UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Structural studies of the ETS domain Donaldson, Logan William Frederick 1996

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

Item Metadata


831-ubc_1996-090728.pdf [ 15.21MB ]
JSON: 831-1.0094704.json
JSON-LD: 831-1.0094704-ld.json
RDF/XML (Pretty): 831-1.0094704-rdf.xml
RDF/JSON: 831-1.0094704-rdf.json
Turtle: 831-1.0094704-turtle.txt
N-Triples: 831-1.0094704-rdf-ntriples.txt
Original Record: 831-1.0094704-source.json
Full Text

Full Text

STRUCTURAL STUDIES OF THE ETS DOMAIN by LOGAN WILLIAM FREDERICK DONALDSON B.Sc, Lakehead University, 1989 Dip. Hons. Standing, The University of Western Ontario, 1990 M.Sc, McMaster University, 1992 A THESIS SUBMITTED LN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY in THE FACULTY OF GRADUATE STUDIES DEPARTMENT OF BIOCHEMISTRY AND MOLECULAR BIOLOGY We accept this thesis as conforming to the required standard THE UNIVERSITY OF BRITISH COLUMBIA January 1996 © Logan William Frederick Donaldson, 1996 In presenting this thesis in partial fulfilment of the requirements for an advanced degree at the University of British Columbia, I agree that the Library shall make it freely available for reference and study. I further agree that permission for extensive copying of this thesis for scholarly purposes may be granted by the head of my department or by his or her representatives. It is understood that copying or publication of this thesis for financial gain shall not be allowed without my written permission. Department The University of British Columbia Vancouver, Canada DE-6 (2/88) Abstract This thesis describes the cloning, overexpression, biophysical and structural analyses of fragments from murine Ets-1 that comprise the ETS domain, a motif responsible for the sequence-specific DNA binding activity of the ets family. Multidimensional nuclear magnetic resonance (NMR) techniques feature prominently throughout this study. From the tertiary structure of a 110 residue Ets-1 fragment, the ETS domain is demonstrated to possess a winged helix-turn-helix fold. As the name suggests, the winged helix-turn-helix motif is comprised of a helix-turn-helix motif and a characteristic (3-sheet component which pack to form a compact domain. In some members of the winged helix-turn-helix family, a mobile wing-like (3-turn supplies base-specific contacts to augment contacts made by the recognition helix. The ETS domain of Ets-1 is flanked by amino- and carboxy-terminal sequences that cooperate to attenuate DNA binding affinity by approximately twenty fold. Through structural analysis of a larger, 142 residue fragment of Ets-1, the flanking inhibitory sequences are shown to form a module comprised of two N-terminally located cc-helices, one C-terminally located ct-helix and one cc-helix from the ETS domain. An allosteric mechanism of repression is proposed as the inhibition module is located on different face of the ETS domain relative to the recognition helix. Dynamic information obtained from 1 5 N relaxation and fast amide hydrogen exchange experiments support this mechanism of Ets-1 intramolecular inhibition. Table of Contents Abstract ii Table of Contents iii List of Tables vii List of Figures viii List of Equations x Abbreviations .. xi Acknowledgements xiii Chapter 1 — The Ets Family of Transcription Factors 1 1.1. A Central Role for DNA Recognition 1 1.2. The Ets Family 5 1.2.1. Introduction 5 1.2.2. Ets Proteins are Important Regulators of Differentiation 8 1.2.3. Regulation of Ets family members 9 1.2.4. Cooperativity Between Ets Proteins and Other Factors 11 1.2.5. Coupling of Ets Protein to the Signal Transduction Pathway 11 1.2.6. Higher Order Complexes Involving Ets Proteins 16 1.2.7. Intramolecular Inhibition of DNA Binding 19 1.3. The Structural Characterization of Murine Ets-1 20 1.3.1. Rationale 20 1.3.2. Thesis Organization 27 1.3.3. Collaborations 28 Chapter 2 — Overexpression and Purification of ETS Domain Proteins 29 2.1 Overexpression Systems 29 2.2. The Necessity of Isotope Labelling 30 2.3. Overexpression of a Minimal ETS Domain Containing Protein 35 2.3.1. Background 35 2.3.2. First Generation Constructs 40 2.3.3. Second Generation Constructs of the Ets-112kDa fragment 40 2.3.4. Alternative Promoter / Strain Combinations 42 2.3.5. Histidine-Tagged Constructs 46 2.3.6. Periplasmic Secretion Signals 48 2.3.7. Fusion Proteins 54 2.3.8. Conclusions 54 2.4. Solution Attributes of Uncomplexed Ets-1 AN331 55 2.4.1. Thiol Titrations 55 2.4.2. Analytical Ultracentrifugation 55 2.5. Ets-1AN331 Has Sequence Specific DNA Activity 56 2.5.1. Ultraviolet Laser Crosslinking 56 2.5.2. Protein/Nucleic Acid Titration Studies 60 2.5.3. Summary 63 2.6. Secondary Structural Characteristics of Ets-1AN331 63 2.6.1. CD Spectroscopy 63 2.6.2. FT-IR Spectroscopy 66 2.6.3. Conclusions 68 Chapter 3 — Assignment and Secondary Structure of the ETS Domain 70 3.1. Theory and Technique 70 3.1.1. Introduction 70 3.1.2. The Building Blocks of a Modern Pulse Sequence 72 3.1.3. The CBCA(CO)NH Experiment 75 3.2. Experimental Methods 78 3.3.1. Expression of Isotopically Labelled Proteins 78 3.2.2. Instrumentation and Software 79 3.2.3. NMR Experiments 79 3.3. Results : ". 80 3.3.1. Preliminary Spectra 80 3.3.2. Assignment of Ets-1AN331 81 3.3.3. Secondary Structure Analysis by NMR 87 3.4. The Winged-Helix-Turn-Helix Motif 96 — iv — Chapter 4 — The Tertiary Structure of the ETS Domain 99 4.1. Introduction 99 4.2. The NMR Approach to Structure Determination 100 4.2.1. Restraints • 101 4.2.2. Structural Generation 102 4.3. Experimental Methods 106 4.3.1. Protein Samples 106 4.3.2. NMR Experiments 106 4.3.3. Structural Restraints 107 4.3.4. Structure Calculations 108 4.3.5. Structure Assessment : 110 4.4. Structural Features • 114 4.5. The Winged Helix-Turn-Helix Motif 118 4.6. Homology Among ETS Domain Sequences 121 4.7. DNA Binding by the ETS domain 127 4.8. The ETS Domain of FIi-1 139 4.9. A Comparison of Three ETS Domains 140 Chapter 5 — Intramolecular Regulation of DNA Binding 145 5.1. Introduction 145 5.2. The Inhibition Module of Ets-1 146 5.3. Structural Modelling of the Inhibitory Module 153 5.4. A Minimal Autoinhibited Ets-1 Polypeptide, Ets-1 AN301 158 5.4.1. Overexpression and Characterization 158 5.4.2. Backbone and Sidechain Assignment of Ets-1 AN301 165 5.4.3. NMR Restraint Analysis and Topology of Ets-1AN301 166 5.4.4 . Paramagnetic Spin-Relaxation 168 5.5. Dynamics of Ets-1AN301 , 171 5.5.1. Introduction 171 5.5.2. Relaxation Theory 171 5.5.3. Methods 176 5.5.4 . Results 179 5.6. A Hydrogen Exchange Study of Ets-1AN301 187 5.6.1. Introduction 187 5.6.2. Theory and Experimental Methods 189 5.6.3. Results and Discussion 192 5.7. Characterization of the Ets-1 AN301 / DNA Complex 195 5.8. A Possible Mechanism for Inhibition 198 5.9. Conclusions ...„......:. : 208 Bibliography 210 Appendix 229 — vi — List of Tables Table 1.1 — Some Prominent Ets Family Members 6 Table 1.2 — Summary of Ets-1 Deletion Mutants 27 Table 2.1 — Natural Abundances of Spin 1/2 Nuclei Found in Proteins 31 Table 2.2 — Strains Tested for Overexpression 39 Table 2.3 — Summary of Ets-1AN331 Biophysical Characteristics 69 Table 4.1 — Restraint Classes 101 Table 4.2 — Restraint Summary for Ets-1AN331 109 Table 4.3 — Structure Generation and Refinement Protcol for X-PLOR 3.1 I l l Table 4.4 — Statistics for the 20 Ets-1AN331 Simulated Annealing Structures 114 Table 4.5 — The Winged Helix-Turn-Helix Family 122 Table 4.6 — Interhelical Angles 143 Table 5.1 — Plasmid/Strain Combinations Assayed for Ets-1AN301 160 Table 5.2 — Summary of the Model-Free Dynamical Parameters 180 Table 5.3 — A Comparison of k R vs. R e x for Helix HI1 of Ets-1 AN301 193 Table 5.4 — Proposed Mutations to Probe Inhibition Domain Function 207 Table A . l — Backbone Chemical Shifts of Ets-1AN331 229 Table A.2 — Hydrogen Exchange Results for Ets-1 AN301 233 Table A.3 — Materials and Methods Quick Reference 237 — vii — List of Figures Figure 1.1 — The Pre-Initiation Complex 2 Figure 1.2 — Location of the ETS Domain in Ets Family Members 7 Figure 1.3 — Chromosomal Rearrangements Lead to Transformation 9 Figure 1.4 — A Survey of Function Regions Mapped to Ets-1 10 Figure 1.5 — Coupling of Ets Proteins to the Signal Transduction Pathway 12 Figure 1.6 — Higher Order Complexes Involving Ets Proteins 16 Figure 1.7 — Deletion Mapping of Ets-1 22 Figure 1.8 — The Structure of Ets-1 AN331 23 Figure 1.9 — The Proposed Structure of Ets-1 AN301 25 Figure 2.1 — A Survey of Selective Labelling Techniques 33 Figure 2.2 — Ets-1 Fragments Defined by Proteolysis 36 Figure 2.3 — First and Second Generation Clones 37 Figure 2.4 — tac Promoter Contructs 38 Figure 2.5 — N-terminal Histidine Tagged Constructs 46 Figure 2.6 — A T7-lac Construct 47 Figure 2.7 — Signal Peptide Containing Constructs 50 Figure 2.8 — The NTCB Reaction 52 Figure 2.9 — UV Crosslinking of Ets-1AN331 57 Figure 2.10 — Protein Titration of Ets-1AN331 with the SCI Duplex 61 Figure 2.11 — CD spectra of Ets-1AN331 65 Figure 2.12 — FT-IR spectrum of Ets-1 AN331 67 Figure 3.1 — Basic Experiments Involving Echoes 73 Figure 3.2 — Uses of Pulsed Field Gradients 74 Figure 3.3 — The CBCA(CO)NH Experiment 76 Figure 3.4 — Three-Dimensional Heteronuclear Experiments 82 Figure 3.5 — One-Dimensional *H Spectrum of Ets-1 AN331 83 Figure 3.6 — One-Dimensional 1 3 C Spectrum of Ets-1AN331 84 Figure 3.7 — Two-Dimensional !H , ^N-HSQC Spectrum of Ets-1 AN331 85 Figure 3.8 — Sequential Assignment of Ets-1AN331 85 Figure 3.9 — Helix Capping Boxes 86 Figure 3.10 — Secondary Structure of Ets-1AN331 98 Figure 3.11 — Anti-Parallel p-sheet of Ets-1AN331 92 Figure 3.12 — Alignment of the ETS domain with other wHTH Members 98 — viii — List of Figures Figure 4.1 — Structure Generation Flowchart 103 Figure 4.2 — Ambiguity in Structure Determination 104 Figure 4.3 — The Ensemble of Ets-1 AN331 Structures 113 Figure 4.4 — Graphical Summary of Restraints 115 Figure 4.5 — Ramachandaran Plot of Ets-1AN331 Average Structure 117 Figure 4.6 — The Winged Helix-Turn-Helix Family 119 Figure 4.7 — Homology Within the ETS Domains 124 Figure 4.8— Electrostatic Profile of the ETS Domain , 126 Figure 4.9 — DNA Footprinting of the ETS domain, CAP and HNF-3y 129 Figure 4.10 — Selective i5N-tyrosine Labelling of Ets-1AN331 130 Figure 4.11 — A Soluble Human Ets-1 Contruct 132 Figure 4.12 — Contacts to DNA Made by the ETS Domain 134 Figure 4.13 — The Structure of the ETS Domain Bound to DNA 136 Figure 4.14 — Chemical Shift Perturbation Analysis of Free/Bound Ets-1 137 Figure 4.15 — Superimposition of Three ETS Domain Structures 139 Figure 5.1 — Deletions of Ets-1 Identify an Inhibitory Domain 147 Figure 5.2 — *H, 1 5N-HSQC Spectra of Ets-1AN280 148 Figure 5.3 — Amalgamation of Deletion Mutant Data 151 Figure 5.4 — Chemical Shift Perturbation Map of the Inhibition Domain 152 Figure 5.5 — The Inhibition Domain Helices Are Amphipathic 154 Figure 5.6 — A Proposed Autoinhibited ETS Domain, Ets-1AN280 156 Figure 5.7 — A Model for Intramolecular Inhibition 157 Figure 5.8 — A Minimally Sized Inhibited Ets-1, Ets-1 AN301 159 Figure 5.9 — *H, 1 5N-HSQC Spectra of Ets-1 AN301 160 Figure 5.10 — Entry of Ambiguous NOE Data 166 Figure 5.11 — The Paramagnetic Spin-Label, TEMPO 168 Figure 5.12 — Spin-labelling of Ets-1AN301 169 Figure 5.13— Monoexponential Curve-Fitting of Ti and T2 Data for Ets-1 AN30 178 Figure 5.14 — Ti and T 2 Plots for Ets-1 AN301 181 Figure 5.15 — Spectral Density Parameters for Ets-1AN301 182 Figure 5.16 — Base Catalysed Proton Exchange 187 Figure 5.17 — Amide Proton Exchange Can Occur Over a Wide Variety of Rates 187 Figure 5.18 — Hydrogen Exchange Summary for Ets-1 AN301 193 Figure 5.19 — Free Energies Associated With Autoinhibition 197 Figure 5.20 — Rates Associated With Autoinhibition 199 Figure 5.21 — The Role of the Basic Patch on the ETS Domain 200 Figure 5.22 — The First Contacts Made to DNA 199 Figure 5.23 — A Sidechain Rearrangement Begins at the H1/H3 Interface 201 Figure 5.24— Accommodating Conformational Changes Upon DNA Binding 202 Figure 5.25 — Accessibility of W338 for DNA Binding 203 Figure 5.26 — A Possible Sidechain Rearrangement 202 Figure 5.27 — Helix HI1 of the Inhibition Domain Unfolds 205 — ix — List of Equations Equation 3.1 — Precession of a Nucleus in an Applied Magnetic Field 70 Equation 3.2 — The Bloch Equations 71 Equation 5.1 — The Autocorrelation Function 172 Equation 5.2 — The Spectral Density Function 172 Equation 5.3 — Ti Relaxation 173 Equation 5.4 — T2 Relaxation 174 Equation 5.5 — The Heteronuclear NOE 175 Equation 5.6 — Spectral Density Expressed in the Model Free Formalism 175 Equation 5.7 — Specral Density for Two Correlation Times 176 Equation 5.8 — A Normalized NOE 177 Equation 5.9 — RMS Error in the NOE Measurements 177 Equation 5.10 — Definition of a Hydrogen