was added when a native Trp residue was not present in the sequence. This facilitated quantification via UV-spectroscopy, and in the case of Rassf1C2338v,extended the C-terminus from the interacting Arg38. 95 Figure 3.3 - Identification of the RassflC838 residues that interacted with DAXX55’44.(A) Superimposed regions of the ‘H-’5N HSQC spectra of 15N-labeled RassflC838 with increasing amounts of unlabeled DAXX55’44.Arrows indicate chemical shift changes from the free to the bound state. (B) The chemical shifts of residues 31-36 of Rassfl C (initially 90 iM in 500 pL) as a function of added DAXX55144 from a 1.3 mM stock. Fitting the titration curve (solid line) yielded a KD value of 210 ± 30 .tM. (C) Chemical shift perturbations resulting from addition of 4:1 DAXX55’44 to‘5N-labeled RassflC838, calculated as .J(zx6H)2 +(O.2MSN) . Although residues 3 1-36 showed the greatest changes upon binding, residues 11-29 were also slightly perturbed. Chapter 3 — Structure of the DAXX/RASSF1C complex __ 120 E z __ 0.24 E LQ 0.16 0.08 B [;i3:1 • E32 w Q33 I V Y34 • F35 • T36 a a B I I V KD=210±30 jiM 1 2 3 4 DAXX:RassflC ratio I •1 10 15 20 25 30 35 RassflC residue number 96 Chapter 3 — Structure of the DAXX/RASSF1C complex I 110 115 120 125 E z 9.5 90 8.5 8.0 7.5 7.0 ‘H (ppm) 9.5 9:0 8.5 8.0 7.5 7.0 ‘H (ppm) 8.2 8.1 8.0 7.9 7.8 • 2 ‘H (ppm) DAXX:RassflC ratio Figure 3.4 - RassflC2338w, but not RassflC821, bound DAXX55’44. (A) ‘H-’5N HSQC spectrum of‘5N-His6DAXX5’4in the absence (red) and presence of a stoichiometric amount of unlabeled RassflC82 (black). The lack of chemical shift changes indicated that Rassf1C82’ did not bind DAXX55’44 measurably. (B) ‘H-’5N HSQC spectrum ofHis6-DAXX55’44in the absence (red) and presence of a stoichiometric amount of unlabeled RassflC2338” (black). Significant shift changes were observed in a number of resonances, thus confirming binding. (C) A section of the complementary ‘H-’5N HSQCs5pectrum of15N-Rassf1C2338’”(125 jiM, red) with progressive increments in unlabeled DAXX5 144 (from a 2 mM stock, cyan to black). The binding event was in the fast exchange regime. (D) The chemical shift of residues 31-37 of RassflC2338”as a function of added DAXX55’44 was used to fit the titration curve (solid line), yielding a KD value of 55 ± 5 jiM. A 0 0 B 0 •0 0 0 0 o0 0 00 0 fl 0 0° 000 0 0 0 0 0 0, 0 . 0 0 0 0 0 0 • V 0 0 0 0o0 0 0 0 0 ç,0 0 0 0 0 , 0% 0 0$ S 0 00 V 000 0 ‘ 00 0 97 Chapter 3 — Structure of the DAXX/RASSF1C complex perturbations, also in the fast exchange regime (Figure 3.4C). The dissociation constant fit for the reverse titration is consistent with this initial result (KD = 55 ± 5 M, Figure 3.4D). These values are about four times lower than those observed with RassflC838 since these titrations were performed under lower ionic strength conditions. Throughout screening experiments, His6-tagged‘5N-labeled DAXX was used. To test whether binding was biased by the presence of this tag, thrombin was added to the solution from the last titration point of 15N-labeled DAXX55’with Rassf1C2338v. Beyond the expected disappearance of ‘H1”-’5N signals corresponding to the His6-tag, no other significant spectral changes were observed. Thus, the His6-tag did not measurably affect the DAXXIRassf1C interaction. Once the His6-tag was removed, it was investigated whether there was indeed an important electrostatic contribution to the DAXX/Rassfl C interaction. Addition of increasing amounts of KC1 produced progressive changes in the ‘H-’5N HSQC spectrum of‘5N-labeled DAXX55’44in the reverse direction, as observed during the initial titration. This clearly demonstrates a decrease in the population of the complex with increasing ionic strength, thereby verifying that electrostatic interactions indeed play a major role in complex formation (Compare Figure 3.5A with Figure 3.5C). Subsequently, it was investigated which residues were involved in the interaction by mapping chemical shift perturbations onto the structure of apo-DAXX551’.Figure 3.6 shows that residues most affected by binding of Rassf1C2338’’form two clusters. The first cluster covered N-terminal residues, including Cys58 and Lys6O. Noteworthy, in previous experiments, it was observed that the downfield shifted amide lHNl5N signal of Lys6O 98 Chapter 3 — Structure of the DAXX/RASSF1C complex is exquisitely sensitive to the salt concentration of the sample, i.e. changing from 9.3 ppm to 8.6 ppm with small increases in ionic strength (not shown). Such changes were attributed to the possible disruption of a hydrogen bond between the Lys6O amide and the Leu 100 carbonyl oxygen, and were not investigated further. Regardless, the chemical shift changes of Cys58 and Lys6O likely reflect subtle sample changes during titrations with of Rassf1C2338’’. In support of this conclusion, the fitted KD values of these two residues are —10 fold higher than those observed for other DAXX55’44residues in the same titration experiment (Figure 3.5E-F; 700 .iM versus 65 !IM). Furthermore, the observation that the chemical shift changes of Lys6O and Cys58 did not reverse as salt was added to disrupt the complex, as observed with other residues (Figure 3.5B-D), argues against a secondary, weak binding site for Rassf1C2338vby these residues. The most predominant cluster of residues showing Rassf1C233Svdependent chemical shift changes (residues 84-85, and 124-130) formed a cleft between helices 2 and 5 (Figure 3.6B,C). Most of these residues are hydrophobic, and importantly are not buried within the core of the DHE domain (Figure 3.6D, green). It is also worth mentioning that these residues have been highly conserved during evolution, as the sequence alignment of DAXX from various organisms shows some variability in the DHB domain sequence, but very little variation in residues 84-85, and 124-135 is observed (Figure 2.1). These data strongly indicate that this region is the binding site for RassflC. 99 Chapter 3 — Structure of the DAXX/RASSF1C complex Figure 3.5 - Electrostatic interactions contributed to the binding of Rassf1C2338”and DAXX55’44.(A-B) Sections of the ‘H-’5N HSQC spectrum of‘5N-His6-DAXX5’44(110 tiM, red) with progressive increments in unlabeled RassflC2338”(from a 4 mM stock, cyan to black). (C-D) The same sections of the ‘H-’5N HSQC spectrum of thrombin treated‘5N-DAXX5 ‘44/Rassflc2338”with increments of NaC1 from 0 mM (black) to 450 mM (blue). The resonances from residues Asn128 and Va185 reversed direction towards the shifts from the free protein, indicating that the complex population decreased with increasing ionic strength. On the other hand, Lys6O continued to shift upfield with added salt, presumably by disrupting a hydrogen bond between its amide and the Leul 00 carbonyl oxygen. (E-F) The chemical shifts of residues 58, 60, 85, and 125-128 of DAXX55’44 as a function of added Rassf1C2338’’were fit to obtain residue-specific dissociation constants. The anomalously high dissociation constant of residues 58 and 60 also suggested that these changes reflect a secondary local event independent of RassflC binding. 100 zz 119.0 119.5 120.0 0.16 0.12 0.08 0.04 8.10 ‘H (ppm) 8.05 1 2 3 4 5 6 RassflC:DAXX ratio Chapter 3 — Structure of the DAXX/RASSF1C complex 123.5 122.5 123.5 124.5 9.4 9.2 9.0 8.8 8.6 ‘H (ppm) 122.5 124.5 V85 D CQ I 0 8.15 E ---. - -- - - A . I .S. SI . • V125 • 127 KL= 65 ± 16 tM • Y126 7 8 RassflC:DAXX ratio 101 Chapter 3 — Structure of the DAXX/RASSF1C complex 0.30 E 0.20 - 0.10 B ciA 60 70 80 90 100 110 DAXX residue number 120 130 140 Figure 3.6 - Identification of the Rassf1C2338’’ binding interface on DAXX55144. (A) Chemical shift perturbations resulting from addition of RassflC2338”to‘5N-labeled DAXX5544, calculated as l()2 +(O.2Aö,N) . (B) Cartoon representation of DAXX55’44 with its 5 helices (al- c5). Residues Va184, Va185, Tyr124, Va1125, 11e127, Asn128, Glu129 and Leul3O, which showed significant spectral perturbations (A > 0.085 ppm), are highlighted in red. Also in red are Cys58 and Lys6O, which experienced spectral changes during titrations that were independent of the binding of Rassf1C2338’”(see text). (C) Surface of DAXX55’44showing how the side chains of the affected hydrophobic residues are solvent exposed, and presumably involved in substrate binding. (D) Cartoon representation of DAXX55144 with non-polar residues shown in green. The amphipathic character of the helices is evident except for a few residues that are solvent exposed. These same residues were the most affected by addition of Rassf1C2338’’. A II I I I tIIIIIIIiIIIIiIIhIiIIIIIIIIII.IIiIII II.iI.i111 Y124 D C 102 Chapter 3 — Structure of the DAXX/RASSF1C complex 3.4 Screen of conditions for the formation of a tighter complex In order to obtain a high resolution three-dimensional structure of a biomolecular complex by NMR spectroscopy it is desirable that the components interact strongly (i.e. a 1:1 complex in slow exchange). Low affinity binding precludes saturation at the stoichiometric ratio, thus making intermolecular NOEs weaker and hence structural calculations more difficult (Nietlispach et al., 2004). As summarized in the previous section, it was found that residues 31-36 of Rassfl C are key to binding with DAXX55144.The optimized construct, RassflC2338’’, bound DAXX55144 with a KD of 65 ± 16 .iM in 10 mM phosphate buffer, pH 6.5 and 25 °C. Thus, under the conditions used for the titration experiments (100 jaM protein in the NMR tube, 5 mM peptide being added), only 45% saturation was reached at the stoichiometric point (i.e. equimolar amounts of each species). Based on mass action, this would improve to only 75% saturation by increasing the concentration of the protein to 1 mM. Therefore, with the aim of optimizing the conditions for the formation of a tighter 1:1 complex, the dissociation constant of the complex was measured as a function of temperature, pH, and buffer type, as well as examining slightly longer RassflC constructs (Table 3.1). From Table 3.1, one can conclude the following points. (1) Longer constructs, such as RassflC838 or RassflC235°produced irreversible precipitation at low ionic strength, or showed much weaker binding at high ionic strength, and hence were not suitable for structural studies. (2) A small extension of the C-terminus in RassflC2344 did not produce significantly tighter binding, and the non-specific proteolysis that the construct underwent makes it not suitable for 103 Chapter 3 — Structure of the DAXX/RASSF1C complex Table 3.1 - Dissociation constants of DAXX55’44 and various RassflC constructs under different conditions. ‘5N-labeled Unlabeled construct / construct / Buffer a p11 Temperature K0 (.tM) concentration concentration C38 DAXX55144 IrreversibleRassfl 10 mM KPO4 6.5 25 °C precipitation90j.tM 1.3mM RassflC838 DAXX55’44 10 mM KPO4, 6.5 25 °C 210 ± 30 90p.M 1.3mM 100rnMKC1 His6-DAXX5’44 RassflC82’ 10 mM KPO4 6.5 25 °C No binding 110 j.tM 2.3mM His6-DAXX5144 Rassf1C2338v 10 mM KPO4 6.5 25 °C 65 * 16 110pM 4.1mM c2338’’His6-DAXX5144 Rassfl 10 mM KPO4 6.5 15 °C 69 ± 14 110J.LM 3.3mM His6-DAXX5144 RassflC233Sw 10 mM MES 5.5 25 °C 48 ± 13 100 0.8mM c2338wHis6-DAXX5144 Rassf 10 mM HEPES b 25 °C 28±7 100 1.8mM His6-DAXX5’44 RassflC2344wc 10 mM KPO4 6.5 25 °C 20-290 100 iM 0.65-3 mM ‘ C235°C IrreversibleHis6-DAXX55’44 Rassfl 10 mM KPO4 6.5 25 °C precipitation100 2.4rnMd a All buffers also included 0.1 mM EDTA. b Resonances from amides from the termini (55, 57, 58, 140, 141, 142) and loops (83, 97, 121) were not detectable at pH 7.5 due to fast hydrogen exchange with the solvent. C Constructs 23-44w and 23-50 were purified using a different plasmid that contained a PreScission cleavage site instead of the thrombin cleavage site like the other constructs. This was needed as RassflC’5°contains an internal thrombin cleavage site at Arg38. d MALDI-ToF-MS showed that construct 23-44w was a mixture of the intact peptide, and a shorter non-specifically proteolysed peptide 23-42 without a C-terminal Trp. The difference in extinction coefficient between both constructs introduced a large error in the calculated concentration of the peptide. RassflC235° also showed the presence of a smaller fragment. e The KD was calculated using either the concentration from an intact 23-44w (20 !IM) or a proteolysed 23-42 (290 tM). 104 Chapter 3 — Structure of the DAXX/RASSF1C complex structural studies (see Materials and Methods for details). (3) Temperature did not seem to play a major role in the interaction, as the dissociation constant of the complex at 15 °C and 25 °C were essentially identical. (4) Moderately tighter binding was observed at different pH values, although this is most likely due to the zwitterionic buffers used, which decreased the ionic strength of the sample with respect to the phosphate buffer. (5) The tightest complex observed was with 10 mM HEPES, pH 7.5, although a number of amide resonances in the termini and loops were not detectable due to fast exchange with the buffer at this alkaline pH value. One interesting observation is that amide signals in this buffer had a comparable signal-to- noise ratio as their counterparts in 10 mM KPO4, pH 6.5, despite obviously undergoing faster hydrogen exchange. This is consistent with recent reports that demonstrate that cryogenic probes perform much better with low conductivity buffers relative to phosphate (Horiuchi et al., 2005; Kelly et al., 2002). With this information at hand, a stable relatively tight complex (KD 20 1iM) suitable for structural studies was prepared (10 mM MOPS, 0.1 mM EDTA, pH 6.5; see section 3.7.4 for details on how the final samples were prepared and how the structure was calculated). 3.5 Determination of the tertiary structure of the complex 3.5.1 Changes in DAXX55’44secondary structure upon binding Upon binding RassflC2338”, the spectra of DAXX55’44 displayed only modest perturbations of a limited number of backbone amide groups 1HN and ‘5N (Figures 3.4B and 3.6A), and of‘3Cc and ‘Hi’ values (Figure 3 .7C,D). This indicated that the tertiary structure of the protein was not significantly affected. Rather, small spectral changes for residues in a localized region of the 105 Chapter 3 — Structure of the DAXX!RASSF1C complex protein likely reflect minor structural changes accompanying RassflC2338” binding. However, one residue stands out in these shift perturbation plots is Va184. In the presence of RassflC2338”, its 13C and ‘H chemical shift values become more typical of residues in an tx—helix. Interestingly, Va184 lies at the N-terminus of Helix-2 (residues 85-93) in the apo-protein, and is integral to the RassflC233 binding interface (Figure 4.6C). As it will be discussed in the next section, Va184 becomes part of a slightly elongated Helix 2 upon binding RassflC. 3.52 Changes in RassflC23.38wsecondary structure upon binding The apo-RassflC2338”’ is predominantly unstructured, as evidenced by its random coil chemical shifts and low SSP scores (not shown). Upon binding DAXX55’44,however, RassflC23 38w exhibits dramatic spectral changes. First, almost all of its amides show IHN and ‘5N chemical shifts changes in ‘H-’5N-HSQC spectra when titrated with DAXX55’44(Figure 3.8A). Second, its ‘H and 13C chemical shift values move upfield and downfield, respectively (Figure 3.7A-B. This is diagnostic of a coil to u-helix transition. Third, the ‘5N {‘H}-NOE values of most amides (0.45) are typical of a moderate degree of order on the nsec-psec time scale. In contrast, the negative NOE values of its N-terminal amides are typical of a peptide that is fully disordered on this timescale (Figure 3.8B). In all of these plots, residues 30-38 undergo the greatest changes when binding DAXX55’44,consistent with the original findings from the RassflC’5°titration (Figure 2C). 106 Chapter 3 — Structure of the DAXX/RASSF1C complex 1.0 0.8 ,— 0.6 E 04 c-) <0.2 0 .0.2 0.05 p i -0.10 -0.15 -0.20 25 30 35 Residue Residue Figure 3.7 - Changes in the secondary structures of RassflC2338’’ and DAXX55’44 upon binding. (A-B) Changes in the corresponding ‘3Cc and ‘H° chemical shift values of RassflC23 38w in its bound versus free states. The upfield and downfield changes in the l3Ca and ‘H chemical shift values, respectively, are consistent with the formation of an a-helix (Wishart and Sykes, 1994; Wishart et a!., 1992). (C-D) Changes in the C and H chemical shift values of DAXX55144 in its bound versus free states. Only Va!84, which is in the binding interface, showed a significant change due to a slight elongation of helix 2. This demonstrates that the protein did not undergo significant structural changes upon binding Rassft C. C UkII I I i .j I I 1.11111 11.111 III I ill1 lb I — I III .... I - 60 80 100 120 140 107 Chapter 3 — Structure of the DAXX/RASSF1C complex 115 I 7 120 125 8.5 8:0 7.5 7.0 II (ppm) :: -0.2 -0.4 - 24 2 32 36 Residue Figure 3.8 - Changes in the amide chemical shifts and relaxation of RassflC2338wupon binding DAXX5514. (A) Annotated ‘H-’5N HSQC spectrum of Rassf1C2338”in its free and DAXX55’44-bound form. Peaks corresponding to the region 3 1-38 underwent major spectral perturbations, indicating significant changes in its secondary and tertiary structure. (B) Heteronuclear 15N {‘H}-NOE values ofDAXX55’44-bound Rassf1C2338”’. The low/negative and low NOE values in residues 22-26 indicate a flexible N-terminus. The moderate values of residues 27-3 8 indicate a more ordered, structured region. The reduced NOE value of Trp39 indicates somewhat increased flexibility at its C-terminus. A dl Q24 : S27 S29 —E25 QQZ4 1)28 E39 A37 W39 ‘‘ L31 108 Chapter 3 — Structure of the DAXX/RASSF1C complex 3.5.3 Tertiary structure of the DAXX551/Rassf1C2338’’complex The structural ensemble of the DAXX55’44/ RassflC2338’ complex was determined from an extensive set of NMR-derived inter and intramolecular distance restraints (Table 3.2; Figure 3.9B; details in section 3.7.4). A ribbon diagram of the DAXX55’44 / RassflC2338”complex (Figure 3 .9A) demonstrates that peptide-bound DAXX55144 retained the secondary and tertiary structure of the free DHB domain (Chapter 2), based upon 5 x-helices (PROMOTIF boundaries: 60-77 red, 84-93 orange, 99-101 green, 103-118 blue, 122-137 purple; Figure 3.9A). This is reflected quantitatively in the low RIvISD between the helices in the lowest energy members of free and bound DAXX55’44structural ensembles (1.08 A [backbone C/N/O atoms only] and 1.75 A [all C/N/U atoms in the helices]; Figure 3.100). Indeed, the most significant change in the backbone conformation occurred with the small elongation of helix 2 to include Va184. This resulted from the side-chain of Va184 being “pushed” back by the RassflC peptide (Figure 3.100). Additionally the packing of aromatic rings of Tyr124 and Phe87 changed significantly to accommodate RassflC2338” in the hydrophobic binding cleft of the DHB domain (Figure 3.1 OD,F). Rassf1 C23 38W lies in the cleft between helices 2 and 5 of the DHB domain (Figure 3.9 A-D). The peptide forms a short amphipathic helix with Leu3 1, Tyr34, Phe35 and Thr36 making most of the intermolecular contacts with Val84, Phe87, Tyr124, and 1le127 of the DAXX DHB (Figure 3.1 OA-E). On the outer side of the amphipathic helix, the hydrophilic 5er29, G1u30, Glu32 and G1n33 face the solvent. 109 Chapter 3 — Structure of the DAXX/RASSF1C complex Table 3.2 - NMR restraints and statistics for the ensemble of the ten lowest energy structures calculated for the DAXX5544/RassflC2338wcomplex. Summary of restraints. NOESa Intra-residue 849 (106) Sequential 385 (99) Medium range (i-j 2-4) 256 (152) Long range (i-j 5) 217 (124) Intermolecular 33 (65) Total 1740 (546) Dihedral angles: 81 81 Deviation from restraints NOE(A) 0.052 ± 0.000 Dihedral restraints (degrees) 0.80 ± 0.13 Deviation from idealized geometry Bonds (A) 0.005 + 0.000 Angles (degrees) 0.63 ± 0.01 Improper angles (degrees) 1.61 + 0.08 Residues in allowed region of the Ramachandran plot (%) 98.1 Mean energiesb (kcal mor’) Ed -310±20 48 ± 2 Eangies 204 ± 7 Eimproper 97 ± 8 ENOE 280+20 Edh 6±2 RMSD from average structureC (A) Residues in a-helices Backbone atoms d 0.35 ± 0.07 Heavy atoms 0.92 ± 0.15 All residues Backbone atoms 0.57 ± 0.22 Heavy atoms 1.13 ± 0.26 a Number of unambiguous restraints, with ambiguous restraints in parentheses. b Final ARIA/CNS energies for van der Waals (vdw), bonds, angles, NOE restraints (NOE), and dihedral restraints (cidh). Residues in a-helices: DAXX (60-77, 84-93, 99-101, 103-118, 123-137), RassflC (29-33). All residues: DAXX(60- 137), Rassfl C (27-39). d “heavy” C/N/O atoms. 110 Chapter 3 — Structure of the DAXXIRASSF1C complex Figure 3.9 - Tertiary structure of the DAXX55144 / RassflC2338w complex. (A) Ribbon diagram of one low-energy NMR-derived structure of the complex. Helices 1-5 of the DHB domain are shown in red, orange, green, blue and purple, respectively. Rassfl C is shown in old. (B) Superimposed backbone atoms from seven low-energy members of the DAXX55 - RassflC2338w structural ensemble. (C) Surface of one low-energy NMR-derived structure of the DHB domain in the complex, colored according to an electrostatic potential (blue positive, red negative, white neutral). A Rassfl C ribbon diagram with selected residues involved in the binding is shown for reference, and the negatively-charged residues 25, 26 and 28 in red. (B) Surface of one low-energy NMR-derived structure of the RassflC2338”’in the complex, colored according to an electrostatic potential. A DAXX55’44 ribbon diagram with selected residues involved in the binding is shown for reference. The view on the left shows a “slab” picture from the inside to out of the DHB domain, highlighting helices 2 and 5, which mediate the binding process. The end of helices 1, 3 and 4 are in transparent gray. The view on the right shows the position of all helices in the complex. All electrostatic potentials were calculated using PDB2PQR with the AMBER forcefield (Dolinsky et al., 2007; Dolinsky et a!., 2004), and plotted using the APBS plug-in from Pymol (DeLano Scientific), with an electrostatic potential gradient from -8.0 to 8.0 kT/e. 111 Ra,’F1( Hi N4’ H2 Chapter 3 — Structure of the DAXX/RASSF1C complex 450 * 180° H’ 90° * 112 B R38 90° +0 Q23 G55 N144 180w P D 45° * 1127 F87’ Chapter 3 — Structure of the DAXXIRASSF1C complex Figure 3.10 - Residues mediating the intermolecular interaction between DAXX55144 and RassflC233Sw. Residues Leu3l, Phe35 and Thr36 in RassflC (A) and residues Va184, Phe87, Tyr124, 11e127 and Asn128 in DAXX55’44 (B) have the largest number of intermolecu]ar NOEs. (C) Diagram representing the intermolecular NOE connections between residues in helices 2 and 5 from the DHB domain (cyan) and the RassflC2338’peptide (gold). (D-E) Interaction surfaces of one low energy NMR-derived structure of the DHB/Rassfl C complex. The two molecules are separated and colored by surface physicochemical properties (hydrophobic green (D), red negative, blue positive, and white polar neutral (E). The arrow shows the hydrophobic cleft from the DHB domain occupied by the hydrophobic cluster from Rassfl C. The default values for calculating the electrostatic potential (at zero ionic strength) from MOLMOL were used. (F) Ribbon diagram of a low-energy NMR structure of the DHB (cyan) - Rassfl C (gold) complex. The side-chain atoms of interacting residues are highlighted. Oxygen atoms are shown in red, nitrogen atoms in blue. (G) Ribbon diagram of superimposed low-energy structures of apo- and Rassfl C-bound DAXX55’44.Changes in the structure of free DAXX55’44(brown) and Rassfl C bound DHB (cyan) are located in the binding surface. The aromatic rings of Y124 and F87 move to accommodate the incoming Rassfl C peptide. Similarly, Va184 is “pushed” back by Leu3 1 in the peptide. 113 o 30 20j z Chapter 3 — Structure of the DAXX/RASSF1C complex I II [I C I)AXX HeIi’ 5 I)AXX.” HeIi 2 26 30 34 Residue 85 120 130 Residue D 1132 N120 1127 V124 l(2’ 5 S27 )2 F32 Q3.i 35 [31, 57 R35 39 5,55, v05 -, E 026 F DHB RassflC DHB RassflC G 114 Chapter 3 — Structure of the DAXX/RASSF1C complex Juxtaposed to the hydrophobic cleft, a cluster of positively-charged residues in DAXX55’44 (Arg9l, Arg94, Lys135, His137, Lysl4O, Lysl4l and Lys142) faces a cluster of negatively- charged residues in RassflC2338” (G1u25, Asp26, and Asp28) (Figure 3.1 OE). This electrostatic interaction is consistent with the observed salt-dependence of binding (Figure 3.5). The final structure completely explains the preliminary experiments (sections 3.2-3.5.2). The peptide occupies the hydrophobic cleft predicted qualitatively from chemical shift perturbation mapping of titration data (Figure 3.6C) (Figure 3.1OD). PROMOTIF and DSSP identified residues 29-33 as the common helix boundaries (with residues 28 and 34 of Rassf1C2338”’helical in some members of the ensemble). This is consistent with chemical shift analyses and NOE data, which predicted that these residues formed an x-helix (Figures 3.7-3.8). The interactive region is close to the C-terminal non-native Trp39 of the RassflC2338’’construct. This resulted in Trp39 giving some intermolecular NOEs to Tyr124 (Figure 3.1OC), intramolecular NOEs to Thr36 (not shown), and a small positive ‘H-’5N heteronuclear NOE value (Figure 3.8B). Attempts to extend the C-terminus with other constructs did not result in any improvement in complex formation, and in some cases led to aggregation (Table 3.1). These results suggest that Trp39 did not contribute significantly to the affinity of the DAXX55’44/ Rassf1C2338”’complex. 3.6 Structural implications 3.6.1 Structural verification that the DAXX DHB and Sin PAll domains differ in binding mechanisms With the structure of the DHB domainlRassfl C complex in hand, one can now conclude definitely that its binding mode is completely different to that of the Sin3 PAll domains (Figure 3.11). Sin3 PAll domains (discussed in Chapter 2) use a large hydrophobic cleft between helices 1 and 2 for binding (Figure 3.1 1B). The “opening” of the interhelical angle between helix 1 and 115 Chapter 3 — Structure of the DAXX/RASSF1C complex 2 exposes hydrophobic residues from helices 3 and 4 to form a binding interface. For example, in the Sin3B PAH2/Mad complex (Spronk et al., 2000), this large 2100 A2 hydrophobic cleft includes residues Phe7, AlalO, hell, 11e17, Tyr28, Phe3l, Leu32, Leu35, Phe59, Leu65, Phe66, Leu72, Leu73, Phe76, Phe79 and Leu8O (Figure 3.11C). In contrast, DAXX DHB domain uses a much smaller cleft along helices 2 and 5 to bind RassflC. The cleft is only 570 A2 in hydrophobic surface area and lined by Val84, Phe87, Tyr124, 11e127 and Cysl3l (Figure 3.1OD). The Mad peptide forms a longer amphipathic helix (11 amino acids) when bound to Sin3B PAH2, whereas the Rassfl C peptide adopts a shorter helix (5 amino acids) in the resulting DHB domain complex. In the Sin3B/Mad complex, the hydrophilic residues of the Mad peptide are negatively-charged and complement the positively-charged residues of helix 2 from Sin3 (Figure 3.11 B), although they do not face each other directly as in the case of the DAXX/Rassf1 C complex (Figure 3.1OE). Additionally, Sin3B/Mad complex produces much tighter binding than the DAXX55 ‘44/RassflC2338”complex. For example, a KD of 15-29 nM for Sin3A/Mad complex was measured by ITC (Brubaker et al., 2000), and a KD of 1.4 jtM for the Sin3B/Mad complex was measured by surface plasmon resonance (van Ingen et al., 2004). These values are one or two orders of magnitude smaller than the best KD -‘20 tM (at low ionic strength) for DAXX55 ‘44/RassflC2338” . However, in the context of the full-length proteins, other domains could also interact and thereby co-operatively enhance the specificity and affinity. Furthermore, post translational modifications on either protein could enhance the interaction. Interestingly, the Ser residues on RassflC2338 are within ideal recognition sites of the CK2 kinase (discussed in Chapter 5). Further experiments are required to clarify these hypotheses. 