Exchange Protection Factor 188 Equation 5.11 — Definition of the Random Coil Hydrogen Exchange Rate 188 Equation 5.12 — Amide Hydrogen Exchange Varies With pH 188 Equation 5.13 — Longitudnal Relaxation Incorporating Hydrogen Exchange 190 Equation 5.14 — Relaxation in the Rotating Frame With Hydrogen Exchange 190 Equation 5.15 — Exclusive Amide Hydrogen Exchange Expression 190 Equation 5.16 — Exclusive Dipole-Dipole Interactions Expression 190 Equation 5.17 — Normalized NOE from Experimental Data 191 Equation 5.18 — Normalized ROE from Experimental Data 191 Equation 5.19 — Definition of ICR, Exchange in the Rotating Frame 191 Equation 5.20 — Definition of 191 Abbreviations Amino-Acid Abbreviations Follow the One-Letter Code: A, alanine; C, cysteine; D, aspartic acid; E, glutamic acid; F, phenylalanine; G, glycine; H, histidine; I, isoleucine; K, lysine; L, leucine; M , methionine; N , asparagine; P, proline; Q, glutamine; R, arginine; S, serine; T, threonine; V, valine; W, tryptophan; Y, tyrosine. Mutation of Alanine-100 to Valine, for example, is abbreviated as A100V. A protein domain, in this text, is defined as a protein fragment (of contiguous sequence) that retains a similar fold and activity in solution as what is found in the parent protein. A protein module defines a protein sequence (or sequences) that provide an activity that is found in the parent protein. Thus, according to this definition, all protein domains are modules. However, protein modules are not necessarily domains. The term, deletion mutant, is used interchangably with protein fragment and has been accepted as a proper term by the scientific journals in which much of this data appears. CBF core binding factor CD circular dichroism DSS 2,2-dimethyl-2-silapentane-5-sulfonate DTT dithiothreitol EBS Ets DNA binding site EDTA ethylenediaminetetraacetic acid ETS E26 transformation specific Ets-1AN280 the N-terrninal protein fragment of murine Ets-1 consisting of residues 280-440 Ets-1AN301 the N-terminal protein fragment of murine Ets-1 consisting of residues 301-440 Ets-1AN331 the N-terminal protein fragment of murine Ets-1 consisting of residues 331-440 ETS Domain the minimal DNA binding domain of the Ets family as defined by homology DQF-COSY double quantum filtered correlation spectroscopy DNTB dinitrothiobenzoic acid ERK extracellular signal regulated kinase FID free induction decay FT-IR Fourier transform infrared Gu»HCl guanidinium hydrochloride INEPT insensitive nuclei enhanced by polarization transfer IPTG isopropyl p-D-thiogalactoside JNK c-Jun N-terminal kinase HSF heat shock factor Abbreviations HSQC heteronuclear single quantum correlation HTH helix-turn-helix MAP mitogen activated protein NOE nuclear Overhauser effect NOESY nuclear Overhauser effect spectroscopy PCR polymerase chain reaction pH* the observed pH meter reading without correction for isotope effects ROE rotating frame Overhauser effect RMS root-mean-square SDS sodium dodecylsulphate SRE serum response element SRF serum response factor TCF ternary complex factor TEMPO 2,2,6,6-tetramethylenepiperidine-l-oxyl TFE trifluoroethanol TOCSY total correlation spectroscopy wHTH winged helix-turn-helix Acknowledgments LMc: Three years ago, I looked around that empty lab with the bright orange desktops and I knew that was in a unique position. There, you allowed me to pursue science on my own terms, always with much encouragement and support. Most of all, I appreciate the many opportunities that you have provided me. At this time, I would also like to thank those who provided me the same opportunities in the past: To Professors Alistair Macdonald, Lada Malek and Ken Edgecombe for letting me spend summers in their labs. To George Mackie for encouragement at Western. To Tom and Donna Jovin for showing me all the good things about a career in science. Much of the work described in this thesis has been in collaboration with Jeannine Petersen and Professor Barbara Graves at the University of Utah. The support of Professor Tom Alber and technical wizardry of Professor Lewis Kay is gratefully acknowledged. From Western to McMaster to UBC and all those days in between when it felt like it was raining anvils, I would have surely went insane without the steady supply of wit from Dave Burk. Philip Johnson, Leigh Plesniak and Manish Joshi made the LMc lab a fun place to be. Deb Sauve, Dave Leggett, Duncan Ho, Patrick Gowdy, Amy Olson, Cynthia Chow, and Brendan Bell made sure that the third floor was the hip floor. My family and that uncle of mine in L.A. have always been cheering for me, and I thank them, above all. C h a p t e r 1 The Ets Family of Transcription Factors The ETS domain is responsible for the sequence-specific DNA binding activity of the Ets family of transcription factors. Sequence comparisons and biochemical analyses suggested that the ETS domain could be a novel DNA binding motif. The initial goal of this study was to overexpress a fragment of murine Ets-1 encompassing the ETS domain and determine its structure using nuclear magnetic resonance (NMR) spectroscopy. The secondary and tertiary structure of a 110 residue fragment demonstrated that the ETS domain constitutes a winged helix-turn-helix motif. From the tertiary structure, the structural analysis was extended to a larger Ets-1 protein fragment containing the ETS domain and a flanking inhibitory activity. NMR secondary structural analysis showed that a 20-fold reduction in DNA-binding affinity was attributed to the packing of three helices onto the ETS domain at the opposite face of the protein to the recognition helix. A possible mechanism for intramolecular inhibition involving a helix-to-coil transition is discussed in light of dynamics data and knowledge of the complexed form. 1.1. A Central Role for D N A Recognition Transcription of eukaryotic genes by RNA polymerase II begins with the recruitment and step-wise assembly of a set of general transcription factors to produce a pre-initiation complex. The binding of general transcription factor TFII-D, itself composed of a TATA-box binding protein and up to seven TAFs (TATA-box associated factors), begins the assembly of the pre-initiation complex, assisted by TFII-A. The binding of TFII-B follows and serves as a platform to recruit RNA polymerase II and TFII-E/F. Transcription commences upon phosphorylation of the RNA polymerase C-terminal domain by TFII-H and other kinases. Formation of the pre-initation complex is the rate-limiting step in transcription. Thus, in the absence of any influences on the assembly of the pre-initiation complex, only one rate of transcription, a basal level, is possible. 1 Chapter 1 — The Ets Family of Transcription Factors 2 Basal level transcription is enhanced or repressed by the sequence-specific binding of various transcription factors to promoter/enhancer sites adjacent to the transcription initiation site. Once a transcription factor binds a promoter/enhancer sequence through its DNA binding activity, its transactivation activity becomes locally available to enhance the stability and recruitment of essential general pre-initiation factors such as TFII-D and TFII-B (Buratowski, 1994; Hori and Carey, 1994; Chi et al, 1995; Roberts et al, 1995). The protein-protein interactions involved are numerous and complex (Figure 1.1). enhancer element pre-initiation complex Figure 1.1: Transcription factors are localized to enhancer sequences by their intrinsic sequence-specific DNA binding activity (DBD). The rate of transcription is enhanced through multiple protein-protein interactions (dashed arrows) between activation domains and the general transcriptional factors that comprise the pre-initiation complex. If each gene required a unique transcription factor bound to its promoter to initiate expression, an unrealistic number of proteins would be required to ensure even a Chapter 1 — The Ets Family of Transcription Factors 3 moderate degree of regulation. Nature achieves the necessary complexity with economy by building proteins and in turn, macromolecular assemblies, from discrete modules well suited to a given task (Frankel and Kim, 1991; Kodadek, 1995). Biological processes are dynamic and associative — temporal effects (being at the right place at the right time) and concentration effects (as there is strength in numbers) introduce a level of regulation and engineering beyond that of any process that could be considered in purely mechanical terms. The ability to recognize the correct promoter sequence is specified by the DNA binding module. If promoter recognition is abolished, the transcription factor cannot be localized to a specific site to perform its function. It is not surprising then, that many of the other modules in the transcription factor are engaged in some respect with the modulation of the DNA binding activity. Transactivation, like DNA binding, is an activity which can be considered autonomous, yet susceptible to regulation. However, the transactivation activity still depends on a DNA binding activity to guide it to the promoter. Biochemically, many DNA binding and transactivation modules are true structural domains since their activities are portable and may be recombined in novel ways. In gene expression studies, the chimeric protein GAL4-VP16 is a popular reagent since it couples a small, specific zinc-finger DNA binding motif to a very potent transactivating activity (Sadowski et al., 1988; Mitchell and Tjian, 1989). The two-hybrid system represents one contemporary use of chimeric transcription factors to identify relationships between protein modules (Chien et al, 1991). The DNA binding domain provides a structural scaffold to which specificity and affinity can be fine-tuned according to the identity of a few side-chains at key positions. For many DNA binding motifs such as the homeodomain, leucine zipper, helix-turn-helix and zinc-finger families, a "recognition helix" reads the edges of the bases in the major groove by hydrogen bonding and hydrophobic interactions (Steitz, 1990; Harrison, 1991). In an classic experiment, Wharton et al. (1994,1995) demonstrated that specificities of two bacterial repressor proteins could be interchanged by swapping residues from the Chapter 1 — The Ets Family of Transcription Factors 4 solvent exposed face of the recogntion helix. This success made the prospect of finding the recognition code for DNA binding motifs that employ a helix to read the major groove seem possible. Suzuki (1994) approached this task by creating a structural database containing residue types that were likely to interact with the two basepairs in two possible orientations. However, the Suzuki method quickly collapses if many hydrogen bonds through water, hydrogen bonds that bridge two bases and the possibility of nucleic acid deformation have to be factored into the prediction. The success of a predictive method directly depends on the completeless of the structural database it was derived from. A combination of ongoing structural characterizations of novel DNA binding motifs and the re-examination of known DNA binding motifs is the most direct route to the identification of a protein-DNA recognition code. With many modules catenated together, the overall size of the average transcription factor puts complete structural elucidation of the protein far beyond the limits of current NMR methods. In the meantime, each module may be characterized individually or in combination if the total molecular weight does not become prohibitive ( < 30 kDa). A reductionist strategy applied to the complete structural characterization of a transcription factor should begin with the DNA binding domain. Just as the DNA binding domain is a scaffold for its own activities, it can also act as a focal point to assay other functions including transactivation, repression, cofactor binding and phosphorylation. In this thesis, I have employed both a reductionist strategy to examine the structural and mechanistic aspects of DNA binding in murine Ets-1, a member of the Ets family of transcription factors and proto-oncoproteins, and a constructionist strategy to examine the regulation of DNA binding by additional modules within the native protein. Chapter 1 — The Ets Family of Transcription Factors 5 1.2. The Ets Family 1.2.1. Introduction The archetypal member of the ets gene family, v-ets, was identified as the gene fragment responsible for the transforming activity of avian erythroblastosis virus E26 protein pl35S ag-my b- e t s (Radke et al, 1982; Nunn et al, 1983, Leprince et al, 1983). The transduced avian gene, c-ets-1, has very similar Drosophila, murine and human counterparts. Further genetic studies over the last decade have resulted in the identification of over twenty ets genes from a variety of higher eukaryotic species (Lautenberger et al, 