116 Chapter 3 — Structure of the DAXX/RASSF1C complex DAXX DAXX + RassflC H1PIR N FAT + Paxillin Figure 3.11 - DAXX and Sin3 PAH domains use different binding modes. Ribbon diagram of a low energy structure of the DAXX55‘4/RassflC2338”complex (A) and the Sin3B PAH2/Mad complex (B; 1 E9 1 .pdb). The helices numbering is shown as reference. (C) Surface of the Sin3B PAH2 in one low-energy structure of the Sin3B PAH2/Mad complex, showing the large hydrophobic cavity used for binding. The Mad peptide is shown in transparent cyan. Hydrophobic, negatively-charged and positively-charged residues of the Sin3B PAH2 domain are colored in green, red and blue, respectively. Other structural relatives of the DHB/RassflC complex. Ribbon diagram of a low energy structure of the DAXX55’44/RassflC2338’’ complex (D), the Huntingtin-interacting protein-i (H1P1R) related (E) and Focal adhesion targeting (FAT) domain/Paxillin complex (F). The FAT/Paxillin complex displays a Paxillin peptide on each side of the protein on the same crystal structure. Helices 1 (red), 2 (orange), 3 (blue), 4 (purple), 5 or intermolecular helix (gold) and short helices in the loops (green) of the helix bundles are highlighted. A B SIN3B C MAD E F C 117 Chapter 3 — Structure of the DAXX/RASSF1C complex 3.6.2 Other structural relatives share helix packing patterns Additional structural relatives of the DHB/RassflC2338” complex were searched through DALI (Holm and Sander, 1995). The best hit was the five-helix bundle THATCH domain from the Huntingtin-interacting protein-i related (HIP 1 R), a protein-trafficking control multi-domain protein (Brett et al., 2006). Although not a complex, the first four helices of the helix bundle follow the same topology as the DHB domain, and the fifth helix packs on the groove of helices two and four, in a very similar way to how the Rassfi C helix packs along the equivalent groove in the DHB domain (Figure 3.11E). A more distant structural relative is the Focal Adhesion Targeting (FAT) domain in complex with Paxillin LD domains (Hoellerer et al., 2003). The FAT domain forms an all-antiparallel right-handed four-helix bundle that binds Paxillin LD peptides. In the structure of the complex, two LD peptides form amphipathic helices that pack on two sides of the FAT domain, each time using a cleft between two helices similar to the Rassf1 C peptide. Therefore the packing of a parallel helix (Rassfl C) along the groove of two other helices of a helical bundle (helices 2 and 5 of the DHB domain) is not uncommon. 3.6.3 Other binding partners of DAXX This work provides the first description of how DAXX interacts at the molecular level with other proteins. The structural characterization of the DHB domain and its interaction with a segment of Rassfl C also provide clues on how DAXX interacts with other proteins. The “dipolar” nature of the DHB domain, in which opposite faces of the protein have opposite electrostatic charges (Figure 2.1 2C), could prove important in mediating its interactions. The work presented in this chapter shows that the positively-charged face of the DHB domain 118 Chapter 3 — Structure of the DAXX/RASSF1C complex interacts with the negatively-charged RassflC2338” peptide. This electrostatic interaction presumably helps to position the hydrophobic residues of RassflC2338v for the formation of a stable complex. Indeed, recent studies have shown that formation of protein complexes involves an “encounter” complex, in which partner proteins display few specific interactions and assume multiple orientations (Ubbink, 2009). Electrostatic interactions dominate the formation of the encounter complexes. Subsequently, the complexes become “productive” with specific interactions leading to a single predominant conformation. It is worth mentioning that the opposite face of the DHB domain is negatively-charged and also contains a hydrophobic cluster formed primarily by residues in the ioops between helices 1 and 2, and 4 and 5 (Figure 2.12 C D). However, this hydrophobic “wall” is much smaller (350 A2) than the one involved in binding RassflC2338’. At this point it can only be speculated that this face might be involved in binding proteins with a different motif to the one encountered in Rassfl C. The RassflC2338’” sequence certainly provides clues on how the DHB domain identifies potential binding partners. It can be hypothesized that other interacting motifs contain a similar negatively-charged region juxtaposed with a short hydrophobic core. Some variability in these putative “DHB-interacting motifs” (DIMs) is expected, as the isolated DAXX’2°(SIM-N) is also able to bind to the DHB domain in the same region as RassflC2338’’ (section 4.5) despite having a very different hydrophobic sequence. Indeed, based on the structure of the DHB/RassflC2338” complex, it can be predicted that the intramolecular binding of these two DAXX regions occurs in a reverse orientation with respect to RassflC. This is because, contrary to the RassflC2338v peptide, the hydrophobic core in SIM-N is N-terminal to the negatively-charged cluster of residues (Table 3.3). Reverse binding of short motives is also observed with SUMO (Reverter and Lima, 2005; Song et a!., 2005), and Sin3B (Brubaker et a!., 2000; Swanson et al., 2004). 119 Chapter 3 — Structure of the DAXX/RASSF1C complex Table 3.3 - Proteins previously reported in the literature to interact with the N-terminus of DAXX. 3 19 RassflC 23 SQEDSDSELEQYFTAR 38USP7 or HAUSP 1-160, Full-length CoIP Hausp 828 QRLNTDPMLLQFFKSQ 843 (Tang et al.,2006)(NP_003461) 347-570 (1102 aa) Hausp 856 YEG LRDLLQFFKPR 871 1-160, RassflC 23 SQEDSDSELEQYFTAR 38 MDM2 1-260, Full-length Mdm2 158 SSSPKEEQLEQVLDKQ 141 (Tangetal.,Co-IP(NP_002383) 260-740, (497aa) Mdin2 296 FEEDPEISLPDYWKCT 313 2006) 157-260 Mdm2 298 EFSDTDSEGAQYV1VQ 283 RassflC 23 SQED---SDSELE-QYFTAR 38GST, (Gostissa etp53 1-188 294-393 p53 39 NDDLMLSPDDIE—QWFTEfl 57 al., 2004)(BAC16799) Y2H. p53 312 SSSPQPKKKPLDQEYFTLQ 331 STAT3 GST, RassflC 23 SQEDSDSELEQYFThR 38 (Muromoto1-240 320-493(NP_644805) Co-IP Stat 458 SHTELDIKLGQHYVET 443 et al., 2006) NHEI 1-130, Y2H, RassflC 23 SQEDSDSELEQYFTR 38 (Jungetal.,567-637(NP 003038) 13 1-400 Co-1P Nhel 565 QERSKEPQLIAFYHKM 580 2008) RassflC 23 SQEDSDSELEQY-FTR 38 ASK-i Askl 662 EEGDCESDLLEYDYEYO 678 (Kitamura70-2 16 Full-length Co-IP Aski 985 VSPDTELKVDPFSFKTR 1001 et al., 2009)(NP_005914) Aski 1305 GTNTEOSELTDw-LRVN 1320 pp7l 43-197 RassflC 23 SQEDSDSELEQY--FTR 38 (HofmannFull length Co-IP(AAA45997) 439-501 pp71 458 TDSDEDGSAEEF’GAFAE 441 et al., 2002) 1-686, 1- 598, 757-Axin 1-197 832, Co-IP No alignment (Li et al.,(AA113172) 2007)1-506, 576-832 HIPK2 1 188 1-520, (Hofiann- GST No alignment(NP 073577) 189-520 et al., 2003) (MuromotoDMAPI 1-240 293-411 Co-IP No alignment et al.,(CA123 197) 2004a) Co-IP: Co-immunoprecipitation of exogenous protein, GST: GST-pulldown of exogenous protein, Y2H: Yeast- two-hybrid. (b) Residues in red, blue and green are negatively-charged, positively-charged or hydrophobic, respectively. (c) Inverted sequence. As mentioned at the beginning of this chapter, the reported binding partners of DAXX have generally been studied with coarsely-defined truncation fragments. Table 3.3 summarizes most proteins that have been reported to bind constructs of DAXX that included the DHB domain. A sequence alignment with RassflC2338 shows that many of them do contain a potential DIM sequence. Moreover, Leu3 1, Tyr34 and Phe35 provide most of the intermolecular NOEs in the DAXX/Rassfl C complex, and most of the DIMs in these molecules have a high degree of conservation of these key residues (Table 3.3). (CAG33366) SIM-N AADEDDDDLVI I SNATAM 1 120 Chapter 3 — Structure of the DAXX/RASSF1C complex Constructs failing to give a good alignment covered regions of DAXX that might include parts of other domains (Chapter 2), or conversely, DAXX might contain recognition sequences in its unstructured regions for recruiting folded domains from other proteins. Further experiments are needed to investigate these possibilities. One DAXX binding partner of interest is the tumor suppressor protein p53. This so-called ‘guardian of the cell” is inactivated by mutation in about half of all known human cancers (Joerger and Fersht, 2007). p53 has been shown to interact with DAXX in a number of studies (Gostissa et al., 2004; Kim et al., 2003; Zhao et a!., 2004). This interaction seems to be crucial to the ability of cells to respond to DNA damage (Gostissa et a!., 2004). DAXX’’80,which contains the DHB domain, was shown to be involved in the interaction with p53 (as well as other DAXX domains). A sequence alignment between RassflC2338 and p53 reveal two possible DIMs within p53, one between residues 39-57 and the other in the region 312-331 (Table 3.3). Consistent with this hypothesis, Gotissa et al. reported that region 294-3 93 of p53 is involved in binding DAXX. However, all of their deletion constructs lacked residues 1-74, and thus they might have missed the first DIM. The N-terminal DIM in p53 might have higher affinity for DAXX as its hydrophobic core is preceded by acidic residues similar to the Rassf1C2338vpeptide, whereas the second DIM contains a number of Lys residues, which would decrease or abrogate the electrostatic interaction. Additionally, the first DIM resides in the unfolded N-terminal transactivation domain of p53, which is known to interact with Mdm2, with components of the transcription initiation complex, and with acetyltransferases p300/CBP (Joerger and Fersht, 2007). The Mdm2 interaction involves residues 15-29 of p53, which also form an amphipathic helix upon binding (Kussie et 121 Chapter 3 — Structure of the DAXX/RASSF1C complex al., 1996). Interestingly, Tang et al. reported a direct interaction between DAXX and Mdm2 by co-immunoprecipitation studies. This direct interaction was mediated by residues 157-260 of DAXX (Tang et al., 2006). It can be hypothesized that p53 could serve as a bridge between Mdm2 and DAXX, enhancing their direct interaction. It will be interesting to investigate this in the future. 3.6.4 Possible new DIM-containing interacting partners of DAXX A BLAST search using the RassflC2338 as a query returned mostly proteins involved in transcription or signaling pathways. However, they have not (yet) been reported to interact directly with DAXX (Table 3.4). Of interest is the general transcription factor IIIB, which regulates RNA polymerase Ill-mediated transcription. This could provide a new mechanism by which DAXX regulates transcription independently of HDACs. It will be exciting to see whether these structural studies reveal new DAXX partners that could play an important role in the numerous functions attributed to DAXX to date. 122 Chapter 3 — Structure of the DAXX!RASSF1C complex Table 3.4 - BLAST search for sequences similar to RassflC2338.(a) Protein Protein size (Accession Alignment with RassflC Protein Function Reference number) Location of possible DIMs (b) 419 aa. TFHLB 23 — SQEDSDSELEQYFTAR — 38 Transcription initiation factor (TeichmannNP 060780 381 — DENISDSEIEQYLRTP — 395 for RNA polymerase Ill. etal..2000) C-terminus. 2641 aa. PDZ domain 23 — SQEDSDSELEQYFTAR — 38 Scaffold protein with three containing 662 - DSLISESELSQYFAHD - 675 PDZ domains, involved in (Tam etal.,protein AJPC transcriptional activation of 2008)AAKO7 661 Immediately after the second PDZ p53. domain. 305 aa Co-repressor of transcriptionAlien 24 — QEDSDSE—-——LE—QYFTAR - 38 factors involved in cell-cycle (PapaioannouAAD30269 18 — SEDSNSEPNVDLENQYYNS — 36 etal.. 2007) regulation and DNA repair. N-terminus. Chromodomain 2715 aa. Helicase DNA Chromatin remodeling enzyme (Lutz et al.,(CHD) binding 24 — QEDSDSELEQYFTAR — 38 involved in transcription 2006)protein 6 487 LDKKSDESLEQYFYSVA -1504 regulation.AA172397 772 aa.Nuclear Factor Transcription factor involved(erythroid 24 — QEDSDSELEQYFTAR - 38 in the induction of genes (Wangetal.,derived 2)-like 1 67 — IHPKSIDLDNYFTAR — 81 encoding detoxifying 2007)NPO 03195 enzymes. N-terminus. (a) Significant hits returned by BLAST (Altschul et a!., 1990) using the RassflC2338 sequence as a query (top sequence) in the non-redundant-protein sequence database, and searching only human proteins. (b) Residues in red, blue and green are negatively-charged, positively-charged or hydrophobic, respectively. 123 Chapter 3 — Structure of the DAXX/RASSF1C complex 3.7 Biological implications of the RassflC-DAXX interaction The molecular mechanism for the interaction between DAXX and Rassfl C presented here is consistent with previous studies by Kitagawa et al. (DAXX’’60/RassflC’220) and Lindsay et al. (DAXX’’42/RassflC50)(Kitagawa et a!., 2006; Lindsay et a!., 2009b). However, as discussed in section 1.4.3 and summarized in Figure 1.6, the two groups proposed different functions for the DAXX/Rassfl C interaction. Kitagawa et al. suggested that the interaction resulted in sequestration of Rassfl C into PML NBs, and Lindsay et al. proposed that the interaction controlled a mitotic checkpoint in a cell- cycle dependent manner. Lindsay et al. failed to reproduce Kitagawa’ s result, and DAXX and Rassf1 C did not interact during interphase, as the proteins were in the nucleus and cytoplasm, respectively. Other groups have also reported that Rassfl C is mostly located in the cytoplasm with RassflA (Estrabaud et al., 2007; Liu et a!., 2005). It is unclear at the moment what other mechanisms were involved in Kitagawa’ s study that resulted in Rassfl C nuclear localization, and DAXX-mediated sequestration into the PML-NBs during interphase. Additionally, Lindsay et al. found that the Rassfl C/DAXX interaction played an important role in the cell cycle progression. Particularly, this proved very important to understand how breast cancer cells can show resistance to taxol (a common agent used for chemotherapy). In short, upon taxol treatment, Rassfl C and DAXX interacted in taxol-sensitive cancer cells, which resulted in a quick exit from mitosis. This quick exit eventually leads to cell death due to the stabilization of microtubules and a net increase in tubulin polymerization caused by taxol. On the other hand, taxol-resistant cells were found to be DAXX-depleted. The lack of the Rassfl/DAXX complex caused cells to arrest in mitosis long enough for taxol to decay. After 124 Chapter 3 — Structure of the DAXX/RASSF1C complex depletion of taxol, these cancer cells exited mitosis and continued to proliferate. Interestingly, Lindsay et al. were able to convert taxol-sensitive cancer cells to taxol-resistant cells by overexpressing GFP-Rassfl C 1-50, which presumably competitively depletes DAXX and abrogates its interaction with endogenous Rassfl C. The role of DAXX in mitosis is just beginning to be investigated and a number of questions remain unanswered. The most obvious question is how the DAXX!RassflC interaction does to affect the progression of mitosis. Lindsay et al. found that RassflC dimerized with RassflA, which has been recently shown to physically interact with and become phosphorylated by the Aurora-A kinase (Rong et al., 2007; Song et al., 2009). The Aurora-A kinase has been highly conserved through evolution, and is carefully regulated (by ubiquitylation and phosphorylation) during mitosis (Fu et al., 2007). Interestingly, Lindsay et al. discovered that inhibitors for the Aurora kinase converted DAXX/RassflA-depleted taxol resistant cells to taxol-sensitive cells (Lindsay, 2009). It is unclear yet how DAXX/RassflA could inactivate the Aurora kinase. Since both proteins do not have any catalytic domain, it is fair to speculate that inactivation would be through recruitment of other complexes with enzymatic activity, or by competitive binding of the active domain of the kinase. The next question concerns the mechanisms that regulate these interactions. The observation that S TM-N and Rassfl C bind at the same site of DAXX55’44 could provide a mechanism in which SUMO binding regulates association of the DHB domain with Rassfl C by relieving intramolecular binding of SIM-N (discussed further in Chapter 5). Indeed, the role of SUMO in mitotic checkpoints is just beginning to be uncovered (Gutierrez and Ronai, 2006). It will be 125 Chapter 3 — Structure of the DAXX/RASSF1C complex definitely exciting to see what further experimentation reveals in this new research area in the near future. 3.8 Materials and methods 3.8.1 Cloning Table 3.1 summarizes the constructs used for this chapter. Unless noted otherwise, all of the Rassfl C constructs were cloned via PCR amplification from a pBABE plasmid containing RassflC’5°kindly provided by Dr. Alexander Ishov (University of Florida). Constructs that were inserted into a PGEX-2T vector were cloned with BamHI and EcoRI recognition sequences. The genes encoding Rassf1C2338’and RassflC82’were cloned using two complementary DNA oligomers encoding the entire Rassfl C fragment sequence, with BamHI and EcoRI sticky ends (PAGE purified, Integrated DNA technologies, Iowa, USA). The DNA oligoniers were dissolved in deionized water to a final concentration of 200 jtM, and mixed in a ratio 1:1. 0.5 tL of this mixture were diluted in 99.5 iL of water and kept at 95 °C for 5 minutes. Then, the samples were cooled slowly to room temperature over 20 minutes. The annealed inserts were ligated into previously cleaved and dephosphorylated PGEX-2T. The Rassf1C2344’’ and RassflC235° genes were cloned into PGEX-6P- 1 (which encodes a PreScission proteolysis-site instead of thrombin site) via PCR (BamHI/EcoRI). RassflC’5°was also inserted into a pET28a vector with NdeI and Hindill restriction enzymes. 3.8.2 Sample purification DAXX55’44 purification was carried out as described in Chapter 2. The purification of RassflC821,RassflC2338”,and RassflC838 followed the same protocol as described in Chapter 4 126 Chapter 3 — Structure of the DAXX/RASSF1C complex with the GST-SIM peptides. The following sections describe the purification of RassflC’50, RassflC234”, and RassflC2350. RassflC15° ExpressedHis6-RassflC’5°formed inclusion bodies. Thus, its purification followed a similar protocol to that described for DAXX55’44in Chapter 2, although 6 M guanidinium hydrochloride was added to the Ni2 binding and elution buffers. The eluted sample was purified by reversed phase HPLC (0.1% TFA, water/acetonitrile gradient) to remove the salt. Preliminary tests showed that His6-RassflC’5°was insoluble at pH values below 8. However, after thrombin treatment to remove the His6-tag, RassflC’5°became soluble at pH 6.5. Prolonged thrombin treatment was found to be problematic, as MALDI-ToF-MS showed the presence of a significant amount of RassflC’38.The latter results from cleaving at a pseudo-thrombin cleavage site between Arg38 and Thr39. For large scale production of‘5N- and13C/’5N-labeled samples, HPLC-purified peptides were re-suspended in Tris buffer (20 mM, 150 mM KC1, 2.5 mM CaC12,pH 8.8) and treated with 1.5 units of thrombin (Novagen) for 1.5 hours. The sample pH was then adjusted to 6.5, and the lack of precipitation showed that the His6-tag removal was complete. Excess DTT (10 mM final) and benzamidine beads (200 iL, Sigma) was added to deactivate thrombin. The samples were then immediately re-purified by HPLC. MALDI-ToF-MS confirmed that the‘5N-RassflC’°was homogenous, with no RassflC’38present. Unfortunately, the‘3C/’5N sample contained a small amount of Rassfl C’38. 127 Chapter 3 — Structure of the DAXX/RASSF1C complex Rassf1C234”’and Rassf1C235° Rassf1C2344’and RassflC235°were purified from a pGEX-6P-1 vector, which contains a PreScission cleavage site between the GST-tag and the peptide. This was used to avoid the heterogeneous samples that resulted from cleavage at the pseudo-thrombin site at Arg3 8. The PreScission enzyme was kindly prepared by members of Dr. Strynadka laboratory. Samples were incubated with 200 tg of GST-tagged PreScission for 48 hours at room temperature. SDS-PAGE showed incomplete cleavage after 24 hours, and complete cleavage after 48 hours. MALDI-ToF MS showed that the RassflC235° sample had a major peak at 3683 Da (expected mass 3682 Da), and a smaller contaminant at 1843 Da, probably RassflC’226 or RassflC2’7.RassflC 23-44w showed a peak 3102 Da (expected 3100 Da) and another peak of same intensity at 2604 Da, which corresponds to RassflC2342. Preliminary tests with these samples showed no significant improvement in binding DAXX55144 (Table 3.1), and thus further efforts to optimize their purification was unnecessary. 3.8.3 NMR-monitored titrations Table 3.1 outlines all the samples used for titrations performed in this chapter, with their respective concentrations and buffers. Concentration of the samples was determined by UV absorption at 280 nm using the following predicted molar absorptivities: RassflC’5° and RassflC838 (8480 M’ cm1); RassflC821,RassflC2338’,and Rassf1C2344”’(6990 M’ cm1); and RassflC235°(1490 M’ cm’). All titrations were performed with 500 tL of the‘5N-labeled species and increasing amounts of the unlabeled species until saturation was reached as determined by plotting the binding curve 128 Chapter 3 — Structure of the DAXX!RASSF1C complex Hon the flyH• All of the NMR-monitored titrations were in the fast-exchange regime, and the dissociation constants were calculated as described in Appendix 4. 3.8.4 Structure of the complex Sample preparation After optimizing conditions, MOPS buffer (10 mM, pH 6.5, 0.1 mM EDTA) was employed for preparing the complex between DAXX55’44 and RassflC2338”. Specifically,‘3C/’5N-labeled RassflC2338V(1.2 mM, 250 iL) was prepared and titrated with unlabeled DAXX55’44 (3.4 mM) to saturation. At a 1.5:1 DAXX5’44:RassflC2338”ratio, -92% saturation was reached (Figure 3.1 2A) as determined by comparison with a previous titration (Figure 3 .4C-D). The estimated KD (25 .iM) was consistent with the result in HEPES buffer at pH 7.5 (Table 3.1). A control 2D-’3C/5Nisotope-filtered NOESY-HSQC experiment of labeled RassflC 23-38w in the absence of protein did not show any cross-peaks (not shown). In contrast, at 92% saturation with DAXX55’44the spectra showed a considerable number of cross peaks (Figure 3.12 B-C). 129 Chapter 3 — Structure of the DAXX/RASSF1C complex Figure 3.12 - Preparation of DAXX55444IRassf1C2338’’complexes with complementary labeling for structural studies. The sections on the left (A-C) show results from the‘3C/’5N- RassflC2338”/ unlabeled DAXX55’44 sample (Rd complex); whereas the sections on the right illustrate results from the‘3C/’N-DAXX54/unlabeled RassflC2338s sample (Dr complex). (A,D) Titration curves show that samples at an unlabeled/labeled ratio of 1.5 were >90% saturated as estimated from the regression curves. The dissociation constant was approximately 20 jiM. (B-C, E-F) Sections of the methyl region (C,F) or amide region (B,E) of a 2D-’3C/5N isotope-filtered NOESY experiment. Cross peaks correspond to intermolecular NOEs. The lines illustrate the diagonal for reference. 130 E 0. S 0. 0. Chapter 3 — Structure of the DAXX/RASSF1C complex ‘3ci5N-DAXx44 liC/J4b.Rassf1C233 rj, 6 ‘3C/’5N—RassfiC2338 2/’4-DAXX’4 0.4 I o. 0.3 0. < 0.2 0.1 DA • I5 • A37 • __—. DAXX:RassflC ratio • i1$ -—- • V125 YI2 1127 /‘ - K-211iM 0.5 1.5 RassflC/DAXX ratio E ;- _-:-‘ 7.5 7.0 F 2. 4. 6 2 4 6 B 0 . - 7.5 C -- 0.3 0.2 0.1 131 2 C, O 4 6 2 1.5 ii RassfIC23 - ‘H (ppm) . — _____________ 1.5 1.0 DAXX ‘H (ppm) Chapter 3 — Structure of the DAXX/RASSF1C complex The reversed titration with‘3C/’5N-DAXX5’44(1.1 mM) and unlabeled Rassf1C2338”’(10.8 mM) was carried out until reaching a similar level of saturation ( 96%) at roughly the same stoichiometric ratio as the previous titration (Figure 3.12D). Consistently, a similar KD (21 1iM) was obtained. The 2D ‘3C/’5N isotope-filtered NOESY-HSQC experiment showed a small number of cross-peaks, especially in the methyl region (Figure 3.12 E-F). Resonance assignments For simplicity, the‘3C/’5N-RassflC2338”/unl beled DAXX55’44 sample will be referred as the “Rd” sample, and the13C/’N-DAXX55’44/unlabeled Rassf1C2338’’ sample as the “Dr” sample. Resonances from DAXX55’44 were assigned using the following experiments on the Dr complex: HCCH-TOCSY, (H)CC(CO)NH-TOCSY, HCC(CO)NH-TOCSY, (HB)CB(CGCD)HD, (HB)CB(CGCDCE)HE, CT-HSQC, aromatic HCCH-TOCSY, ‘3C- and ‘5N-NOESY-HSQC (see section 2.4.1). Assignments for the apo-protein were used as a starting point since only resonances from residues near the binding site showed any significant changes. Resonances from Rassf1C2338vwere assigned using the following experiments on the Rd complex: (H)CC(CO)NH-TOCSY, HCC(CO)NH-TOCSY, (HB)CB(CGCD)HD, (HB)CB(CGCD)HD-TOCSY, CT-HSQC, TOCSY-HSQC, ‘3C- and‘5N-NOESY-HSQC. Amide resonances on the free peptide were obtained from HNCACB and (HB)CBCA(CO)NH experiments and were transferred to the complex following their shifts on the fast exchange titration. From these spectra, 96% of all non-labile nuclei in the complex were assigned (Appendices 8 and 9). 132 Chapter 3 — Structure of the DAXX/RASSF1C complex 3.8.43 Structure calculation Simultaneous three-dimensional ‘3C- and‘5N-NOESY-HSQC, and ‘3C/’5N isotope-filtered NOESY-HSQC spectra for the Rd and the Dr samples (with 100 ms and 150 ms mixing times, respectively) were collected over two weeks. Control ‘H-15N HSQC spectra before and after the three-dimensional experiments showed that the samples were stable. All spectra were processed with NMRPipe and analyzed with Sparky (Delaglio et a!., 1995; Goddard and Kneller, 2004). Peaks in the eight spectra were manually picked and partially assigned. For automatic NOE assignments and structural calculations, ARIA 2.2 was used (Habeck et al., 2004), specifying that filtered experiments contained intermolecular NOEs. Chemical shift tolerances are an important parameter in automatic NOE assignments. If they are too narrow, ARIA will likely miss the correct assignment for an NOE, and if they are too wide, too many possibilities allow the program discard the peak early in the calculation (Fossi et al., 2005). Thus, relatively wide tolerances (0.08-0.15 ppm) for the filtered experiments were used for the initial run. The initial structure was used to manually fix violations and further assign peaks. This process was repeated eight times, until no violations> 0.5 A were used for the calculation. The majority of long range intramolecular NOEs were automatically assigned, whereas all of the intermolecular NOEs were manually assigned. The quality of the final ensemble was analyzed with Procheck_NMR and by its statistics (Table 3.2) 133 Chapter 4 SUMO-interacting motifs (SIM) in DAXX In this chapter, the identification of two SUMO-interacting motifs in DAXX, one at its N- terminus (SIM-N) and one at its C-terminus (SIM-C) is presented. Their interactions with SUMO- 1 and SUMO-2 were examined using NMR spectroscopy, and the following properties discovered. (1) Binding was mediated by both hydrophobic and electrostatic interactions. (2) SIM-N had a 4 fold higher affinity towards both SUMO paralogues than SIM-C. The difference in affinity is attributed to the difference in charged residues that flank their identical hydrophobic core sequences. (3) SIM-N bound SUMO-1 with one predominant orientation, whereas SIM-C adopted multiple binding modes. (4) Both SIMs were found to bind SUMO-1 and SUMO-2 with similar affinities under most conditions. In addition SIM-N was found to bind intramolecularly to the DHB domain and SIM-C to mediate the interaction of DAXX with the sumoylated Ets-1 transcription factor. Importantly, the latter did not involve any direct contacts between DAXX and Ets- 1, but rather derived from the non-covalent binding of the DAXX SIM-C to SUMO- 1, which in turn was covalently linked via an isopeptide bond in a Theads-on-a-string’ fashion to the unstructured N-terminal segment of Ets- 1. These results provide new insights into the binding mechanisms and hence biological roles of SUMO-interacting motifs. 134 Chapter 4— SUMO interacting motifs in DAXX 4.1 DAXX UMO-interacting motifs (SIM-N and SIM-C) 4.1.1 Identification of SIM-N and SIM-C As discussed in Chapter 1, DAXX has been reported to bind and repress a number of transcription factors, many of which are sumoylated (Lindsay et a!., 2008; Shih et al., 2007). For example, the transcription factor Smad4 is repressed by DAXX, and mutations that prevented Smad4 from being sumoylated completely relieve this effect (Chang et al., 2005). The hypothesis is that DAXX is able to bind sumoylated proteins via possible SIMs. In order to identify the SIMs within DAXX, the consensus sequence established by Song et al. (Val/Ile-X-Val/Ile Val/Ile) was searched (Song et a!., 2004). Close inspection of the DAXX sequence allowed the identification of two segments that contain such consensus motifs, one at the N-terminus (denoted herein as SIM-N) and one at the C-terminus (SIM-C) (Figure 2.2). Indeed, while this work was in progress, two groups independently identified these SIMs and confirmed their ability to mediate binding of DAXX to SUMO. Lin et a!. first reported the presence of a SIM-C within DAXX (Lin et a!., 2006). They used a yeast two-hybrid system to show that deletion of residues 73 2-740 or mutations within the hydrophobic core of the SIM(73311VL6) impaired the binding of DAXX to SUMO-l. Moreover, the presence of SIM-C in DAXX was crucial for its ability to be targeted to PML bodies and to repress several sumoylated transcription factors, including Smad4 and AR. More recently, Santiago et al. reported the existence of SIM-N (Santiago et al., 2009). In their study, SIM-N was shown to bind both SUMO-l and SUMO-2/3 by a yeast two-hybrid system. Indeed, the authors observed that it was necessary to remove both SIM-N and SIM-C from DAXX to completely abrogate its SUMO binding capacity. This stands in contrast to the 135 Chapter 4— SUMO interacting motifs in DAXX work of Lin et al. in which drastic changes in the function of DAXX occurred upon disruption of only SIM-C (with SIM-N being unrecognized at the time). One interesting observation mentioned by Santiago et al. is that DAXX is able to bind SUMO-1, SUMO-2 and SUMO-3 through either of its SIMs, but not SUMO-4. They speculated that the presumed inability of SUMO-4 to be covalently attached to substrates accounts for their result, i.e. that DAXX cannot bind free SUMO, but it rather associates preferably with sumoylated proteins. 