1992; Degnan et al, 1993; Laudet et al, 1993). A representative sampling of some Ets family members is listed in Table 1.1. The reader is also directed to three recent reviews (Macleod et al, 1992; Janknecht and Nordheim, 1993; Wasylyk et al, 1993). Al l Ets family members are characterized by a homologous region of 85 amino acids termed the ETS domain (Karim et al, 1990; Janknecht and Nordheim, 1993) that confers a specific monomeric DNA binding activity toward promoters containing a core sequence of 5'-GGA(A/T)-3' (Woods et al, 1992; Nye et al, 1992). Flanking bases are variable among family members but still provide important contact points evidenced by chemical protection and interference studies that extend outward to 16 bp (Nye et al, 1992). There appears to be no dependency of DNA binding on the position of the ETS domain in Ets proteins. This suggests that the ETS domain is a true protein domain (Figure 1.2; see definition in the Abbreviations). The minimal expressed ETS domain has a very high affinity for its binding sites, ca. 10'1 1 M (Nye et al, 1992; Fisher et al, 1994). The combination of a unique footprint from chemical interference and protection studies and no apparent sequence homology to other DNA binding motifs lead to the assumption that the ETS domain represented a novel DNA binding motif (Karim et al, 1990; Klemsz et al, 1990; Nye et al, 1992). Even within the family, there is considerable variation at the level of sequence in the ETS domain ranging from 96% identity between Ets-1 and Ets-2 to as low as 39% identity between Ets-1 and PU.l. Though the degree of identity between Ets-1 and PU.l may appear to be rather low, Pastore and Lesk (1991) 6 T a b l e 1.1: Identification of some prominent Ets family members Ets member Means of identifcation Reference v-Ets transforming activity of Avian Leukosis Virus E26 [1,2] Ets-1 cellular counterpart of v-Ets [3,4] E^g^ \ sequence homology to ETS domain |gj Elf 1 / degenerate oligonucleotide PCR [7,8] £ r p / low stringency cDNA probing [9] E74 \ [10] PEA3 y isolation of a sequence specific DNA binding activity [11] Erf / [12] PU.l Spleen Focus Forming Virus integration site [13] Fli-1 Friend Murine Leukemia Virus integration site [14] Elk-1 x [15,16] SAP-1 / ternary complex factors at serum response element [17] Net / [18,19] GABPa counterpart of GABPP through ankyrin repeats [20] Tel identification of a transcript coding for a fusion [21] protein with platelet derived growth factor receptor tyrosine kinase References: [1] Leprince et al., 1983. [2] Nunn et al., 1983. [3] Watson et al, 1988 [4] Gunther et al, 1990. [5] Boulukos et al, 1988. [6] Reddy et al, 1987. [7] Hipskind et al, 1991. [8] Thompson et al, 1992. [9] Lopez et al, 1994. [10] Burtis et al, 1990. [11] Xin et al, 1992. [12] Sgouras et al, 1995. [13] Moreau-Gachelin et al, 1988. [14] Ben-David et al, 1991. [15] Hipskind et al, 1991. [16] Dalton and Triesman, 1992. [17] Dalton et al, 1994. [18] Giovane et al, 1994. [19] Price et al, 1995. [20] LaMarco et al, 1991. [21] Golub et al, 1994. 7 jmu-Ets-1 100% mu-Ets-2 I Dr-Pointed •nmmm hu-Erg mu-Fli-1 moGABPa /T7£>ER81 mu-ER71 mi>-PEA3 3D/--E74 !mu-PU.1 i hu-Elk-1 S3/7U-SAP DmoElf-1 iDr-Yan 96% 95% 71% 69% 67% 65% 65% 63% 62% 59% 54% 53% 47% 39% 85 aa E T S d o m a i n Figure 1.2: The variable location of the conserved ETS domain within the Ets family suggests that this 85 residue DNA binding region represents an autonomous module. Shown beside the schematic of each of the fifteen Ets family members is the degree of sequence similarity to murine Ets-1, the subject of this study. Chapter 1 — The Ets Family of Transcription Factors 8 note that structural similarity in the order of < 1A root-mean-square main-chain deviation can occur among proteins with as little as 50% sequence identity. Furthermore, for a given set of proteins that display a 20%-50% degree of sequence identity, considerable structural similarity may still be present. Many secondary structure predictions have been reported for the ETS domain. All predictions tend to describe a motif involving a recognition helix (Seth et al., 1990; Reddy and Rao, 1991; Wang et al, 1992; Laget et al., 1993; Landsman and Wolffe, 1995; Swindells, 1995). The conspicuous spacing of three essentially invariant tryptophans spaced 18-19 residues apart is reminicent of the DNA binding domain of the Myb proto-oncoprotein which adopts a helix-turn-helix motif (Anton and Frampton, 1988; Ogata et al., 1992; Ogata et al, 1995). The PHD secondary structure prediction algorithm predicts a helix-strand-strand-helix-helix-strand-strand motif for the ETS domain (Rost and Sander, 1993; Rost and Sander, 1994). The accuracy of the PHD algorithm is greatly enhanced when several homologous sequences are given; thus, the ETS domain represents a good test case of its predictive ability. 1.2.2. Ets Proteins are Important Regulators of Differentiation and Cellular Growth The ability of Ets proteins to transform cells highlights their central role in differentiation, notably in blood and immune cell lineages (Bories et al., 1995; Muthusamy et al., 1995). Disruption of the murine Ets-1 gene results in the production of immature B-lymphocytes and decreased T-lymphocyte survival (Bories et al., 1995; Muthusamy et al, 1995). Chromosomal translocation causing the fusion of the Fli-1 and Erg ETS domains to the EWS gene is responsible for Ewing's sarcoma (Zhang et al., 1993; Sorensen et al., 1994; Mao et al, 1994). Similarly, myelomonocytic leukemia is caused by a fusion of the Tel activation domain with the tyrosine kinase activity of the platelet derived growth factor receptor P-subunit (Golub et al, 1994). Integration of Friend murine leukemia virus in the vicinity of the fli-1 gene enhances Fli-1 activity, ultimately leading to transformation. Ets binding sites are found in the Chapter 1 — The Ets Family of Transcription Factors 9 enhancer sequences of collagenase and stromelysin; hence, deregulation of Ets proteins many indirectly support conversion of transformed cells to a metastatic state (Buttice and Kurkinen, 1993; Higashino et al, 1995). The presence of Ets binding sites in the LTR's of human T-cell lymphotrophic virus type 1, herpes simplex virus type 1, mouse moloney tumor virus and the polyoma virus enhancer suggest a strong dependency on Ets proteins to maintain or influence a differentiated state suitable for viral growth (Gegonne et al, 1993; Thompson et al, 1991; Gunther et al., 1990; Xin et al, 1992). chr. 11 chr. 5 EWS Fli-1 (ets) activation DNA binding TEL (ets) PDGFR-p activation tyrosine kinase chr. 22 chr. 12 Figure 1.3 : Chromosomal rearrangements with Ets proteins can result in transformation. 1.2.3. Regulation of Ets Family Members Since Ets family members all bind relatively similar promoter sequences containing the consensus motif of 5'-GGA(A/T)-3', other mechanisms of regulation must exist to ensure gene expresssion is restricted to the target gene. Alone, or in combination, Ets proteins have been observed to use alternative splicing, phosphorylation, intermolecular associations and intramolecular associations to achieve specificity. In contrast, two homedomain proteins, Eve and Ftz, simply use high level transient expression to directly compete for similar binding sites (Walter et al, 1994; Kodadek, 1995). Though interesting regulatory aspects of Ets family members will be covered in this section, particular attention is given to Ets-1 as it is the subject of my disseration. A survey of functional regions mapped to Ets-1 is illustrated in Figure 1.4. 10 28 68 135 161 244 331 370 1 IN | IV 1 V | : •..,.VI..,v, | VII |v ml ix 440 ETS domain v-Ets mutations Intramolecular Inhibition c-Jun association ATF/CREB association CBP (Runt Domain) association Pointed Domain Figure 1.4: A survey of functional regions mapped to Ets-1. The human ets-1 gene is composed of eight exons (I, HI, IV, V, VI, VII, Vffl and IX) of which one (exon VII) is alternatively spliced (Koizumi et al, 1990; Jorcyk et al, 1991). The ETS DNA binding domain is located near the C-terrninus of the Ets-1 gene product and is flanked by a bipartite inhibition region. Three different regions have been mapped which are involved in protein-protein interactions with c-Jun, ATF/CREB and the Runt domain of the core binding protein (CBP) family. The Pointed domain is a conserved sequence of unknown function common to a subset of ets genes {ets-1, ets-2, pointedP2 and yan). Several transcriptional activation domains exist but have been difficult to map. N F K B has been observed to interact directly with the Elf-1 ETS domain (John et al, 1995). A similar observation has not yet been reported for Ets-1. Chapter 1 — The Ets Family of Transcription Factors 11 1.2.4. Cooperativity Between the Ets Proteins and Other Transcription Factors In many promoter and enhancer sequences, the conspicuous clustering of Ets binding sites adjacent to other transcription factor binding sites suggests that many Ets proteins rely on cooperativity to promote gene expression (Figure 1.5). Direct protein-protein interactions do not necessarily have to be involved in all cases. Synergism can be accomplished simply by having a high local concentration of activation domains assembled at an enhancer. A sampling of partnerships include (i) Ets-1/SP1 at the HTLV-1 LTR (Gegonne et al, 1993), (ii) Ets-l/PU.l at the IgM promoter (Nelson et al, 1993), (Hi) Ets-1/AP-1 at the polyoma virus enhancer (Wasylyk et al, 1990) and at the tumor necrosis factor promoter (Kramer et al, 1995), (iv) Ets-2/Myb at the mim-1 promoter (Dudek et al, 1992), (v) GABPcc/GABPp* at the HSV-1 LTR (LaMarco et al, 1991), (vi) Ets-1/GATA-1 at several megakaryocyte specific enhancers (Lemarchandel et al, 1993) and the c-mpl proto-oncogene promoter (Alexander and Dunn, 1995), (vii) Elf-1 / N F - K B at the interleukin-2 receptor cc-chain promoter (John et al, 1995), (viii) Elf-1/retinoblastoma protein (Wang et al, 1993) and (ix) PU.l/TFII-D (Hagemeier et al, 1993). 1.2.5. Coupling of Ets Proteins to the Signal Transduction Pathway A link has been established between Ets proteins and the signal transduction pathway in some of the better characterized partnerships. In each of the four cases illustrated in Figure 1.5, a signal is passed on to an Ets family member to inititate an immediate-early type response. An immediate-early response is defined by the absence of a protein synthesis requirement in order to receive and transmit a signal. For Ets proteins, the signal takes on the form of a phosphorylation by one of several protein kinases. In response, the activated (or derepressed) Ets protein induces the expression of additional transcription factors that will initiate an event such as cell division or differentiation. 12 U x C3 Ut O Cu < CO cn re X o> X C o Cu 3 Cu a re cj 0) re Cu c _o CJ 3 T3 co C (0 i s "53 c 60 0) X C o cn -x! >^  Cu C 'C re cj £ £ 0 J-i2 cn 01 bO « u <0 cn ai X C cu > 01 1-1 a, ai -3 re c 3 fa 60 5 01 c re Cu oi X (50 _g (J re X 0) c 01 bO CN W TJ CU .w (0 C '35 0) T3 01 X w l-l OI X 73 01 o X Cu cn O X CU cn O g re * i cj cn C re (0 s w cn (0 C (0 T3 C re CN Cu c Cu, Oi XI ' G £ « cj 01 Cu cn .52 G 6 0 "S _cn C T3 C < 2 Q >> re co c 3 O) c 0 co co 01 i-, Cu 01 — Cu . h > cj re CO C re CO cn C Cu cn W 0 c 01 01 Ol XI <5J CD ca w '53 bo g g CO CO cn 0) m " ~ T3 C re Ol X Ol x re '33 cn Cu < X C o to re _^ cn CQ C >_ '53 £ o C C cn 01 u C re X c Ol .2 g 13 w Ol Cu o 0 CJ co 01 Ol T 3 CO H, ° ^ x CU co T3 O re CU & u re _2 cu < ^ cn *Z 01 ^ c X bo cu o CU Q X c re c c .2 x QH O <-l T-H CO I C M 2 w c ° .2 ^ .2 CO o 2 x cu Oi X c c 01 £ 3 u, Ol co X re _c •a c re c 01 £ Oi cn C 0 Cu CO 01 2 o- £ C £ .2 « > Cu -x ' c 2 cj re c 2 re x Cu CO O X UJ x oi 3^ X Cu oi cj C o> CO X re Ol co X * '1 O ? re <j g •£ 2 x; c > U Ol — Ol X w _ CSS Cu cn 01 X T3 g X O CU CSS co O re X re U oi cn re C 3 T3 O re o re W oi X oo oi C Ol cn C o W re CSS Cu .5 ^ o X Cu co O X Cu 01 cu cn T3 C re Ol u C re X c Ol y bo Ol X c 01 cj re re CO Ol cn re g 12 g 'aJ — re ^ CJ C 13 .2 1 0 re 2 cu 2 CJ re c _o .2" 'C cj CO C re Chapter 1 — The Ets Family of Transcription Factors 14 At the ets-2 promoter, Erf acts as a repressor until it is phosphorylated by mitogen-activated protein kinase (MAPK) or Cdc-2 kinase at threonine-526 in a portable C-terminal transcriptional repressor domain (Sgouras et al., 1995, Cano and Mahadevan, 1995). As the phosphorylation load on Erf varies according to cell cycle phase, Erf may be an important link between the extracelluar signal to divide and the response in the nucleus. Antagonism between Ets proteins is evident in Drosophila photoreceptor cell development (Scholz et al, 1993). To produce an ommatidium, the basic unit of a Drosophila eye, eight cells must differentiate in a precise order. The penultimate cell to differentiate in the ommatidium is R7, which receives its notice to mature from adjacent R8 cell. If the signal from R8 to R7 is not successfully passed on, a "sevenless" (R7 cell deficient) phenotype is observed. Upon receiving a signal from the boss membrane protein on an R8 cell, the sevenless receptor tyrosine kinase in a neighbouring R7 cell transmits a signal through the ras pathway to two competing Ets proteins, Pointed-P2 and Yan (Brunner et al., 1994; O'Neill et al., 1994; reviewed in Dickson, 1995 and Shokat, 1995). The specific signal is a number of phosphorylations on both Pointed-P2 and Yan by MAPK. In the R7 cell, and what is also similarly observed in many other unrelated cell-types, c-Jun acts as potent inducer of the differentiation response. Like Erf at the Ets-2 gene, Yan operates as a repressor to prevent binding of c-Jun. Once phosphorylated, the Yan repressor protein leaves the promoter and allows an activated Pointed-P2 to bind. Pointed-P2, unlike Yan, does not occlude c-Jun from binding at its adjacent site. Together, Pointed-P2 and a c-Jun induce the expression of two transcription factors, sina and phyllopod, to continue the process of differentiation. Cells can be forced into a quiescent, undifferentiated state under conditions of serum-starvation. Following application of serum, a cascade of gene expression occurs that ultimately leads to cell division and differentiation. At the serum response element (SRE), cooperativity between Ets proteins and serum response factor (SRF) initiate expression of c-Fos (Figure 1.5C; Dalton and Treisman, 1992; Hill et al, 1993; Karin and Chapter 1 — The Ets Family of Transcription Factors 15 Hunter, 1995). Like c-Jun, c-Fos is a member of the AP-1 transcription factor family. The leucine zipper motifs of c-Fos and c-Jun preferentially heterodimerize to produce an active transcription factor that can further stimulate the expression of other downstream transcription factors in a complex cascade of events. At the serum response element, the B-box region of the Ets protein, Elk-1, mediates protein-protein contacts with the unusual tripartite DNA binding domain of dimeric SRF (Janknecht et al., 1994; Shore et al, 1995; Pelligrini et al, 1995). When the B-box is not interacting with SRF, it serves a dual role as an effective intramolecular inhibitor of Elk-1 DNA-binding. The structural details of the interaction between the B-box and the ETS domain of Elk-1 are currently unknown. Derepression of Elk-1 DNA binding and the concominant interaction with SRF to produce a partership that is capable of facilitating gene expression is initiated by phosphorylation of the B-box motif by MAP kinase and phosphorylation of SRF by calmodulin kinase (Whitmarsh et al, 1995; Deng and Karin, 1995; Gille et al, 1995). The recent identification of at least five additional Ets family members that mimic Elk-1 suggest that the interplay between Ets proteins and SRF to tightly regulate c-Fos is very complex (Lopez et al, 1994; Giovane et al, 1994; Pingoud et al, 1994). Casein Kinase II, located in both the cytoplasm and the nucleus, phosphorylates the Ets protein, PU.l , at four positions in vivo (Pongubala et al, 1993; Moreau-Gachelin, 1994). Serine-148 appears to be the most important position since its phosphorylation allows another transcription factor, NF-EM5, to bind at site adjacent to PU.l . The resulting partnership stimulates IgK gene expression (Figure 1.5D). Clearly, direct coupling of Ets proteins to several signal transduction pathways provides much of their requisite regulation. The combinatorial benefits of partnerships can be extended through higher order complexes. Chapter 1 — The Ets Family of Transcription Factors 16 1.2.6. Higher Order Complexes Involving Ets Proteins Higher order complexes are defined as a cooperativity among partnerships of transcription factors to provide a higher degree of control and number of regulatory options at a given enhancer. In some cases, a special type of transcription factor with the ability to induce bends into D N A is required to create an architecture that wil l faciliate the coupling of several transcription factor bound at an enhancer. The promoter architecture in four examples shown in Figure 1.6 is mediated by HMG-box family of D N A binding proteins. This family includes HMG-1, YY1, LEF-1, SRY and IHF. To date, protein-DNA complexes of LEF-1 and SRY have been characterized (Love et ah, 1995; Werner et ah, 1995a). The structural data reveals that in order to bind the minor groove of its target site, the H M G domain produces a considerable bend (ca. 120°) in the DNA. Using the three examples in Figure 1.6 as a guide, partnerships can be loosely divided into two categories depending on whether or not an HMG-box D N A binding protein is necessary to facilitate an interaction between an Ets family member and a transcription factor from another family. Figure 1.6 : Higher order complexes involving Ets proteins (TCRoc - Giese et ah, 1995; TCRP - Wotton et ah, 1994; HIV-1 - Sheridan et ah, 1995). An additional higher order complex involving an Ets family member [Elf-1], an architectural factor [HMG-I(Y)]; and an auxilary factor (NFKB) has been described by John et al. (1995) but not at a similar level of detail as those depicted above. Chapter 1 — The Ets Family of Transcription Factors 17 Members of the core binding factor family (CBFa/CBF(3) and the E-box family do not require DNA bending to facilitate their respective partnerships with the Ets proteins as the interactions occur at adjacent sites in the enhancer. The core binding factor (CBF) family binds DNA as a CBFa/CBFP heterodimer through the Runt DNA binding motif located in the CBFa subunit. The Runt domain is also responsible for heterodimerization (Kagoshima, 1993). Proximal partnerships observed between CBF and Ets-1 do not appear to require HMG-box protein dependent DNA bending based upon three lines of evidence: (i) Ets/CBF interactions have been detected in the absence of DNA. This has allowed the regions responsible for the protein-protein interactions in each protein to be identified (Figure 1.2; Giese et al., 1995) (ii) Experiments that highlight differences in electrophoretic mobility of various protein-DNA complexes demonstrate that ternary complexes are more stable than binary complexes formed by either factor alone (Giese et al, 1995). (Hi) A selected and amplified binding assay is used to identify, a set of strong binding sites from a pool of random oligonucleotide complexes. This assay performed on the Ets/CBF complex revealed a wide variety of spacing and orientations between the consensus sites (Sun et al, 1995) — an observation supported by the presence of both Ets/CBF orientations in the TCRP enchancer (Wotton et al., 1994). Though Ets/CBF is stable enough to observe by bandshift analysis, it is too unstable to DNase I footprint or to enhance the level of transcription above the basal level (Giese et al., 1995). This suggests that the Ets/CBF partnership requires the intervention of additional factors to stabilize the partnership. Addition of the HMG-box protein, LEF, to transcription assays promotes protein-protein interactions between the Ets/CBF complex and an upstream transcription factor, ATF. The bending function of LEF in the TCRa enhancer can also be partially restored by a distantly related HMG-box protein, SRY (Giese et al, 1995). Chapter 1 — The Ets Family of Transcription Factors 18 Two transcription factors brought together by bending cannot fully interact unless the sites are properly phased (Natesan and Gilman, 1993). When basepairs are added in between the ATF and Ets sites of the TCRa enhancer, transcription is diminished. The complex set of transcription factor partnerships occurring at the HIV-1 LTR may be dissected by reconstitution experiments. In other words, each factor is assayed alone, and in combination with other transcription factors, to determine the minimum complement of factors that will support transcription. At the site in the HIV-1 LTR occupied by LEF-1, Ets-1, USF (an E-box binding protein), N F K B and SP1, the minimum complement of transcription factors that will support transcription consists of LEF-1, Ets-1, USF and SP1 (Sheridan et al., 1995). Only a higher order complex could supply a sufficient level of regulation needed to orchestrate HrV-1 gene expression. Substitition of full-length LEF-1 with a deletion mutant of LEF-1 encoding only the HMG-box DNA binding motif also will support transcription, albeit at a reduced level. This suggests that in addition to DNA bending, other portions of LEF-1 serve important roles. When similar assays were performed on the HIV-1 LTR and the TCRa enhancer in vivo, the DNA binding domain alone of LEF-1 was insufficient to support transcription (Carlsson et al, 1993; Sheridan et al, 1995). The combined studies of the TCRa enhancer, the TCR(3 enhancer and the HIV-1 LTR demonstrate the complexity of the interactions among Ets proteins and other transcription factor families. In all three cases, Ets proteins will not avidly bind their enhancer element until they are stablilized by protein-protein interactions with other transcripton factors. Experiments performed on the TCRa enhancer and HIV-1 LTR demonstrate that a specific architecture mediated by proteins capable of DNA bending is necessary to achieve a high level of transcription. Structural studies of the HMG-box family of proteins made it possible to begin visualizing aspects of molecular architecture among transcription factors assembled at an enhancer. The recent crystal structure of the ternary complex of the general Chapter 1 — The Ets Family of Transcription Factors 19 transcription factors, TFII-D, TFII-B and DNA dramatically illustrates how an atomic representation can provide a conceptual framework for all the biochemical and molecular biological data that preceeds it (Nikolov et al., 1995). 1.2.7. Intramolecular Inhibition of DNA Binding Ets-1 has been shown to contain inhibitory sequences both amino and carboxy terminal to the ETS domain. Deletion of either region results in as much as 10 to 20 fold increase in affinity for DNA (Hagman and Grosschedl, 1992; Lim et ai, 1992; Nye et ai, 1992; Wasylyk et al, 1992; Fisher et al, 1994; Petersen et ai, 1995; Jonsen et al., 1996). Several natural variants of Ets-1, differing in either the amino or carboxy sequences flanking the ETS domain, also enhance DNA binding. An alternatively spliced isoform of Ets-1, p42, lacking exon VII, binds to DNA with high affinity (Jorcyk et al, 1991; Wasylyk et al., 1992; Fisher et al., 1994). Exon VII encodes the entire amino inhibitory sequence. Phosphorylation has been identified as a potential mediator of the full-length p51 E t s" 1 activity. Treatment of p51 E t s _ 1 with calcium-dependent calmodulin kinase II in vitro results in a further 14-fold decrease in DNA-binding affinity beyond the initial 10-fold difference observed between p42Ets~1 and p S l ^ " 1 (Fisher et al., 1992). Exon VII specific phosphorylation in vivo has been attributed to a different calcium-dependent kinase, myosin light chain kinase (Koizumi et al, 1990; Fleischman et al, 1993). The phosphorylation load of human Ets-1 is coincident with the cell-cycle (Fleischman et al, 1993). These observations together lead to the conclusion that phosphorylation serves a important role in modulating the activity specified by exon VII. However, a precise set biological consequences has yet to be determined. The oncogenic viral form of Ets-1, v-Ets, contains a 16 amino acid substitution within the carboxy terminal inhibitory sequence that also results in increased DNA binding affinity (Leprince et al, 1983, 1994,1993; Nunn et al, 1983,1989; Hagman and Grosschedl, 1992; Lim et al, 1992; Hahn and Wasylyk, 1994). This substitution is predicted to disrupt a helix that is an essential component of the bipartite inhibitory module (Donaldson et al, Chapter 1 — The Ets Family of Transcription Factors 20 1994; Petersen et al., 1995; Donaldson et ai, 1996). The predicted helical disruption has not been confirmed by direct structural characterization; however carboxy-terminal deletions into the helix of c-Ets-1 abrogate the inhibitory activity (Hahn and Wasylyk, 1994; Jonsen et al, 1996). 1.3. The Structural Characterization of Murine Ets-1 1.3.1. Rationale Biochemical studies of Ets proteins have shown that phosphorylation, intermolecular and intramolecular interactions all contribute to a complex regulatory palette. Structural studies provide the necessary molecular detail to complement the current body of biochemical knowledge and to provide a molecular mechanism for uncovering new research paths. The broad objective of my thesis, therefore, is to present a compilation of data from biophysical, biochemical and nuclear magnetic resonance spectroscopy measurements to begin examining functional and regulatory aspects of murine Ets-1 at the molecular level. 