4.1.2 Role of DAXX SIMs in its localization As described in Chapter 1, DAXX localization is crucial to its function. In the studies described above, both groups showed that recruitment of DAXX into PML-NBs depends greatly on the ability of DAXX to bind sumoylated PML (Lin et a!., 2006; Santiago et al., 2009). A group in France initially identified Lys 160 as the sumoylation site in PML that controls DAXX recruitment (Zhu et al., 2005). More recently, the same group reported that PML is primarily isopeptide-linked to SUMO-2 at this lysine, with SUMO-1 being added at Lys65 or Lys490. (Lallemand-Breitenbach et al., 2008). Therefore, paralogue selectivity from each SIM might have biological importance in determining the localization of DAXX. 4.1.3 Research objectives The proposed model of Santiago et a!. in which DAXX SIMs cannot bind free SUMO is at odds with various reports in which SIMs of other proteins have bound free SUMO (Duda et al., 2007; Hecker et al., 2006; Sekiyama et al., 2008; Song et a!., 2005). The first objective was to determine the functionality of DAXX SIMs. Can they bind free SUMO? Is this binding independent of the protein to which SUMO is covalently linked? 136 Chapter 4— SUMO interacting motifs in DAXX It is peculiar that DAXX possesses a SIM at each of its termini. It is tempting to speculate that they might have different functions. Indeed, a recent report used a yeast-two hybrid assay to suggest that the SIM-N and SIM-C have preference towards SUMO-2 and SUMO-1, respectively (Chen et a!., 2006). Since it is established now that the specific SUMO paralogue on a protein can have distinct functions (discussed in section 1.3.4), the second objective was to determine the relative affinities of DAXX SIM-N and SIM-C towards SUMO-1 and SUMO-2. What are the structural and thermodynamic bases for differences in affinity/specificity of the SUMO paralogues? Section 4.2 explores these two questions. The available structures of SUMO-l with SIMs show two possible binding modes: one with a SIM adopting a 13—strand conformation to from a short parallel 13—sheet with strand 2 of SUMO-1 (Song et a!., 2005), and one at the same site but with an antiparallel orientation (Reverter and Lima, 2006) (Figure 4.5A). The third objective was to determine the binding mode of SUMO-1 and SUMO-2 with the two DAXX SIMs. Which binding mode do DAXX SIMs adopt to bind SUMO? In sections 4.3 and 4.4 an answer to this question is found. It is desirable to expand the knowledge of the biological function of SIM-N and SIM-C. SIMs can be involved in regulation of protein sumoylation, conformation, localization and interaction (discussed in section 1.3.4) (Kerscher, 2007). Until now all SIMs reported in the literature are specific for SUMO-binding. The fourth objective was to resolve whether DAXX SIMs can only bind SUMO or whether they are also involved in other mechanisms. Section 4.5 discusses how SIM-N is involved in intramolecular binding with the DHB domain, which opens up possibilities of a new regulatory function of SIMs involving competitive binding. 137 Chapter 4— SUMO interacting motifs in DAXX Finally, it has been reported that DAXX repression is mediated by DAXX SIMs (Lin et a!., 2006; Shih et a!., 2007). The last objective was to investigate the role of DAXX SIMs in mediating the interaction with transcription factors. Is there any cooperative binding involving other regions of DAXX with the model transcription factor Ets- 1? Section 4.6 investigates how surnoylation of the transcription factor Ets-l affects its ability to bind DAXX. 4.2 NMR-monitored titrations of DAXX-SIM-N and SIM-C with SUMO-1 and SUMO-2 4.2.1 Preamb1e SIM binding by SUMO-1 and SUMO-2 SUMO-l and SUMO-2 share 50% sequence identity and adopt highly similar ubiquitin-like structures (Figure 4.1). Not surprisingly, each utilizes the analogous cleft to bind SIMs (Hecker et al., 2006). This cleft, which is between the strand 2 and helix 1, consists mostly of hydrophobic residues surrounded by positive charges from a number of Lys and Arg residues (Figure 4.1). This “key lock” matches the complementary negative charges from Glu and Asp residues within SIMs. However, the exact position of the Lys/Arg residues does differ between SUMO-l and SUMO-2, suggesting that electrostatic interactions could dictate a paralogue preference towards various SIMs. Indeed, one of the first SIMs that showed higher affinity towards SUMO-2 could be converted to binding SUMO-1 preferably by changing the charges flanking the hydrophobic core of the SIM (TTRAP, Figure 4.1 C) (Hecker et a!., 2006). However, a recent study suggested that the SIM hydrophobic residues alone can determine paralogue specificity (USP25, Figure 4.1C) (Meulmeester et al., 2008). Interestingly, the SIM used for that particular case contained an unusual Val-Ile-Asp-Leu core, so presumably electrostatics still played a role in determining binding specificity. 138 Chapter 4— SUMO interacting motifs in DAXX Figure 4.1 - Differences between SUMO-1 and SUMO-2. (A) Sequence alignment of SUMO 1 and SUMO-2. Identical and similar residues are highlighted in black and gray, respectively. Secondary structure elements are shown as arrows (13-strands) and cylinders (ct-helices). (B) Comparison of the surfaces of SUMO-1 (1A5R.pdb) and SUMO-2 (1WM2.pdb), showing the side containing the SIM binding cleft (dashed circle). Positively-charged residues are shown in blue, negatively-charged in red, and non-polar residues in gray (electrostatic surface calculated with the “simplecharge” method of MOLMOL using default parameters). Cartoon representations of the molecules are shown for reference. (C) Sequence alignment of SIMs reported to have SUMO-paralogue specificity (top) and selected SIMs that showed no SUMO paralogue preference (middle). The sequence of the studied SIM-N and SIM-C peptides (bottom). The sequence of the human SIM-C, identical to the murine version in its SIM, is shown for reference, although it was not used for these studies (gray). The hydrophobic core is shown in green, the Asp/Glu residues in red. 139 Chapter 4— SUMO interacting motifs in DAXX A 10 20 30 40 50 31 132 I ccl S UMO- 1 SQPI SUMO- 2 AIEKPIE 131 Ø11 132 Øii ccl 1 10 20 30 40 50 60 70 80 90 133 ____ ci2 135 _____ ____ SUMO-1 NS SUMO-2 RQ I •PN f33 134 a2 60 70 80 90 Protein RANBP2 TTRAP wr TTRAP mutant USP25 U5P25 MCAF1 ELM [.]DNEKEC RCGGLPNN RCCGLPNN GSQADTN TN G SSESE Preference SUMO-1 SUMO-2 SUMO-1 SUMO-3 SUMO-3 SUMO-3 SUMO-2 Reference Tatham et al. 2005 Hecker et al. 2006 Hecker et al. 2006 Meulmeester et al. 2008 Meulmeester et al. 2008 Sekiyama et al. 2008 Zhu et al. 2008 hDAXX SIM-N mDAXX SIM-C hrAXX SIM-C SUMO- 1 / 2 SUMO-1 or SUMO-1/2 SUMO-1 SUMO-2B C Sequence Cliv WEKKPTVEK?U<ADTLKLP [. I 11W WEFLGKPKHCQYTWDTQ 111W WEFEDEPKHCQYTWDT VIDL TGDDRDDL VIDL T VIOL TMDDEE QIDL TEEQKDDSEWLSSD VC1 DDGPI PIAS1 KNKK VEVI DLTIOSSSDEEEEEP SUMO-1/2 Hecker et al. 2006 SENAThXIN SRGQ VIII SDSODDDDE SUMO-1/2 Hecker et al. 2006 SP100 QASD IIVI SSEOSEGSTDVD SUMO-1/2 Hecker et a1. 2006 TOPORS RSPV VITI OSDSDKDSEVKGD StJMO-1/2 Recker et al. 2006 ZNF237 0000000 vvFI ESI SUMO—1/2 Recker et al. 2006 MATS IIVL DDDDEDEAR 19 718 IYKTSVATQCDPEE IIVL SDSD 739 719 TCITS’JEE IIVL SDSD 740 140 Chapter 4— SUMO interacting motifs in DAXX In fact, all of the peptides reported to bind preferentially to SUMO-2 contain an Asp at the third position of the V/I-V/I-X-V/I SIM (MCAF 1, TTRAP, USP25 and BLM Figure 4.1 C) (Hecker et al., 2006; Meulmeester et a!., 2008; Sekiyama et al., 2008; Tatham et a!., 2005; Zhu et al., 2008). Only the section containing the so-called 1R2 (internal repeat 2) of RANBP2 and the mutant TTRAP have been reported to bind preferentially to SUMO-1 in vitro (Figure 4.1C) (Hecker et al., 2006; Tatham et al., 2005). Overall, SIMs reported to show specificity for a particular paralogue have the negatively- charged residues mixed with neutral amino acids. In contrast, most known SIMs have Asp/Glu stretches mixed only with Ser residues, and show no paralogue predilection (selected examples from (Reeker et a!., 2006) on Figure 4.1C). Thus, SIMs with apparent paralogue specificity have one of two characteristics: an asymmetric hydrophobic core containing a charged residue; and/or juxtaposed charged residues mixed with non-polar residues. In contrast, most SIMs with canonical hydrophobic cores and contiguous juxtaposed charged residues have no apparent paralogue specificity. It is interesting that both SIMs from DAXX contain the identical hydrophobic core Ile-Ile Val-Leu, which is rather symmetric and non-polar. Thus, any specificity would have to reside on the juxtaposing charged residues. SIM-N has a stretch of seven Asp/Glu residues following the hydrophobic core (Figure 4.1 C), which makes it an unlike candidate for exhibiting paralogue specificity. SIM-C, on the other hand, has much shorter Asp/Glu stretches on both sides of the hydrophobic core, mixed with only a Pro residue on one side and two Ser residues on the other (Figure 2.1). SIM-C might be a better candidate for having paralogue specificity. To address this question, NMR spectroscopy was used to characterize the structural and thermodynamic bases 141 Chapter 4 — SUMO interacting motifs in DAXX for the interaction of isolated peptides containing the DAXX SIM-N and SIM-C with purified SUMO-1 and SUMO-2. 4.2.2 NMR characterization of SIM-N, SIM-C, SUMO-1 and SUMO-2 An expression system to obtain 2O aa peptides encompassing SIM-N (residues 1-19 of hDAXX) and SIM-C (residues 718-739 of mDAXX, identical in its SIM to hDAXX) was designed (Figure 4.1C, see materials and methods for details). The ‘H-’5N HSQC spectra of labeled SIM-N and SIM-C show little dispersion in the amide hydrogen dimension, indicating the lack of any predominant secondary structure (Appendix 3). Using HNCACB and CBCA(CO)NH experiments (Figure 2.5A), the signals from the backbone ‘H, ‘3C, and ‘5N nuclei of both peptides was assigned. Similarly, the assignments of SUMO- 1 and SUMO-2, reported previously in the literature, were mostly confirmed, although some errors in the literature assignments were detected and corrected (Appendix 3). 4.2.3 NMR-monitored interactions b The interactions of the DAXX SIM peptides and the SUMO paralogues were monitored by following changes in ‘H-’5N HSQC spectra of each‘5N-labeled species upon titration with the unlabeled species. Importantly, both isolated SIMs bound both purified SUMOs in vitro as shown by spectral changes. Chemical shifts are exquisitely sensitive to structural perturbations accompanying ligand binding. Although interpreting these shift changes in terms of explicit direct or indirect structural changes is difficult, a qualitative, low resolution view of the binding b To calculate the concentration of the samples via UV-spectroscopy, samples used for NMR-monitored titrations had a non-native Trp residue at either the N-terminus (SIM-C”, SUMO1” and SUMO2v) or at the C terminus (SUMO-N”). These constructs are marked with a superscript “w”. Control experiments with SUMO 1 !SIM-C (without Trp) showed identical results to SUMO- 1 w/ experiments (not shown), indicating that the non-native Trp did not interfere with binding. See materials and methods for details. 142 Chapter 4— SUMO interacting motifs in DAXX interface can be obtained by mapping residues showing the largest spectral perturbations onto the sequence or structure of the protein. Mapping of the binding interface Chemical shift perturbations confirm that the binding of SIM-N”’ and SIM-C”’ occurred at the cleft between strand 2 and helix 1 of both SUMOs (Figure 4.2). Conversely, the hydrophobic core, Ile-Ile-Val-Leu, of SIM-N” and SIM-C” showed the greatest chemical shift perturbations when bound to SUMO-1”’ and SUMO-2’’. This is in agreement with reported structures of SUMO/SIM complexes (Duda et a!., 2007; Hecker et a!., 2006; Sekiyama et a!., 2008; Song et al., 2005). In one particular titration, SUMO-l”/SIM-C”, peaks broadened beyond detection both on SUMO-l and SIM-C (Figure 4.2C, 4.2F). As explained in section 4.3, this appears to be due to conformation exchange between multiple binding orientations. SIM-N binding to SUMO-1 and SUMO-2 Upon titration with unlabeled SIM-N’’, the lHNl5N signals from many amides in‘5N-labeled SUMO-1’ and SUMO-2’’ showed small shift changes and progressive loss of intensity with the concomitant appearance of new peaks arising from the resulting complex (no added KCI, 17°C). This behavior is indicative of relatively high affinity binding at the edge of the slow exchange regime (kex < Av) on the chemical shift timescale (see Appendix 4 for an explanation on titration curves). For those amides showing fast exchange behavior (i.e. with small chemical shift changes M changes so kex > Av, Appendix 6), plots of A versus SIM concentration indicated essentially stoichiometric binding under the experimental conditions and hence KD < 5tM (Table 4.1; Appendix 5). 143 Chapter 4— SUMO interacting motifs in DAXX Figure 4.2 - Mapping of the binding interface of SUMO-1 and SUMO-2 with SIM-N and SIM-C. (A,C) Amide chemical shift perturbations (calculated as ,J(AoH)2 + (O.2A615N))of SUMO-l”’ and (B,D) SUMO-2” upon binding SIM-N” (A,B) and SIMCv (C,D). Also shown are SIM-N’1” (E,G) and SIM-C” (F,H) amide chemical shift changes upon binding SUMO-1’ (E,F) and SUMO2w (G,H). The signals of residues marked as * were broadened beyond detection in the complex. (I) Residues with chemical shift changes greater than 0.2 ppm upon binding SIM-N and SIM-C are shown in red on the cartoon structures of SUMO-l (1A5R.pdb) and SUMO-2 (1WM2.pdb). Arg39 of SUMO-1 broadened beyond detection in the complex with SIM-C (blue). 144 Chapter 4— SUMO interacting motifs in DAXX 20 40 60 SIM-N’ + 9 SUMO-1” SUMO-2” D 0.6 C JJIJ ÷ CM + CM S 0.5 0.4 & 0.3 0.2 0.1 0.8 S 0.6 < 0.4 0.2 0.8 S . 0.6 • 0.4 0.2 80 SIM-C” 0.4 0.2 0.6 0.4 0.2 F - --- _II:*II I H YKTSVATQCDPEEI IVLSDSD 0.6 0.4 0.2 I + CM C Residue SUMO-1 719 724 729 734 Residue SUMO-2 739 +SIM-N -i-SIM-C ÷SIM-N ÷SIM-C 145 Chapter 4— SUMO interacting motifs in DAXX As described above, it was hypothesized that the binding would be dependent on the overall sample ionic strength (salt concentration) due to the electrostatic contribution of the SIM-SUMO interactions. Indeed, as the KC1 concentration was increased from 0 mM to 100 mM to 200 mM, the binding progressively weakened in all four cases (Figure 4.3). This allowed the determination of KD values from the dependence of Aö on added ligand concentration (for details see Appendix 4). This result confirms the importance of the electrostatic contributions to SIM SUMO binding. Indeed, in a separate experiment, weak binding of a SIM-N mutant lacking completely the hydrophobic core was observed (section 4.4), showing that electrostatic interactions are sufficient for binding at very low ionic strengths. Inspection of Table 4.1 reveals that SIM-N” bound to SUMO-1”’ and SUMO-2”’ with similar affinity at all tested salt concentrations. Thus SIM-N” did not display paralogue specificity between SUMO-1 and SUMO-2. This is attributed to the symmetric composition of the hydrophobic core of SIM-N, together with the lack of non-polar residues on the acidic stretch of SIM-N (Figure 4.1C). SIM-C binding to SUMO-1 and SUMO-2 The NMR-monitored titrations of 15N-labeled SIM-C’ with unlabeled SUMO-1’’ and SUMO-2” showed ‘3-4 fold weaker binding than observed with SIMNv under similar conditions (100 and 200 mM KC1; Table 4.1). The signals of the hydrophobic core, IIVL, also exhibited significant broadening due to conformational exchange (section 4.3), which progressively worsened at higher salt concentrations (Figure 4.4). (For a detailed discussion on 146 Chapter 4— SUMO interacting motifs in DAXX z 125 2 126 z ‘5N SIM-N’ + SUMO1v 1116 Eli A18, A20 - 18A40 - V9 17 0 A 19 17 A19 ‘5N SIM-N” + SUMO-2’ 1116 E17 Ml A18’b A2 - 018 --- 1)16 Elf 17 A19 0 A2 0 A4 0 V9i 2 ‘5N SUMO-1’ + SIM-N” 123 124 ‘5N SUMO-2” + SIM-N B — - • ,v• E13 A L24 K37 R70 M59 E89 H43 F87 124 126 128 F64 K1l 126 127 1224 l24 8.9 8.8 8.7 Ml 8.7 8.6 8.5 8.4 8.3 8.2 A1870 128. l3OJ 122E 1116 E17 Ml A19 0 Ml 2 C.) 2 124J B 126 128 130 OLIO 8.6 8.4 8.2 8.0 ‘H (ppm) ‘H (ppm) Figure 4.3 - NMR-monitored titrations demonstrated that SIMNw bound both SUMO1w and SUMO2w. (A-B) Superimposition of a section of the ‘H-’5N HSQC spectra of SUMO-1’ and SUMO-2’ in the absence (red) and presence of increasing amounts (cyan to black) of unlabeled SIM-N” at 0 mM KC1. (C-F) Superimposition of a section of the ‘H-’5N HSQC spectra of SIM-N” in the absence (red) and presence of increasing amounts (cyan to black) of unlabeled SUMO-l” and SUMO-2’” at 100 mM and 200 mM KC1. 147 Chapter 4— SUMO interacting motifs in DAXX + SUMO-1” SIMCv + SUMO2sv Figure 4.4 - NMR-monitored titrations demonstrated that SIM-C’’ bound both SUMO-1’’ and SUMO-2’’. Superimposition of a section of the ‘H-’5N HSQC spectra of SIM-C” in the absence (red) and presence of increasing amounts (cyan to black) of unlabeled SUMO- 1” and SUMO-2” at O mM, 100 mM and 200 mM KC1. 127 ,1733 128 _-‘ A724 / 129i, Z / D739 B V734 D739 ((4 C V734 .j 1733 L735 ‘ °A724 D739 @ 127 I 128 129 127 I 128 129 1733 D739 0 % c-) E c-) c.) E C FE V734 1733 0 D739L735 A724 V734 1733 8.6 8.4 8.2 8.0 ‘H (ppm) o D’7 8.6 8.4 8.2 8.0 ‘H (ppm) 148 Chapter 4— SUMO interacting motifs in DAXX lineshape changes due to ligand binding see Appendix 6). Furthermore, at 0 mM salt, SIM-C exhibits —3 fold higher affinity towards SUMO- 1’’ than SUMO-2’. This preference is reduced at 100 mM and 200 mM salt concentrations, enforcing the hypothesis that charges on the SIMs are crucial in determining paralogue specificity (Table 4.1). Table 4.1 - Dissociation constants between SIMNw and SIM-C’’ with SUMO1w and suMo2i.a SIM-N’’ KD(1iM) SIM-C’’ KD (jiM) [KCI] (mM) SUMO-1’’ SUMO-2’’ SUMO1v SUMO2w Stoichiometric Stoichiometric0 4.5±1.4 16±7(<5) (<5) 100 8.1± 1.7 8.7±0.9 30±7 24±6 200 36±3 38±6 110±14 170±24 a pH 6.5, and 25 °C. Reported KD values and errors correspond to the mean and standard deviation of the individually fit titration curves of at least 3 residues. 4.3 Comparison with values from the literature Table 4.2 summarizes the dissociation constants between SUMO and SIMs in the literature. At low ionic strength (20 mM phosphate buffer), the reported KD values are all between 1-10 jiM, which is similar to the KD values of DAXX SIMs (10 mM phosphate buffer, < 5 jiM for SIM-N’, 5-15 jiM for SIM-C”). Only Sekiyama et al. studied KD, values as a function of ionic strength, and did not interpret their data as showing any significant changes in the KD values (Table 4.2) (Sekiyama et a!., 2008). This is surprising, since the SIM peptide in that study (G TMDDEE) contained a stretch of charged residues, which when they mutated, dramatically disrupted binding to both SUMO-1 and SUMO-3. Closer inspection of their ITC binding curves reveals that the data became progressively noisier at higher salt concentrations (due to smaller heat release), and thus their reported KD values carry a greater error. Indeed, at 300 mM NaCl, SUMO-3 did not show any significant binding to the MCAF1 peptide (Sekiyama 149 Chapter 4— SUMO interacting motifs in DAXX et al., 2008). Thus, despite their interpretation, ionic strength seemed to play a significant role in the interaction affinities. As summarized in the previous section, NMR-monitored titrations allowed a more accurate measurement of KD values as a function of ionic strength. In contrast to the previous study, the progressive weakening of the interaction between DAXX SIMs and SUMO- 1 and SUMO-2 at higher salt concentrations (Table 4.1) is consistent with electrostatic forces playing a major role in SUMO/SIM interactions. Table 4.2 - Summary of reported dissociation constants of SIMs/SUMO complexes. (a) SUMO SIM KD Buffer Referenceparalogue (Song et a!.,SUMO-1 PIASX-Pa 7 20 mM Tris, pH 7.6, 30 C 2004) (Song et a!.,SUMO-1 PIASX-N 9 20 mM Tris, pH 7.6, 30 °C 2004) ADDINSUMO-3 PIASX-N 5 20 mM Tris, pH 7.6, 30 C 2004) SUMO-1 PIASX-Pb 6 20 mM phosphate, pH 7.4, 30 °C (Song et a!.,2005) (Sekiyarna etSUMO-1 MCAFI 14 20 mM phosphate, pH 6.8, 25 °C al., 2008) SUMO-1 MCAF1 12(c) 20 mM phosphate, pH 6.8, 25 °C (Sekiyama et150 mMNaC1 a!., 2008) SUMO-1 MCAF1 8(c) 20 mM phosphate, pH 6.8, 25 °C (Sekiyama et300 mM NaCI al., 2008) SUMO-3 MCAF1 1 20 mM phosphate, pH 6.8, 25 °C (Sekiyama et a!., 2008) SUMO-3 MCAF1 1(c) 20 mM phosphate, pH 6.8, 25 °C (Sekiyarna et150 mMNaC1 al., 2008) SUMO-3 MCAF1 N.D. d 20 mM phosphate, pH 6.8, 25 °C (Sekiyama et300 mM NaC1 al., 2008) (a) All reported dissociation constants were measured with ITC. (b) PIASX-Pa: KVDVIDLTIESSSDEEEEDPPAKR; PIASX-N: VDVIDLTIE; PIASX-Pb: KVDVIDLTIE MCAF-1: GVIDLTMDDEE(c) Noisy binding curve. (d) Not determined due to no significant heat release. 150 Chapter 4— SUMO interacting motifs in DAXX 4.4 Two orientation binding of DAXX SIM-C to SUMO-1 To date, five structures of SIMs with SUMO-l have been reported (Baba et al., 2005; Duda et a!., 2007; Reverter and Lima, 2005; Sekiyama et al., 2008; Song et a!., 2005). In all cases, the SIM adopts a n-strand conformation, pairing with strand 2 of SUMO-1. However, this pairing can either be in a parallel or anti-parallel orientation (Reverter and Lima, 2006; Song et al., 2005) (Figure 4.5A). It has been suggested that the juxtaposition of charged groups relative to the hydrophobic core is responsible for determining the binding mode of a given SIM (Kerscher, 2007). However, SIM-C contains acidic residues on both sides of a rather symmetrical hydrophobic core (Figure 4.1 C), thus precluding an obvious prediction of its binding mode with SUMO-1. One way of answering this question is to measure intermolecular NOEs between SIM-C complexed with SUMO-1, as these would be very different in each binding mode. However, upon titration with unlabeled SIM-C”, several amide signals in the‘5N-HSQC spectrum of ‘5N- labeled SUMO-1’ surprisingly broadened beyond detection upon reaching saturation (Figures 4.2C,4.2F). The same behavior occurred during the reciprocal titration of‘5N-labeled SIM-C” with unlabeled SUMO-1’’. This of course precluded the measurement of the desired intermolecular NOEs. 151 Chapter 4— SUMO interacting motifs in DAXX Figure 4.5 - SIM-C bound SUMO-1 in both parallel and anti-parallel orientations. (A) SUMO-l is known to bind SIMs either in a parallel (left, SUMO-1:PIASx, 2ASQ.pdb) or an anti-parallel orientation (right, SUMO-1 :RanBP2,1Z5S.pdb). It has been proposed that the charged residues juxtaposed with the hydrophobic core of SIMs are involved in dictating the orientation. (B) Three unlabeled peptides with a cysteine residue at a different position were TEMPO nitroxide spin labeled. (C) PREs are measured as the spectral differences for a sample in the presence of the paramagnetic nitroxide and the reduced, diamagnetic hydroxylamine. (B) ‘H-’5N HSQC spectrum of SUMO-1 saturated with TEMPOSIMC74OC(left). Upon reduction of the spin label, peaks that had been broadened beyond detection re-appeared (right, red boxes). (E) Residues that showed the greatest PRE change in 1HN transverse relaxation rate, AR2 (> 30 s ‘,blue) or that re-appear (AR2> 100 s1, red) upon reduction of the spin label from TEMPO-SIM C, TEMPOSIMCs73scand TEMPOSIMC74ocare located on both sides of the SUMO-1 binding surface. (F,G) AR2 for TEMPOSIMCs73sc, from which triangulation of the PRE derived distances between the spin label (balls) and individual amides (G1u33, G1n55, Met59, green; Val38, Leu44, Leu47, red) show that the peptide bound SUMO-l in two possible orientations. Similar results were obtained with the other two spin-labeled SIM-C peptides (see Appendix 10 for details on the calculations). 152 Chapter 4— SUMO interacting motifs in DAXX WT (C727) GSIYKTSVATQCDPEEIIVLSDSD 740C GSIYKTSVATQADPEEIIVLSDSDC S738C GSIYKTSVATQADPEEIIVLSDCD A + PIASx + RANBP2 B C E/ 121 ((p) - 3/ D 122 123 124 8.7 NH NH -r > II&--CH.- S—-S-- H ‘ ) ir ‘ 8.6 8.58.4 78 - 5 ‘H (ppm) ‘H (ppm) +TEMPO- +TEMPO- +TEMPO SIM-C SIMC74oc S1MC738c 80 60 40 20 G F I I ii i I I 1,1 20 40 60 Residue •N(). 80 153 Chapter 4— SUMO interacting motifs in DAXX It was hypothesized that msec-Ilsec timescale conformational exchange between 2 or more binding modes within the SIM-C/SUMO-1 complex led to the observed spectral behavior. To confirm this hypothesis, paramagnetic relaxation enhancement (PRE) measurements using a nitroxide spin label placed on one end or the other of the SIM-C peptide were undertaken. PRE originates from a dipolar interactions between an unpaired electron of a paramagnetic center (spin label) and the nucleus of an observed ‘H (Su and Otting, 2009). Similar to the NOE, a PRE is proportional to the inverse sixth power of the distance between the spin label and detected nucleus (r6). However, due to the large magnetic moment of the unpaired electron, the effect is measureable over distances much larger than possible for a NOE (up to 20 A vs. -5 A) (Clore et al., 2007). A wild-type cysteine residue (Cys727) was used to place a TEMPO nitroxide spin label N-terminal to SIM-C. This residue was also mutated to an Ala and added a new single cysteine at the C-terminus of SIM-C at either position 738 or 740 to place the spin label on the other side of SIM-C (Figure 4.5B). PREs were measured qualitatively and quantitatively via‘5N-HSQC spectra of‘5N-labeled SUMO- 1 titrated separately with the three SIM-C derivatives in their paramagnetic nitroxide forms and then subsequently reduced to their diamagnetic hydroxylamine forms (Figure 4.5 C). As expected, upon reaching saturation, a number of peaks in the spectra of SUMO-1 were broadened (often beyond detection) due to the bound nitroxide, yet reappeared upon addition of reducing agent (Figure 4.5D). Qualitatively, SUMO- 1 amides experiencing the largest intensity changes were closest to the SIM-C-binding site. However, recall that several amides are undetectable in the unmodified SIM-C/SUMO-1 complex, thus complicating the analysis of these spectra. 154 Chapter 4— SUMO interacting motifs in DAXX To extract distance restraints necessary for positioning the spin labeled SIM-C relative to SUMO- 1, the differences in transverse relaxation rates, z\R2, of corresponding amide 11N nuclei in the presence of the label in its paramagnetic and diamagnetic states were also measured (Figure 4.5 F). Based on these PRE R2 values (and calculated distances), it is clear that SIM-C must be bound in both parallel and anti-parallel modes in order to account for strong PRE effects on both ends of the SIM-binding site of SUMO-1 (Figure 4.5E,G). This result strongly suggests that the broadening observed in the titration is due to conformational exchange between two detectable binding modes that inter-converted on the chemical shift time scale (Appendix 6). This agrees with the model in which the juxtaposing acidic residues determine the binding mode of the peptide. When they are asymmetrically placed with respect to the rather symmetric hydrophobic core of the DAXX SIMs, the peptide binds in one mode (see for example SIM-N, next section). When they are symmetrically placed, the peptide can adopt two bound orientations. 4.5 Determination of the binding mode of DAXX SIM-N with SUMO-1 In contrast to SIM-C’’, SIM-N” bound SUMO-1’’ and SUMO-2” with higher affinity and without exchange broadening effects under saturating conditions. It was thus hypothesized that, consistent with its asymmetric charge distribution, SIM-N” binds SUMO in only one preferential orientation. To verify this hypothesis, ‘3C/’5N filtered-edited NOESY-HSQC experiments were used to measure intermolecular NOEs between unlabeled SIM-N’’ and labeled SUMO-1 and unlabeled SIM-N’’ (Table 4.3). 155 Chapter 4— SUMO interacting motifs in DAXX Table 4.3 - Intermolecular NOEs between unlabeled SIM-N and 13C/’5Nlabeled SUMO-1. 13C/5N SUMO-1 unlabeled SIM-N 134 18 Ha, H, H; V9 H H35 HN 18 Ha,H2 1135 Ho’, H2 18 H, fl72 F36 H1,H2 LlO H62 F36 (or 35) aromatic LlO H, H2; V9 W”2 8 H K37 RN LlO H, H62 K37 H1,H2 V9 W”29; LlO 1162 K37 LiO H62 V38 116(10r2) LiO 1161, 1162 V39 H Dii H2 V39 H2 LiO 1131, H32 T42 LiO H82;DllH H43 H” Dii a 112 L47 HN LlO H61,H82 L47 H LiO 61 H82 L47 H61 LlO 1162 S50 HN LiO He’, H82 R54 H1 18 fl61 R54 H2 17 or 18 I-F2 R54 H61 18 H61 The intermolecular NOEs on the structure of SUMO-1 and the sequence of SIM-N were then mapped (Figure 4.6A). 11e8 gave NOE interactions with Arg54 and 11e34, His35 and Phe36, whereas Aspi 1 gave NOEs to Thr42, His43 and Lys39. This clearly demonstrates that the SIM N” peptide binds in a parallel conformation, i.e. similar to that observed with PIASx (Figure 4.5A, left). This situates the negative charge of the acidic residues on the positive side of SUMO 1 (Figure 4.6B). 156 Chapter 4— SUMO interacting motifs in DAXX Figure 4.6 - SIM-N bound SUMO-1 in a parallel orientation. (A) Six representative ‘H-’H strips from a 3D‘3C/’5N filtered-edited NOESY-HSQC spectrum showing intermolecular NOEs between unlabeled SIM-N and three residues from strand 2 and three from helix 1 in‘3C/’5N- labeled SUMO- 1. (B) Mapping of intermolecular NOEs to the structure of SUMO- 1. Strand 2 and the helix are shown in a cartoon of SUMO- 1 in purple (left) and as isolated elements, with cyan highlighting the residues that receive intermolecular NOEs from SIM-N. Each line represents the presence of at least one intermolecular NOE. Positively-charged and negatively charged residues are shown in blue (left) and red (right), respectively. (C) NMR-monitored titration of SUMO-l with SIM-N. Arrows show peaks shifting as the concentration of SIM NGS increases. (D) Comparison of chemical shift perturbation (calculated as,J(M,i)2 +(O.2AöN) ) from SUMO-l residues upon addition of SIM-N (top) and SIMNGS (bottom). (E) Mapping of residues of SUMO- 1 showing the greatest chemical shift perturbation (<0.1 2ppm) upon binding SIM-N. 157 Chapter 4— SUMO interacting motifs in DAXX A B iS H611 I8Hy2 LIOH62 0 IS H311 IS fly2 2 LIOHO2 (.10 H011 DII HI’ L10 1102 ,- LII) 1162 LII) Hol 2 2 DII HO 5, 3 1M 2A 3T 4A 5N 6S 71 . c Dl1H 2. - .7 8.9 7.5 3.1 K39 H2 H43 HN S50 uN R54 110 0 2 2 2 3 3 3 4 4 4 . iSilco L10H -J 0.6 9,1, 134 HyZ 1(37 uN SuM-N” C Q94 K37 124 \ 124 I . V26 •(S, SI2 5M59 114327 ‘H (ppm) E 0.6 0.5 0.4 1,0 ((.2 0.! ((II (15 0.4 J0.3 0.2 0.1 0,0 + SIM-N .1 .fl 20 4!) Residue 158 L,Jt.iih.,1 611 ((0 Chapter 4— SUMO interacting motifs in DAXX If the charge of the juxtaposing acidic residues in SIM-N is solely responsible for establishing the binding mode of SIM-N on SUMO-l, it would be expected that increasing the ionic strength of the solution would weaken the electrostatic contribution of those residues and perhaps lead to multiple binding modes as with SIM-C. Indeed, it was observed that the lHNl5N signals from several amides in the hydrophobic core of SIMNv were broadened beyond detection in the ‘H-’5N HSQC spectra of the SUMO-1 complex recorded in the presence of 200 mM KC1 (11e8, Va19 and LeulO, Figure 4.3E), similar to SIM-C at 100 mM KC1 (section 4.3). Curiously, under these conditions SIM-N’ bound SUMO-1 with a KD similar to that between SUMO-1’’ and SIM-C” at 100 mM KC1 (Table 4.1). Broadening of peaks could also result from being in intermediate exchange between the free and bound states under non-saturating conditions (Appendix 6). However, the titration curves of peaks that moved in fast exchange indicated that saturation was reached. Thus, the broadening is most likely the result of conformational exchange in the complex. Interestingly, SUMO2v in 200 mM KC1 produced broadening on the peaks of SIM-N” as well, albeit not beyond detection (Figure 4.3F), despite having similar kex and Axo (Appendix 6, compare Figure 4.3C with F4.3D). This suggests that the binding mode has not reversed at that ionic strength yet, despite having a very similar KD to SUMO-1” (Table 4.1). This phenomenon was also observed with SIM-C’’, where SUMO-2’’ showed broadened but detectable peaks when SUMO1w did not (Figure 4.4). It appears that the different hydrophobic and charge distribution between the paralogues is enough to favor one detectable orientation on SUMO-2’’ whereas SUMO-1’ accommodates multiple orientations. 159 Chapter 4— SUMO interacting motifs in DAXX Although flanking charged residues dictate the orientation of binding, the closest SIM-N’’ SUMO- I contacts, as detected by intermolecular NOE interactions, involved the hydrophobic residues of the peptide. Accordingly, it was hypothesized that mutation of these residues would completely disrupt binding. To test this, a construct with the hydrophobic core, Ile-Ile-Val-Leu, mutated to Gly-Ser-Gly-Ser (SIM-N) was produced. Surprisingly, the mutant was still able to bind SUMO-l at low ionic strength, although understandably, weakly and in the fast exchange limit (Figure 4.6C). The dissociation constant of the mutant was determined to be 440 ± 30 tiM, compared to the < 5iM for the wild type (Table 4.1). The chemical shift perturbation with SIMNGS was much less than with the wild type (Figure 4.6D), suggesting that SIM-N°5produced a smaller perturbation to the structure of SUMO- 1. Interestingly, residues showing the greatest changes when binding SIMNGS are clustered on the binding cleft (Figure 4.6E), particularly on the part rich in basic residues (Figure 4.6B). This result highlights the contribution of the charged residues of SIM-N in binding SUMO-1. 4.6 Intramolecular binding of SIM-N to the DHB domain With the goal of investigating the behavior of SIM-N in a more “native context”, i.e. in the presence of the DHB domain, NMR spectroscopy was used to characterize DAXX1’44,which contains both SIM-N and the DHB domain. Surprisingly, the ‘H-’5N HSQC spectrum of this species in 100 mM KCI indicated extensive aggregation (Figure 4.7A). This behavior was not alleviated with addition of excess DTT to ensure complete reduction of all four cysteine residues. 160 Chapter 4— SUMO interacting motifs in DAXX 2 z 130 110 I z 120 130 110 I 120 ‘H (ppm) ‘H (ppm) [KCIJ t SIM-N(j ° SIM-NO 0 --—-‘O Figure 4.7 - DAXX’144 self-association was ionic strength dependent. (A) The 1H-5N HSQC spectrum of DAXX’44 shows broad and missing peaks at low salt concentrations. As the KC1 concentration was increased, lHNI5N peaks became sharper and reappeared. (B) At low salt concentration, and the high concentration of DAXX144 (100 1iM) intermolecular binding of multimers was presumably favored, while at high salt concentration, the monomer was favored. A 110 120 OmMKC1 . .# 300 mM KC1 4 . a Co 100 mM KC1 . . o . •0 : ; 400 mM KCI :. • 0.0 O0* 200 mM KC1 0 oe0 . . 0 00 0 0 9 00 0 0 9 . 0 130 500 mM KCI 0 0.0 0 0 0 0 , . 0 : 10 9 8 7 B 10 9 8 7 [-.1 0 -® K 161 Chapter 4— SUMO interacting motifs in DAXX Progressively increasing the sample ionic strength to 500 mM KC1 yielded progressively sharper ‘H-’5N HSQC signals, indicative of monomeric DAXX’’44 (Figure 4.7A). A simple explanation for this behavior is an ionic strength-dependent self-association of DAXX’144 due to intermolecular cross-linking interactions of the added N-terminal residues with the DHB domain (Figure 4.7B). In order to map such an interaction,‘5N-DAXX’6was purified and changes in its 1H-5N HSQC spectrum upon addition of unlabeled DAXX55’44 were monitored. Binding did occur and the residues of‘5N-DAXX’6most affected corresponded to those of SIM-N (Figure 4.8A). Since at pH 6.5 SIM-N is negatively-charged, and DAXX55’44 net positive, this is consistent with an electrostatically-driven interaction. This interaction was confirmed through the reverse titration of‘5N-DAXX5’44 with DAXX’56 (Figure 4.8B), and with a titration of‘5N-DAXX5’44with SIMN? (not shown). Following the peaks in the fast exchange regime (100 mM KC1), the dissociation constant was determined to be 170 ± 60 j.iM (Figure 4.8C). However, in the context of the full-length protein this value will likely be lower as the linker between SIM-N and the DHB domain might improve binding. Mapping residues showing the largest chemical shift perturbations due to DAXX’56 binding onto the structure of DAXX55’44 showed that, apart from the C-terminus (His 137, Ala 139, Lys 140, and Lysl4l), the loop between helix 1 and helix 2 (Thr78, Ala79, Asp8O), and helices 2 and 5 (Phe87, Leu88, Gln92, Tyr124, Asn128, Glu129, Thr132, A1a136) were most affected by the interaction (Figure 4.8D-E). Clearly this region is very positively charged (Figure 4.8F), 162 Chapter 4— SUMO interacting motifs in DAXX Figure 4.8 - DAXX’56 interacted with DAXX55144 through SIM-N. (A) Superimposition of the 1H-5N HSQC spectra of DAXX’56 in the absence (150 !IM in 500 iL, red) and presence of unlabeled DAXX55’44 (added 250 iL of 1.4 mM stock, black). Residues showing the most significant changes correspond to those of SIM-N (labeled). (B) Titration of‘5N-DAXX5’44 (140 .iM in 500 iL) with unlabeled DAXX’56 confirmed weak binding (added 210 iL of 2.2 mM stock). Arrows indicate peaks undergoing changes upon addition of DAXX’56. (C) A representative peak, K 141, used for the determination of the dissociation constant for the resulting complex. The dissociation constant (170 ± 60 tiM) was calculated using chemical shift changes in five amides. (D) Chemical shift perturbations of DAXX55’44, calculated as /5)2 +(O.2&55N)2, upon addition of excess DAXX’56.(E) Mapping of residues involved in binding DAXX’56 on the structure of the DHB domain. Residues showing the greatest changes (>0.045 ppm) are shown in red. (F) Proposed intramolecular binding orientation. The hydrophobic residues of SIM-N face the hydrophobic residues of DAXX55’44 (green) and the negatively-charged residues of SIM-N (red) face the positively charged residues of DAXX55l (blue). 163 Chapter 4— SUMO interacting motifs in DAXX B 0 0 I) e 8 • G f2 1 t 0 — 8 0 0 — V - 0 0A 0 0 -D13 1)12 0 1)14 1)16 D11/E13 17 - 3.20 E17 - -- Al9 F • Oc AI8’1 o 18 L10 3 110 115 Z 120 125 122.8 123.0 123.2 123.4 123.6 123.8 C 9.0 8.5 8.0 7.5 7.0 ‘H (ppm) 110 115 z 120 125 0.14 0.12 11.1 0.08 0.06 0.04 002 8.6 8.4 8.2 8.0 7.8 8.15 8.11) 8.05 II (ppm) ‘H (ppm) D K,,=i70605M 0.12 — 004 50 100 150 204) Volume(1L) E A79 T132 _.a— AIi6..15KI4() 55 144 F Iii id... 56 61 66 71 76 79 85 90 93 % 11 106 109 114 119 125 128 133 13.) 1)3 Residue 19 164 Chapter 4— SUMO interacting motifs in DAXX confinning the previous result that the interaction is mainly electrostatic. As in the RassflC2338” titration (section 3.3.3), Lys6O was affected but displayed a dissociation constant two orders of magnitude larger than the other residues (KD = 13 mM), suggesting that its chemical shift change was independent from the binding to SIM-N. One interesting observation is that helices 2 and 5 were also involved in binding RassflC (Chapter 3). However, despite being of a similar size as RassflC2338”, the residues affected by SIM-N binding are less clustered than with RassflC2338”(compare Figure 4.8 with Figure 3.6). This could be explained if SIM-N did not form a helix, but instead it bound in an extended conformation. It is hypothesized that the negatively-charged residues at the C-terminus of SIM-N (residues 11-17) faced the positively-charged face of the DHB domain similar to RassflC2338’, but the hydrophobic residues at its N-terminus (residues 2-10) had contacts with the hydrophobic cavity of Phe87/Ty 124 (filled by RassflC2338”), as well as with the hydrophobic cluster between helices 1 and 2 (Thr78/Ala79; Figure 4.8F). Lastly, SIM-N binding to the DHB domain produced significant broadening in the signals, whereas RassflC2338’’ binding did not produce such broadening effect (compare Figure 4.8B with Figure 3.4B). This suggests that the dynamics of the two complexes were significantly different. It is possible that the helix formation, undergone by RassflC2338”upon binding the DHB domain to fill the well defined hydrophobic cavity of Va184, Phe87, Tyr124 and 11e127, prevented big conformational changes, whereas an extended conformation of SIM-N was free to undergo conformational exchange. A full relaxation study of the two complexes could provide some clues on the viability of these proposed mechanisms. 165 Chapter 4— SUMO interacting motifs in DAXX The use of the same binding interface of the DHB domain with SIM-N and RassflC2338’ leads to a tantalizing model where a competing intramolecular SIM-N and DHB interaction mediates the intermolecular interaction of DAXX with RassflC and SUMO-l/2 (Chapter 5). This model will be tested through further NMR-monitored titration experiments and mutational studies. 4.7 Characterization of the interaction between sumoylated Ets-1 and DAXX SIM-C A prevalent function of DAXX is to repress transcription factors. One such factor is Ets-1, which contains a DNA-binding domain at its C-terminus (residues 331-415), and a helical bundle PNT domain (residues 42-135) that mediates MAPK docking and binding to the general transcriptional co-activator CBP/p300. At its N-terminus, Ets-1 contains an unstructured region (residues 1-42) where it can be regulated via sumoylation and/or phosphorylation (Lee et a!., 2005; Lee et a!., 2008; Macauley et a!., 2006; Mackereth et a!., 2004). Ets- 1 was shown through a yeast two-hybrid screen and a luciferase reporter assay to interact and be repressed by DAXX, respectively (Li et a!., 2000b). Li et a!. identified residues 567-740 of DAXX to be responsible for the interaction with residues 1-139 of Ets-1, which they named the “DAXX interaction domain (DID)”. However, attempts to reproduce that interaction in vitro were unsuccessful, as the 1H-’5N HSQC spectrum of‘5N-Ets’138 did not show any perturbations upon addition of excess unlabeled mDAXX566739 (Figure 4.9D). Recently, our group and others showed that Ets- 1 is sumoylated at Lys 15 and Lys227 (Ji et al., 2007; Macauley et al., 2006; Nishida et al., 2006). Furthermore, our group also demonstrated that Lysl5-sumoylated Ets-1 can be described as “beads-on-a-string” with the Ets-l and SUMO 166 Chapter 4— SUMO interacting motifs in DAXX 1 components behaving independently and simply linked via a flexible isopeptide bond (Macauley et al., 2006). Since the DAXX-Ets interaction was identified in a two-hybrid screen in yeast, where sumoylation is known to occur (Panse et al., 2004), it was hypothesized that the interaction reported by Li et al. was actually due to DAXX binding via SIM-C to a SUMO covalently linked to Ets- 1. In the simplest model, the interaction between SUMO-1 and mDAXX566739 would be independent of the presence of Ets’’38. As discussed in Chapter 2, mDAXX566739 also known as the SPT-domain, was intrinsically unstructured, and binding to Ets-1 would likely occur without any cooperative effects involving all three species (SPT-domain, SIM-C and Etsi 1-139, as in Figure 1 .4A). To corroborate that this is indeed the case, it was first checked whether SUMO 1 binds mDAXX567739 only through its SIM-C. Indeed, the ‘H-15N HSQC spectrum of SUMO-1 showed identical changes when titrated with mDAXX566739 or SIM-C (Figure 4.9A-B), yet no changes when titrated with mDAXX566727 (which lacks SIM-C) (Figure 4.9C). Then, the effect of mDAXX566739 on sumoylated Ets’’38 was checked. The latter was prepared from unlabeled SUMO-1 and‘5N-labeled-Ets38using an in vitro sumoylation reaction (Macauley et al., 2004). Addition of mDAXX566739 did not cause any spectral changes, showing that the binding of its SIM-C to SUMO-1 does not perturb Ets’’38 (Figure 4.9D). These results therefore demonstrate that there is no cooperative binding between the PNT domain and the SPT region of DAXX (Figure 4.9E). 167 Chapter 4— SUMO interacting motifs in DAXX Figure 4.9 - Sumoylated Ets-1 bound SIM-C of mDAXX567739 via the covalently attached SUMO. Superimposition of the ‘H-’5N HSQC spectra of‘5N-labeled SUMO-1 in the absence (red) and presence of unlabeled (A) mDAXX567739,(B) DAXX 729-740, or (C) DAXX 567-727 (black). Chemical shift perturbations demonstrate that SUMO-1 bound only to DAXX truncation fragments containing SIM-C (residues 728-740). (D) Superimposition of the ‘H-’5N HSQC spectra of sumoylated 15N-Ets-1 1-138 in the absence (red) and presence of unlabeled mDAXX567 (black). The lack of any detectable chemical shift perturbations in the amide signals of Ets- 1’ 138 demonstrates that the Ets-1 and DAXX fragments did not bind with any appreciable affinity. Note that the SUMO moiety of this covalently-linked species was unlabeled. (E) Model for the binding of mDAXX567739 to sumoylated Ets38 showing that there is no interaction between the PNT domain containing fragment of Ets-1 and the SPT region of DAXX. (F) Current model of Ets-l repression by DAXX. Sumoylation mediates the interaction of Ets-1 with DAXX. This leads to repressed transcription, possibly through the recruitment of histone deacetylases (HDACs), which cause chromatin remodeling. 168 Chapter 4— SUMO interacting motifs in DAXX 120 122 124 z 126 128 I29 . ..I4 37 1144 ‘7’ 0 , 120 122 124 z 126 128 120 122 124 z 126 128 073/029 V3 Ci:C 0 091 L1/3 I36 R63 1•5I 1 K37o1 L44 V26 E89 l43 I/IS 0 127 E z 110 115 120 125 9.6 9.4 9.2 9.0 8.8 ‘H (ppm) D a • G.0 0 0 * 0 0 0 0 o 0 * o a 0* a ___ a 0 a, 0. 9.0 8.0 7.0 ‘H (ppm)9.6 9.4 9.2 9.0 8.8 ‘H (ppm) E F SPT 740 HDACS Repression 169 Chapter 4— SUMO interacting motifs in DAXX 4.8 Summary and significance In this chapter, two SIMs in DAXX, which are able to bind SUMO- 1 and SUMO-2 in vitro, have been identified. It was found that there is no significant SUMO paralogue specificity for either SIMs. However, SIM-N binds both SUMO-1 and SUMO-2 with higher affinity than SIM C. Also, SIM-N binds SUMO-l in one detectable orientation whereas SIM-C binds SUMO-1 in multiple modes. Additionally, SIM-N is the first SIM demonstrated to bind a domain other than SUMO. It was discovered that SIM-N also bound the DHB domain of DAXX, although with ‘-20 fold lower affinity than SUMO under similar conditions. Even more interestingly is the observation that the interface used by the DHB domain to bind SIM-N was the same as the one utilized to bind Rassfl C (Chapter 3). This opens a new model to explain how DAXX is regulated. It is possible that SUMO-1 or -2 relieve the intramolecular binding between SIM-N and the DHB domain so that it can interact with other partners such as Rassfl C. Such a model is discussed in Chapter 5. The data suggest that SUMO- 1 functions as an independent tag on Lys 15 of Ets- 1, and by inference other transcription factors, recruiting DAXX through SIM-C (and likely SIM-N). Such recruitment of DAXX can then bring HDACs (Hollenbach et al., 2002), which have been shown to associate with DAXX, and which in turn modify the tails of histones producing chromatin remodeling and gene repression (Kuo et al., 2005) (Figure 4.9F). It is clear that the data presented in this chapter bring up many new questions. In Chapter 5 some of them and future experiments are discussed. 170 Chapter 4— SUMO interacting motifs in DAXX 4.9 Materials and methods 4.9.1 Constructs Genes encoding residues 1—97 of human SUMO-1 (NP_003343), residues 1-93 of human SUMO-2 (P61956), residues 1-138 of murine Ets- 1 (AAN3 8317), full-length human UBC-9 (P63279), residues 567-740 and 567-727 of murine DAXX (NP 03 1855) were cloned via PCR into the pET28a expression vector (Novagen) using NdeI and XhoI restriction enzyme sites. The resulting constructs contained an N-terminal His6-tag with a thrombin cleavage site such that a Gly-Ser-His extension remained after proteolytic processing. New constructs were created with a Trp residue between the Gly-Ser-His and Met-i of SUMO-1 (SUMO-1’’) and SUMO-2 (SUMO 2j to allow quantification via UV absorbance. Genes encoding residues 1-19 (SIM-N) of human DAXX (CAG33366), residues 1-19 with 7IIVL’° mutated to 7GSGS’° (SIM-N GSGS), residues 718-739 of murine DAXX (SIM-C or SIM-C WT), residues 718-739 of murine DAXX with C728A and S738C mutations (SIM C5738c), residues 718-739 of murine DAXX with C728A mutation and an additional cysteine residue at the C-terminus (SIMC74OC) were cloned via PCR into pGEX-2T expression vector (GE Healthcare) using BamHI and EcoRI restriction enzyme sites. A Trp residue was added at the C-terminus of DAXX’’9 and at the N-terminus of DAXX718739.The resulting constructs contained an N-terminal Glutathione-S-Transferase (GST) at the N-terminus with a thrombin cleavage site such that a Gly-Ser extension remained after proteolytic processing. All clones were verified by DNA sequencing. DAXX55’44 was cloned and purified as described in Chapter 2. The clone of the yeast El activating enzyme (AOS1/UBA1) was generated as described in (Bencsath et al., 2002). A 171 Chapter 4— SUMO interacting motifs in DAXX peptide containing residues 729-740 of human DAXX was synthesized and HPLC purified by EZBiolab (Westfield, USA) and its purity was verified by mass spectrometry and NMR spectroscopy. 4.9.2 Protein expression Freshly transformed E. coil BL21 (2DE3) cells were grown overnight in 25 mL of Luria Broth, collected by centrifugation, and transferred to 1 L of minimal M9 medium enriched with 1 g of (99%)-15NH4C1(Spectral Stable Isotopes) as the sole nitrogen source for uniform ‘5N labeling, or to 1 L of Luria Broth (for unlabeled proteins). When cells reached an 0D600 0.6, protein expression was induced with 1 mM IPTG. Cells were then grown overnight (for labeled proteins) or for 5 hours (for unlabeled proteins) at 30 °C, harvested by centrifugation, and frozen at -70 °C prior to further purification. 4.9.3 Protein purification Thawed E. coil pellets were resuspended in binding buffer (5 mM imidazole, 50 mM HEPES, 500 mM NaCl, 5% (v/v) glycerol, pH 7.5) with a protease inhibitor cocktail tablet (Roche), and lysed by passage through a French press (1000 psi) followed by sonication (duty cycle 50%, 10 mm). After centrifugation at 15,000 rpm in a Sorval SS32 rotor for 30 mm, the supernatant was passed through a 0.8 pm filter (Millex). For pET28a-encoded proteins, the lysate was applied to a 5 mL His-Trap metal affinity column (Amersham Biosciences) connected to an AKTA chromatography system (GE Healthcare). The column was washed with 200 mL of washing buffer (binding buffer plus 60 mM imidazole), and the His6-tagged proteins eluted with a gradient to 500 mM imidazole. Fractions containing the protein, as detected by absorption at 280 nm, were pooled and buffer 172 Chapter 4— SUMO interacting motifs in DAXX exchanged with a 3K MWCO Amicon concentrator into a thrombin cleavage buffer (20 mM Tris-HC1, 150 mM NaC1, 2.5 mM CaC12,pH 8.4). The His6-tag was removed by incubation with thrombin ( 1.4 units, Novagen) at room temperature for -5 hours. Proteolysis was monitored by MALDI-ToF-MS. The reaction was terminated by incubation with 200 1iL p-aminobenzamidine beads (Sigma) for 1 hour. To further purify the sample and remove the His6-tag, the supernatant was diluted 10 fold in buffer without salt, applied to an ion exchange column (HiPrep Q FF 16/10), and eluted with a salt gradient (20 mM Tris, pH 8.0, 2 mM DTT, 0 — 1 M NaC1). Fractions absorbing at 280 nm were evaluated by SDS-PAGE and those containing the sample were concentrated using a 3K MWCO Amicon centricon (Millipore). In each case, the protein purity was judged to be >95% by SDS-PAGE, and the protein identity confirmed by MALDI ToF-MS. For GST-tagged peptides, the lysate was applied to four 5 mL GSTrap HP affinity columns (GE Healthcare) connected to an AKTA chromatography system (GE Healthcare). The lysate was kept on ice-cold water to prevent proteolytic cleavage and slowly loaded to the room temperature column at 0.7 ml/min (GST binding to the column is slow and temperature dependent). Once loaded, the column was washed with 100 mL of PBS/EDTA buffer (137 mM NaC1, 2.7 mM KC1, 4.3 mM Na2HPO4,1.4 mM KH2PO4,5 mM EDTA, pH 7.3), and the GST tagged peptides were eluted with glutathione buffer (10 mM Tris, 20 mM glutathiorie, 5 mM DTT, pH 8.0), prepared fresh. Fractions containing the GST-tagged peptides, detected by absorption at 280nm, were pooled and buffer exchanged with a 10K MWCO Amicon concentrator into a thrombin cleavage buffer (20 mM Tris-HC1, 150 mM NaC1, 2.5 mM CaC12, pH 8.4). The GST was removed by incubation with thrombin (- 1.4 units, Novagen) at room temperature for —5 hours. Proteolysis was monitored by MALDI-ToF-MS. The reaction was 173 Chapter 4— SUMO interacting motifs in DAXX terminated by incubation with 200 tL p-aminobenzamidine beads (Sigma) for 1 hour. The supernatant was passed through the 10K MWCO Amicon filter, which retained the GST and uncleaved GST-peptide in the concentrate. The cleaved peptides were collected in the flow- through. Concentrates were washed with 40 mL of thrombin cleavage buffer. The flow-through was injected into a C18 HPLC column, washed thoroughly with 0.1% TFA water, and eluted with anH20:acetonitrile gradient (0.1% TFA). Fractions absorbing at 280nm were evaluated by MALDI-MS, and those containing the desired peptide were lyophilized. Protein concentrations were determined by UV light absorption spectroscopy using predicted 278 values (Gasteiger et al., 2005) of 9970 M’cm’ for SUMO-1, 6990 M1cm’ for SUMO-2 and DAXX718739, 5500 M’cm’ for DAXX’’9,23000 M’cm’ for Etsl-138, 29700 M’cm’ for UBC9, and 67500 M’cm’ for Aosl/Uba2. 4.9.4 NMR spectral assignments NMR data were collected using a 600 MHz Varian Inova NMR spectrometer and processed using NMRpipe (Delaglio et al., 1995) and analyzed by SPARKY (Goddard and Kneller, 2004). 1H’ and ‘5N resonance assignments for the amide groups of SIM-N and SIM-C were determined from HNCACB and CBCA(CO)NH experiments of‘3C/’5N-labeled samples. 4.9.5 NMR titrations All NMR-monitored titrations were carried out in NMR sample buffer (5% D20, 10 mM K2HPO4buffer, 0.1 mM EDTA, 10 mM DTT, pH 6.5 buffer with 100 mM KC1 unless otherwise stated). SUMO-1 and SUMO-2 were exchanged into this buffer by concentrating/diluting 1:10 with a 3K MWCO Amicon Centricon at least 3 times. Lyophilized peptides were dissolved in NMR buffer, with small amounts of NaOH added to re-adjust pH back to 6.5. 174 Chapter 4— SUMO interacting motifs in DAXX Sensitively-enhanced1H-’5N HSQC spectra were recorded at 25 °C using an Inova 600 MHz spectrometer. 500pL samples of the‘5N-labeled species were titrated with increasing amounts of the unlabeled species. Table 4.4 summarizes the sample conditions used for all the titrations presented in this chapter. Table 4.4 - Sample conditions used for titrations presented in Chapter 3. Final point Labeled- Concentration Concentration volume (j.L); ratio ‘5NUnlabeled species (m1 species :unlabeledspecies (jiM) species SUMO1lv SIM-N130 3 32;l:l.5OmMKCI OrnMKC1 SUMO2v SIM-N100 3 27;l:l.6OmMKC1 OmMKCI SIM-N SUMO-l’’120 2.5 130;l:5.5100mMKC1 IOOmMKCI SIM-N SUMO-2’’120 2.5 130;1:5.5100mMKC1 100mMKC1 SIM-N SUMO1#90 2.3 112;l:5.8200 mMKC1 200 mMKC1 SIM-N SUMO-2”140 8.7 120;1:15.5200 mM KCI 200 mM KC1 SIM-C SUMO-1’’110 1.7 142;l:4.4OmMKCI OmMKC1 SIM-C SUMO-2’’110 1.7 l42;l:4.4OmMKCI OmMKCI SIM-C SUMO1#130 2.2 220;l:7.8100 mMKCI 100 mMKCI SIM-C SUMO2v130 1.7 250;l:6.8100 mMKCI 100 mMKC1 SIM-C SUMO-1’’90 2.3 240;l:12200 mMKCI 200 mMKC1 SIM-C SUMO-2”130 8.7 180;1:24200 mMKC1 200 mMKC1 175 Chapter 4— SUMO interacting motifs in DAXX 4.9.6 Dissociation constants Dissociation constants (KD) for the complexes with 1HN and ‘5N signals in the fast exchange regime were obtained from least-squares fitting of chemical shift changes as a function of total ligand concentration (Johnson et al., 1996) using Sigma Plot (see Appendix 4 for details). 4.9.7 Spin labeling To ensure full reduction of all Cys residues, 20 mM DTT was maintained through the purification of the SIM-C, SIMC738C and SIMC74OC until HPLC purification. Peptides were HPLC purified, lyophilized, re-suspended in 50 mM phosphate buffer, pH 7.4; and subsequently reacted for 24 hours in the dark at room temperature with MTSL (Toronto Research Chemicals), previously dissolved in acetone, at a peptide:MTSL ratio of at least 1:5. MALDI-ToF-MS showed the signal for the reaction to be -90% complete. Peptides were re-purified by HPLC to remove any unreacted peptide and side products. Spin-labeled peptides are denoted as TEMPO SIM-C, TEMPOSIMC738cand TEMPOSIMC74oc,respectively. 4.9.8 Paramagnetic relaxation enhancement measurements 13C/’5N-labeled SUMO-l was dialyzed into non-reducing NMR buffer with 100 mM KC1, and the TEMPO-modified peptide was titrated in to saturation (determined from previous titrations, as the spin label absorbs at 280nm, thus precluding accurate concentration measurement by UV-absorbance). A series of relaxation experiments were carried out to determine the R2 of the amide IHN (Donaldson et al., 2001). Then, a mixture ascorbic acid (in NMR sample buffer, 2 mM final), and 10 mM DTT for SIMC74oc, were added to reduce the spin label and the same measurements repeated. 176 Chapter 4— SUMO interacting motifs in DAXX The reported average global correlation time of 8.6 ns for SUMO-1) (Macauley et al., 2004) was used to calculate the effective distance (reff) from the amide proton to the paramagnetic center according to the following equation (Johnson et al., 1999): 6 K( 3r. ‘ 1+a)i K is 1.23 x 1 032 cm6s for a nitroxide radical, r is the distance between the electron and the proton, and 0H is the Larmor frequency of the proton (3.8 x 1 0 rad s’). The calculated distance was used to triangulate the location of the spin labels using three residues on each side of SUMO- 1 by solving three equations of a sphere (with the amide proton at the center and the radius equal to the effective distance) with the program Maple. For example, the position of the 1H” of Va138, Leu44, and Leu47 (Figure 4.5G, red) in the PDB file (model 1, 1 A5R.pdb), and their effective distances to the nitroxide radical were determined as: X z ref f LEU 44 81.420 —5.838 —0.168 13.3 A LEU 47 83.481 —2.193 —4.234 13.1 A VAL 38 79.324 4.386 —5.251 15.4 A Appendix 1 OA shows a sample sheet from the program Maple solving the three resulting sphere equations, which provided two possible solutions, one of which could be discarded as it was located on top of other SUMO atoms (Appendix 1 OB). 