1.3.2. Thesis Organization The data presented in this thesis is organized according to the milestones encountered during structural studies of any biomolecule. Specifically, these milestones include (i) the design of highly optimized expression and purification protocols to obtain milligram quantities of protein, (ii) the assignment of the polypeptide backbone by heteronuclear NMR experiments to derive a precise secondary structure, and (Hi) the subsequent assignment of side chain resonances leading, finally, to (iv) the calculation of a tertiary structure. By homology, the ETS domain proper spans a region of 85 residues in the C-terminal portion of murine Ets-1. My research began with a gene fragment provided by the Graves laboratory encoding a 110 residue polypeptide encompassing the ETS domain and 25 amino-acids extending to the native C-terminus of Ets-1, denoted at Ets-1AN331 Chapter 1 — The Ets Family of Transcription Factors 2.1 in Figure 1.7. The C-terminal tail was required to maintain solubility of the ETS domain containing fragments, and fortuitously played an important role in the inhibition of Ets DNA binding. Ets-1AN331 therefore became a desirable candidate for analysis by multidimensional heteronuclear NMR methods. Chapter Two summarizes the cloning strategies used to overexpress the 110 residue ETS domain containing fragment in E. coli grown in isotopically enriched media. Characterization of the recombinant protein by circular dichroism spectroscopy, Fourier transform infrared spectroscopy and ultraviolet laser crosslinking is also presented. Since the structure of the ETS domain was presumed to be novel, a comprehensive description of the secondary structure using FTIR, CD and NMR represented the next milestone to overcome. In Chapter Three, I present evidence based upon the secondary structure of the Ets-1AN331 deletion mutant demonstrating that the ETS domain is not a novel motif. Rather, it is a member of the diverse winged helix-turn-helix (wHTH) superfamily of DNA binding domains. A wHTH motif is characterized by a helix-turn-helix buttressed by an anti-parallel (3-sheet. This key discovery provided immediate insights into the mechanism of DNA binding by the ETS domain. The essence of Chapter Three was published in Donaldson et al. (1994). The NMR-derived tertiary structure of Ets-1AN331 presented in Chapter Four conclusively demonstrates that the ETS domain is a wHTH member (Figure 1.8). The results of this chapter were published in Donaldson et al. (1996). The Ets-1AN331 structure is discussed in light of sequence conservation among other ETS domains and the structure of the related DNA binding domain of Ets member, Fli-1 (Liang et al, 1994b). The chapter ends with a discussion on the mode of DNA binding based upon results provided by the laboratories of Drs. G.M. Clore and A.M. Gronenborn in advance of publication (Werner et al., 1995b). 22. Bipartite Inhibitory Region ETS domain Ets-1 H H H S S H H S S H 280 440 ETS domain Ets-1 AN280 301 440 ETS domain Ets-1 AN301 331 440 ETS domain Ets-1 AN331 Figure 1.7 : Three N-terrninal deletion mutants of murine Ets-1 are featured in this thesis. A schematic of full-length murine Ets-1 is shown for comparison. The secondary structures found in the region spanning residues 280-440 are denoted as H = cc-helix, S = fi-strand under the Ets-1 schematic. All mutants bind DNA in a sequence specific manner. Experiments leading to the tertiary structure of the minimal ETS domain containing protein, Ets-1AN331, form the bulk of the data discussed in this thesis (Chapters 2, 3 and 4). The ETS domain possesses a winged helix-turn-helix fold. Secondary structure boundaries obtained from Ets-1AN280 analysis prompted the construction and structural characterization of a minimal autoinhibited Ets-1 deletion mutant, Ets-1AN301 (Chapter 5). Since both the N-terminal and C-terrninal flanking inhibitory regions must be present to form an inhibition module, Ets-1AN331 is not autoinhibited. 21 Figure 1.8 : The tertiary structure of the murine Ets-1 ETS domain places it in the winged helix-turn-helix family of D N A binding proteins, oc-helices are designated by tubes and the letter H . B-strands are designated by arrows and the letter S. The secondary structure of the ETS domain follows the order H l - S l -S2-H2-H3-S3-S4. Helix H3 is the recognition helix that binds the major groove of D N A . Helix H4 constitutes the C-terminal half of the inhibition module. Chapter 1 — The Ets Family of Transcription Factors 24 Sequences flanking both the amino and carboxy termini of the Ets-1 ETS domain form an inhibitory module that is capable of attenuating the DNA binding affinity of the ETS domain by at least one order of magnitude (Section 1.2.7). An amino terminal Ets-1 deletion mutant, Ets-1AN280, embodies the "intramolecularly inhibited" or "autoinhibited" DNA binding characteristics of the full-length protein. This deletion mutant therefore served as a starting point in Chapter Five for the characterization of the structure of the inhibition module. The backbone atom (H, N , C a , C ) assignments of Ets-IA N280 by heteronuclear NMR methods provided an avenue for a precise derivation of its secondary structure (Skalicky et al, 1996). This NMR study showed that the bipartite inhibition module of Ets-1 was composed of two oc-helices in the amino-terminal portion and one oc-helix in the carboxy-terminal portion (Figure 1.7). The extreme amino-terminal 20 residues of Ets-1AN280 formed a continuous stretch of extended conformations and were predicted to be non-essential for the inhibitory activity. The residue boundaries of the carboxy-terminal oc-helix in Ets-1AN280 were similar to the boundaries those observed in the structure of Ets-1AN331 depicted in Figure 1.8. NMR chemical shift perturbation analysis defined a surface on the Ets-1AN331 domain that interacted with the amino-terminal sequences of Ets-1AN280. This surface was adjacent to, rather than coincident with, the surface specified by the recognition helix of the ETS domain. By not directly occluding the major DNA binding surface, an allosteric mechanism for autoinhibition was proposed. Supported by proteolysis studies (Petersen et al, 1995; Jonsen et al, 1996), progressive deletion analysis (Wasylyk et al, 1992) and the NMR secondary structure determination of Ets-1AN280 (Skalicky et al, 1996), I cloned and overexpressed a minimal deletion mutant of Ets-1, Ets-1AN301, that retained both the inhibitory and DNA binding activities (Figure 1.7). Following a similar strategy used to determine the structure of Ets-1AN331, I obtained enough structural information to propose a global fold for Ets-1AN301. Two helices in the N-terminal inhibitory region combine with one helix in the Chapter 1 — The Ets Family of Transcription Factors 25 C-terminal inhibitory region and one helix (HI) of the ETS domain to form an anti-parallel four helix bundle-like module(Figure 1.9). ETS Domain Figure 1.9 : A model for the inhibition module of murine Ets-1 based on data obtained from chemical shift difference analysis between Ets-1AN331 and Ets-1AN280, and data obtained from an ongoing structural analysis of Ets-1AN301. (Panel A) Two oc-helices (HI1/HI2; in green) from the N-terminal inhibitory region are assembled with one a-helix from the C-terminal inhibitory region (H4; in green) and one a-helix (HI; in red) from the ETS domain to form an anti-parallel four helix bundle. Helices H2 and H3 comprising the helix-tum-helix motif of the ETS domain are coloured red. The four-stranded anti-parallel fi-sheet of the ETS domain is coloured blue. A black arrow points to the inhibition module, Helix HI, that unfolds upon DNA-binding. (Panel B) N-terminal and C-terminal inhibitory sequences (green) immediately flank the ETS domain (red). Helix HI of the ETS domain is also an integral part of the inhibition module. Therefore, it likely represents a important link in the structural communication that must occur between the inhibition module and the ETS domain. Werner et al. (1995b) showed that the indole ring of a tryptophan (W338) located in Helix H I intercalates into the D N A duplex. From N M R studies of the Ets-1AN301 deletion mutant, I observed that residues Chapter 1 — The Ets Family of Transcription Factors 26 surrounding W338 in Helix HI and more importantly, the inhibition module, must undergo a conformational change in order to allow W338 to become available for DNA binding. An unexpected loss of secondary structure in the first helix of the inhibition module upon formation the DNA complex was observed for the deletion mutant, Ets-1AN280, by circular dichroism spectroscopy and hypersensitivity to proteolysis(Petersen et al., 1995). Following the model of the inhibition module depicted in Figure 1.9, helical unfolding provides a mechanism for releasing W338 from its associations with the inhibition module to enable DNA intercalation. Furthermore, increased off-rate kinetics observed by Jonsen et al. (1996) may be rationalized by the drive of the unfolded inhibition helix to refold, and thus, release the ETS domain from the DNA. In Chapter Five, I present an amide hydrogen exchange and a 1 5 N relaxation study of Ets-1 AN301, the minimally sized, autoinhibited fragment of Ets-1. In a stable helix, the hydrogen bond between the carbonyl oxygen of residue (i) and the amide hydrogen of residue (i+4) is not readily broken; therefore, the amide hydrogen exchanges with the solvent at a slow rate. From this amide hydrogen exchange study, a higher exchange rate was observed for amide hydrogens throughout the first inhibition helix, relative to the other five helices in Ets-1 AN301. This finding, combined with the observance of a number of fast motions from the relaxation study, suggests that the first inhibition helix is inherently unstable in order to accommodate the folded and unfolded states. Ideally, a mechanistic description of intramolecular inhibition should incorporate structural data from the four possible permutations — the uncomplexed-uninhibited, the uncomplexed-inhibited, the complexed-uninhibited and the complexed-inhibited forms (Table 1.2). To date, structural data is available for all but the complexed-inhibited form (Liang et al., 1995; Werner et al, 1995b; Donaldson et al, 1996). With the inclusion of dynamical data and the knowledge that an inhibition helix must unfold, a mechanism is 27 Table 1.2 : Summary of Ets-1 deletion mutants discussed in this dissertation* Protein Size DNA binding activity Conclusions Ets-1AN331 110 aa. derepressed (1) The tertiary structure of ETS domain defines it as a winged helix-turn-helix motif. (2) A helix is identified in the C-terminal tail spanning residues 415-440. This helix packs against the first helix of the ETS domain. (3) The C-terminal tail is believed to interact with sequences immediately flanking the N-terminal side of the ETS domain to form an inhibitory module. The proximity of the C-terminal tail, particularly the C-terminal helix, to the N-terminal residues observed in the Ets-1AN331 structure provides the first structural proof for the proposed interaction. Ets-1AN280 160 aa. repressed (1) The secondary structure of Ets-1AN280 identifies two helices in the N-terminal portion of the inhibitory module. (2) Chemical shift perturbation analysis between Ets-1AN331 and Ets-1AN280 identifies the contact surface of the N-terminal helices to be localized to the C-terminal helix and the first helix of the ETS domain. Ets-1AN301 139 aa. repressed (1) The N-terminal boundary of this deletion mutant is the N-terminal boundary of the first helix of the N-terminal inhibitory flank. This deletion mutant represents the minimal intramolecularly inhibited fragment of the Ets-1 protein. (2) A full complement of NMR experiments were measured to determine the structure of the intramolecularly inhibited ETS domain. The inhibitory module is defined as a bundle of four anti-parallel a-helices. (3) Dynamic characterization of the Ets-1 AN301 amides reveals that the first, most N-terminal helix of the inhibition module is unstable relative to the other helices in the inhibition module. This observation is key to the development of an allosteric mechanism of intramolecular inhibition that involves the unfolding of this helix upon DNA binding by the ETS domain. *Much of this work, to date, has been published. The reader is referred to Donaldson et al. (1994), Petersen et al. (1995), Skalicky et al. (1996) and Donaldson et al. (1996). Chapter 1 — The Ets Family of Transcription Factors 28 proposed in Chapter Five based on a molecular switch that highlights the involvement of W338. 