177 Chapter 5 Concluding remarks and future experiments In this thesis, the first structural characterization of DAXX has been presented. The modular structure of DAXX, with both folded and intrinsically disordered regions, is consistent with its role as a scaffold protein that mediates transcription repression, apoptosis regulation and mitosis control via specific protein-protein interactions. The structure of the DHB domain (Chapter 2) and the DHB/RassflC2338” complex (Chapter 3) defined how DAXX interacts with flexible DAXX interaction motifs (DIMs). The study of the SIMs in DAXX (Chapter 4) provided important clues to how DAXX binds SUMO and sumoylated transcription factors. Together, these provide a structural foundation for understanding protein partnerships involving DAXX and prompt new ideas on how DAXX could be regulated. This final chapter summarizes the proposed structural configuration of DAXX. Furthermore, a model for the regulation of DAXX is presented based on the observations from previous chapters. Finally, important remaining questions that require further experimentation are discussed. 178 Chapter 5 — Concluding remarks and future experiments 5.1 Structure of DAXX The overall goal of this thesis was to present a structural characterization of the domains of DAXX and their role in mediating protein-protein interactions. Secondary structure prediction algorithms, presented in Chapter 2, suggested the presence of at least two well-defined and mostly c-helica1 structural domains at the N-terminus of DAXX (residues 60-145 and 175-400), joined by an unstructured linker. A long intrinsically disordered C-terminal segment of the protein was also predicted (Figure 2.1). Importantly, these predictions agreed with the sequence conservation of DAXX orthologues found in different vertebrates (Figure 2.2). Figure 5.1 A summarizes the truncation constructs presented in this thesis, and Figure 5.1 B shows the proposed modular organization of DAXX. In this model, the first half of DAXX contains at least two helical domains. The first is the DHB domain (residues 55-144), whose structure and function in mediating protein-protein interactions were presented in Chapters 2 and 3. The second, a putative DAXX helical domain (DHD, residues 175-400), could not be expressed in soluble form in E. coli. Smaller fragments within this region, studied in Chapter 2 (DAXX’61342 and DAXX347420),partially confirmed the existence of folded elements, but further studies are needed to determine their exact structure and function. These experiments could include the production of full-length DAXX using an insect cell expression system (Zhao et al., 2004)), followed by partial proteolytic analysis and mass spectrometry to identify structural domain boundaries. Alternatively, different eukaryotic expression systems, where protein modification and processing patterns are more similar to those in vertebrates, could render this domain soluble (Hunt, 2005; Schmidt, 2004). 179 Chapter 5 — Concluding remarks and future experiments Figure 5.1 - Domain structure of DAXX. (A) Summary of the principal constructs studied in this thesis: SIM-N and SIM-C (red), the DHB domain (green), DAXX’61342 (sand), DAXX34242° (orange), and mDAXX566739 (magenta). A putative DAXX helical domain (DHD) between residues 175-400 is shown in cyan. (B) Model of the structural organization of DAXX. DAXX contains the characterized DHB domain, along with a second putative cs-helical DHD. A model for the second DHD, created by Robetta is shown (with same color scheme as diagram on the top) (Kim et al., 2004). The intrinsically disordered regions, including two SIMs (red with side- chains), are to scale with the folded domains, and shown in an arbitrary extended conformation. In reality, they would sample a much smaller average radius of gyration. (C) List of reported DAXX binding partners, along with coarsely mapped regions of DAXX mediating the interaction. References: (Boellmann et al., 2004; Emelyanov et al., 2002; Gostissa et al., 2004; Hofmann et al., 2002; Hofmann et al., 2003; Hollenbach et al., 1999; Jung et al., 2008; Lalioti et al., 2002; Li et al., 2000a; Li et al., 2007; Li et al., 2000b; Lin et al., 2003; Lin and Shih, 2002; Murakami et al., 2006; Muromoto et a!., 2004a; Muromoto et al., 2004b; Ohiro et al., 2003; Penman et al., 2001; Pluta et al., 1998; Song et a!., 2008; Tang et al., 2006; Tang et al., 2004; Wethkamp and Klempnauer, 2009; Yang et al., 1997; Zhao et al., 2004). 180 AQO 42 0 72 4 74 0 Th I [ D H B D II I)’ 1 17 55 14 4 16 1 24 3 B SJ M -C S I M - N D H B C In te ra ct or D A X X bi nd in g re gi on R ef er en ce In te ra ct or D A X X bi nd in g re gi on R ef er en ce M SP 5S 1- 25 0, 25 0- 50 1 (L in an d Sh ih ,2 00 2) Et s- 1 56 7- 74 0 (L ie t al ., 20 00 b) H IP K 2 1- 18 8 {H ofr na nn et al. ,2 00 3) Fa s 62 6- 74 0 (Y an g e t al. , 19 97 ) p p 7 1 43 -1 97 , 4 39 -5 01 (H ofr na nn et al. ,2 00 2) PM L 67 1- 74 0 (L i e t al .. 20 00 a) D M A P 1 1- 24 0, 24 1- 49 2 (M uro rno to et a L 20 04 a) Tf 3R I 1 48 9- 74 0 (P erl rna n et aL ,2 00 1) TS G IO I 1- 24 0, 24 1- 49 2 (M ur om oto et al ., 20 04 b) Pa x3 63 5- 74 0 (H oll en ba ch et al ., 19 99 ) A TR X 1- 16 0 (T an ge tal ..2 00 4) CE N P- C 63 6- 74 0 (P lu tac taL ,1 99 8) p5 3 1- 50 1 (Z1 ’iao et a! .. 20 04 ) G LU T4 66 1- 74 0 (L ali oti et aL .2 00 2) p5 3 1- 18 8, 18 9- 40 0. 60 0- 74 0 (G os tis sa et a! ., 20 04 ) Pa xS 62 6- 74 0 (E me lya no v et a! ., 20 02 ) LA N A 1- 50 0 (M ura ka mi et aL .2 00 6) p5 3 43 3- 49 3, 62 5- 74 0 (O hir oe t al. ,2 00 3) NI -T El 1- 13 0, 13 1- 40 0 (Ju ng et aL ,2 00 8) G R 50 1- 74 0 (L in et aL .2 00 3) A xi n (L ie t al. , 2 00 7> H SF 1 56 0- 74 0 (B oe llm an n et al ,, 20 04 ) R as sf lA 1- 24 0 (S on ge t a! ., 20 08 ) C/ EB PJ T 19 0- 40 0 (W eth ka rnp an d K le rn pn au er , 2 00 9) H A U SP 1- 16 0, 34 7- 57 0 (T an ge t al ., 20 06 ) M D M 2 15 7- 26 0 (T an g e t al ., 20 06 ) 0 -t I cL I. Chapter 5 — Concluding remarks and future experiments In addition to the linker between the observed and proposed DHB and DHD domains, intrinsically disordered regions of 50 and 340 residues are also predicted at the N- and C- terminal segments of DAXX. Analysis of DAXX1’44 and mDAXX566739 confirmed the intrinsic disorder of these regions. Each of these disordered regions contains a SIM, characterized in Chapter 4. Despite the lack of structural elements, numerous proteins have been reported to interact with the Ser-Pro-Thr rich C-terminal DAXX (Figure 5.1 C). Indeed, intrinsically disordered regions have been shown to serve as “linear peptide recognition motifs” in a number of proteins, and are often the site of post-translational modifications that regulate the function of a protein (Uversky et al., 2008). For example, DAXX contains a nuclear localization signal in this region (Yeung et al., 2008), at least two phosphorylation sites (Ecsedy et al., 2003; Song and Lee, 2003), and multiple sumoylation sites (Jang et al., 2002; Lin et al., 2006), so the flexibility of the C-terminus could play a role in facilitating access to modifying enzymes and interacting partners. At the same time, we also speculate that, as with the case of Ets 1, many of these reported interactions may not be direct but instead may be mediated by an unrecognized intervening SUMO tag covalently attached to the partner protein (see below). 5.2 Significance for future truncation constructs of DAXX Figure 5.1 C summarizes a number of studies that used truncation constructs of DAXX to partially define regions mediating interactions with each partner. In most cases, these constructs are rather coarse and mostly likely arbitrarily defined, e.g. based on convenient restriction sites for cloning, poor knowledge of the domain structure of DAXX, etc. Most investigators have assumed that DAXX contains two PAH domains, as originally proposed by Hollenbach et al. a decade ago based on sequence comparisons with Sin3 (Hollenbach et al. predicted a PAH1 182 Chapter 5 — Concluding remarks and future experiments domain within residues 64-108 and a PAH2 domain within residues 192-240 (Hollenbach et al., 1999)). This original proposal was put forward before any structural characterization of Sin3 PAH domains and the DHB domain was done, and it is now clear that a new way of thinking for the structural domain boundaries of DAXX is needed. For example, the structure of the DHB domain (DAXX55’44)presented in this thesis significantly differs to Sin3 PAH domains both in overall fold and function, as demonstrated in Chapters 2 and 3. Furthermore, the originally proposed DAXX PAH2 domain resides on a putative coiled-coil region, where secondary and tertiary structure prediction algorithms do not support the existence of a four-helix bundle similar to the Sin3 PAH domains or the DHB domain (see Figure 2.2 and 5.1B). Instead, a much longer DHD covering residues 175-400 is proposed here. Although further experiments are needed to characterize this structural region, it is clear that the original hypothesis of the existence of two PAH domains within DAXX was not correct. Additionally, constructs of the C-terminus of DAXX studied in the literature included SIM C, and it is unclear if the reported interaction was direct or indirect via SUMO, as it was demonstrated for Ets-1 (Chapter 4). Indeed, GLUT4, PML, p53, GR, HSF1, CENP-C are reported to be sumoylated (Chung et al., 2004; Giorgino et a!., 2000; Gostissa et a!., 1999; Hong et al., 2001; Sternsdorf et al., 1997; Tian et a!., 2002), so a SUMO-dependent interaction with DAXX is certainly possible. Further experiments with similar DAXX truncation constructs lacking SIM-C are needed to clarify whether their interaction is indeed SUMO-mediated. Furthermore, Figure 5.1C shows that a number of studies reported simultaneous interactions of various proteins with multiple truncation constructs of DAXX. It is likely that the DHD and 183 Chapter 5 — Concluding remarks and future experiments DHB domains interact with different parts of some of these proteins. Indeed, multiple domain interactions could increase the overall affinity and specificity of the binding. Therefore, the structure of the DHB domain might be only part of the puzzle for many interactions, and more studies are needed to determine its role in vivo. These studies include the functional characterization of truncation constructs lacking specifically the putative DHD and/or the DHB domain. Alternatively, now that the one interaction surface of the DHB domain is known, point mutants could be studied. For example, the hydrophobic core (Val84, Phe87, Tyr124, 1le127), or the positively-charged surface (Arg9l, Arg94, Lys 135, Lysl4O) could be mutated. Such mutations could disrupt specific DHB controlled interactions and provide important clues of the scaffolding function of DAXX. Overall, the work presented in this thesis will be the foundation for future structural experiments and at the same time, it will facilitate the design of better constructs in future protein-protein interaction studies of DAXX. 5.3 Proposed mechanism for activation of the DHB domain DAXX5544 bound RassflC2338”via electrostatic and hydrophobic interactions along helices 2 and 5 (Chapter 3, Figure 3.10). Surprisingly, this region was also involved in intramolecular binding to SIM-N (Chapter 4, Figure 4.8). This raises the question of whether the DHB domain and S TM-N also interact within the context of the full-length protein. It is very tempting to hypothesize that an auto-regulatory mechanism for DAXX results from the intramolecular association of the DHB domain and the SIM-N. This could competitively sequester either or both regions of DAXX from their binding partners (i.e. RassflC and SUMO, respectively). 184 Chapter 5 — Concluding remarks and future experiments Figure 5.2 summarizes a possible mechanism for such auto-inhibitory regulation of DAXX. In this “competitive/co-operative coupled equilibria” (CCCE) model, DAXX exists in conformation equilibrium between SIM-N/DHB associated or separated (Figure 5.2A). The former is inactive due to competitive binding of SIM-N to the binding interface of DIMs (like RassflC during mitosis). The latter is free to bind either SUMO (Figure 5.2B), which shifts the equilibrium to DHB-free DAXX and enhances RassflC-binding (Figure 5.2D); or RassflC (Figure 5.2C), which shifts the equilibrium to SIM-N-free DAXX and enhances SUMO-binding (Figure 5.2E). In this model, interaction of DAXX with SUMO enhances RassflC binding and vice versa, making their overall interaction co-operative. Cells could control this mechanism by regulating the available concentration of each protein (by gene expression, localization, turnover, etc.) or by post-translational modifications of either DAXX or RassflC. For example, phosphorylation of Ser23, Ser27 or Ser29 of RassflC would make the DIM more negative and presumably would increase the affinity of Rassfl C towards the positively-charged interaction surface of the DHB domain. Interestingly, a phosphorylation prediction server returned very high scores (>0.97, 1 being a perfect phosphorylatiori site) for Ser23 as a putative phosphorylation site for casein kinase II, DNA-Protein Kinase and ATM kinase, and Ser27 and Ser29 as putative phosphorylation site for casein kinase II (Blom et al., 1999; Blom et al., 2004). Alternatively, post-translational modifications on DAXX could alter its binding affinity as well. For example, very recently Lys122 has been reported to be an ubiquitylation site in DAXX (Fukuyo et al., 2009). Certainly, ubiquitin at Lys 122 would produce steric hindrance for RassflC binding and SUMO binding. 185 Chapter 5 — Concluding remarks and future experiments Figure 5.2 - Proposed auto-regulatory mechanism of the DHB domain. The competitive/co operative coupled equilibria (CCCE) model. In the context of the full-length protein, the DHB domain and SIM-N are in equilibrium between free and bound states (A). Such equilibrium can be shifted by competitively binding either SUMO (B) or a DIM-containing protein, such as RassflC during mitosis (C). SUMO binding would increase the amount of DHB-free DAXX, thereby enhancing RassflC binding (D). Similarly, RassflC binding would increase the amount SIM-N free DAXX, thereby increasing the SUMO binding (E). The folded domains and unstructured regions are to scale. (*) The dissociations constants are those measured with minimum isolated constructs (DAXX55 44 DAXX’’ [SIM-N] and RassflC838, 100 mM KC1). However, in the context of larger constructs they might change due to the linker between the DHB domain and SIM-N. 186 Chapter 5 — Concluding remarks and future experiments * ®1L KD = 170 tM* V DHB 160-740 187 Chapter 5 — Concluding remarks and future experiments Future experiments are needed to study this CCCE model. For example, the KD values for each step need to be measured in the context of larger fragments, since the values on Figure 5.2 are from isolated minimum constructs. In larger constructs, the SIM-N would be covalently attached to the DHB domain, as in the native protein, and could thereby bind it with higher affinity. Future NMR experiments studying the ability of DAXX’’44 to bind SUMO and/or Rassf1C2338’will be carried out to test this model. At the biological level, point mutations could provide clues to the veracity of the CCCE model. Ideally, a mutant disrupting one interaction and not the other would be ideal. For example, the C-terminus of the DHB domain (Lys 140 and Lysl4l) is more disrupted by the intramolecular binding of SIM-N than by Rassfl C binding. Thus, point mutations at these flexible residues are good candidates for disrupting intramolecular binding without disrupting the structure of the DHB domain, and without significantly affecting Rassfl C-binding. With those mutants, the CCCE model would be shifted towards free DHB domain, enhance RassflC binding, and a phenotype change of this mutant during mitosis could be observed. Alternatively, it was demonstrated that mutations at the hydrophobic core of SIM-N decreased drastically its binding to SUMO. However, such mutations might disrupt the electrostatically-driven intramolecular binding of SIM-N differently. If this is indeed the case, such mutant would shift the CCCE model towards SIM-N-bound DHB domain, reducing Rassfl C binding, and producing a different phenotype. To date, no SIM has been reported to display such multiple functions, and it will be exciting to see if the experiments suggested above lead to the discovery of new SUMO-mediated cell cycle mechanisms. 188 Chapter 5 — Concluding remarks and future experiments 5.4 Future experiments in defining DHB domain interacting motifs (DIMs) The work presented in Chapter 3 is the first in defining the physicochemical properties of a DIM. Sequence alignment of RassflC2338’with proteins reported to interact with the N-terminal region of DAXX suggests the presence of a common DIM (Table 3.3). Expression and purification of these putative DIMs is needed to corroborate this hypothesis and answer important questions. For example, do they bind the DHB domain in vitro? What are the dissociation constants? What are the structural bases for binding? Is reverse binding indeed possible? Good candidates for such experiments are Mdm2296313,Hausp828843,Hausp856871,and Stat44348,which show a high degree of conservation with RassflC23.38N. Fascinatingly, a number of proteins involved in transcription, but not yet reported to interact with DAXX, have also been identified through BLAST using RassflC DIM as bait (Table 3.4). Until now, the transcription repression mediated by DAXX has been suggested to occur through the recruitment of HDACs (Figures l.4B and 3.9F). However, the first candidate in Table 3.4 is the transcription initiation factor IIIB (TFIIIB). TFIIIB activates RNA polymerase III, which catalyzes transcription of a variety of short RNA molecules with essential functions in cellular metabolism. TFIIIB activity is carefully regulated during the cell-cycle, and is believed to be activated by phosphorylation by ERK and other kinases, although the detailed mechanisms for its regulation are still unknown (Mauger and Scott, 2004). Protein-protein interaction between DAXX and TFIIIB could act as a repression mechanism as proposed in Figure 1 .2B. Further in vivo and in vitro experiments are needed to investigate whether DAXX and TFIIIB proposed interaction is real and functional. 189 Chapter 5 — Concluding remarks and future experiments Finally, further experiments are needed to investigate if other DIMs exist. A yeast-two hybrid screen using DAXX55’44as bait could provide clues on whether Rassfl C-defined DIM is unique or one among many. 5.5 DAXX SIMs It is interesting that DAXX contains two SIMs, one at its N-terminus and the other at the C- terminus (Chapter 4). The distance between them suggests that they function independently, and not in synchrony as it is the case for RNF4, which binds polySUMO-2/3 chains with adjacent SIMs (Tatham et al., 2008). Indeed, there are significant differences between them. There is overwhelming evidence that SIM-C is active in vivo, and mediates DAXX-dependent repression to sumoylated transcription factors (Lin et al., 2006). Excitingly, a new paradigm for regulating SIM interactions has emerged very recently. Stehmeier et al. have discovered that the casein kinase 2 (CK2), an essential kinase in numerous signaling pathways, can regulate a number of SIMs by phosphorylating serine residues adjacent to the hydrophobic core (Stehmeier and Muller, 2009). SIM-C (and Rassfl C) contains a perfect recognition sequence for CK2(737SDSD40),which has been conserved in all the organisms that contain DAXX (Figure 2.2). Chapter 4 has shown the importance of electrostatic interactions in SUMO/SIM interactions, and it is expected that phosphorylation at either serine would change the ability of DAXX to recognize sumoylated proteins significantly. First of all, SIM-C can bind SUMO in multiple orientations due to the symmetric distribution of negatively-charged residues in SIM-C. Phosphorylation of the serines would likely affect this, inducing a single binding mode like in the SIM-N/SUMO interaction. Second, SIM-C phosphorylation will definitely affect the dissociation 190 Chapter 5 — Concluding remarks and future experiments constant between SUMO and SIM-C, most likely decreasing it. Future NMR-monitored titration experiments of SUMO with phosphorylated versions of SIM-C will help to clarify these hypotheses. On the other hand, the role of SIM-N in mediating transcription repression is unclear, since only DAXX’740, and not DAXX’732,was unable to repress Smad4-, AR-, or CBP-mediated transcription (Lin et a!., 2006). This is despite the fact that its dissociation constants for SUMO-l and SUMO-2 are 3-4 times smaller than those of SIM-C (Table 4.1). The CCCE model provides a mechanism to explain this result, that is, SIM-N is unable to substitute SIM-C since it can only be active under special circumstances due to its intramolecular binding to the DHB domain. Interestingly, its sequence does not contain a CK2 recognition site, but RassflC does. It is plausible that CK2 could regulate activation of SIM-N via activation of Rassfl C (as discussed above). Future experiments investigating the effect of point-mutations on Rassfl C CK2 sites (Ser23, Ser27 and Ser 29) are needed to corroborate this hypothesis. 5.6 Remaining questions about the DAXX/RassflC interaction As mentioned in the introduction, members of the Ras-association domain family are also scaffold proteins. The most characterized member of the Ras-association domain family is Rassfl A, due to its important role as tumor suppressor. Rassfl C shares most of its sequence with RassflA, differing only in its 50 N-terminal amino acids (Figure 1.5). The function of these unique residues is just emerging, and no structural characterization had been done on this region. In light of the results presented in Chapter 3, it is clear that these residues are unstructured, and mediate protein-protein interactions through small recognition sequences such as the DIM. The structure of the DHB/RassflC2338’’ complex suggests that electrostatic and hydrophobic 191 Chapter 5 — Concluding remarks and future experiments interactions are essential in DIMs. Although it is still unclear how these interactions are regulated, conspicuous CK2 phosphorylation sites within the RassflC DIM provide possible avenues for regulation (discussed in section 5.2). Chapter 3 has provided a structural mechanism for the direct binding of RassflC2338 with the DiTB domain of DAXX. Based on the biological significance of this interaction (discussed in section 3.7), there are several hypotheses to be examined. For example, will expression of GFP RassflC2338 return similar results to those found with GFP-RassflC’50? Furthermore, will point mutations within the binding interface of the DHB domain disrupt Rassfl C association and hence render cells taxol-resistant? Or, are there interactions from other domains of Rassfl C and DAXX that increase affinity/specificity? It will also be interesting to investigate whether gene therapy in taxol-resistant cells is viable. If DAXX expression were reinstated, it could make the cells taxol-susceptible. If indeed so, then a minimum construct of DAXX including the DHB domain could be sufficient to render the cells sensitive to taxol. Overall, the DHBfRassf1C2338’’ complex has provided the molecular basis for the function of specific domains in DAXX and RassflC. 5.7 Concluding remarks Taken as a whole, DAXX is a very interesting protein. It is a scaffold that brings together diverse complexes that fulfill specific functions in specific places and times. Despite being involved in important processes such as transcription regulation, apoptosis control, mitosis management, and other cellular mechanisms, the molecular basis for its scaffolding properties and regulation are still unknown. The structure of the DHB domain, its interaction with Rassfl C, and the properties of its SIMs are the first stones in the structural knowledge needed to decipher 192 Chapter 5 — Concluding remarks and future experiments this enigmatic protein. There are still a number of important unanswered questions that will require further structural and functional studies in this exciting field. 193 References References Agathanggelou, A., Cooper, W.N., and Latif, F. (2005). Role of the Ras-association domain family 1 tumor suppressor gene in human cancers. Cancer Res 65, 3497-3508. Akterin, S., Cowburn, R.F., Miranda-Vizuete, A., Jimenez, A., Bogdanovic, N., Winblad, B., and Cedazo-Minguez, A. (2006). Involvement of glutaredoxin-1 and thioredoxin-l in beta amyloid toxicity and Alzheimer’s disease. Cell Death Differ 13, 1454-1465. Alberts, B. (2002). Molecular biology of the cell, 4th edn (New York: Garland Science). Altschul, S.F., Gish, W., Miller, W., Myers, E.W., and Lipman, D.J. (1990). Basic local alignment search tool. J Mol Biol 215, 403-4 10. Avruch, J., Xavier, R., Bardeesy, N., Zhang, X.F., Praskova, M., Zhou, D., and Xia, F. (2009). Rassf family of tumor suppressor polypeptides. J Biol Chem 284, 11001-11005. Ayaydin, F., and Dasso, M. (2004). Distinct in vivo dynamics of vertebrate SUMO paralogues. Mol Biol Cell 15, 5208-52 18. Baba, D., Maita, N., Jee, J.G., Uchimura, Y., Saitoh, H., Sugasawa, K., Hanaoka, F., Tochio, H., Hiroaki, H., and Shirakawa, M. (2005). Crystal structure of thymine DNA glycosylase conjugated to SUMO-1. Nature 435, 979-982. Bencsath, K.P., Podgorski, M.S., Pagala, V.R., Slaughter, C.A., and Schulman, B.A. (2002). Identification of a multifunctional binding site on Ubc9p required for Smt3p conjugation. J Biol Chem 277, 47938-47945. Bergink, S., and Jentsch, 5. (2009). Principles of ubiquitin and SUMO modifications in DNA repair. Nature 458, 461-467. Bernardi, R., and Pandolfi, P.P. (2007). Structure, dynamics and functions of promyelocytic leukaemia nuclear bodies. Nat Rev Mol Cell Biol 8, 1006-1016. Bernardi, R., Papa, A., and Pandolfi, P.P. (2008). Regulation of apoptosis by PML and the PML NBs. Oncogene 27, 6299-63 12. Blom, N., Gammeltoft, S., and Brunak, 5. (1999). Sequence and structure-based prediction of eukaryotic protein phosphorylation sites. J Mol Biol 294, 135 1-1362. Blom, N., Sicheritz-Ponten, T., Gupta, R., Gammeltoft, S., and Brunak, S. (2004). Prediction of post-translational glycosylation and phosphorylation of proteins from the amino acid sequence. Proteomics 4, 1633-1649. 194 References Boddy, M.N., Howe, K., Etkin, L.D., Solomon, E., and Freemont, P.S. (1996). Plc 1, a novel ubiquitin-like protein which interacts with the PML component of a multiprotein complex that is disrupted in acute promyelocytic leukaemia. Oncogene 13, 97 1-982. Boellmann, F., Guettouche, T., Guo, Y., Fenna, M., Mnayer, L., and Voelimy, R. (2004). DAXX interacts with heat shock factor 1 during stress activation and enhances its transcriptional activity. Proc Nati Acad Sci U S A 101, 4100-4105. Booth, V., Koth, C.M., Edwards, A.M., and Arrowsmith, C.H. (2000). Structure of a conserved domain common to the transcription factors TFIIS, elongin A, and CRSP7O. J Biol Chem 275, 31266-3 1268. Bossis, G., and Meichior, F. (2006). Regulation of SUMOylation by reversible oxidation of SUMO conjugating enzymes. Mol cell 21, 349-357. Brett, T.J., Legendre-Guillemin, V., McPherson, P.S., and Fremont, D.H. (2006). Structural definition of the F-actin-binding THATCH domain from HIP1R. Nat Struct Mol Biol 13, 121-130. Brubaker, K., Cowley, S.M., Huang, K., Loo, L., Yochum, G.S., Ayer, D.E., Eisenman, R.N., and Radhakrishnan, 1. (2000). Solution structure of the interacting domains of the Mad-Sin3 complex: implications for recruitment of a chromatin-modifying complex. cell 103, 655-665. Chang, cc., Lin, D.Y., Fang, H.I., chen, R.H., and Shih, H.M. (2005). Daxx mediates the small ubiquitin-like modifier-dependent transcriptional repression of Smad4. J Biol Chem 280, 10164-10173. Chang, H.Y., Nishitoh, H., Yang, X., Ichijo, H., and Baltimore, D. (1998). Activation of apoptosis signal-regulating kinase 1 (ASK1) by the adapter protein Daxx. Science 281, 1860- 1863. Charette, S.J., and Landry, J. (2000). The interaction of HSP27 with Daxx identifies a potential regulatory role of HSP27 in Fas-induced apoptosis. AnnN Y Acad Sci 926, 126-13 1. Chen, A., Wang, P.Y., Yang, Y.C., Huang, Y.H., Yeh, J.J., Chou, Y.H., Cheng, J.T., Hong, Y.R., and Li, S.S. (2006). SUMO regulates the cytoplasmonuclear transport of its target protein Daxx. J Cell Biochem 98, 895-911. Chen, L.Y., and Chen, J.D. (2003). Daxx silencing sensitizes cells to multiple apoptotic pathways. Mol Cell Biol 23, 7108-7121. Cheng, J., Randall, A.Z., Sweredoski, M.J., and Baldi, P. (2005). SCRATCH: a protein structure and structural feature prediction server. Nucleic Acids Res 33, W72-76. Chou, J.J., Gaemers, S., Howder, B., Louis, J.M., and Bax, A. (2001). A simple apparatus for generating stretched polyacrylamide gels, yielding uniform alignment of proteins and detergent micelles. Journal of Biomolecular Nmr 21, 377-382. 195 References Chung, T.L., Hsiao, H.H., Yeh, Y.Y., Shia, H.L., Chen, Y.L., Liang, P.H., Wang, A.H., Khoo, K.H., and Shoei-Lung Li, S. (2004). In vitro modification of human centromere protein CENP-C fragments by small ubiquitin-like modifier (SUMO) protein: definitive identification of the modification sites by tandem mass spectrometry analysis of the isopeptides. J Biol Chem 279, 39653-39662. Clore, G.M., Tang, C., and Iwahara, J. (2007). Elucidating transient macromolecular interactions using paramagnetic relaxation enhancement. Cuff Opin Struct Biol 17, 603-6 16. Comerford, K.M., Leonard, M.O., Karhausen, J., Carey, R., Colgan, S.P., and Taylor, C.T. (2003). Small ubiquitin-related modifier-i modification mediates resolution of CREB dependent responses to hypoxia. Proc Natl Acad Sci U S A 100, 986-991. Cornilescu, G., Delaglio, F., and Bax, A. (1999). Protein backbone angle restraints from searching a database for chemical shift and sequence homology. J Biomol NMR 13, 289-3 02. Cuff, J.A., and Barton, G.J. (2000). Application of multiple sequence alignment profiles to improve protein secondary structure prediction. Proteins 40, 502-511. Daragan, V.A., and Mayo, K.H. (1997). Motional model analyses of protein and peptide dynamics using C-13 and N-15 NMR relaxation. Progress in Nuclear Magnetic Resonance Spectroscopy 31, 63-105. Delaglio, F., Grzesiek, S., Vuister, G.W., Zhu, G., Pfeifer, J., and Bax, A. (1995). NMRPipe: a multidimensional spectral processing system based on UNIX pipes. J Biomol NMR 6, 277- 293. Dellaire, G., and Bazett-Jones, D.P. (2004). PML nuclear bodies: dynamic sensors of DNA damage and cellular stress. Bioessays 26, 963-977. Deyrieux, A.F., Rosas-Acosta, G., Ozbun, M.A., and Wilson, V.G. (2007). Surnoylation dynamics during keratinocyte differentiation. J Cell Sci 120, 125-136. Dolinsky, T.J., Czodrowski, P., Li, H., Nielsen, J.E., Jensen, J.H., Klebe, G., and Baker, N.A. (2007). PDB2PQR: expanding and upgrading automated preparation of biomolecular structures for molecular simulations. Nucleic Acids Res 35, W522-525. Dolinsky, T.J., Nielsen, J.E., McCammon, J.A., and Baker, N.A. (2004). PDB2PQR: an automated pipeline for the setup of Poisson-Boltzmann electrostatics calculations. Nucleic Acids Res 32, W665-667. Donaldson, L.W., Skrynnikov, N.R., Choy, W.Y., Muhandiram, D.R., Sarkar, B., Forman-Kay, J.D., and Kay, L.E. (2001). Structural characterization of proteins with an attached ATCUN motif by paramagnetic relaxation enhancement NMR spectroscopy. J Am Chem Soc 123, 9843-9847. 196 References Donehower, L.A., Harvey, M., Slagle, B.L., McArthur, M.J., Montgomery, C.A., Jr., Butel, J.S., and Bradley, A. (1992). Mice deficient for p53 are developmentally normal but susceptible to spontaneous tumours. Nature 356, 215-221. Dosset, P., Hus, J.C., Blackledge, M., and Marion, D. (2000). Efficient analysis of macromolecular rotational diffusion from heteronuclear relaxation data. J Biomol NMR 16, 23-28. Drohat, A.C., Amburgey, J.C., Abildgaard, F., Stanch, M.R., Baldisseri, D., and Weber, D.J. (1996). Solution structure of rat apo-S100B(beta beta) as determined by NMR spectroscopy. Biochemistry 35, 11577-11588. Duda, D.M., van Waardenburg, R.C., Borg, L.A., McGarity, S., Nourse, A., Waddell, M.B., Bjomsti, M.A., and Schulman, B.A. (2007). Structure of a SUMO-binding-motif mimic bound to Smt3p-Ubc9p: conservation of a non-covalent ubiquitin-like protein-E2 complex as a platform for selective interactions within a SUMO pathway. J Mol Biol 369, 6 19-630. Dundr, M., and Misteli, T. (2001). Functional architecture in the cell nucleus. Biochem J 356, 297-310. Ecsedy, J.A., Michaelson, J.S., and Leder, P. (2003). Homeodomain-interacting protein kinase 1 modulates Daxx localization, phosphorylation, and transcriptional activity. Mol Cell Biol 23, 950-960. Elcock, A.H. (2001). Prediction of functionally important residues based solely on the computed energetics of protein structure. J Mo! Biol 312, 885-896. Emelyanov, A.V., Kovac, C.R., Sepulveda, M.A., and Birshtein, B.K. (2002). The interaction of Pax5 (BSAP) with Daxx can result in transcriptional activation in B cells. J Biol Chem 277, 11156-11164. Estrabaud, E., Lassot, I., Blot, G., Le Rouzic, E., Tanchou, V., Quemeneur, E., Daviet, L., Margottin-Goguet, F., and Benanous, R. (2007). RASSF1C, an isoform of the tumor suppressor RASSF1A, promotes the accumulation of beta-catenin by interacting with betaTrCP. Cancer Res 67, 1054-1061. Everett, R.D., Earnshaw, W.C., Pluta, A.F., Sternsdorf, T., Ainsztein, A.M., Carmena, M., Ruchaud, S., Hsu, W.L., and Orn, A. (1999). A dynamic connection between centromeres and ND1O proteins. J Cell Sci 112 (Pt 20), 3443-3454. Farrow, N.A., Muhandiram, R., Singer, A.U., Pascal, S.M., Kay, C.M., Gish, G., Shoelson, S.E., Pawson, T., Forman-Kay, J.D., and Kay, L.E. (1994). Backbone dynamics of a free and phosphopeptide-complexed Src homology 2 domain studied by 15N NMR relaxation. Biochemistry 33, 5984-6003. Fossi, M., Linge, J., Labudde, D., Leitner, D., Nilges, M., and Oschkinat, H. (2005). Influence of chemical shift tolerances on NMR structure calculations using ARIA protocols for assigning NOE data. J Biomol NMR 31, 2 1-34. 197 References Fu, J., Bian, M., Jiang, Q., and Zhang, C. (2007). Roles of Aurora kinases in mitosis and tumorigenesis. Mol Cancer Res 5, 1-10. Fukuyo, Y., Kitamura, T., Inoue, M., Horikoshi, N.T., Higashikubo, R., Hunt, C.R., Usheva, A., and Horikoshi, N. (2009). Phosphorylation-dependent Lys63-linked polyubiquitination of Daxx is essential for sustained TNF-{alpha}-induced ASK1 activation. Cancer Res 69, 75 12- 7517. Garcia-Dominguez, M., and Reyes, J.C. (2009). SUMO association with repressor complexes, emerging routes for transcriptional control. Biochim Biophys Acta. Gasteiger, E., Hoogland, C., Gattiker, A., Duvaud, S., Wilkins, M.R., Appel, R.D., and Bairoch, A. (2005). John M. Walker (ed): The protein protocols handbook (Humana Press). Gay, D.L., Ramon, H., and Oliver, P.M. (2008). Cbl- and Nedd4-family ubiquitin ligases: balancing tolerance and immunity. Immunol Res 42, 51-64. Geiss-Friedlander, R., and Meichior, F. (2007). Concepts in sumoylation: a decade on. Nat Rev Mo! Cell Bio! 8, 947-956. Geoffroy, M.C., and Hay, R.T. (2009). An additional role for SUMO in ubiquitin-mediated proteolysis. Nat Rev Mol Cell Biol 10, 564-568. Geourjon, C., and Deleage, G. (1995). SOPMA: significant improvements in protein secondary structure prediction by consensus prediction from multiple alignments. Comput Appl Biosci 1], 681-684. Giorgino, F., de Robertis, 0., Laviola, L., Montrone, C., Perrini, S., McCowen, K.C., and Smith, R.J. (2000). The sentrin-conjugating enzyme mUbc9 interacts with GLUT4 and GLUT1 glucose transporters and regulates transporter levels in skeletal muscle cells. Proc Natl Acad SciU SA97, 1125-1130. Goddard, T.D., and Kne!ler, D.G. (2004). Sparky 3. (University of California, San Francisco). Golebiowski, F., Matic, I., Tatham, M.H., Cole, C., Yin, Y., Nakamura, A., Cox, J., Barton, G.J., Mann, M., and Hay, R.T. (2009). System-wide changes to SUMO modifications in response to heat shock. Sci Signal 2, ra24. Gostissa, M., Hengstermann, A., Fogal, V., Sandy, P., Schwarz, S.E., Scheffner, M., and Del Sal, G. (1999). Activation of p53 by conjugation to the ubiquitin-like protein SUMO-1. EMBO J 18, 6462-6471. Gostissa, M., Morelli, M., Mantovani, F., Guida, E., Piazza, S., Collavin, L., Brancolini, C., Schneider, C., and Del Sal, G. (2004). The transcriptional repressor hDaxx potentiates p53- dependent apoptosis. J Biol Chem 279, 480 13-48023. 198 References Greger, J.G., Katz, R.A., Ishov, A.M., Maul, G.G., and Skalka, A.M. (2005). The cellular protein daxx interacts with avian sarcoma virus integrase and viral DNA to repress viral transcription. J Virol 79, 4610-4618. Grigoryan, G., and Keating, A.E. (2008). Structural specificity in coiled-coil interactions. Curr Opin Struct Biol 18, 477-483. Grzenda, A., Lomberk, G., Zhang, J.S., and Urrutia, R. (2009). Sin3: master scaffold and transcriptional corepressor. Biochim Biophys Acta 1789, 443-450. Guermeur, Y., Geourjon, C., Gallinari, P., and Deleage, G. (1999). Improved performance in protein secondary structure prediction by inhomogeneous score combination. Bioinformatics 15, 413-421. Guo, D., Li, M., Zhang, Y., Yang, P., Eckenrode, S., Hopkins, D., Zheng, W., Purohit, S., Podolsky, R.H., Muir, A., et a!. (2004). A functional variant of SUMO4, a new I kappa B alpha modifier, is associated with type 1 diabetes. Nat Genet 36, 837-841. Gutierrez, G.J., and Ronai, Z. (2006). Ubiquitin and SUMO systems in the regulation of mitotic checkpoints. Trends Biochem Sci 31, 324-332. Habeck, M., Rieping, W., Linge, J.P., and Nilges, M. (2004). NOE assignment with ARIA 2.0: the nuts and bolts. Methods Mol Biol 278, 379-402. Han, J.H., Batey, S., Nickson, A.A., Teichmann, S.A., and Clarke, J. (2007). The folding and evolution of multidomain proteins. Nat Rev Mol Cell Biol 8, 319-330. Hardeland, U., Steinacher, R., Jiricny, J., and Schar, P. (2002). Modification of the human thymine-DNA glycosylase by ubiquitin-like proteins facilitates enzymatic turnover. EMBO J 21, 1456-1464. Harjes, E., Harjes, S., Wohlgemuth, S., Muller, K.H., Krieger, E., Herrmann, C., and Bayer, P. (2006). GTP-Ras disrupts the intramolecular complex of Cl and RA domains of Norel. Structure 14, 88 1-888. Hay, R.T. (2005). SUMO: a history of modification. Mol Cell 18, 1-12. Hecker, C.M., Rabiller, M., Haglund, K., Bayer, P., and Dikic, I. (2006). Specification of SUMO1- and SUMO2-interacting motifs. J Biol Chem 281, 16117-16127. Hesson, L.B., Cooper, W.N., and Latif, F. (2007a). Evaluation of the 3p2l.3 tumour-suppressor gene cluster. Oncogene 26, 7283-7301. Hesson, LB., Cooper, W.N., and Latif, F. (2007b). The role of RASSF1A methylation in cancer. Dis Markers 23, 73-87. Hicke, L., Schubert, H.L., and Hill, C.P. (2005). Ubiquitin-binding domains. Nat Rev Mol Cell Biol 6, 610-621. 199 References Hietakangas, V., Anckar, J., Blomster, H.A., Fujimoto, M., Palvimo, J.J., Nakai, A., and Sistonen, L. (2006). PDSM, a motif for phosphorylation-dependent SUMO modification. Proc Nat! Acad Sci U S A 103, 45-50. Hoellerer, M.K., Noble, M.E., Labesse, G., Campbell, I.D., Werner, J.M., and Arold, S.T. (2003). Molecular recognition of paxillin LD motifs by the focal adhesion targeting domain. Structure 11, 1207-1217. Hofmann, H., Sindre, H., and Stamminger, T. (2002). Functional interaction between the pp7l protein of human cytomegalovirus and the PML-interacting protein human Daxx. J Virol 76, 5769-5783. Hofmann, T.G., Stollberg, N., Schmitz, M.L., and Will, H. (2003). HIPK2 regulates transforming growth factor-beta-induced c-Jun NH(2)-terminal kinase activation and apoptosis in human hepatoma cells. Cancer Res 63, 827 1-8277. Hollenbach, A.D., McPherson, C.J., Mientjes, E.J., Iyengar, R., and Grosveld, G. (2002). Daxx and histone deacetylase II associate with chromatin through an interaction with core histones and the chromatin-associated protein Dek. J Cell Sci 115, 3319-3330. Hollenbach, A.D., Sublett, J.E., McPherson, C.J., and Grosveld, G. (1999). The Pax3-FKHR oncoprotein is unresponsive to the Pax3-associated repressor hDaxx. EMBO J 18, 3 702- 3711. Hoim, L., and Sander, C. (1995). Dali: a network tool for protein structure comparison. Trends Biochem Sci 20, 478-480. Hong, Y., Rogers, R., Matunis, M.J., Mayhew, C.N., Goodson, M.L., Park-Sarge, O.K., and Sarge, K.D. (2001). Regulation of heat shock transcription factor 1 by stress-induced SUMO 1 modification. J Biol Chem 276, 40263-40267. Hoofnagle, A.N., Resing, K.A., and Aim, N.G. (2003). Protein analysis by hydrogen exchange mass spectrometry. Annu Rev Biophys Biomol Struct 32, 1-25. Horiuchi, T., Takahashi, M., Kikuchi, J., Yokoyama, S., and Maeda, H. (2005). Effect of dielectric properties of solvents on the quality factor for a beyond 900 MHz cryogenic probe model. Journal of Magnetic Resonance 174, 34-42. Huang, L., Xu, G.L., Zhang, J.Q., Tian, L., Xue, J.L., Chen, J.Z., and Jia, W. (2008). Daxx interacts with HIV- 1 integrase and inhibits lentiviral gene expression. Biochem Biophys Res Commun 373, 24 1-245. Huang, Y.S., and Shih, H.M. (2009). Daxx positively modulates beta-cateninlTCF4-mediated transcriptional potential. Biochem Biophys Res Commun 386, 762-768. Hunt, I. (2005). From gene to protein: a review of new and enabling technologies for multi parallel protein expression. Protein Expr Purif 40, 1-22. 200 References Hutchinson, E.G., and Thornton, J.M. (1996). PROMOTIF--a program to identify and analyze structural motifs in proteins. Protein Sci 5, 2 12-220. Hwang, E., Ryu, K.S., Paakkonen, K., Guntert, P., Cheong, H.K., Lim, D.S., Lee, J.O., Jeon, Y.H., and Cheong, C. (2007). Structural insight into dimeric interaction of the SARAH domains from Mstl and RASSF family proteins in the apoptosis pathway. Proc Natl Acad Sci US A 104, 9236-9241. Hwang, J., and Kalejta, R.F. (2009). Human cytomegalovirus protein pp7i induces Daxx SUMOylation. J Virol 83, 659 1-6598. Ishov, A.M., Sotnikov, A.G., Negorev, D., Viadimirova, O.V., Neff, N., Kamitani, I., Yeh, E.T., Strauss, J.F., 3rd, and Maul, G.G. (1999). PML is critical for ND1O formation and recruits the PML-interacting protein daxx to this nuclear structure when modified by SUMO- 1. J Cell Biol 147, 221-234. Ishov, A.M., Viadimirova, O.V., and Maul, G.G. (2004). Heterochromatin and ND1O are cell- cycle regulated and phosphorylation-dependent alternate nuclear sites of the transcription repressor Daxx and SWI/SNF protein ATRX. J Cell Sci 117, 3807-3820. Jang, M.S., Ryu, S.W., and Kim, E. (2002). Modification of Daxx by small ubiquitin-related modifier-i. Biochem Biophys Res Commun 295, 495-500. Ji, Z., Degerny, C., Vintonenko, N., Deheuninck, J., Foveau, B., Leroy, C., Coil, J., Tulasne, D., Baert, i.E., and Fafeur, V. (2007). Regulation of the Ets-1 transcription factor by sumoylation and ubiquitinylation. Oncogene 26, 395-406. Joerger, A.C., and Fersht, A.R. (2007). Structure-function-rescue: the diverse nature of common p53 cancer mutants. Oncogene 26, 2226-2242. Johnson, E.S. (2004). Protein modification by SUMO. Annu Rev Biochem 73, 355-382. Johnson, P.E., Brun, E., MacKenzie, L.F., Withers, S.G., and McIntosh, L.P. (1999). The cellulose-binding domains from Cellulomonas fimi beta-i, 4-glucanase CenC bind nitroxide spin-labeled cellooligosaccharides in multiple orientations. J Mol Biol 287, 609-625. Johnson, P.E., Tomme, P., Joshi, M.D., and McIntosh, L.P. (1996). Interaction of soluble cellooligosaccharides with the N-terminal cellulose-binding domain of Cellulomonas fimi CenC 2. NMR and ultraviolet absorption spectroscopy. Biochemistry 35, 13895-13906. Jones, M.C., Fusi, L., Higham, J.H., Abdel-Hafiz, H., Horwitz, K.B., Lam, E.W., and Brosens, J.J. (2006). Regulation of the SUMO pathway sensitizes differentiating human endometrial stromal cells to progesterone. Proc Natl Acad Sci U S A 103, 16272-16277. Jung, Y.S., Kim, H.Y., Kim, J., Lee, M.G., Pouyssegur, J., and Kim, E. (2008). Physical interactions and functional coupling between Daxx and sodium hydrogen exchanger 1 in ischemic cell death. J Biol Chem 283, 1018-1025. 201 References Jung, Y.S., Kim, H.Y., Lee, Y.J., and Kim, E. (2007). Subcellular localization of Daxx determines its opposing functions in ischemic cell death. FEBS Lett 58], 843-852. Junn, E., Taniguchi, H., Jeong, B.S., Zhao, X., Ichijo, H., and Mouradian, M.M. (2005). Interaction of DJ- 1 with Daxx inhibits apoptosis signal-regulating kinase 1 activity and cell death. Proc Natl Acad Sci U S A 102, 969 1-9696. Juo, P., Kuo, C.J., Yuan, J., and Blenis, J. (1998). Essential requirement for caspase-8/FLICE in the initiation of the Fas-induced apoptotic cascade. Cuff Biol 8, 1001-1008. Kabsch, W., and Sander, C. (1983). Dictionary of protein secondary structure: pattern recognition of hydrogen-bonded and geometrical features. Biopolymers 22, 2577-2637. Kamtekar, S., and Hecht, M.H. (1995). Protein Motifs. 7. The four-helix bundle: what determines a fold? FASEB J 9, 10 13-1022. Karunakaran, S., Diwakar, L., Saeed, U., Agarwal, V., Ramakrishnan, S., Iyengar, S., and Ravindranath, V. (2007). Activation of apoptosis signal regulating kinase 1 (ASK1) and translocation of death-associated protein, Daxx, in substantia nigra pars compacta in a mouse model of Parkinson’s disease: protection by alpha-lipoic acid. FASEB J 21, 2226-2236. Kay, L.E., Keifer, P., and Saarinen, T. (1992). Pure Absorption Gradient Enhanced Heteronuclear Single Quantum Correlation Spectroscopy with Improved Sensitivity. Journal of the American Chemical Society 114, 10663-10665. Kay, L.E., Torchia, D.A., and Bax, A. (1989). Backbone dynamics of proteins as studied by 15N inverse detected heteronuclear NMR spectroscopy: application to staphylococcal nuclease. Biochemistry 28, 8972-8979. Kelly, A.E., Ou, H.D., Withers, R., and Dotsch, V. (2002). Low-conductivity buffers for high- sensitivity NMR measurements. J Am Chem Soc 124, 12013-12019. Kerscher, 0. (2007). SUMO junction-what’s your function? New insights through SUMO interacting motifs. EMBO Rep 8, 550-555. Kim, D.E., Chivian, D., and Baker, D. (2004). Protein structure prediction and analysis using the Robetta server. Nucleic Acids Res 32, W526-531. Kim, E.J., Park, J.S., and Um, S.J. (2003). Identification of Daxx interacting with p’73, one of the p53 family, and its regulation of p53 activity by competitive interaction with PML. Nucleic Acids Res 31, 5356-5367. Kim, J.H., and Baek, S.H. (2009). Emerging roles of desumoylating enzymes. Biochim Biophys Acta 1792, 155-162. Kiriakidou, M., Driscoll, D.A., Lopez-Guisa, J.M., and Strauss, J.F., 3rd (1997). Cloning and expression of primate Daxx cDNAs and mapping of the human gene to chromosome 6p2l .3 in the MHC region. DNA Cell Biol 16, 1289-1298. 202 References Kitagawa, D., Kajiho, H., Negishi, T., Ura, S., Watanabe, T., Wada, T., Ichijo, H., Katada, T., and Nishina, H. (2006). Release of RASSF1C from the nucleus by Daxx degradation links DNA damage and SAPK/JNK activation. EMBO J 25, 3286-3297. Kitamura, T., Fukuyo, Y., Inoue, M., Horikoshi, N.T., Shindoh, M., Rogers, B.E., Usheva, A., and Horikoshi, N. (2009). Mutant p53 disrupts the stress MAPK activation circuit induced by ASK1-dependent stabilization of Daxx. Cancer Res 69, 768 1-7688. Kneller, D.G., Cohen, F.E., and Langridge, R. (1990). Improvements in protein secondary structure prediction by an enhanced neural network. J Mo! Biol 214, 171-182. Kovacs, H., O’Ddonoghue, S.I., Hoppe, H.J., Comfort, D., Reid, K.B., Campbell, D., and Nilges, M. (2002). Solution structure of the coiled-coil trimerization domain from lung surfactant protein D. J Biomol NMR 24, 89-102. Kuo, H.Y., Chang, C.C., Jeng, J.C., Hu, H.M., Lin, D.Y., Maul, G.G., Kwok, R.P., and Shih, H.M. (2005). SUMO modification negatively modulates the transcriptional activity of CREB-binding protein via the recruitment of Daxx. Proc Nat! Acad Sci U S A 102, 16973- 16978. Kussie, P.H., Gonna, S., Marechal, V., Elenbaas, B., Moreau, J., Levine, A.J., and Pavietich, NY. (1996). Structure of the MDM2 oncoprotein bound to the p53 tumor suppressor transactivation domain. Science 274, 948-953. Lalioti, V.S., Vergarajauregui, S., Pulido, D., and Sandoval, I.V. (2002). The insulin-sensitive glucose transporter, GLUT4, interacts physically with Daxx. Two proteins with capacity to bind Ubc9 and conjugated to SUMO1. J Biol Chem 277, 19783-19791. Lallemand-Breitenbach, V., Jeanne, M., Benhenda, S., Nasr, R., Lei, M., Peres, L., Zhou, J., Zhu, J., Raught, B., and de The, H. (2008). Arsenic degrades PML or PML-RARalpha through a SUMO-triggered RNF4/ubiquitin-mediated pathway. Nat Cell Biol 10, 547-555. Laskowski, R.A., Rullmannn, J.A., MacArthur, M.W., Kaptein, R., and Thornton, J.M. (1996). AQUA and PROCHECK-NMR: programs for checking the quality of protein structures solved by NMR. J Biomol NMR 8, 477-486. Lee, G.M., Donaldson, L.W., Pufall, M.A., Kang, H.S., Pot, I., Graves, B.J., and McIntosh, L.P. (2005). The structural and dynamic basis of Ets-1 DNA binding autoinhibition. J Biol Chem 280, 7088-7099. Lee, G.M., Pufall, M.A., Meeker, C.A., Kang, H.S., Graves, B.J., and McIntosh, L.P. (2008). The affinity of Ets- 1 for DNA is modulated by phosphorylation through transient interactions of an unstructured region. J Mo! Biol 382, 10 14-1030. Lehembre, F., Muller, S., Pandolfi, P.P., and Dejean, A. (2001). Regulation of Pax3 transcriptional activity by SUMO- 1-modified PML. Oncogene 20, 1-9. 203 References Li, H., Leo, C., Zhu, J., Wu, X., OrNeil, J., Park, E.J., and Chen, J.D. (2000a). Sequestration and inhibition of Daxx-mediated transcriptional repression by PML. Mo! Ce!! Bio! 20, 1784- 1796. Li, J., Wang, F., Protopopov, A., Ma!yukova, A., Kashuba, V., Minna, J.D., Lerman, M.I., K!ein, G., and Zabarovsky, E. (2004). Inactivation of RASSF1C during in vivo tumor growth identifies it as a tumor suppressor gene. Oncogene 23, 594 1-5949. Li, Q., Wang, X., Wu, X., Rui, Y., Liu, W., Wang, J., Liou, Y.C., Ye, Z., and Lin, S.C. (2007). Daxx cooperates with the AxinIHIPK2/p53 complex to induce cell death. Cancer Res 67, 66- 74. Li, R., Pei, H., Watson, D.K., and Papas, T.S. (2000b). EAP1/Daxx interacts with ETS1 and represses transcriptional activation ofETS1 target genes. Oncogene 19, 745-753. Lin, D.Y., Huang, Y.S., Jeng, J.C., Kuo, H.Y., Chang, C.C., Chao, T.T., Ho, C.C., Chen, Y.C., Lin, T.P., Fang, H.I., et a!. (2006). Role of SUMO-interacting motif in Daxx SUMO modification, subnuclear localization, and repression of sumoylated transcription factors. Mo! Cell 24, 34 1-354. Lin, D.Y., Lai, M.Z., Ann, D.K., and Shih, H.M. (2003). Promyelocytic leukemia protein (PML) functions as a glucocorticoid receptor co-activator by sequestering Daxx to the PML oncogenic domains (PODs) to enhance its transactivation potential. J Biol Chem 278, 15958- 15965. Lin, D.Y., and Shih, H.M. (2002). Essential role of the 58-kDa microspherule protein in the modulation of Daxx-dependent transcriptional repression as revealed by nucleolar sequestration. J Biol Chem 277, 25446-25456. Lindsay, C.R. (2009). DAXX and Rassfl define novel mitotic stress checkpoint that is critical for cellular taxol response. (Univeristy of Florida), p. 89. Lindsay, C.R., Giovinazzi, S., and Ishov, A.M. (2009a). Daxx is a predominately nuclear protein that does not translocate to the cytoplasm in response to cell stress. Cell Cycle 8, 1544-1551. Lindsay, C.R., Giovinazzi, S., Morozov, V.M., Escobar-Cabrera, E., McIntosh, L.P., Summers, M.K., Han, H.S., and Ishov, A.M. (2009b). Regulation of mitosis and taxane response by Daxx and Rassfl. Oncogene submitted. Lindsay, C.R., Morozov, V.M., and Ishov, A.M. (2008). PML NBs (ND1O) and Daxx: from nuclear structure to protein function. Front Biosci 13, 7 132-7142. Lindsay, C.R., Scholz, A., Morozov, V.M., and Ishov, A.M. (2007). Daxx shortens mitotic arrest caused by paclitaxel. Cell Cycle 6, 1200-1204. Linge, J.P., ODonoghue, S.I., and Nilges, M. (2001). Automated assignment of ambiguous nuclear overhauser effects with ARIA. Methods Enzymol 339, 7 1-90. 204 References Liu, B., and Shuai, K. (2008). Regulation of the sumoylation system in gene expression. Cuff Opin Cell Biol 20, 28 8-293. Liu, B., Yang, Y., Chernishof, V., Loo, R.R., Jang, H., Tahk, S., Yang, R., Mink, S., Shultz, D., Bellone, C.J., et al. (2007). Proinflammatory stimuli induce IKKalpha-mediated phosphorylation ofPIAS1 to restrict inflammation and immunity. Cell 129, 903-914. Liu, L., Vo, A., and McKeehan, W.L. (2005). Specificity of the methylation-suppressed A isoform of candidate tumor suppressor RASSF1 for microtubule hyperstabilization is determined by cell death inducer Cl 90RF5. Cancer Res 65, 1830-1838. Lutz, T., Stoger, R., and Nieto, A. (2006). CHD6 is a DNA-dependent ATPase and localizes at nuclear sites of mRNA synthesis. FEBS Left 580, 585 1-5857. Macauley, M.S., Errington, W.J., Okon, M., Scharpf, M., Mackereth, C.D., Schulman, B.A., and McIntosh, L.P. (2004). Structural and dynamic independence of isopeptide-linked RanGAP1 and SUMO-1. J Biol Chem 279, 4913 1-49137. Macauley, M.S., Errington, W.J., Scharpf M., Mackereth, C.D., Blaszczak, A.G., Graves, B.J., and McIntosh, L.P. (2006). Beads-on-a-string, characterization of ETS-1 sumoylated within its flexible N-terminal sequence. J Biol Chem 281, 4164-4172. Mackereth, C.D., Scharpf, M., Gentile, L.N., MacIntosh, S.E., Slupsky, C.M., and McIntosh, L.P. (2004). Diversity in structure and function of the Ets family PNT domains. J Mol Biol 342, 1249-1264. Mahajan, R., Deiphin, C., Guan, T., Gerace, L., and Melchior, F. (1997). A small ubiquitin related polypeptide involved in targeting RanGAP1 to nuclear pore complex protein RanBP2. Cell 88, 97-107. Marsh, J.A., Singh, V.K., Jia, Z., and Forman-Kay, J.D. (2006). Sensitivity of secondary structure propensities to sequence differences between alpha- and gamma-synuclein: implications for fibrillation. Protein Sci 15, 2795-2804. Mauger, E., and Scott, P.H. (2004). Mitogenic stimulation of transcription by RNA polymerase III. Biochem Soc Trans 32, 976-977. McDonnell, A.V., Jiang, T., Keating, A.E., and Berger, B. (2006). Paircoil2: improved prediction of coiled coils from sequence. Bioinformatics 22, 356-358. McGuffin, L.J., Bryson, K., and Jones, D.T. (2000). The PSIPRED protein structure prediction server. Bioinformatics 16, 404-405. Mcintosh, L.P., Brun, E., and Kay, L.E. (1997). Stereospecific assignment of the NH2 resonances from the primary amides of asparagine and glutamine side chains in isotopically labeled proteins. Journal of Biomolecular Nmr 9, 306-3 12. 205 References Meulmeester, E., Kunze, M., Hsiao, H.H., Urlaub, H., and Meichior, F. (2008). Mechanism and consequences for paralog-specific sumoylation of ubiquitin-specific protease 25. Mol Cell 30, 610-619. Michaelson, J.S. (2000). The Daxx enigma. Apoptosis 5, 217-220. Michaelson, J.S., Bader, D., Kuo, F., Kozak, C., and Leder, P. (1999). Loss of Daxx, a promiscuously interacting protein, results in extensive apoptosis in early mouse development. Genes Dev 13, 1918-1923. Michaelson, J.S., and Leder, P. (2003). RNAi reveals anti-apoptotic and transcriptionally repressive activities of DAXX. J Cell Sci 116, 345-352. Minty, A., Dumont, X., Kaghad, M., and Caput, D. (2000). Covalent modification of p73alpha by SUMO- 1. Two-hybrid screening with p73 identifies novel SUMO- 1-interacting proteins and a SUMO-1 interaction motif. J Biol Chem 275, 363 16-36323. Moore, A.D., Bjorklund, A.K., Ekman, D., Bornberg-Bauer, E., and Elofsson, A. (2008). Arrangements in the modular evolution of proteins. Trends Biochem Sd 33, 444-45 1. Morozov, V.M., Massoll, N.A., Viadimirova, O.V., Maul, G.G., and Ishov, A.M. (2008). Regulation of c-met expression by transcription repressor Daxx. Oncogene 27, 2 177-2186. Murakami, Y., Yamagoe, S., Noguchi, K., Takebe, Y., Takahashi, N., Uehara, Y., and Fukazawa, H. (2006). Ets-1-dependent expression of vascular endothelial growth factor receptors is activated by latency-associated nuclear antigen of Kapos?s sarcoma-associated herpesvirus through interaction with Daxx. J Biol Chem 281, 28113-28121. Muromoto, R., Nakao, K., Watanabe, T., Sato, N., Sekine, Y., Sugiyama, K., Oritani, K., Shimoda, K., and Matsuda, T. (2006). Physical and functional interactions between Daxx and STAT3. Oncogene 25, 213 1-2136. Muromoto, R., Sugiyama, K., Takachi, A., Imoto, S., Sato, N., Yamamoto, T., Oritani, K., Shimoda, K., and Matsuda, T. (2004a). Physical and functional interactions between Daxx and DNA methyltransferase 1-associated protein, DMAP1. J Immunol 172, 2985-2993. Muromoto, R., Sugiyama, K., Yamamoto, T., Oritani, K., Shimoda, K., and Matsuda, T. (2004b). Physical and functional interactions between Daxx and TSG1O1. Biochem Biophys Res Commun 316, 827-833. Nietlispach, D., Mott, H.R., Stott, K.M., Nielsen, P.R., Thiru, A., and Laue, E.D. (2004). Structure determination of protein complexes by NMR. Methods Mo! Biol 278, 255-288. Nilges, M. (1995). Calculation of protein structures with ambiguous distance restraints. Automated assignment of ambiguous NOE crosspeaks and disulphide connectivities. J Mo! Biol 245, 645-660. 