1.3.3. Collaborations Much of the data collected this thesis would not have been possible without a combined collaborative effort with the laboratories of Drs. Barbara Graves (Cellular and Molecular Biology Department, University of Utah) and Tom Alber (Department of Biochemistry and Molecular Biology, University of California at Berkeley). Results from studies by B. Graves and other researchers in the Ets field will be incorporated into the discussion of this thesis. However, unless specifically stated, all of the data presented herein represents research that I personally performed during my graduate studies. C h a p t e r 2 Overexpression and Purification of ETS Domain Proteins Nearly any protein can be envisioned a potential structural project. However, there is one prerequiste which can often eliminate an idea from the very start — the availability of milligram quantities of pure, soluble and active protein. If the structure of only a domain is sought, additional techniques must be employed to delineate the boundaries of the domain. In this chapter, I will begin by discussing the strategies employed to express a variety of protein fragments containing the DNA binding domain of murine Ets-1, termed the ETS domain. For cases involving fusion proteins, new purification strategies were necessary. Sequencing, mass spectrometry and immunoassays were routinely used to confirm the identity of each overexpressed protein. Lastly, methods of isotopic labelling, a prerequiste for a heteronuclear NMR study, will be presented. 2.1 Overexpression Systems Central to any structural study is the requirement for homogenous, milligram quantities of soluble and active material. Many smaller proteins, that are destined for homonuclear NMR or crystallographic studies, may be purified from their original biological source. More often than not, however, expression in a suitable bacterial, fungal or mammalian host is required to obtain adequate quantities of material for structural characterization. The term "overexpression" is defined as a significantly higher level of protein production than that of a typical protein in the expression host. A level in excess of 10% of total protein mass is considered an arbitrary threshold for sucessful overexpression. In some instances, expression levels of 50% (75 mg/L) have been achieved. Moderate expression is attributed to a level in the 1% range. For a typical 1 liter bacterial culture consisting of 10 1 2 cells, a 10% level of overexpression corresponds to approximately 15 mg of protein (New England Biolabs Catalog, 1993). 29 Chapter 2 — Overexpression and Purification of ETS Domain Proteins 30 The molecular biology of Escherichia coli is well understood and, consequently, many overexpression systems are available for this host. Methods leading to the optimization of these systems are numerous and generally exploit every aspect of protein biosynthesis. For example, potent, inducible promoters and efficient transcriptional terminator sequences ensure that large amounts of mRNA are produced in a tightly regulated manner. Non-coding sequences are optimized to enhance the stability of the mRNA and to provide an optimal ribosome binding site. In addition to vector optimizations, the use of protease deficient straints may provide additional insurance against unwanted degradation of the expressed protein. Rather than summarize the biochemical merits of modern expression systems, I will restrict my discussion in this chapter to those specifically employed for Ets-1 overexpression and refer the reader to an excellent review by Gross (1989). A summary of commerical systems up to 1992 may be consulted in Donaldson (1992). 2.2. The Necessity of Isotope Labelling The complexity of homonuclear proton NMR spectra increases rapidly with molecular weight. The severe degeneracy in the spectra observed for proteins > 80 residues can be reduced to some extent by homonuclear three-dimensional experiments. However, it was only when the technology arrived to make isotopically labelled proteins and to collect multidimensional heteronuclear did NMR become a serious contributor to the field of macromolecular structure determination (Bodenhausen and Ruben, 1980; Fesik and Zuiderweg, 1988; Fesik and Zuiderweg, 1990). The effective sensitivity of spin 1/2 heteronuclei is expressed as the product of their natural abundance and the cube of their gyromagnetic ratios (Equation 2.1). effective sensitivity - (abundance) • ( y3) [2.1] Chapter 2 — Overexpression and Purification of ETS Domain Proteins 31 averaging cannot completely allieviate the sensitivity issue as the number of transients, N , can only provide a N 1 / 2 increase in signal-to-noise. Table 2.1: Natural abundances of spin 1/2 nuclei found in proteins0 Spin 1/2 isotope Natural Abundance (%) Relative Sensitivity a Absolute Sensitivity^ lH 99.98 1.00 1.00 1 3 C 1.11 ' l .6x l0- 2 1.8 xlO" 4 1 5 N 0.37 1.0 x IO"3 3.8 x IO"6 3 1 P 100.00 6.6x10-2 6.6 x 10-2 aRelative sensitivity varies with y3. bAbso lute sensitivity is the product of the natural abundance and the relative sensitivity. It can be seen that to observe 1 3 C and 1 5 N , it is necessary to enrich their abundance by isotopic labelling. Even at 100% abundance, the sensitivity of * 3 C and ^ N is still much lower than ^H. c adapted from Wiithrich, 1986 Since the gyromagnetic ratio is a constant, sensitivity gains are achieved by enriching the isotopic abundance of 1 3 C and 1 5 N . From Table 2.1, isotopic enrichment results in a gain in sensitivity of ca. 100-fold and 300-fold respectively for 1 3 C and 1 5 N . Bacterial expression systems offer the easiest route to isotope labelling as the media can be easily tailored to a specific application. For example, if uniform labelling of 1 3 C is desired, a minimal medium is used in which the sole carbon source is 13Ci-6-glucose, 1 3 Ci / 2-acetate, 13Ci_3_glycerol or 1 3 C ^-succinate. Uniform labelling of 1 5 N is achieved with a minimal medium in which an ammonium salt such as 1 5NH4C1 serves as the sole nitrogen source. Chapter 2 — Overexpression and Purification of ETS Domain Proteins 32 minimal medium in which an ammonium salt such as 1 5Nri~4Cl serves as the sole nitrogen source. Selective labelling exploits well-characterized biosynthetic pathways to isotopically label individual residues or classes of related residues (Mcintosh and Dahlquist, 1990). In practice, selective labelling is achieved by using a synthetic rich medium containing every appropriately unlabelled/labelled amino acid in a vast excess over the metabolic requirements of the bacteria. In this manner, import of labelled materials is favored over de novo biosynthesis. The synthetic rich medium additionally contains nucleotides, metals and vitamins. For some amino acids, such as tyrosine, that lie at the end of metabolic pathways for amino acid biosynthesis and are regulated enzymatically and genetically by feedback inhibition, the effect of isotopic dilution and/or metabolic redistribution into other amino acids is low. Amino acids or a related metabolite, which are themselves precursors in biosynthetic pathways, may result in more than one residue being labelled. For example, addition of 15N-glutamate will produce a protein containing 15N-glutamate, 15N-glutamine and several other amino acids, depending on the exact composition of the growth medium. To prevent isotopic dilution, it is then desirable to use auxotrophic bacteria with mutations in genes encoding enzymes for amino acid biosynthesis. For example, an auxotrophic £. coli strain deficient in aspartyl, aromatic, branched-chain and alanine-valine transaminases (aspC, tyrB, ilvE, avtA) cannot transfer the oc-15N of glutamate to aspartate, valine, isoleucine, leucine, phenylalanine or tyrosine. Much of the pioneering work related to the use of auxotrophic strains is attributed to D. M . LeMaster. An early account describing the utility of selective labels to classify residue type in thioredoxin, a ca. 100 residue protein, is found in LeMaster and Richards (1985, 1986). Today, selective labelling is not restricted to 1 5 N labelling. A variety of 1 3 C and 2D isotopically labelled amino acids are commercially available. Either alone or in combination, selective labels are powerful tools for simplifying NMR spectra. 33 A 1 2 C H 3 1 2 C H 3 \ V ' 2 CH I V 12c I 1 2 C H 3 0 I II 1 2 C H 2 0 I II 1 1 2 C H 2 O I II I II 1 5 N 12C —12C — 1 1 1S N _ 1 2 C _ 1 2 C — 1 1 1 II 1 5 N — 1 2 C — 1 2 C g 1 1 1 1 H 1H 1 1 'H • 'H t 1H 1H ala leu asp B <2CH3 1 2 C H 3 O O" \ / 1 2 CH I 1 2 c I 1 2 C H 3 • I O ' 2 C H 2 O II I II i 1 2 C H 2 O I II 14N I I — 12C — 1 II i II 12C 15N —12C — 12C 1 I 14N I 1 II_12C —12C | I <H I *H I 1 "H 1H I m I 1H ala leu asp O D D D proteo-tyr deutero-phe deutero-trp Figure 2.1 : Examples of Selective Labelling. (Panel A) An amide ( 1 5N, 1 H pair) in a uniformly 15N-labelled sample is an effective residue marker as there is typically only one 1 5 N per residue. Selective labelling is achieved by growing bacteria in a synthetic rich medium. To produce the labelling pattern in Panel B, all amino acids that were added to the medium were unlabelled except for leucine (as 15N-leucine). Selective amide labelling is very useful during the initial stages of backbone resonance assignment. (Panel C) Selective deuteration is also effective at simplifying spectra. In the example shown, addition of deuterated phenylalanine and tryptophan at the ring positions to a synthetic rich medium containing proteo-tyrosine and all other amino acids results in a simplification of the aromatic resonances. Only the and H e aromatic resonances of tyrosine residues will be observed. Chapter 2 — Overexpression and Purification of ETS Domain Proteins 34 NMR experiments involving 1 3 C and 1 5 N selective labels work well for the same reasons that preclude observance of these heteronuclei at natural abundance. Relative to the isotopic abundance of the intended amino acid, any spurious background labelling is generally negligible and therefore below the signal-to-noise threshold dictated by the NMR experiment. On experimental terms, heteronuclei offer two important advantages. First, heteronuclei offer a greater chemical shift dispersion in proteins (180 ppm in 1 3 C and 60 ppm in 1 5 N versus 12 ppm in : H). Second, NMR relaxation mechanisms of a proton directly bonded to a heteronucleus can be derived mathematically and approached experimentally. As NMR relaxation is sensitive to the intrinsic motions of the protein, valuable dynamics data on the protein backbone may be obtained via the isolated 1 5 N - 1 H amide bond spin system. The deuteron, 2 H , is also a very effective isotope label. Incorporation of deuterated amino acids renders that residue and others in the same metabolic pathway "invisible"; thus, any spectrum can be simplified. For example, the aromatic region of the 1 H spectrum that spans a narrow region of 7.0 - 8.0 ppm can be simplified by including deuterated tyrosine, phenylalanine or tryptophan into the synthetic rich medium. A similar method may be used to simplify the methyl region (0.0 - 1.5 ppm in 1 H) by deuterating one or more of valine, leucine or isoleucine. The relaxation rate of a heteronucleus bound to 2 H is much slower than the corresponding relaxation rate of a heteronucleus bound to 1 H . In practical terms, a reduced relaxation rate means that a greater amount of signal will be available to acquire after application of a pulse sequence. A bacterial growth medium containing 70% D2O / 30% H2O provides an adequate amount of deuteration on the side chains to reduce the rate of 1 3 C relaxation while still ensuring that there is enough 1 H available to detect. The ability to uniformly and selectively label biomolecules with a variety of isotopes is central to modern NMR experiments. Isotope labels not only can greatly simplify Chapter 2 — Overexpression and Purification of ETS Domain Proteins 35 interpretation of a spectrum but also can enhance the data content. Examples of labelling discussed throughout this section are summarized in Figure 2.1. There is a strong impetus to optimize overexpression systems when using isotopically labelled media as the cost can quickly become prohibitive. Optimization of overexpression is evident when isotopic labelling in 1 3 C as 1 litre of minimal medium consumes $1500 of 13C-glucose ( 3 g required; at the time of writing glucose was listed at $500/gram ; Isotec, Miamisburg, OH) and 1 litre of rich-media consumes $4000 of algal extract (10 g required at $400/gram; Isotec). For adequate signal-to-noise, at least a 1 mM protein sample in a 500 |iL volume is desired. Thus, for a typical 15 kDa protein, a moderate degree of overexpression (7.5 mg/liter) from a 1 liter fermentation is required to produce an NMR sample. In the following discussion of the strategies used to produce labelled ETS domain proteins, it will be seen that this optimal case could not be realized, therefore significantly elevating the cost of preparing protein samples for NMR characterization. 2.3. Overexpression of a Minimal ETS Domain Containing Protein 2.3.1. Background As defined by homology among over thirty ets gene family members, the ETS domain spans a region of 85 residues and is responsible for the DNA binding activity of this class of transcription factors and proto-oncoproteins (Karim et al, 1992). Previously, Gunther et al. (1990) screened a Xgtll cDNA library for cellular proteins that bind the Mouse Moloney Tumor Virus (MMTV) long terminal repeat (LTR). One candidate was the murine homolog of the human ets-1 gene. The 440 residue murine Ets-1 protein displays 100% sequence identity to its human counterpart in the ETS domain. Southwestern blotting and transactivation assays showed that the murine Ets-1 protein was capable of driving transcription from a 5'-GGA-3' core motif in the MMTV LTR, thus reinforcing its role as an important cellular regulator. Chapter 2 — Overexpression and Purification of ETS Domain Proteins 36 ETS domain 280 312 330 330 440 I Ets-1 Ets-1 AN280 Ets-1AN312 Ets-1 AN331 416 Ets-1AN331AC416 Figure 2.2: Murine Ets-1 deletion analysis performed by members of the Graves laboratory (Petersen et al, 1995; Jonsen et al, 1995) to identify a minimally sized fragment of Ets-1 that still retained sequence specific DNA binding activity. Ets-1AN280 (18 kDa) and Ets-1AN312 (14 kDa) were identified as metastable protein fragments from a limited proteolysis study. Ets-1AN331 (12 kDa) begins at the ETS domain and extends to the native C-terminus. A deletion mutant corresponding to the ETS domain (10 kDa) defined by homology among Ets family members had poor solubility and expression. Ets-1AN331, therefore, is the minimally sized, soluble protein fragment of Ets-1 that retains full sequence-specific DNA binding activity. Chapter 2 — Overexpression and Purification of ETS Domain Proteins 37 pAED4-10kDa pAED4-ets ETS-domain [331-415] T7 cys term T7 ETS-domain [331-440] i cys —r-cys Bam HI 1 - = term pAED4-ala pAED4-asn pET3-ser ETS-domain [331-440] T7 ala —r— ala ETS-domain [331-440] T7 ala ETS-domain [331-440] T7 asn ser J-cm— term term Figure 2.3 : First (Panel A) and second (Panel B) generation clones assayed for expression and suitability for NMR structural analysis. Clones in the pAED4- and pET3- series are T7 polymerase driven, and were transformed into E.coli BL21::DE3 cells. Cotransformation of the plasmid, pLysS, allows the constitutive low-level expression of T7 lysozyme which may enhance the level of protein production by reducing T7 polymerase activity, thus allowing translational processes to catch up (Moffatt and Studier, 1987). In all of the constructs discussed in this chapter, the suffix ets denotes the wild-type gene sequence. The suffixes, ala and asn and ser, denote Ets-1 double mutants (C350A, C416A), (C350A, C416N) and (C350N, C416S), respectively. A transcriptional termination sequence is represented as a term box in the figure. Plasmids pAED4-10kDa, pAED4-ets, pAED4-ala and pAED4-asn were obtained from Dr. Barbara Graves (University of Utah). The plasmid, pET3-ser, was subcloned from a plasmid harbouring a synthetic Ets-1 gene fragment. Chapter 2 — Overexpression and Purification of ETS Domain Proteins pCW-ets pCW-ala pCW-asn pCW-ser tec tac tac tac Nde[ Bam H l l _ T " cys Ets-domain [331-440] "1)-cys Ets-domain [331-440] % ——I 1 ala Ets-domain [331-440] *~g}-ala Ets-domain [331-440] g| i ser Hindm _J Bg/ll Bg/ll term Figure 2.4 : The 12 kDa Ets-1AN331 construct was recloned into a new overexpression context based upon the tac promoter. Plasmids are named according to the convention described in Figure 2.3. 39 Table 2.2 : Strains tested for Ets-1 expression Standard Cloning Strains: JM101 F+ endAl, supE, recA+, lac, lactf DH5a F', endAl, hsdRl7 (rk; mk+), supE44, thi-1, recAl, gyrA (Nalr) relAl AdacZYA' argF)U169, deoR (<p80AlacA(lacZ)M15) T7 Polymerase Specific: BL21 F+ ompT, Ion; hsdS (rB"' mB"), A.DE3 Miscellaneous Strains: MV1190 endAl; supE,recA+, lac, lacla KS476 F", degP, lac; AphoA CAG597 F + , tpoH165 (am), zhgr.TnlO, lacZ (am), supC (ts), lac (am), trp (am), pho (am), mat (am), rpsL Y1090 F", A(lac)U169, lotrlOO, araD139, rpsL(Str), supF, mcrA, trpC22::TnlO(pMC9; TetrAmp1) TBI F', araA(lac-proAB), rpsL (Strr) [<j)80 AlacA(lacZ)M15), hsdR(rk; mk+) NM522 F", lacflA(lacZ)M15, proA+B+, supE, thi, A(lac-proAB), A(hsdMS-mcrB) (rk; mk; McrBC') Wild-Type Strains: R1360 F+ endAl Topp-1 F", proAB, lacflZAMlS, TnlO, teF, riF Topp-2 F", proAB, ladlZAMW, TnlO, tetr, riF Topp-3 F", proAB, ladaZAM15, TnlO, tetr, riF, kanr Topp-4 F", proAB, lactfZAM15, TnlO, teF, riF Topp-5 F", proAB, lacttZAMVo, TnlO, tetr, riF Topp-6 F", proAB, lacIaZAM15, TnlO, teF, riF Chapter 2 — Overexpression and Purification of ETS Domain Proteins 40 2.3.2. First Generation Constructs In a T7 polymerase driven bacterial overexpression system (Studier and Moffat, 1986; Rosenberg et al, 1987) a 10 kDa fragment of Ets-1 encompassing the 85 residue ETS domain demonstrated sequence specific DNA binding activity. However, the protein was expressed at a low level and was relatively insoluble (B. Graves, personal communication). Given that sequence homology does not necessarily define the absolute bounds of a domain, the full-length Ets-1 protein was partially proteolysed with trypsin and chymotrypsin to generate fragments dictated by the protein fold (Petersen et al, 1995; Jonsen et al, 1996). The fragments containing ETS domain sequences were tracked by Western blotting with an ETS domain specific antibody. Among two proteolytic fragments (Ets-1AN280, Ets-1AN312) and one deletion mutant (Ets-1AN331; Figure 2.2), Ets-1AN331 was chosen as the lead candidate for crystallographic and NMR analyses owing to its minimal size and acceptable solubility. The corresponding first-generation clone provided by Dr. Barbara Graves (University of Utah) is designated in Figure 2.3 as pAED4-ets. Extracts of Ets-1AN331 from pAED4-ets retained full sequence specfic DNA binding activity. EtSrlAN331 was considerably more soluble than the minimal expressed domain. From a purified sample, encouraging data was obtained from a low-concentration, one-dimensional lH jump-return NMR experiment described in Chapter 3. Expression levels of pAED4-ets were ca. 2-4 mg/L culture. At the time, it was decided that there was still opportunity to improve expression levels and solubility of this 12 kDa fragment of Ets-1. 2.3.3. Second Generation Constructs of the Ets-112kDa fragment Attempts were made to improve the pAED4-ets clone in two key areas, namely by mutating cysteine residues and by designing a bacterially codon-biased synthetic gene. In Ets-1AN331, there are two cysteines, one of which is conspicuously, but not absolutely, conserved across the family. DNA footprint analysis (Nye et al, 1992) strongly suggested that Ets-1 bound to the major groove of DNA as a monomer. Hence, the free thiols did not likely serve a structural role by promoting intermolecular dimer Chapter 2 — Overexpression and Purification of ETS Domain Proteins 41 formation. Moreover, if a disulfide linkage was present it would likely be absolutely conserved across the family. Of the two cysteines in Ets-1AN331, only one cysteine (C350) is found in the ETS domain proper. Free thiols are generally undesirable since multimers can form if the protein sample oxidizes over time. Precautions, such as purging the oxygen out the sample with inert gases, followed by treatment with DTT and EDTA and subsequent storage in an inert atmosphere are technically difficult to achieve and maintain. Thiol oxidation and multimer formation are hazardous to any structural study since they can introduce mosaicity into a crystal and degrade the quality of NMR data. To eliminate this potential problem, mutations were introduced into the wild-type gene using site-directed mutagenesis to produce the thiol-free (C350A/C416A) and (C350A/C416N) double mutant proteins. Alanine was chosen as an aliphatic side chain substitute. In an alignment of Ets family members, aspargine and serine are found at similar positions as the cysteines in other members. No differences could be detected in the footprint or binding affinity of the mutant proteins, relative to the wild-type protein (B. Graves, unpublished data). However, neither the expression nor the solubility of Ets-1AN331 was improved by these mutations; therefore, the wild-type protein fragment was chosen for structural studies. Overexpression of mammalian genes in E. coli may place an excessive demand on pools of rare tRNAs (Robinson et al, 1984). To determine the effect of codon bias on Ets-1 protein expression, a synthetic gene was made in the Graves laboratory as the ligation product of five overlapping oligonucleotide duplexes. At every residue position in the synthetic gene, the selected codon corresponded to the highest abundance bacterial tRNA. The new gene also included a double mutation at the cysteines (C350N/C416S) and a new ochre (TGA) stop codon. Amber (TGA) stop codons should be avoided since most strains of E. coli are decended from K12 and possess the supE mutation (Bachman, 1972). This lesion results in the expression of a small pool (ca. 1 %) of glutamine amber-Chapter 2 — Overexpression and Purification of ETS Domain Proteins 42 suppressing tRNAs. Often, read-through events are negligible, but for the special case of an overexpressed VP16 activation domain, amber suppression accounted for up to 25 % expression of a longer secondary product (Donaldson and Capone, 1992). Relative to the original construct, the expression level of the T7 driven synthetic gene in plasmid pET3-ser (Figure 2.4) did not differ significantly from that of the wild-type gene. This outcome suggested that the difficulty in expression was due to factors other than codon usage, such as: (i) misfolding and subsequent protein degradation in the bacterium, (ii) a non-optimal choice of the N-terminal residues (Bachmair and Varshavsky, 1986), and (Hi) leakage of the promoter which may have been enhancing (iv) the inherent toxicity of Ets-1 AN331. Regardless of the expression level, all of the T7 polymerase based constructs of Ets-1AN331 had to be discounted for NMR studies. In a minimal M9 medium, growth was very poor and upon induction, ceased. Therefore, different promoter and strain combinations were tested. Additionally, a different Ets-1 construct was designed. 2.3.4. Alternative Promoter/Strain Combinations Dramatically Improve Expression trpflac (tac) hybrid promoters offer an alternative to T7 polymerase based systems (DeBoer et al., 1983). Rather than induce an integrated foreign polymerase controlled by the lac repressor, a strong trp bacterial promoter is coupled to lac repressor control on a plasmid. To ensure that the lac site on the plasmid is occupied at all times, a tac-based expression vector harbours a lacV gene allowing for constitutive expression of lac repressor. The key advantage of a tac system is that every component of the overexpression system is plasmid-borne and transportable to any bacterial strain. To change strains in a T7-based system, a phage A.DE3 lysogen would have to be produced to introduce the T7 polymerase gene into the bacterial chromosome. Taking the de-based system one step further, a sterically-repressed promoter, srp, may be considered Chapter 2 — Overexpression and Purification of ETS Domain Proteins 43 (Ezaz-Nikpay et al, 1994). Here, the lac operator site is placed within the RNA polymerase binding site, and strictly enforces the uninduced state. The gene fragments corresponding to the Ets-1AN331 gene were ported to a tac system by inserting an Ndel/BamHI fragment from pAED4-ets into the Ndel/Bgl II sites of vector pCW (Gegner et al, 1992) creating the plasmid, pCW-ets. The synthetic Ets-1AN331 gene fragment was cloned via an Ndel/Hindlll insertion. In pCW, Bglll sites flank a cassette containing the transcriptional termination sequence for the vector. As a result of the subcloning, the termination sequence was deleted in plasmids, pCW-ets, pCW-ala and pCW-asn (Figure 2.4) with no apparent decrease in expression relative to the construct, pCW-ser, which retained its transcriptional termination sequence. Since the tac system affords a number of strains to be examined, ten strains listed in Table 2.2 were assayed for expression in the context of the wild-type construct, pCW-ets. The ensemble constituted a mix of standard cloning strains, protease-deficient strains and proprietary non-K12 strains. E. coli Topp 1-6 were purchased from Stratagene (La Jolla, CA). Other strains were provided by the Cellulase Research Group at UBC. A General Assay for Ets-1 Expression A convenient assay for expression was developed as follows: From a 2 mL overnight culture grown in either 2YT, M9, synthetic rich media (Mcintosh and Dahlquist, 1990), 500 [iL was inoculated into a 30 mL master culture at 30°C or 37°C in the presence of 50 jj.g/mL carbenicillin and/or 34 |j.g/mL chloramphenicol for selection. Ampicillin is depleted by secreted B-lactamase during growth. If the Ets-1 gene is toxic and some protein is expressed due to promoter leakage, the bacteria may expel the vector under relaxed selection conditions. Carbenicillin, a non-metabolizable analog of ampicillin, reduces the likelihood of this event from occurring. Cells were induced with 1 mM IPTG at an optical density of 0.8 at 600 nm. At induction, and several times post-induction, 1 mL of cells was removed and pelleted. 