206 References Nilges, M., and O’Donoghue, S.I. (1998). Ambiguous NOEs and automated NOE assignment. Progress in Nuclear Magnetic Resonance Spectroscopy 32, 107-139. Nishida, T., Terashima, M., and Fukami, K. (2006). PIASy-mediated repression of the Ets-1 is independent of its sumoylation. Biochem Biophys Res Commun 345, 1536-1546. Nomura, M., Uda-Tochio, H., Murai, K., Mori, N., and Nishimura, Y. (2005). The neural repressor NRSF/REST binds the PAH1 domain of the Sin3 corepressor by using its distinct short hydrophobic helix. J Mol Biol 354, 903-915. Ohiro, Y., Usheva, A., Kobayashi, S., Duffy, S.L., Nantz, R., Gius, D., and Horikoshi, N. (2003). Inhibition of stress-inducible kinase pathways by tumorigenic mutant p53. Mol Cell Biol 23, 322-334. Okura, T., Gong, L., Kamitani, T., Wada, T., Okura, I., Wei, C.F., Chang, H.M., and Yeh, E.T. (1996). Protection against Fas/APO-1- and tumor necrosis factor-mediated cell death by a novel protein, sentrin. J Immunol 157, 4277-4281. Orengo, C.A., Michie, A.D., Jones, S., Jones, D.T., Swindells, M.B., and Thornton, J.M. (1997). CATH--a hierarchic classification of protein domain structures. Structure 5, 1093-1108. Ottiger, M., Delaglio, F., and Bax, A. (1998). Measurement of J and dipolar couplings from simplified two-dimensional NMR spectra. Journal of Magnetic Resonance 131, 373-378. Owerbach, D., McKay, E.M., Yeh, E.T., Gabbay, K.H., and Bohren, K.M. (2005). A proline-90 residue unique to SUMO-4 prevents maturation and sumoylation. Biochem Biophys Res Commun 337, 5 17-520. Panse, V.G., Hardeland, U., Werner, T., Kuster, B., and Hurt, E. (2004). A proteome-wide approach identifies sumoylated substrate proteins in yeast. J Biol Chem 279, 41346-41351. Papaioannou, M., Melle, C., and Baniahmad, A. (2007). The coregulator Alien. Nucl Recept Signal 5, e008. Pelton, J.G., Torchia, D.A., Meadow, N.D., and Roseman, S. (1993). Tautomeric states of the active-site histidines of phosphorylated and unphosphorylated IIIGlc, a signal-transducing protein from Escherichia coli, using two-dimensional heteronuclear NMR techniques. Protein Sci 2, 543-558. Pennisi, E. (2003). Human genome. A low number wins the GeneSweep Pool. Science 300, 1484. Perlman, R., Schiemann, W.P., Brooks, M.W., Lodish, H.F., and Weinberg, R.A. (2001). TGF beta-induced apoptosis is mediated by the adapter protein Daxx that facilitates INK activation. Nat Cell Biol 3, 708-7 14. 207 References Pluta, A.F., Earnshaw, W.C., and Goldberg, I.G. (1998). Interphase-specific association of intrinsic centromere protein CENP-C with HDaxx, a death domain-binding protein implicated in Fas-mediated cell death. J Cell Sci 1]] (Pt 14), 2029-2041. Poleshko, A., Palagin, I., Zhang, R., Boimel, P., Castagna, C., Adams, P.D., Skalka, A.M., and Katz, R.A. (2008). Identification of cellular proteins that maintain retroviral epigenetic silencing: evidence for an antiviral response. J Virol 82, 23 13-2323. Pollastri, G., and McLysaght, A. (2005). Porter: a new, accurate server for protein secondary structure prediction. Bioinformatics 2], 1719-1720. Presnell, S.R., and Cohen, F.E. (1989). Topological distribution of four-alpha-helix bundles. Proc Nat! Acad Sci U S A 86, 6592-6596. Reverter, D., and Lima, C.D. (2005). Insights into E3 ligase activity revealed by a SUMO RanGAP1-Ubc9-Nup358 complex. Nature 435, 687-692. Reverter, D., and Lima, C.D. (2006). Structural basis for SENP2 protease interactions with SUMO precursors and conjugated substrates. Nat Struct Mo! Biol 13, 1060-1068. Richter, A.M., Pfeifer, G.P., and Dammann, R.H. (2009). The RASSF proteins in cancer; from epigenetic silencing to functional characterization. Biochim Biophys Acta. Rodriguez, M.S., Dargemont, C., and Hay, R.T. (2001). SUMO-1 conjugation in vivo requires both a consensus modification motif and nuclear targeting. J Biol Chem 276, 12654-12659. Roeder, R.G. (2005). Transcriptional regulation and the role of diverse coactivators in animal cells. FEBS Lett 579, 909-9 15. Rong, R., Jiang, L.Y., Sheikh, M.S., and Huang, Y. (2007). Mitotic kinase Aurora-A phosphorylates RAS SF1 A and modulates RASSF 1 A-mediated microtubule interaction and M-phase cell cycle regulation. Oncogene 26, 7700-7708. Rost, B. (1996). PHD: predicting one-dimensional protein structure by profile-based neural networks. Methods Enzymol 266, 525-539. Rost, B. (2001). Review: protein secondary structure prediction continues to rise. J Struct Biol 134, 204-2 18. Sahu, S.C., Swanson, K.A., Kang, R.S., Huang, K., Brubaker, K., Ratcliff, K., and Radhakrishnan, I. (2008). Conserved themes in target recognition by the PAH1 and PAH2 domains of the Sin3 transcriptional corepressor. J Mol Biol 375, 1444-1456. Saitoh, H., and Hinchey, J. (2000). Functional heterogeneity of small ubiquitin-related protein modifiers SUMO-1 versus SUMO-2/3. J Biol Chem 275, 6252-625 8. Salomoni, P., and Khelifi, A.F. (2006). Daxx: death or survival protein? Trends Cell Biol 16, 97- 104. 208 References Salomoni, P., and Michod, D. (2009). End of the Daxx diatribe? Cell Cycle 8, 1467-1468. Sandström, J. (1982). Dynamic NMR spectroscopy (London ; New York: Academic Press). Santiago, A., Godsey, A.C., Hossain, J., Zhao, L.Y., and Liao, D. (2009). Identification of two independent SUMO-interacting motifs in Daxx: evolutionary conservation from Drosophila to humans and their biochemical functions. Cell Cycle 8, 76-87. Sattler, M., Schleucher, J., and Griesinger, C. (1999). Heteronuclear multidimensional NMR experiments for the structure determination of proteins in solution employing pulsed field gradients. Progress in Nuclear Magnetic Resonance Spectroscopy 34, 93-158. Schmidt, F.R. (2004). Recombinant expression systems in the pharmaceutical industry. Appi Microbiol Biotechnol 65, 363-372. Schmitz, M.L., and Herrmann, H. (2008). Functional architecture of the cell nucleus. Biochim Biophys Acta 1783, 2041-2043. Sekiyama, N., Ikegami, T., Yamane, T., Ikeguchi, M., Uchimura, Y., Baba, D., Ariyoshi, M., Tochio, H., Saitoh, H., and Shirakawa, M. (2008). Structure of the small ubiquitin-like modifier (SUMO)-interacting motif of MBD 1-containing chromatin-associated factor 1 bound to SUMO-3. J Biol Chem 283, 35966-35975. Senn, H., Werner, B., Messerle, B.A., Weber, C., Traber, R., and Wuthrich, K. (1989). Stereospecific Assignment of the Methyl H-i -Nmr Lines of Valine and Leucine in Polypeptides by Nonrandom C-13 Labeling. Febs Letters 249, 113-118. Shen, T.H., Lin, H.K., Scaglioni, P.P., Yung, T.M., and Pandolfi, P.P. (2006). The mechanisms of PML-nuclear body formation. Mo! Cell 24, 331-339. Shen, Z., Pardington-Purtymun, P.E., Comeaux, J.C., Moyzis, R.K., and Chen, D.J. (1996). UBL 1, a human ubiquitin-like protein associating with human RAD5 1 /RAD52 proteins. Genomics 36, 271-279. Shiels, C., Adams, N.M., Islam, S.A., Stephens, D.A., and Freemont, P.S. (2007). Quantitative analysis of cell nucleus organisation. PLoS Comput Biol 3, e138. Shih, H.M., Chang, C.C., Kuo, H.Y., and Lin, D.Y. (2007). Daxx mediates SUMO-dependent transcriptional control and subnuclear compartmentalization. Biochem Soc Trans 35, 1397- 1400. Sikorski, T.W., and Buratowski, S. (2009). The basal initiation machinery: beyond the general transcription factors. Curr Opin Cell Biol 21, 344-351. Silverstein, R.A., and Ekwall, K. (2005). Sin3: a flexible regulator of global gene expression and genome stability. Cuff Genet 47, 1-17. 209 References Song, J., Durrin, L.K., Wilkinson, T.A., Krontiris, T.G., and Chen, Y. (2004). Identification of a SUMO-binding motif that recognizes SUMO-modified proteins. Proc Nati Acad Sci U S A 101, 14373-14378. Song, J., Zhang, Z., Ru, W., and Chen, Y. (2005). Small ubiquitin-like modifier (SUMO) recognition of a SUMO binding motif: a reversal of the bound orientation. J Biol Chem 280, 40122-40129. Song, J.J., and Lee, Y.J. (2003). Role of the ASK1-SEK1-JNK1-HIPK1 signal in Daxx trafficking and ASK1 oligomerization. J Biol Chem 278, 47245-47252. Song, J.J., and Lee, Y.J. (2004). Tryptophan 621 and serine 667 residues of Daxx regulate its nuclear export during glucose deprivation. J Biol Chem 279, 30573-30578. Song, M.S., Song, S.J., Kim, S.Y., Oh, H.J., and Lim, D.S. (2008). The tumour suppressor RASSF1A promotes MDM2 self-ubiquitination by disrupting the MDM2-DAXX-HAUSP complex. EMBO J 27, 1863-1874. Song, S.J., Song, M.S., Kim, S.J., Kim, S.Y., Kwon, S.H., Kim, J.G., Calvisi, D.F., Kang, D., and Lim, D.S. (2009). Aurora A regulates prometaphase progression by inhibiting the ability of RASSF1A to suppress APC-Cdc2O activity. Cancer Res 69, 2314-2323. Spronk, C.A., Tessari, M., Kaan, A.M., Jansen, J.F., Vermeulen, M., Stunnenberg, H.G., and Vuister, G.W. (2000). The Madl-Sin3B interaction involves a novel helical fold. Nat Struct Biol 7,1100-1104. Stehmeier, P., and Muller, 5. (2009). Phospho-regulated SUMO interaction modules connect the SUMO system to CK2 signaling. Mol Cell 33, 400-409. Sternsdorf, T., Jensen, K., and Will, H. (1997). Evidence for covalent modification of the nuclear dot-associated proteins PML and SpiOO by PlC 1 /SUMO- 1. J Cell Biol 139, 1621-1634. Stieglitz, B., Bee, C., Schwarz, D., Yildiz, 0., Moshnikova, A., Khokhlatchev, A., and Herrmann, C. (2008). Novel type of Ras effector interaction established between tumour suppressor NORE1A and Ras switch II. EMBO J 27, 1995-2005. Strasser, A., Jost, P.J., and Nagata, S. (2009). The many roles of FAS receptor signaling in the immune system. Immunity 30, 180-192. Su, X.C., and Otting, G. (2009). Paramagnetic labelling of proteins and oligonucleotides for NMR. J Biomol NMR. Swanson, K.A., Knoepfler, P.S., Huang, K., Kang, R.S., Cowley, S.M., Laherty, C.D., Eisenman, R.N., and Radhakrishnan, I. (2004). HBP1 and Madi repressors bind the Sin3 corepressor PAH2 domain with opposite helical orientations. Nat Struct Mol Biol 11, 73 8-746. Tam, C.W., Liu, V.W.S., Leung, W.Y., Yao, K.-M., and Shiu, S.Y.W. (2008). The autocrine human secreted PDZ domain-containing protein 2 (sPDZD2) induces senescence or 210 References quiescence of prostate, breast and liver cancer cells via transcriptional activation of p53. Cancer Letters 271, 64-80. Tang, J., Qu, L.K., Zhang, J., Wang, W., Michaelson, J.S., Degenhardt, Y.Y., El-Deiry, W.S., and Yang, X. (2006). Critical role for Daxx in regulating Mdm2. Nat Cell Biol 8, 855-862. Tang, J., Wu, S., Liu, H., Stratt, R., Barak, O.G., Shiekhattar, R., Picketts, D.J., and Yang, X. (2004). A novel transcription regulatory complex containing death domain-associated protein and the ATR-X syndrome protein. J Biol Chem 279, 20369-20377. Tatham, M.H., Geoffroy, M.C., Shen, L., Plechanovova, A., Hattersley, N., Jaffray, E.G., Palvimo, J.J., and Hay, R.T. (2008). RNF4 is a poly-SUMO-specific E3 ubiquitin ligase required for arsenic-induced PML degradation. Nat Cell Biol 10, 53 8-546. Tatham, M.H., Kim, S., Jaffray, E., Song, J., Chen, Y., and Hay, R.T. (2005). Unique binding interactions among Ubc9, SUMO and RanBP2 reveal a mechanism for SUMO paralog selection. Nat Struct Mol Biol 12, 67-74. Tavalai, N., and Stamminger, T. (2008). New insights into the role of the subnuclear structure ND 10 for viral infection. Biochim Biophys Acta 1783, 2207-2221. Teichmann, M., Wang, Z., and Roeder, R.G. (2000). A stable complex of a novel transcription factor JIB- related factor, human TFIIIB5O, and associated proteins mediate selective transcription by RNA polymerase III of genes with upstream promoter elements. Proc Natl Acad Sci U S A 97, 14200-14205. Thomas, M.C., and Chiang, C.M. (2006). The general transcription machinery and general cofactors. Crit Rev Biochem Mo! Biol 41, 105-178. han, S., Poukka, H., Palvimo, J.J., and Janne, O.A. (2002). Small ubiquitin-related modifier-i (SUMO-1) modification of the glucocorticoid receptor. Biochem J 367, 907-911. Toni, S., Egan, D.A., Evans, R.A., and Reed, J.C. (1999). Human Daxx regulates Fas-induced apoptosis from nuclear PML oncogenic domains (PODs). EMBO J 18, 6037-6049. Ubbink, M. (2009). The courtship of proteins: understanding the encounter complex. FEBS Lett 583, 1060-1066. Ulrich, H.D. (2008). The fast-growing business of SUMO chains. Mol Cell 32, 301-305. Uversky, V.N. (2009). Intrinsically Disordered Proteins and Their Environment: Effects of Strong Denaturants, Temperature, pH, Counter Ions, Membranes, Binding Partners, Osmolytes, and Macromolecular Crowding. Protein J. Uversky, V.N., Oldfield, C.J., and Dunker, A.K. (2008). Intrinsically disordered proteins in human diseases: introducing the D2 concept. Annu Rev Biophys 37, 2 15-246. 211 References van der Weyden, L., and Adams, D.J. (2007). The Ras-association domain family (RASSF) members and their role in human tumourigenesis. Biochim Biophys Acta 1776, 5 8-85. van Ingen, H., Baltussen, M.A., Aelen, J., and Vuister, G.W. (2006). Role of structural and dynamical plasticity in Sin3: the free PAH2 domain is a folded module in mSin3B. 3 Mo! Biol 358, 485-497. van Ingen, H., Lasonder, E., Jansen, J.F., Kaan, A.M., Spronk, C.A., Stunnenberg, H.G., and Vuister, G.W. (2004). Extension of the binding motif of the Sin3 interacting domain of the Mad family proteins. Biochemistry 43, 46-54. Vavvas, D., Li, X., Avruch, J., and Zhang, X.F. (1998). Identification ofNorel as a potential Ras effector. J Biol Chem 273, 5439-5442. Vertegaal, A.C., Andersen, J.S., Ogg, S.C., Hay, R.T., Mann, M., and Lamond, A.I. (2006). Distinct and overlapping sets of SUMO-l and SUMO-2 target proteins revealed by quantitative proteomics. Mo! Cell Proteomics 5, 2298-2310. Vogel, C., Bashton, M., Kerrison, N.D., Chothia, C., and Teichmann, S.A. (2004). Structure, function and evolution of multidomain proteins. Cuff Opin Struct Biol 14, 208-216. Wang, H., Clark, I., Nicholson, P.R., Herskowitz, I., and Stiliman, D.J. (1990). The Saccharomyces cerevisiae SflT3 gene, a negative regulator of HO, contains four paired amphipathic helix motifs. Mol Cell Biol 10, 5927-5936. Wang, W., Kwok, A.M., and Chan, J.Y. (2007). The p65 isoform of Nrfl is a dominant negative inhibitor of ARE-mediated transcription. J Biol Chem 282, 24670-24678. Wang, Z.G., Ruggero, D., Ronchetti, S., Zhong, S., Gaboli, M., Rivi, R., and Pandolfi, P.P. (1998). PML is essential for multiple apoptotic pathways. Nat Genet 20, 266-272. Wei, W., Yang, P., Pang, J., Zhang, S., Wang, Y., Wang, M.H., Dong, Z., She, J.X., and Wang, C.Y. (2008). A stress-dependent SUMO4 sumoylation of its substrate proteins. Biochem Biophys Res Commun 375, 454-459. Wethkamp, N., and Klempnauer, K.H. (2009). Daxx is a transcriptional repressor of CCAAT/enhancer-binding protein beta (C/EBPbeta). J Biol Chem. Wilkins, M.R., Gasteiger, E., Bairoch, A., Sanchez, J.C., Williams, K.L., Appel, R.D., and Hochstrasser, D.F. (1999). Protein identification and analysis tools in the ExPASy server. Methods Mol Biol 112, 53 1-552. Wishart, D.S., and Sykes, B.D. (1994). The 13C chemical-shift index: a simple method for the identification of protein secondary structure using 13C chemical-shift data. J Biomol NMR 4, 171-180. 212 References Wishart, D.S., Sykes, B.D., and Richards, F.M. (1992). The chemical shift index: a fast and simple method for the assignment of protein secondary structure through NMR spectroscopy. Biochemistry 31, 1647-165 1. Wohigemuth, S., Kiel, C., Kramer, A., Serrano, L., Wittinghofer, F., and Herrmann, C. (2005). Recognizing and defining true Ras binding domains I: biochemical analysis. J Mo! Biol 348, 741-758. WUthrich, K. (1986). NMR of proteins and nucleic acids (New York: Wiley). Xue, Y., Gibbons, R., Yan, Z., Yang, D., McDowell, T.L., Sechi, S., Qin, J., Zhou, S., Higgs, D., and Wang, W. (2003). The ATRX syndrome protein forms a chromatin-remodeling complex with Daxx and localizes in promyelocytic leukemia nuclear bodies. Proc Natl Acad Sci U S A 100, 10635-10640. Yang, S.H., Galanis, A., Witty, J., and Sharrocks, A.D. (2006). An extended consensus motif enhances the specificity of substrate modification by SUMO. EMBO J 25, 5083-5093. Yang, X., Khosravi-Far, R., Chang, H.Y., and Baltimore, D. (1997). Daxx, a novel Fas-binding protein that activates INK and apoptosis. Cell 89, 1067-1076. Yeh, E.T. (2009). SUMOylation and De-SUMOylation: wrestling with lifes processes. J Biol Chem 284, 8223-8227. Yeung, P.L., Chen, L.Y., Tsai, S.C., Zhang, A., and Chen, J.D. (2008). Daxx contains two nuclear localization signals and interacts with importin alpha3. J Cell Biochem 103, 456-470. Zhang, J., Cado, D., Chen, A., Kabra, N.H., and Winoto, A. (1998). Fas-mediated apoptosis and activation-induced T-cell proliferation are defective in mice lacking FADDIMort1. Nature 392, 296-300. Zhao, L.Y., Liu, J., Sidhu, G.S., Niu, Y., Liu, Y., Wang, R., and Liao, D. (2004). Negative regulation of p53 functions by Daxx and the involvement of MDM2. J Biol Chem 279, 50566-50579. Zheng, G., and Yang, Y.C. (2004). ZNF76, a novel transcriptional repressor targeting TATA binding protein, is modulated by sumoylation. J Biol Chem 279, 42410-4242 1. Zheng, N., Wang, P., Jeffrey, P.D., and Pavietich, N.P. (2000). Structure of a c-Cbl-UbcH7 complex: RING domain function in ubiquitin-protein ligases. Cell 102, 533-539. Zhu, J., Zhou, J., Peres, L., Riaucoux, F., Honore, N., Kogan, S., and de The, H. (2005). A sumoylation site in PML/RARA is essential for leukemic transformation. Cancer Cell 7, 143- 153. Zhu, J., Zhu, S., Guzzo, C.M., Ellis, N.A., Sung, K.S., Choi, C.Y., and Matunis, M.J. (2008). Small ubiquitin-related modifier (SUMO) binding determines substrate recognition and paralog-selective SUMO modification. J Biol Chem 283, 29405-29415. 213 Appendices Appendices Appendix 1 Circular dichroism (CD) spectra of DAXX34742° Appendix 2 Expression tests of DAXX constructs Appendix 3 Annotated ‘H-’5N HSQC spectra Appendix 4 Calculation of dissociation constants in NMR-monitored titrations Appendix 5 Dissociation constant plots from SIM/SUMO NMR-monitored titrations Appendix 6 Exchange regimes Appendix 7 Assignment of the DHB-domain Appendix 8 Assignment ofRassflC2338”’in the DHB/RassflC2338”complex. Appendix 9 Assignment of the DHB domain in the DHB/RassflC2338”complex Appendix 10 Triangulation of the spin label 214 Appendix 1 Circular dichroism (CD) spectra of DAXX347-42° Appendix 1 — CD of DAXX34742° S I. 1 190 200 210 -2 -7 A I S C.? — C.? -12 47 -22 I. B Wavelength (nm) -8 -8.5 220 230 240 250 260 -9 -9.5 -10 -10.5 -11 10 20 30 40 50 60 70 80 90 Temperature (°C) (A) CD spectrum of DAXX34742°(10 tiM, 10 mM phosphate, pH 6.5, 100 mM KC1, 0.1 mM EDTA, 25 °C, path-length 0.1 cm). The spectrum demonstrates that this construct was predominantly disordered, yet may adopt a small fraction of secondary structural elements. In particular, the CD signal at —222 nm is diagnostic of x-helical conformations. (B) Thermal denaturation of DAXX34742°monitored by CD spectropolarimetry. The loss of CD signal at 222 nm with increasing temperature also confirmed the presence of a small fraction of helical secondary structure. 215 Appendix 2— Expression test of DAXX constructs Appendix 2 Expression tests of DAXX constructs Summary of all DAXX constructs used for expression tests. The solubility results are summarized in Figure 2.3. The red star highlights the resulting DAXX fragment. In cases where multiple colonies of the same construct were expressed, only the colony used for DNA sequencing is labeled. The protein markers (M) had the following masses: 14 kDa, 18 kDa, 25 kDa, 35 kDa, 45 kDa, 66 kDa, 116 kDa. The lanes are labeled with the following code: un = uninduced 3h = three hours after induction with 1 mM IPTG. 5h = five hours after induction with 1 mM IPTG. 9h = nine hours after induction with 1 mM IPTG. on = overnight (approximately sixteen hours after induction). t = total (soluble + insoluble). s = soluble. i insoluble. EtOH = 5% glycerol, 2.5% ethanol added to the media. The methodology for preparing samples of each lane is described in section 2.6.2. 216 Appendix 2— Expression test of DAXX constructs Identifier A B C I) E F C H I J K NI N 0 P Q R S T I-I 1 w x 1-160 46-160 46-420 55-243 161-243 161-342 161-420 181-335 181-365 181-385 181-407 *151.365 *151.407 206-335 206-365 206-385 206-407 246-342 261-420 261-342 279-342 347-420 rn566-739 47-560 4—, +— > ÷—* < . 4-> 4-> 4-—, ( 740 4. <- 217 Appendix 2— Expression test of DAXX constructs — ‘—I - - — a - — — - — - --- - — — un M 6ht 6hs 6hi ont un 6ht 6hs 6hi ont A 1-16030°C B 46-16030°C un M 3ht 3hs 3hi C 46-420 —— — -r ———“—, —=—=----- un Sht M 5hs ont ons oni Shi 5hs ont ons oni ont ons oni C 46-42022 C C 46-420 15 °C C 46-420 15°C EtOH —:_. — S——I—— fA —=-‘i un Sht Shs M ont ons oni Sht Shs ont ons oni ont ons oni G 161-42022 °C G 161-420 15 °C G 161-420 15°C EtOH a — s. • .--- 4I_*: . I l’L r un M Sht Shs 5h1 un Sht Shs Shi un Sht 5hs Shi E 16 1-243 30°C D 55-24330°C F 161-34230°C -Iti M on ont ons oni ont ons oni F 161-342 18°C F 161-34222°C un 3ht 3hs 3hi M G 161-42030°C 218 Appendix 2 — Expression test of DAXX constructs - — I — —— •1 — uii 9ht M 9hs 9ht 9hs 91i1 9h 91i1 9hs ‘Jul ‘JIts 9ht 9hs P 206-385 30 C — un M 9ht 9hs 9ht 9hs 9ht 9h 9ht 9hs 9ht 9hs 9ht 9hs J 181-385 30 °C K 181-407 30°C N 206-335 30°C 0 206-36530 °C 91i( 9Iis 1 ‘liii gins SI \I—’ ,nsIrIIi 9ht Ohs 9h1 9ht 9I I. 151-365 I 151-407 I) 2116-407 I’ 2116-355 311 ( 30 ( 311 311 219 Appendix 2— Expression test of DAXX constructs ont OflS Ofli tin ont Ons Ofli ont ons Ofli T 261-342 22°C D 55-243 18 “C D 55-243 22 “C — — — - * un M Sht Shs Shi un Sht Shs Shi un Sht Shs 5h1 R 246-342 30°C T 261-342 30 “C U 279-34230 “C W - * un Sht M Shs Shi un Sht 5Iis Shi V 347-42030 “C S 261-42030 “C 4—--- — ::: iz Iz ‘:‘ -* * — — ‘T un M ont ons oni T 261-342 18°C — e M tin 3ht 3hs 3hi X 47-56030°C : ont un ons M oni ont ons S 261-420 18 “C S 261-420 22 “C 220 Appendix 3 — Annotated spectra of SUMO/SIMs Appendix 3 Annotated1H15N HSQC spectra (A) SUMO-1 (155 j.iM, 17°C) (B) SUMO-2 (1.1 mM, 25 °C) (C) SIMNw (120 jiM, 25 °C) (D) SIM-C’’ (125 jiM, 25 °C). All samples were in NMR sample buffer (10 mM phosphate, 100 mM KC1, 0.1 mM EDTA, pH 6.5). The signals from Asn and Gln side-chain ‘5NH2 groups are connected by horizontal lines. Aliased peaks are shown in red. 221 I z 105 Appendix 3 — Annotated spectra of SUMO/SIMs A A72 110W 1 115 z 120. 125 ‘G68 I66 Q29 - R63 K25 M40 C-T76 , DI 0G14 Q92 / .1, K16t4.E$ 0G19 G56N74 F64 Y51 - /. 21(17K39— MV S31 D< E11 QG96 V57— 1.24 / 134 1--Y9i’ L13 G28 N60 •S61 0 !?.‘ -0 T98 K78\...G97 ••\Q55 cJL8O tj OT’o ‘sso izz OS9os2 V87 0C52 S32 M82 0 K48 E5 k54 V9O 0117569 -OT42 E79 L62 L.L47 0 ;..L6S 0 ci 0 L44 MR* ,. -....; E89 (1435 E93 0A6 - :1143 l27 T4l 0 E67 110 115 (.27 T7’B -j.G24 2 G52 @D63 .G9 .A23 C92 -2-.H37 0Q51 -. % R56 -- — - p ..S54 _.__ . — QG93 158 S41J1 0Q57 01311 ,)2(.._fI2 - —iM•78 _______ C48. 74fl)7_S L76 1’8 01)80 ..R36 - 1(45 L$3 V86 ‘ vio 1F60 y47 K21 -. N14 1)16 R K42 —162 - N15 . .-.. - F31Q25 . :Rs9 0 -Q(. K61 C 42 Q)i8 MI 0 446 - L4 L40.. OQ9*_j3” (1)7J -N68 cGU Q89 .F87 —A2 Q31 0 KU 0 184 V30 120 125 9.5 9.0 8.5 8.0 7.5 7.0 6.5 ‘H (ppm) 222 Appendix 3 — Annotated spectra of SUMO/SIMs C ON5N-H 115 NS I 120 D13+—D12Z 9D14 1)1I.E15 A19 125 0W20 :A44 . 18 ‘vQ LIO 8.5 8.0 7.5 7.4) Q724N-IF 0 Q726N.FfD . 115 S738 T721 0 S736 S722 120 0 E730E QC727 ‘7Z3 0E731 Q726 -1732 V719 125 V734 1733 .JA72.l L735 c 1)739 8.0 7.5 7.0 ‘H (ppm) 223 Appendix 4 — Dissociation constants Appendix 4 Calculation of dissociation constants in NMR-monitored titrations For a protein binding a ligand, the equilibrium can be written as: P+LPL (1) where the dissociation (KD) and association (Ka) constants are defined as: — [P][L] — [PL] KDli Ka (2) [P], [L], and [PL] indicate the molar concentrations of the unbound protein, unbound ligand, and bound protein-ligand complex. In titrations that occur in the fast exchange regime (Appendix 6), the chemical shift (ö1) of a given nucleus at each titration point (I) is the population-weighted average of its chemical shifts in the free () and fully bound (sb) species. Thus, its chemical shift can be expressed in terms of the fraction of the bound species (fbI, where fbI + ffi 1) as: fbl = (3) or = fbi (b — f) + (3a) This fraction can also be expressed in terms of concentrations of the free ([Pf]) or bound ([PL]) species: — [PL] fbi — [Pf]+[PL] (4) 224 Appendix 4 — Dissociation constants Experimentally, we know the total, rather than free or bound, concentrations of protein and ligand. The total amount of protein ([Pt]) and ligand ([Li]) can be expressed as the sum of the free and bound species: [Pt] = [Pr] + [PL] (5) [Lb] = [L1] + [PL] (6) The above equations can be combined to derive a quadratic equation for the fraction of bound species at each titration point: from equation 2: [PLI = [P1[LIKa Ka[Pf][Lf] Ka[Lfl [Lf] substituting into equation 4 f = = = 1 (7)[Pf]+Ka[Pf][Lf] 1+Ka[Lf] —+[Lf] from equation 6 [Lf] = [Lb] — [PL] substituting into equation 7 fbi = ±[Lb[PL1 (8) substituting equation 5 into equation 4 fbi = or [PL] = fbi [Pt] substituting into equation 8 fbi = = 1 [Ltlfbi[Pt] = Ka([Ltbfbi[Ptl) +[Lt]fbi [Pt] (1+Ka[LtbKafbj[Pt]) 1+Ka[Lt]Kafbj[Pt] which can be rearranged into the quadratic equation: fbi + fbiKa[Ltl — Kafbi2[Pt] = Ka([Lt] — fbi[Pt]) —fbi — fbi’<a[’t] + Kafbi2[Pt] + Ka[LtbfbiKa[Ptl = 0 225 Appendix 4 — Dissociation constants ([Pt])fb2 + (—KD — [Lv] — [Pt])fb, + [Lb] = 0 (9) Quadratic equations have the solutions: ax2 + bx + c 0 —b±gb2—4ac x= (10) 2a Using the solution that will give a number between 0 and 1: — —(—KD—[Lt]—[Pt])—,J(—KD—[Lt]—[Pt])24([Lt] [Pt]) fbi — 2[Pt] Substituting into equation 3a: — KD+ [Lt]+[Pt]—lJ(KD+[Lt]+ [Pt])2—4([Ltl [Pt]) ,.- — 2[P] b5’f)+( (12) Finally, to account for dilution during the titrations, [Pt] is substituted by PV/(x+V), [Li] by Lx/(x+V), where V is the initial volume (in j.tL) in the NMR tube, P is the initial concentration of the analyte solution in the NMR tube (in M), L is the initial concentration of the titrant (in M), and x is the volume of titrant added (in jiL) to give the following equation: = (öb — + (13) x+V For the KD calculations reported in Chapters 3 and 4, a plot of ö versus volume (x, in iL) was used. In all plots, ö was calculated using J(L8IH )2 + (O.2AöI5N)2 so was zero. Sigma Plot was used solve equation 13 for KD and i non-linear least squared fit using initial concentrations 226 Appendix 4 — Dissociation constants of protein (P), peptide (L) and initial volume (V). Initial estimates for the variables KD and b were provided by inspection. The R2 of the fit was usually greater than 0.95. The reported errors are standard deviations of independent fitting of signals. 227 Appendix 5 — Dissociation constant plots for SIM/SUMO Appendix 5 Dissociation constant plots from SIM/SUMO NMR-monitored titrations The dissociation constants reported in Table 4.1 were derived using data from the titrations shown in Figures 4.3 and 4.4. Residues used for the calculation of each KD are summarized in the following plots. Equation 13 from Appendix 3 was used for each regression, using protein and peptide concentrations summarized in Table 4.3. The plots for ‘5N SUMO-1” and SUMO-2’’ with unlabelled SIM-N’’ showed stoichiometric binding, so only an upper estimate for the KD was obtained (Table 4.1). 228 Appendix 5 — Dissociation constant plots for SIM/SUMO SIMCw V - A-- - A Ov A -- 4 8 12 16 20 24 SUMO-2” :SIM-C” ratio + SUMO-V + SUMO-2’ 1 2 3 4 1 2 3 4 S 0.3 S ‘& 0.2 0.1 2 4 6 8 2 4 6 • 0739 V D737 G E730 c.) E — E (•l • 0739 I V 0737 • S736 L735 A E730 0728 0.1 V A V V A 3 6 9 12 SUMO-1”:SIM-C ratio 229 Appendix 5 — Dissociation constant plots for SIM/SUMO SUMO-F’ +SIM-N’ SUMO-2’ +SIM-N’ VV V • 27 v H35 O C52 O H75 A K48V • V30 I I v F87 I I I] C48 II <> E81 I I A 184 I • N68 A A • 0 A 0 I V V U V0 V SIM-N”:SUMO-1’ ratio + SUMO-1” SIM-N’ SIM-N” :SUMO-2” ratio + SUMO-2” 0.15 E 0.10 0.05 0.3 E 0- 0.2 0- 0.1 0.3 0.2 0.1 c.) E c-) E C-) E 2 4 2 4 2 4 4 8 12 SUMO-i”:SIM-N” ratio SUMO-2’:SIM-N” ratio 230 Appendix 6— Exchange regimes Appendix 6 Exchange regimes Equation 1 from Appendix 4 can also be written in terms of rate constants: PL (14) where the dissociation (KD) and association (Ka) constants can be re-written in terms of rate constants as: L’ k0ff , k01 on off The apparent exchange rate is defined as: kex = [L] k011 + k0ff (16) In a typical NMR-monitored titration experiment, only the protein or the ligand is labeled. As an example, assume that the labeled species is the protein, and that only one signal is being observed. In this case there will be an observed resonance frequency for the nucleus in the free protein (v1) and a different chemical shift for the same nucleus in the bound species (Vb). The difference in frequencies between the two protein states is: VVb—Vf (17) The exchange regimes of binding events observed by NMR are defined as: slow exchange if kex< öv intermediate exchange if kex fast exchange if kex> öv 231 Appendix 6— Exchange regimes The lineshape of the peak resonance is different for each exchange regime. The equation describing the lineshape of a resonance undergoing a 2-site exchange process is described in (Sandström, 1982): + T-+--+ QR\Tzf T2b1 1(v) = C0 P2+R (18) where 1 \TzfT2b J T2 Tzb Q =r(2v—v(ff—fb)) R = IXv =--—v 2 = = kon(KrJ+Lf) Co is a normalization constant proportional to the total protein concentration, T2f and T2b are the apparent transverse relaxation times for the nucleus in the free and bound protein states, respectively. ff and fb are the fraction of free and bound protein, respectively. Thus, fb = (equation 11) 2[Pt] ff=l—fb (19) Lf=Lt—fbPt (20) 232 Appendix 6— Exchange regimes K=1nM k = 1e8 M’s’ P,= 100 tM T,1=0.01 S T,,= 0.01 S Vr = 100 Flz v0 = 160 Flz ThO0 0000 2000 K,, = 1.M k,, = 1e8 M’s’ P,= 100 tM T,0= 0.01 S T,0.01 s = 100 Hz v0=l6OHz K1, = 100 jiM k, = 1e8 Ms P=lOOjiM T00ls r0= 0.01 S v=lOOHz 0=l6OHz Intermediate Fast Using equation 18 with Co = -1 e6, and assuming k0 to be diffusion controlled (1 x 108 M’s’), the exchange regime is determined by the lifetime of the bound state (as reflected by k0ff) and v. The interplay of these two terms is shown in the following simulations (using KD = k0ff/k = k0ff/1 X 108 M’s’). Different KD values: Slow 280 300 233 Appendix 6— Exchange regimes Different v: 10000 7500 5000 I) 2500 K{)= 1pM k,= le8M’s’ P= 100pM T2f= 0.01 s T,= 0.01 S V1 = 100 Hz = 180 Hz K0= 1pM k0,, = 1e8 M’s’ P= 100pM T,1= 0.01 S T7, 0.01 S v1=lOOHz v1= 160 Hz K0= 1pM = 1e8 Ms’ = 100 pM T,1=0.01 s T2,= 0.01 S 100 Hz Vb = 120 Hz 280 300 260 280 300 280 300 234 Appendix 6— Exchange regimes In addition, T2 values typically decrease in complexes due to slower correlation times for molecular motions. The effect of changing T2 values is shown below for fast exchange: l0000 7500 5000 0 0 K0= 100 tM k0 = 1e8 M’s’ P1 = 100 tM T,1= 0.01 S T2b= 0.01 S v1 = 100 Hz 160 Hz K0 = 100 jiM 1(00 = 1e8 M’s P, = 100 jiM 1,1=0.01 s T,1,= 0.005 s v. = 100 Hz V5 = 160 Hz K0= 100 jiM k, = 1e8 M’s’ P1= 100 jiM T,1= 0.01 s T, = 0.0025 S v1 = 100 Hz ‘i5=l6OHz --- 200 300 10000 7500 s000 2500 0 2043 1-’ 300 200 235 Appendix 6— Exchange regimes The above simulations are for a simple two-state conformation equilibrium. In the case of conformational exchange involving both ligand binding and two possible bound orientations (PL and PU), a general three-state equilibrium can be written as: P + L (jk) PU (21) If the ligand is in sufficient excess to ensure complete saturation, only the resonances from the nucleus in the PL and PL’ states will be observed. The lineshape of these signals is described by equation 18 as well (i.e. 2-state exchange between bound conformations). However, the apparent exchange rate is now independent of ligand concentration (kexki+ki). Assuming a SvpLpL’ = 100 Hz, T2pL=T2pL’=0.05s, PpL=PpL’=O.5 (thenk1), the following graph shows the effect of different values for k1. 50000 >40000 •k=lOs 30000 ik=lOOs” 20000 • k, = 200 ‘ 10000 • k = 400 s • k = 1000 s • k, = 10000 s 0 0 280 236 Appendix 6— Exchange regimes Therefore, if the interconversion between bound conformations occurs in the intermediate exchange regime with kex övpLpL’, the signals from the given nucleus in the saturated complex can be extremely broadened and thus difficult to detect. 237 (J (JI U (Il J1 (J (JI (JI (Il fJ I U i U i U i U i U i U i U i U i U i (1. U i U i U i U i U i U i (I l U i U i U i U i a o C V i V i U i U i ii . . . . . . X X X X C )) C )C ) X X C )C ) X X X X X C )C )C )C )C )X X X X X C ) C ) C ) X C ) t r j X 2 : : O > Z Q C > O C )t i X . C 0 — — - 00 - . ) . 00 00 — — V i 00 c 00 k ) - — . 1 L ) V i \ U I V i U i V i \ U I V i U I \ X t!j V i X ni * U I c X U I ‘ .c X * V i 00 V i (‘I 00 X X x z . U i 00 C) D U I 00 C) nil * U I 00 p 00 U I 00 C) * p 00 . C) . — V i U I — U I 00 C) > V i J 00 V i X 2: 2: U I — i — I X * : - c 3 U I — i X > V i — 1 C) 0 — U i — I C) X PC I’ J U I — C) > J . V i — — a — I t’. ) V i — V i . c i-a 00 2: U I - . l o c — a X C) V i - 4 C) 00 I’ ) , ‘ a 00 00 tJ I - 4 X rn U i - 4 V i Q X nii U I — I - - a V cM — I X 0 * V i - 4 X * 0’ i-a c L. ) V i 0 V i c V i LM X V i c I-a J 00 C) 0 00 cJ . U I 0’ C) C) b LM c j— a — C) tn c - e C) %1 - a — a - a C) n-a —a Vi — C) > c, 2: X 0’ C) 0 X 2: c — C) C) X C) * a — I-a V i C) — I — a — c a i— a 00 0 C) c a - a C ’ I— ) c I— ) a a 2: C) C- ’ — C- ’ a , - ‘ - C V i p-’ . ) I’ ) * a c c 00 a X * I-a 00 V i — I 0 — a la C © X C I- a 9’ t-a ‘-C C ’ X t- ) C I-- a . > C ’ 9 ‘-C C- ’ © a -’ C) 0 a- ’ 00 i-a C) C) 00 ‘ ‘ - C C) i-a a- ’ 9 00 ‘-C C) a- ’ C) 9 ‘-C . C ’ C) > c M — I C ’ — a i— a C ’ © C ’ 2: C ’ © X C) a- ’ = — a ‘ c ‘ - C . X C) i-- a - 4 . i-- a a- ’ C ’ nil * C- ’ i-- a a’ Vi = D ‘— - 0 — . GO — 0 -o M - . 0 — . — I _ j l J i o p ; GO ‘ — ‘ CD . D CD GO ,z G O C GO Z - . ( D • - i - I ‘ - ‘ - ) — 0 ( Cl ’ G O i 0 c _ C CD C )-— E . © C l GO - ‘ p - - I’- ) 00 V i C ’ X . - a — a lJ I ‘ - X V i 00 C ’ © X I— a a ‘- C ‘ - C > - C ‘ - 0 U I 0 U C C) 0 l- ) C) C) C) nil 1— U) ‘ - C ‘ - C C) 0 cM 00 C) — a i-rn a C) GO C ’ cc C a c0 00 2: — a 00 i- a C ’ I— ) V i C a - l— ) I-- a c j I-- a C — a J a — I ‘ - C C ’ c0 C I-- a 9’ . ‘-0 00 - C C ’ C ’ C ’ a a a - C ’ C ’ C ’ C ’ C ’ C ’ a C ’ C ’ C ’ C ’ C ’ C ’ C ’ C ’ a C ’ C ’ C ’ C ’ C ’ C ’ C ’ C ’ C ’ C ’ C ’ C ’ C ’ a U I V i V i V i V i V i V i Li i V i - . (. I. e _ i i-e t- a I4 14 14 X X C )C )C )C )C )C ) X X X X C )C )C )C ) 2 : 2 : X X > 0 C ) t i l D Z C 2 C ) — I i- ) — a ‘ o — — a c — 00 - i- ) - C - a 00 00 00 c C C — a t- — — a - . a - CD C ) GO GO ii. 00C‘I,C Cl)Czrj,CCCl)CSC-C Cl)SC SCCCl)SCl)Cl)N-e c N N C C — SC N 00 N — N c c i c s c 00 S c SC S c 00 C — ‘ r SC c c c c S c — N — . . . . N S c C l c . S c N C l Ifl Il) If) If) SQ SQ SQ SQ SQ SQ SQ SQ S c SO SQ SQ SQ N N N N N N N N N N N N N N 00 00 00 00 00 00 00 00 00 Q N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N .SQQ .c Sr ccN N00r c c ri .Clr’ Nr00 (1l)cc .<C) 00., C) C’) - LC) Scc()c’ CC) -Clci < -NC 4 Nc ’ N *, - *c z N- SCSQ N tN N00 <C) t”N .N C.) tN C) SQ N r . C) ScClN rr e’I N LC.) N cC’) Sc CC) eqN NrSCc4 .‘)ccN * r ’ N eqN c .. NC ’) * I)N *r ’ Nor r ’ N SCc’ L’x N z cr -SC e .)N t’N .SC <C) t ,N SQrC’) C) N C)Cc CC) N Nr’i . < C l c c N SCScc.’ x SQ00 .00N SC N z cNC ’) N 00.l- ccc. c00 00SQ 00Scr’) .N C) . ‘)N .NCcN SQ C) - . N N . .C)r’ C) - . N Clc’ ScNr. C) c ’ c z .N z c:C) ..r)r’ .N CC) . . N C- <C) x S c 00 =C) C ’) c ’ * InN cC) N ‘l- NSC x InN CC) SQ00C’) InN If) c c m < 1)x InN cC - C ’) x U) 00SQ If) N x c c C ’) NN If) N II) Nc c C-q c -I- 0SQ 00SQ c,x InN OSO z N IS) N zx O \ SQ SCC)C” InN cSO 00SC InN <C) 0’ SQ 0’r Sc 14) N C-) C ’) C)Sc C ’ SQ If) N *C) C ’ SQ C ’) C)SC *C l) C) C ’ 50 SC00 CC) C ’ SQ c c S c NC-) ScIlC” 0’SQ <x .NNIf) * N .C). C ’ SQ * NNIl) C” N r - x N C)C’)(‘4 *Cl) 0000ScC” If) C ’) ccN C”U) N ©N N z NIf) C ’) N ©N 00SQ N . C) <C-) N C ” N N 00IfC C) N cC) C ’) Cl (‘4CC) N C”C)cc C”C” CC) ©N N <x N ©N x *Cx N N *C”Cx N z c:’x z —N N N N N —N N N C ’) 00 C ’) 00 C ’) N N — N (‘1 c c If) Sc C ’) • C ’) If) 00 N (‘1 c c — C 00 If) — C ’ If) — (‘1 C-I N C C ’) C’i — c c f) c s c c . Cl C c c C l C l . (‘4 — — N C -I — — C C — c 0 ) C ’ ) N C - I ’ C ’ ) N N N N c 0 N — N 1 r r 4 N JN N 00 00 00 SQ Sc SQ SQ SQ SQ 00 00 00 00 00 SQ SQ SQ Sc Sc 00 — C - — c N N C N N i— N .C C N - C . 00 N N 0 — 00 C ) C ‘ r • N n 0’ 00 C) N 0’ 8 e 8 8 © — — — — — — — — — — — — r ci ci c i c i ci ci ci r 0’ 0’ 0’ 0’ 0’ 0’ 0’ 0’ 0’ 0’ 0’ 0’ 0’ 0’ 0’ 0’ 0’ 0’ 0’ 0’ 0’ 0’ 0’ 0’ 0’ 0’ 0’ 0’ 0’ 0’ 0’ 0’ 0’ 0’ 0’ 0’ 0’ 0’ 0’ 0’ 0’ 0’ ..<r—00C—C-iCD00 rN00\O00’t00 00rr)I00C—0zt,00 crN*C - 00.\C\C<C).00 0000C’) C)00 1 ) C—NCDC).00 0’C’) C l C l CDC)00 00cC- CC)00 c0 . C ’) .<e00 N00 n00<C)0000C’)*CDx.00 -N.C)0000NCl*NCD.00 .QNC)0000C’)00N 0’\ONNC)0000 00 NCl00NCDC)0000C’)C0’\C<C)If)00 C000NCC)0000.0NC)If)00 .C<0000C—) CC l CDC)I,)00 CN000000Ifl NNCDC)If)00 N000000N00C’) <xI,)00 If) *x0000C’)CNIf)00 C ’) *N0000000. C*CDIn00 r—00CD00GOIf)00C*ClCDIn00 .0000GO00GozIn00 ‘ .000Nz0000 r-—If) N.0C)0’00c00If)c<C)00 00C’) C)0’00NCC’)C)00 -C’) CC) *C)0’00N0’.C)o00 N00*C)a00If)0):;NCDC)\00 0’cNNCC)0’00‘.00’0NCC)‘.000 If) , . 0’00 NIl-i --<‘.000 CC l CC) 0’00 C l )-I‘.000 NC0’00 0 ’ 0‘ .000 If) ClN*0’00000.C’)‘.000 If) ClNN0’000’If)C’)Cx‘.000 0000‘.*0’00 ‘ .0ClNCDx‘.000 0000‘.0N0’00 ‘ .0NCDx‘.000 00000’00 ‘ .0.<C)N00 .‘ . 00z0’GOIf)0’C’).C)NGO -000If) <C)©0’C)C—iC’)*CC)NGO C ’) 0000C’) C)0’NC’)*C)N00 NC ’) ‘.6NCDC)0’C00NCC)N00 NC - NCC)=0’00N0NNC)N00 ‘ .0C’) NNC—N00 00CN*NGO 00CNNi;NGO CIf) NNGO ‘ .0C’? NNC - GO 0.c zN00 00)Nz00 00’zIn00 C)1)CGo I CC.0GdCInS0.C. .0GdSC‘IInSC.C. .0GdSC0?In0?SC.C. .0GdSC0?•0In0? 0’ 0’ 0’ 0’ C) © C) © C) C) — — — N N N N GO 00 00 00 00 00 90 00 00 00 ‘ .0 0’ C r ’) 00 — If) ‘ .0 ‘ .0 C ’) 0 0 C l C If) C ’ 00 C l — 0’ — — C 00 N N ‘ . . ‘ .c ‘0 00 N 0 O C C l 0 O ‘ ) r ’ ) r ’ ) N N C -I N O < Z O < Z 00 00 00 00 ci c i c i N C l C l C l N C l C4 CC) CC) C’) C’) C’) C’) 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 Appendix 7— Assignments of the DHB-domain Residue Atom Shift (ppm) Residue Atom Shift (ppm) 93 HB* 2.06 93 HE22 6.78 93 HG* 2.22 93 RN 8.08 93 N 115.46 93 NE2 114.03 94 CA 55.49 94 CB 31.26 94 CD 43.47 94 CG 27.70 94 CO 175.41 94 HA 4.31 94 HB 2.11 94 HB’ 1.63 94 HD* 3.12 94 HG’ 1.76 94 HN 7.26 94 N 115.56 95 CA 51.85 95 CB 19.93 95 CO 176.10 95 HA 4.50 95 HB* 1.39 95 HN 7.17 95 N 123.09 96 CA 57.80 96 CB 32.29 96 CD2 118.46 96 CEI 139.17 96 CO 177.66 96 HA 4.34 96 RB 3.15 96 RB’ 3.07 96 HD2 7.21 96 HEI 7.84 96 HN 8.68 96 N 119.26 97 CA 62.47 97 CB 62.47 97 CO 177.24 97 HA 3.95 97 HB* 3.94 97 RN 8.90 97 N 124.00 98 CA 58.12 98 CB 41.40 98 CD! 24.84 98 CD2 23.78 98 CG 27.18 98 CO 180.40 98 HA 4.21 98 RB 1.79 98 RB’ 1.62 98 HD1* 0.98 98 HD2* 0.94 98 HG 1.76 98 HN 9.40 98 N 123.67 99 CA 61.40 99 CB 39.07 99 CD* 130.75 99 CE* 130.42 99 CO 179.68 99 CZ 128.57 99 HA 4.29 99 RB 3.31 99 HB’ 2.90 99 HD* 6.55 99 HD2 6.55 99 HE* 6.40 99 RE2 6.38 99 RN 7.36 99 HZ 6.63 99 N 119.57 100 CA 56.99 100 CB 41.42 100 CDI 25.72 100 CD2 22.74 100 CG 26.42 100 CO 177.90 100 HA 3.81 100 RB 1.87 100 RB’ 1.51 93 HE2 1 7.42 Residue Atom Shift (ppm) Residue Atom Shift (ppm) 100 HD1* 0.83 100 RD2* 0.54 100 HG 1.58 100 RN 8.03 100 N 118.54 101 CA 51.33 101 CB 20.21 101 CO 176.47 101 HA 4.36 101 RB* 1.68 101 RN 7.28 101 N 121.39 102 CA 59.45 102 CB 66.93 102 CO 175.57 102 HA 4.61 102 HB* 4.38 102 RN 7.50 102 N 115.85 103 CA 54.04 103 CB 18.16 103 CO 179.74 103 HA 2.82 103 RB* 1.14 103 RN 8.92 103 N 126.17 104 CA 59.86 104 CB 29.20 104 CG 36.37 104 CO 179.07 104 HA 3.95 104 HB 2.15 104 RB’ 1.92 104 HG* 2.36 104 RN 8.41 104 N 118.25 105 CA 57.54 105 CB 37.31 105 CD* 131.57 105 CE* 130.89 105 CO 175.65 105 CZ 128.23 105 HA 4.31 105 RB 2.50 105 RB’ 1.42 105 HD* 7.07 105 RD2 7.06 105 HE* 7.23 105 HE2 7.23 105 RN 7.91 105 HZ 6.89 105 N 119.32 106 CA 63.46 106 CB 27.75 106 CO 177.98 106 HA 3.73 106 RB 3.35 106 RB’ 2.56 106 RN 7.21 106 N 120.97 107 CA 56.47 107 CB 38.33 107 CG 175.84 107 CO 177.96 107 HA 4.33 107 RB 2.87 107 HB’ 2.80 107 RD21 7.55 107 HD22 6.89 107 RN 8.10 107 N 118.46 107 ND2 112.62 108 CA 65.19 108 CB 38.91 108 CD! 14.94 108 CG1 29.42 108 CG2 16.90 108 CO 177.80 108 HA 3.72 108 RB 1.97 108 RD1* 1.12 108 RGI* 1.23 108 HG2* 0.99 108 RN 7.94 241 Appendix 7— Assignments of the DHB-domain Residue Atom Shift (ppm) Residue Atom Shift (ppm) 108 N 121.32 109 CA 57.55 109 CB 42.87 109 CDI 25.92 109 CD2 24.07 109 CG 27.09 109 CO 178.35 109 HA 3.62 109 RB 1.54 109 HB 0.97 109 HD1* 0.24 109 HD2* 0.57 109 HG 0.77 109 RN 8.57 109 N 120.59 110 CA 62.61 110 CB 62.66 110 CO 176.94 110 HA 4.02 110 HB* 3.86 110 RN 8.46 110 N 113.19 111 CA 59.34 111 CB 30.06 Ill CD 43.36 111 CG 27.56 111 CO 178.45 111 HA 4.10 111 HB 1.90 111 HD* 3.19 111 HG 1.65 111 HG’ 1.43 111 RN 7.69 111 N 122.40 112 CA 66.64 112 CB 31.20 112 CGI 21.75 112 CG2 22.20 112 CO 177.18 112 HA 3.40 112 RB 1.89 112 HGI* 0.58 112 RG2* 0.78 112 RN 8.51 112 N 118.55 113 CA 58.63 113 CB 42.49 113 CDI 24.92 113 CD2 25.43 113 CG 28.08 113 CO 178.02 113 HA 4.01 113 HB* 1.67 113 HD1* 0.97 113 HD2* 0.96 113 HG 1.53 113 RN 8.29 113 N 120.65 114 CA 61.43 114 CB 62.80 114 CO 178.49 114 HA 4.11 114 HB* 3.97 114 RN 7.53 114 N 111.65 115 CA 59.03 115 CB 28.84 115 CD 43.70 115 CG 27.69 115 CO 177.85 115 HA 3.90 115 RB 1.33 115 RB’ 1.03 115 RD 3.01 115 HD’ 2.95 115 RG* 1.64 115 RN 8.09 115 N 121.53 116 CA 55.02 116 CB 19.37 116 CO 179.10 116 HA 3.99 116 HB* 1.54 116 RN 8.43 Residue Atom Shift (ppm) Residue Atom Shift (ppm) 116 N 121.51 117 CA 58.99 117 CB 30.62 117 CD 43.63 117 CG 29.33 117 CO 178.43 117 FIA 4.04 117 HB* 1.84 117 RD* 3.11 117 RG* 1.62 117 RN 7.80 117 N 113.44 118 CA 60.29 118 CB 64.14 118 CO 175.14 118 HA 4.44 118 HB 4.00 118 RB’ 3.95 118 FIN 7.60 118 N 112.65 119 CA 53.78 119 CB 30.68 119 CD 43.93 119 CG 27.32 119 HA 4.95 119 RB* 2.01 119 RD 3.32 119 RD 3.26 119 HG 1.71 119 HG’ 1.65 119 RN 8.22 119 N 121.29 120 CA 65.30 120 CB 31.66 120 CD 50.31 120 CG 27.36 120 CO 179.17 120 RA 4.54 120 RB 2.51 120 RB’ 2.05 120 RD 3.73 120 HD’ 3.41 120 HG 2.11 120 HG’ 2.06 121 CA 54.40 121 CB 18.77 121 CO 179.76 121 HA 4.36 121 RB* 1.51 121 RN 8.58 121 N 119.99 122 CA 54.93 122 CB 32.21 122 CD 29.06 122 CE 42.28 122 CG 25.66 122 CO 175.78 122 HA 4.53 122 RB 1.81 122 HB’ 1.77 122 RD* 1.66 122 HE* 2.95 122 HG 1.51 122 HG’ 1.26 122 RN 7.98 122 N 116.67 123 CA 59.45 123 CB 42.25 123 CD1 23.83 123 CD2 26.78 123 CG 26.85 123 CO 177.83 123 HA 4.05 123 RB 1.88 123 RB’ 1.57 123 RDI* 0.90 123 HD2* 0.67 123 HG 1.31 123 RN 7.52 123 N 120.16 124 CA 62.77 124 CB 36.79 124 CD* 132.94 124 CE* 118.10 242 .0CCCC))C ‘0 0’ — N 0’ N 00 0’ 00 N C f 00 00 — C N ‘ N C ‘0 ‘ 0’ — 0’ — C C C 00 — N 00 N 00 C . . . . C 0 ’ 0 O N C . 7Q L ) Q L ) U L ) X Z c ) L ) c Z Il) i) I)) Il) If) If) If) If) If) If) If) If) ‘0 ‘0 ‘0 ‘0 ‘0 ‘0 ‘0 N N N N N N N N N N N N 00 00 00 00 00 00 00 00 ) 0) 0) 0) 0) 0) 0) 0) 0) 0) 0) 0) 0) 0) 0) 0) 0) 0) — — — — — — — — — — — — — — — — — — — — — — — — — * — — — — — — — — — — — — — — — * N00C. z©0)—CN=Nr’I CN‘0<c-)—0)—.C*xN..4 NNNr)—0)—NNQxNCl— ,)NNCC)—0)—c:NCl c.<—0)—00*NctxNCl ‘0r)—0)—sc’000NCl N0 . Nx—0)—N00CzNCl— NQ00—0)—0’NN.L)00Cl CN.oz—0)—N00mL)00Cl CN‘0<L)Cl0)‘0Nc)L)00Cl ‘0N00‘0L)Cl0)—Nr00NCc-)00Cl— NNcL)Cl0)—If.<x00Cl— CN‘0NCL)Cl0)—Na.N*x00Cl 1 - 0C4C l 0)—CNNNx00Cl ‘0.xCl0)—-C‘CNNC00Cl 00N*NcxCl0)—r)0000Cl— 0’N Cz00Cl Nn00Cl 0)— -00If‘0<(-)0)0)—NrC‘0<L)0’Cl 00C. L)0)0)—00C’NL)0’Cl— CrNc:(-)0)0)—f)6CcC)aCl N00NNNc::’C)0)0)—c)N0’NCC)a’Cl rIf)odNCC)0)0)—Cx0’Cl C-<0)0)—N0)rx0’Cl CN0)0)—.NNx0’Cl— .nN*ctx0)0)—..r)Ncx0’Cl— Nc ’ *NC)0)0)—‘0NC)x0’Cl 0)000)0)—C‘t00za’Cl If)00<C-)e0)—C’NNIf)<C-)0) ‘000-C)e0)—NN.C)0) 00hIINCC)0)—N-NCC)0) .0)NNCC).0)—-C‘0NNCC)=0) ‘0NNC)C)0)—.NNC)C)0) I1C00CC),0)—If)0’NCC)=0) 000)0)00<x0) 0000x0)—00c.x0)— ‘00)—0)x0)— .00*C.0)—N00*C0) If- ‘0*NCx.0)—N00C*NC0) If)00C).0)—‘0NC)x0)— NC000)—0)r00zx0) 000000z00Cl N‘000zCl0)— NC)c.Nz0)0)—000)Nz0)Cl C - CC))ErM0? CCC)) I C0 . -CC))C0?=01 C — N ‘0 — N 00 — 0) 0) N C C C 0) ‘0 0’ — 00 C N N N C ’ ‘) N C 00 ‘0 ‘0 00 N ‘0 ‘0 — — — 0’ 0) C C N ‘0 o d C C 0 ’ N N 0 0 0 0 I f . . . C ’ 0 0 ’ 0 0 0 0 . o d o d ’ 0 0 ) C 0 ’ 0 ’ 0 0 0 0 — . N 0 ) N N N s C ’ 0 0 0 — N c ’ ) — C C N N — N 0) 0 ) O ’ 0 ’ 0 ’ 0 v t . If) If) If) If) If) If) C l C l C l C l C l C l C l C l C l C l C l C l C l C l C l — — — — * — — — — — — — — * — If) If) If) If) ‘0 ‘0 ‘0 ‘0 ‘0 ‘0 ‘0 ‘0 N N N N N C-I C l C l C l C l C l C l C l C l C l C l C l C l C l C l C l C l — — — — — — — — — — — — — — — — — Appendix 7— Assignments of the DHB-domain Residue Atom Shift (ppm) Residue Atom Shift (ppm) 140 HG* 1.44 140 HN 8.36 140 N 120.19 141 CA 56.31 141 CB 33.32 141 CD 29.29 141 CE 42.21 141 CG 24.80 141 CO 176.17 141 HA 4.28 141 HB* 1.78 141 HD* 1.69 141 HE* 2.98 141 HG* 1.40 141 RN 8.22 141 N 122.94 142 CA 56.31 142 CB 33.22 142 CD 29.29 142 CE 42.21 142 CG 24.81 142 CO 176.19 142 HA 4.32 142 HB 1.83 142 HB 1.76 142 HD* 1.70 142 HE* 3.01 142 HG* 1.43 142 RN 8.35 142 N 123.49 143 CA 55.25 143 CB 42.56 143 CD1 25.06 143 CD2 23.44 143 CG 27.03 143 CO 176.15 Residue Atom Shift (ppm) Residue Atom Shift (ppm) 139 CA 53.08 139 CB 19.25 139 CO 177.65 139 HA 4.27 139 HB* 1.35 139 RN 8.86 139 N 127.01 140 CA 56.19 140 CB 33.08 140 CD 29.25 140 CE 42.24 140 CG 24.87 140 CO 176.42 140 HA 4.31 140 HB 1.84 140 HB 1.74 140 HD* 1.70 140 HE* 3.00 143 HA 4.36 143 HB* 1.61 143 HD1* 0.90 143 HD2* 0.84 143 HG 1.63 143 RN 8.39 143 N 124.78 144 CA 54.72 144 CB 40.37 144 CO 178.33 144 HA 4.41 144 HB 2.75 144 HB’ 2.70 144 HD21 7.52 144 HD22 6.80 144 RN 7.93 144 N 124.00 144 ND2 112.79 244 Appendix 8 — Assignments of RassflC2338” in the DHB/RassflC2338”complex Appendix 8 Assignment of RassflC23-3$win the DHB/RassflC23-38wcomplex. Sample concentration of the Rd complex was 1 mM, NMR sample buffer (10 mM MOPS , 0.1 mM EDTA, pH 6.5, 25 °C). Hydrogen atoms with degenerate shifts are renamed as Q. Residue Atom Shift (ppm) Residue Atom Shift (ppm) Residue Atom Shift (ppm) Residue Atom Shift (ppm) 22 CA 43.63 22 HA 3.87 23 CA 58.75 23 CB 63.96 23 HA 4.50 QB 3.93 24 CA 56.60 24 CB 29.22 24 CG 34.00 24 HA 4.30 24 HB2 2.09 24 HB3 2.01 24 HE2I 7.60 24 HE22 6.84 HN 8.68 24 N 122.17 24 NE2 112.69 24 QG 2.35 25 CA 56.93 25 CB 30.05 25 CG 36.49 25 HA 4.29 25 H132 2.05 25 HB3 1.99 25 - HN 8.45 25 N 121.09 25 QG 2.29 26 CA 54.71 26 CR 41.22 26 HA 4.62 26 HB2 2.72 26 HB3 2.65 26 HN 8.28 26 N 120.97 27 CA 58.92 27 CB 63.87 27 HA 4.41 27 HB2 3.89 27 HB3 3.85 27 HN 8.20 27 N 115.98 28 CA 55.00 28 CD 41.23 28 HA 4.65 28 HB2 2.75 28 HB3 2.67 HN 8.46 28 N 122.60 29 CA 59.55 29 CB 63.78 29 HA 4.34 29 HB2 3.90 29 HB3 3.85 HN 8.23 29 N 116.31 30 CA 57.37 30 CB 30.12 30 CG 36.61 30 HA 4.23 I-IN 8.39 30 N 122.39 30 QB 2.04 QG 2.28 31 CA 56.04 31 CB 42.10 31 CD1 25.21 31 CD2 24.16 31 CG 27.20 31 HA 4.11 31 HB2 1.55 31 HB3 1.51 31 HG 1.51 HN 7.97 31 N 120.78 31 QD1 0.83 31 QD2 0.77 32 CA 58.00 32 CB 29.97 32 CG 36.53 32 HA 3.98 HN 8.10 32 N 119.61 32 QB 1.95 32 QG 2.22 33 CA 56.77 33 CB 29.04 33 CG 33.71 33 HA 4.07 33 HE21 7.37 33 HE22 6.78 33 HN 8.00 33 N 118.77 33 NE2 112.12 33 QB 1.84 33 QG 2.07 34 CA 58.68 34 CB 38.50 34 CD1 132.60 34 CEI 117.92 34 HA 4.27 34 HB2 2.69 34 HB3 2.77 34 HN 7.84 34 N 119.62 34 QD 6.76 34 QE 6.61 35 CA 58.18 35 CB 38.33 35 CD1 131.38 35 CE1 131.15 35 CZ 129.12 35 HA 4.45 35 HB2 3.04 35 HB3 2.90 HN 7.90 35 HZ 7.07 35 N 118.80 35 QD 7.10 QE 7.16 36 CA 62.20 CB 69.79 36 CG2 21.77 36 HA - 4.06 36 HB - 4.06 36 HN 7.80 36 N 113.86 36 QG2 1.07 CA 52.59 37 CB 19.40 37 HA 4.08 1-IN - 7.90 N 125.39 37 QB 1.19 38 CA 55.83 38 CR 30.97 38 CD 43.25 CG 27.03 38 HA 4.16 38 HB2 1.61 38 HB3 - 1.44 FIN 7.83 38 N 119.54 38 QD - 2.94 QG 1.35 39 CA 58.39 39 CR 30.23 39 CD1 126.44 39 CE3 121.08 39 CH2 123.88 39 CZ2 114.17 39 CZ3 121.27 39 HA 4.44 245 Appendix 8 - Assignments of RassflC2338”in the DHB/RassflC2338’’complex Residue Atom Shift (ppm) 39 HB2 3.23 39 HB3 3.04 HD1 7.05 HE1 9.94 39 HE3 7.50 39 HH2 7.04 39 HN 7.51 39 HZ2 7.29 39 HZ3 6.97 39 N 126.54 39 NEI 128.44 246 Appendix 9— Assignments of the DHB-domain in the DHB/RassflC2338”complex Appendix 9 Assignment of the DHB domain in the DHB/RassflC2338wcomplex Sample concentration of the Dr complex was 1 mM, NMR sample buffer (10 mM MOPS , 0.1 mM EDTA, pH 6.5, 25 °C). Hydrogen atoms with degenerate shifts are renamed as Q. Residue Atom Shift (ppm) Residue Atom Shift (ppm) Residue Atom Shift (ppm) Residue Atom Shift (ppm) 53 CA 56.38 53 CB 30.53 53 CD2 119.93 53 CEI 138.14 53 HA 4.66 53 HD2 7.04 53 HEI 7.95 54 CA 55.58 54 CB 32.65 54 CE 16.87 54 CG 31.92 54 HA 4.45 54 HB2 2.11 54 HB3 1.98 54 HG2 2.51 54 HG3 2.45 54 HN 8.40 54 N 122.01 54 QE 2.05 55 CA 45.29 55 HN 8.42 55 N 110.06 55 QA 3.94 56 CA 56.20 56 CR 33.30 56 CD 29.30 56 CE 42.20 56 CG 24.92 56 HA 4.33 56 HB2 1.79 56 HB3 1.66 56 RN 8.13 56 N 121.05 56 QG 1.38 57 CA 56.39 57 CR 33.52 57 CD 29.08 57 CE 42.21 57 CG 24.93 57 HA 4.23 57 HE2 2.71 57 HE3 2.67 57 HG2 1.17 57 HG3 1.01 57 RN 8.34 57 N 123.08 57 QB 1.63 57 QD 1.49 58 CA 59.54 58 CB 29.39 58 HA 4.55 58 I-IN 8.20 58 N 124.09 58 QB 2.87 59 CA 59.50 59 CB 37.83 59 CD1 133.09 59 CEI 117.96 59 HA 4.58 59 HB2 3.15 59 HB3 2.34 59 RN 8.39 59 N 126.79 59 QD 7.11 59 QE 6.80 60 CA 61.94 60 CB 33.89 60 CE 42.35 60 CG 24.94 60 HA 3.96 60 HB2 1.87 60 HB3 1.83 60 I-IN 9.13 60 N 123.91 60 QG 1.25 61 CA 58.03 61 CB 41.31 61 CD1 24.91 61 CD2 23.69 61 CG 27.43 61 HA 4.15 61 HB2 1.70 61 HB3 1.55 61 HG 1.66 61 RN 8.40 61 N 118.16 61 QD1 0.94 61 QD2 0.87 62 CA 60.20 62 CB 29.26 62 CG 36.48 62 HA 3.94 62 HB2 2.12 62 HB3 1.90 62 HG2 2.41 62 HG3 2.24 62 RN 8.63 62 N 122.19 63 CA 55.22 63 CB 37.44 63 HA 4.36 63 HB2 3.07 63 HB3 2.42 63 HD21 6.87 63 HD22 6.34 63 RN 8.95 63 N 118.69 63 ND2 103.30 64 CA 60.27 64 CB - 29.34 64 CG 37.03 64 HA - 4.11 64 HB2 - 2.28 64 HB3 2.11 64 HG2 2.66 64 HG3 2.28 64 RN 8.28 64 N - 123.77 65 CA 59.36 65 CB 32.17 65 CE 42.16 65 CG 24.89 65 HA - 4.15 65 HG2 1.59 65 HG3 1.48 65 UN 7.82 65 N 120.51 65 QB 2.00 66 CA 57.86 66 CB 43.77 66 CD1 - 27.38 66 CD2 - 23.09 66 CG 27.38 66 HA 4.09 66 HB2 1.92 66 HB3 1.16 66 HG 1.86 66 RN - 8.51 66 N 120.06 66 QD1 0.88 66 QD2 0.77 67 CA 62.57 67 CB - 39.23 67 CD1 - 131.66 247 C/DC(I, rJ)CU,C.C.SCC’,OeCC. (IDC=C’, 0 0 ‘4-b ‘ .‘D C ‘0 0 N ‘C 0 0 CO — N N c r— — Co — ‘C CO ‘ N N — N c N — N r’) N CC X X u ( ) X X L )L )X X 00 00 00 00 C — — — — — ‘ -4 N N N N N !) N N N N N N N N N N 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 C ’ m N CO . N N . i,-, ‘C . N N r N N - C O N - N N N N II Ij If) If) If) If) If) If) If) If) If) If) If) CC ‘C ‘C ‘C CC CC CC ‘C ‘C ‘C ‘C ‘C N N N N N N N N N N N N 00 00 00 00 N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N N — C CO m CO r C ’ r 0 - ‘tD C ’ N N N — N C ’ r , N 0 N ‘0 ‘0 — N N c — N — N — N N c _ , . , - c < C Z < L.) L) L ) L) C.) X X C L . C.) C.) X X X ‘ C.) X C.) C.) — — — — — — — — — — — — N N N N N N N N r) r) C’) C’) C’) C’) C’) C’) C’) C’) C’) C’) If) If) N N N N N N N N N N N N IN N N N N N N N N N N N N N N N N N N N N N N N N N N N N N r C ’ CO CO CO — C ’ rD N - CC I — — CO N C ’ ‘IC N ‘ N N N C ’ CO 00 ID r ’ N i CO 0ç N r r . ‘ o á N - ‘C ‘C ‘C ‘C ‘C ‘C ‘C ‘C ‘C ‘C ‘C ‘C ‘C ‘C ‘C ‘C ‘C ‘C ‘C ‘C CC ‘C CC ‘C ‘C ‘C ‘C ‘C N N N N N N N N N N N N N 2CD2CSC . CiDSC‘I,SC.C.2CSC.C. C,) C 00 \C ‘ r - C t 00 00 ‘ ‘ C 0 C C 00 00 in in in in in in a c t- r— r - — r— t oc oo oe 00 00 00 00 00 00 00 00 00 00 0 0 0 0 r r . — — •1- ir ‘ .c co r- i- 00 . . . . 8 z z z 8 © — — — — — — — — — — — e ri ri o o o o o , o o o o o — 0’ 00 — \C — 00 — — — 0’ C C ’ r.) 00 C — 0’ ç r r . . . t 0— 0 0 . z z t F t r 00 00 00 00 00 00 00 00 00 00 00 00 00 0’ 0’ 0’ 0’ 0’ 0’ 0’ 0’ 0’ 0’ 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 0’ 0’ 0’ 0’ 0’ 0’ 0’ 0’ 0’ SC SC m — r - 00 ‘ r l 00 r— — SC — SC r — C — . t m 0 0 0 ’ r — r 1 c0o o ° h z z z 8 ) . • • in in in in in in in in in in SC SC SC SC S C SC SC SC S C SC SC 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 00 SC Appendix 9—Assignments of the DHB-domain in the DHB/RassflC2338wcomplex Residue Atom Shift (ppm) Residue Atom Shift (ppm) Residue Atom Shift (ppm) Residue Atom Shift (ppm) 99 CA 61.41 99 CB 39.15 99 CD1 130.50 99 CE1 130.19 99 CZ 128.38 99 HA 4.24 99 HB2 3.26 99 HB3 2.88 99 HN 7.34 99 HZ 6.60 99 N 119.44 99 QD 6.49 99 QE 6.36 100 CA 57.10 100 CB 41.57 100 CDI 25.70 100 CD2 22.83 100 CG 26.56 100 HA 3.79 100 HB2 1.84 100 HB3 1.49 100 HG 1.54 100 1-IN 8.01 100 N 118.56 100 QDI 0.80 100 QD2 0.51 101 CA 51.34 101 CB 20.31 101 HA 4.34 101 I-IN 7.24 101 N 121.24 101 QB 1.66 102 CA 59.57 102 CB 66.89 102 HA 4.58 102 HN 7.47 102 N 115.80 102 QB 4.39 103 CA 54.06 103 CB 18.31 103 HA 2.81 103 RN 8.93 103 N 126.10 103 QB 1.12 104 CA 60.10 104 CB 29.20 104 CG 36.54 104 HA 3.94 104 HB2 2.12 104 HB3 1.90 104 RN 8.45 104 N 118.24 104 QG 2.34 105 CA 57.54 105 CB 37.28 105 CD1 131.23 105 CE1 130.30 105 CZ 128.09 105 HA 4.27 105 HB2 2.47 105 HB3 1.40 105 RN 7.88 105 HZ 6.86 105 N 119.21 105 QD 7.04 105 QE 7.20 106 CA 63.48 106 CB 27.82 106 HA 3.70 106 HB2 3.31 106 HB3 2.52 106 RN 7.18 106 N 120.92 107 CA 56.56 107 CB 38.44 107 HA 4.29 107 HB2 2.82 107 HB3 2.75 107 HD2I 7.52 107 HD22 6.88 107 RN 8.08 107 N 118.47 107 ND2 112.54 108 CA 65.23 108 CB 38.99 108 CDI 15.06 108 CG1 29.46 108 CG2 17.04 108 HA 3.67 108 HB 1.92 108 RN 7.90 108 N 121.11 108 QD1 1.09 108 QG1 1.17 108 QG2 0.95 109 CA 57.54 109 CB 42.96 109 CD1 26.04 109 CD2 24.18 109 CG 27.20 109 HA 3.59 109 HB2 1.50 109 HB3 0.94 109 HG 0.73 109 RN 8.54 109 N 120.40 109 QD1 0.21 109 QD2 0.54 110 CA 62.82 110 CB 62.78 110 HA 4.01 110 RN 8.42 110 N 113.11 110 QB 3.83 111 CA 59.40 111 CB 30.11 111 CD 43.32 111 CG 27.56 111 HA 4.06 111 HB2 1.89 111 HB3 1.83 111 HG2 1.60 111 HG3 1.41 111 RN 7.66 111 N 122.44 111 QD 3.14 112 CA 66.67 112 CB 31.26 112 CGI 21.95 112 CG2 22.29 112 HA 3.37 112 HB 1.88 112 RN 8.51 112 N 118.41 112 QG1 0.57 112 QG2 0.76 113 CA 58.63 113 CB 42.53 113 CD1 24.95 113 CD2 25.48 113 CG 28.12 113 HA 3.99 113 HG — 1.51 113 RN 8.30 113 N - 120.51 113 QB 1.65 113 QD1 0.93 113 QD2 0.94 114 CA - 61.48 114 CB 62.77 114 HA 4.09 114 RN - 7.51 114 N — 111.63 114 QB 3.95 115 CA 59.16 115 CB 28.76 115 CD 43.62 115 CG 27.85 115 HA 3.86 115 RB2 1.30 115 HB3 1.02 115 HD2 3.00 115 RD3 2.91 115 RN 8.04 250 CD C CD = C - - - - - - - — - - - - - - - - - - - - - — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — te . a • 3 — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — e — — — — — c C 00 00 00 00 00 00 00 - c i - - c i c i - c i c i c i (Il C) C) C) C ) X X X X X C) C) C) C) X - - ‘ z Z z ) > Z > C ) > - C ) c ) c i 0 — 0’ L I 00 C — c i — c i _ ) t) tJ 00 C — t- . — . — c — c — — I — — I. i a C) — I. i — I-. i C) > — J ‘ C L’) = > C B a k— i ‘ C a -. i z a a X C) a ,J X — t’ J X — iJ X > a C) C) — — , J - . e C) t-J w p c w C) c ) c, 00 C) X — - . i 9 0 00 a C) > 9 00 00 a e - J 9 o a I- z c i L1 a - ) a I- Is i X C) ‘ L I. 00 a a X C) I’) 00 00 a a X (-h) a — X X L’J a a X p 00 . p c a - J a C) C) t’ 3 I cc — - J — a c i C) e - t- J L I a IN ) c i a IN ) = C) L I C - — IN ) c i a IN C) X a N . a IN ) c i C) > a IN ) 9N L I a IN ) — J ‘ C a IN ) = c i \C I a IN ci — IN ) z - ) — IN — 1 a IN ) ON — IN ci a IN ) = X >- — a ON — IN ci C) X 00 I’ ) a IN ci z C) - LI . X C) ) I’ ) LI . L I L I — IN ) c i a IN ) N X C) I X C) L’J . IN .) IN .) ( a IN ) aN X X C) I\) I- ) C “ C — IN ) ON X . ) IN. ) X IN.) LI . c i c i a IN ) ON X X X I a IN ) ON C) C) IN-.) X IN.) 00 — IN ) ON C) C) X > C “C ON a IN ) ON C) 00 (I. . C) a IN ) ON C) . ) LI . a IN ) ON 00 00 00 a IN ) ON ‘ C X LI . . . C ci a IN ) ON ‘ C 00 00 “C IN .) IN .) a IN ) U I IN .) 00 00 a IN ) U I IN .) . ) a IN ) U I I . ) a N “ C cc ci a IN ) U I X IN-.) J c i . a IN ) U I z . 00 00 IN. .) - 00 C) tj a IN ) U I IN .) C “C .) a IN ) U I C) : 00 C LI . IN .) a IN U I C) . ) 00 00 00 a IN ) U I C) >- ON ci , ON a IN ) . ‘ C C) IN-.) a IN ) U I 9N ci LI . ‘ C C) N LC LI . a IN ) . z IN .) ON a IN ) . . N C — a X a IN ) . ) IN .) a IN ) 3 J c i C) C) IN.) a IN ) . a IN ) : ON IN .) C) C) 90 LI . IN .) a IN ) . C) w èo — I C) >- a IN ) . ) — a IN .) C ‘ C nj LI . :‘ C c i ‘ C 9 ci 00 2: 9 ci LI. IN .) ) . X L IN- .) 00 C X :- ..l 00 . ) LI . c i C) nj IN .) - a . ) C) IN .) 00 — a ON 9N C . 9N c i L ) — a C ON . ) . 00 - 00 IN .) 00 IN- .) 00 .) C L J a “ C . w IN .) LI . LI . a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a a !) ) IN ) IN ) IN ) IN ) IN ) IN ) IN ) IN ) IN ) IN ) IN ) IN ) IN ) a — a a a a LO 00 00 00 00 00 00 00 00 00 00 c i c i z 2 9 z 2 z “ C “ C - C — a 00 IN .) c i ON LC — - cc - — — ‘ C “ C “ C “ C - C ON LI . 00 — 00 C IN .) W ON LI . LI . L J - ) — L I L — “ C c i . ) L I. 0 0 C C - LI . — c i ‘ . ON C, , C (I) C c— i) C B C B C/) - € r;I ) Cd ) = L, J C C 0 Cd ) CD I — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — — C _ c j CM CM CM CM CM CM CM C i t- J IJ b- .) — X X - - z z c . ) r I > ) 0 0 C 0 0 ‘ — — C • L ) C J — - C 00 — C_ . C • — ( — . — . = — . — — . © — — = — — . © — d — Cd \ 2 — c d — — (- c- — c i (Th 1j — d 00 n — d 00 (-J — t 00 C) > a Cd d 00 z a 00 00 C a C. e 00 a J 00 . (- — c — 00 C-k ) C 00 — a a (J — c -: z - : 00 a .- ‘ j 00 x , — Ci d - t’ J ‘ c 3 — - J , - c ) C — Ci d — x c p C c — Cd — — a Th > :1 C C z — (.d — . — c . — k) . 00 00 c , — c tn - c w C — Cd d t) c — C - — d x -J — a L1 > a (.i i t 00 C a (.. J a (Th tn c C — i CM (- t’J - C J c — d CM Q L, C r > c ) 90 00 w 00 t- a 2 C . 00 C- , — a x Th 00 —- a C— ) ‘ - a C — a C) D 00 — a t- ) — a C L3 00 3 — a Q 0 - a — a i— a C J — a — a C C . — a , . c - a c , eCo --a o a a — a a — . — — — — a a — — — — a a a — a a — a a a — a a a a — — . . ‘ . . . . . . . . . . d d . J tJ d lc — — — a a a — a a a a a a © Z Z 00 00 V C. ) C C — J J — 3 00 - C — a — — a 00 3 . , - L’ 3 00 C 00 C — — — - — — 00 - 00 C - . — — a — — a a a a . t c — a . C — 3 g — — a ‘ .c 00 - d Q Appendix 10 — Triangulation of the spin label Appendix 10 Triangulation of the spin label A > restart; > eql:= > eqi (x— 79.3241 + (y4.386)2 + (+ 5.251 )2 = 237.16 eq2:= (x_83.481)A2+(y+2.193)2+(z+4.234)A2=13.1A2; eq2:=(.v—83.481)++2.193)+(2+4.2341=171.61 eq3:= (x_81.420)A2+(y+5.838)A2+(z+O.168)A213.3A2; eq3 := (.v— 81.420)2 + (v+ 5.838)2+ (z + .168)2 = 176.89 solve({eql,eq2,eq3},{x,y,z}); = 74.80499681. c = -11.24742530. v = -9.059500877 1. = 6.543513866. v = 2.712728785 .x = 89.08359047 B Ii ‘1.’ 14.4 L47 IHN V38 IHN L44HN/O ,.•‘. NO• (A) Sample of a Maple worksheet used for solving three equations of a sphere with the co ordinates of Va138, Leu44 and Leu47 1HN as centers, and calculated reff as radius. (B) Position of the two solutions on the structure of SUMO-1. One solution is impossible as it overlaps with other atoms in the molecule. The only acceptable solution gives an approximate location of the spin label. 253


Citation Scheme:


Citations by CSL (citeproc-js)

Usage Statistics



Customize your widget with the following options, then copy and paste the code below into the HTML of your page to embed this item in your website.
                            <div id="ubcOpenCollectionsWidgetDisplay">
                            <script id="ubcOpenCollectionsWidget"
                            async >
IIIF logo Our image viewer uses the IIIF 2.0 standard. To load this item in other compatible viewers, use this url:


Related Items