200 [iL of a lx strength SDS-PAGE treatment buffer solution was added followed by sonication at 30 W for 10 seconds. Samples were Chapter 2 — Overexpression and Purification of ETS Domain Proteins 44 incubated at 90°C for 1 minute prior to application on a 15 % polyacrylamide gel with 0.1% sodium dodecylsulfate. The gel was then Western blotted and probed with a 20000x diluted polyclonal antibody generated against a C-terminal region in Ets-1 that resides just outside the ETS domain but is present in all constructs for enhanced solubility. Upon application of a secondary horseradish peroxidase conjugated anti-rabbit IgG, visualization was achieved using a chemiluminescence kit (Dupont; Wilmington, DE). The anti-Ets-1 antibody used was very specific and did not crossreact with any bacterial proteins. For this reason, this specificity allowed a slot-blot assay to be performed directly on the SDS solubilized bacterial extract. Based on calibrations with purified Ets-1 protein standards, as little as 5 ng could be readily detected (data not presented). ETS Domain Expression is Strain Dependent Of the ten strains assayed, all overexpressed protein; however, only the Stratagene wild-type strains Topp-2, Topp-3 and Topp-4 produced appreciable levels that could be directly observed as a band on a Coomassie Blue stained gel. Of the Topp strains, Topp-2 was selected as the optimum strain. As the precise genotypes of the Topp strains are proprietary, an explanation for the success of these strains remains uncertain. Most significantly, the success of the Topp strains also applied to growth in M9 minimal media, although at reduced levels relative to rich media. A means of producing sufficient quantities of isotopically labelled Ets-1AN331 was now available. Large Scale Expression and Purification Scale-up was straightforward. Typical cultures involved 2-6 L of media portioned into 1.5 litre batches in 2.8 L Fernbach flasks at 37°C. A 2 L fermenter was also used, but it did not improve yields. Cells were harvested by centrifugation at 2.5-4.0 hours post-induction. The pellet was reconstituted in chilled lysis buffer (20 mM sodium phosphate, pH 6.4, 0.15 M NaCI, 1.0 M urea, 1 mM EDTA, 1 mM PMSF, 20 mM DTT) and sonicated Chapter 2 — Overexpression and Purification of ETS Domain Proteins 45 at 125 W for a total of 4 minutes. The presence of urea in the mixture helped to prevent non-specific binding to sheared bacterial DNA (J. Omichinski, personal communication). Cellular debris was cleared by centrifugation in an SS34 rotor at 17000 rpm for 30 minutes. The supernatant was clarified further by passage through a 0.8 um syringe filter. The amount of bacterial DNA was reduced in the preparation by applying the filtrate to a DEAE sepharose (Pharmacia) column or alternatively through a Q15 anion exchange filter (Sartorius). The highly basic Ets-1 proteins were resolved by cation-exchange FPLC chromatography. Typically, a 16 cm x 1.6 cm column of S-sepharose (Pharmacia) was used accompanied by a linear gradient of 0.15 - 0.70 M NaCI over 300 mL. Ets-1 proteins eluted at ca. 0.5 M NaCI and were determined to be 80 - 95 % homogeneous by SDS-PAGE and reverse-phase chromatography. In most cases, the purity was sufficient to perform NMR experiments. If further purification was required, Sepharose-12 gel filtration chromatography and hydroxylapatite chromatography was applied. Yields in isotopic labelling media (M9, Synthetic Rich) were typically at 2.5 mg/L culture, and thus, these further purification steps were generally avoided. This level was sufficient for NMR studies, though improvement was still sought. Quality Control Protein samples destined for analysis by mass spectrometry or peptide sequencing were purified to homogeneity by HPLC on a Vydac TP1010 C-18 reverse-phase column (Separations Group). Al l ETS domain proteins discussed in this chapter elute at ca. 50% acetonitrile/50 % water in 0.1 % trifluoroacetic acid. A doublet was observed on gels and during chromatography of Ets-1AN331 proteins. Peptide sequencing (Alberta Peptide Institute, Edmonton AB) demonstrated that the N -terminal methionine was incompletely processed by the bacterial methionine aminopeptidase. The ratio of processed-to-unprocessed Ets-1AN331 was variable possibly due to differences in the growth media and whether reverse-phase HPLC was used. Electrospray mass spectrometry (NRC Protein Facility, Ottawa ON) was used to confirm the identity of all expressed proteins in the study. Chapter 2 — Overexpression and Purification of ETS Domain Proteins 46 2.3.5. Histidine-Tagged Constructs A series of histidine-tagged constructs were made to explore the possible effects of a new N-terminal sequence on the stability of the 12 kDa protein fragment and to utilize other methods of purification. Al l of the contructs are based on the pET16 vector series (Invitrogen) and are summarized in Figure 2.5: pET16-ets pET16-asn pET6H-ets T7 lac T7 lac T 7 lac 11x his 11x his 6x his — i — cys Ets-domain[331-440] S cys ala Ets-domain [331-440] cys cys Ets-domain[331-440] < term term term Figure 2.5 : The 12 kDa Ets-1 fragment was N-teiminally tagged with either 11 or 6 histidines and assayed for expression. Plasmids are named according to the convention described in Figure 2.2. The vector, pET6H-ets, was constructed by replacing the Ncol/Ndel fragment of pET16-ets with a oligonucleotide cassette specifying only six rather than eleven histidines. In all constructs, a lac repressor binding site is present downstream of the promoter to further decrease the chances of low-level Ets-1 protein expression before induction. Proteins were expressed in E. coli BL21::DE3(pLysS) in both rich and minimal media, suggesting that either the histidine tag or altered promoter was enhancing expression. A follow-up construct, pT71ac-ets (Figure 2.6), lacking the histidine tag, also supported growth in minimal media but not to the extent of the pCW-ets/Topp-2 vector-strain combination. Indeed, it would appear that addition of a lac repressor site in a different vector context (pAED4 vs. the pET22 series) was sufficient for expression of Ets-1AN331 in minimal media. Chapter 2 — Overexpression and Purification of ETS Domain Proteins 47 pT7lac-ets T7 lac Ets-domain [330-440] term cys cys Figure 2.6 : A new T7 polymerase based vector incorporated a lac repressor site. This construct was made by collapsing the Nde I sites in the pET22-ets. The pET22- series is discussed in section 2.3.4. The histidine tagged proteins were purified using chelating nickel ion affinity chromtography. N i 2 + - N T A agarose (Qiagen) was specifically used due to its high binding capacity and tolerance to B-mercaptoethanol in the elution buffers. Although DTT would have been preferable to use in the solutions, DTT has enough reducing potential to strip the N i 2 + from the column. B-mercaptoethanol was included at 5 mM in all buffers to keep the cysteines in Ets-1AN331 reduced. NaCI was also included at 0.25 M to solubilze Ets-1AN331 and minimize non-specific binding as N i 2 + - N T A agarose is an otherwise excellent ion exchange column. Although Ets-1AN331 had a relatively high calculated pi of 8.0, it was particularly prone to aggregation below a solution pH of 5.0. Generally, histidine-tagged proteins are bound at pH 8.0, above the pK a of the imidazole moieties and eluted from the column at pH 4.5. Since a low pH elution could not be performed, the proteins were eluted by competition with a 10 mM to 250 mM gradient of free imidazole at pH 7.0. Successive washes of 10 mM imidazole before gradient elution were very effective at minimizing non-specific binding. The buffer containing the eluted protein was exchanged over a disposable G-25 gel filtration column (Pharmacia), concentrated and dialysed for NMR analysis. Mass spectrometry and partial DNA sequencing confirmed the identity of the histidine tagged proteins. Chapter 2 — Overexpression and Purification of ETS Domain Proteins 48 NMR samples of 1 5 N labelled hexahistidine-tagged protein derived from pET6H-ets were used to assess the suitability of Ets-1AN331 for a full structural analysis. The primary concern at this early stage was to ensure that there was adequate dispersion of the amide resonances, as they form the basis for higher-dimensional experiments (Figure 2.1). Under the conditions used (20 mM sodium-phosphate, pH 6.2, 0.5 M NaCI, 10 mM DTT), the amide resonances of the six N-terminal histidines were not apparent in a 1 H -1 5 N HSQC spectrum due to their rapid exchange with the solvent. 2.3.6. Periplasmic Secretion Signals Up to this point in the overexpression of Ets-1AN331, some success had been obtained with the pCW series using a proprietary wild-type E. coli strain and a tac promoter. However, the low to moderate level of expression suggested that Ets-1AN331 may be toxic to the bacterium. To address the question of cytoplasmic toxicity, a series of constructs were made with N -terminal periplasmic signal sequences. Periplasmic signals vary in length and composition (von Heijne, 1985). Typically, they are 18-25 residues long, loosely divided into basic, hydrophobic and variable regions. Signal peptide-mediated translocation is facilitated by a complex of a membrane-associated ribosome and secretory pathway proteins. The nascent polypeptide must exist in a translocation competent form in order to successfully traverse the membrane. Upon export, a signal peptidase cleaves the signal peptide. The first-generation clones in this series depicted in Panel A of Figure 2.7 rely upon the periplasmic signal sequences derived from the E. coli ompA and pelB genes (Power et al, 1992). In addition to two signal sequences, two promoters (T7lac, tac) and several Ets-1AN331 protein sequences (wild-type, C350A/C350N and C350N/C350S) were assayed in combination. Ncol/Ndel and Hindlll/Nde I oligonucleotide adaptors coupled the Ets-1 gene fragment to the vector sequences and were designed to introduce as few foreign Chapter 2 — Overexpression and Purification of ETS Domain Proteins 49 residues as possible. Downstream from the adaptor in the pFLAG (Kodak/IBI Biochemicals) series, additional sequences specifying the octapeptide, DYKDDDDK, are found. A monoclonal antibody affinity purfication system is available which recognizes this hydrophilic epitope. A polyhistidine tag is found 3' to the multiple cloning site in the pET22- series (Invitrogen). However, owing to the stop codon present in the cloned fragment, the polyhistidine tag was not expressed. T7 polymerase based constructs were transformed into E. coli BL21::DE3 and BL21::DE3(pLysS) strains; tac based constructs were transformed into the previously successful E. coli Topp-2 strain. Compared with the overexpression assay discussed in section 2.3, unexpected results were obtained. Expression levels in excess of 30 mg/L culture were observed for pFLAG/Topp-2 and pET22/BL21::DE3 vector/strain combinations. Protein production stablized at 4 hours post-induction at 37°C. No expression was observed for the pET22/BL21::DE3(pLysS) vector/strain combination. From molecular weight estimates on denaturing polyacrylamide gels relative to Ets-1DN331 produced in the pCW system, it was suspected that the signal peptide was not being cleaved and that the protein was accumulating in the cells as insoluble inclusion bodies. Inclusion body formation was confirmed by electrophoretic analysis of crude extract and pellet fractions from a sonicated cell preparation. Western analysis of extracts at various times post-induction failed to identify any lower molecular weight species arising due to processing by the bacterial signal peptidase. Furthermore, no protein was found in periplasm extracts produced by a chloroform treatment (pET series manual; Invitrogen). Inclusion bodies not only offer the advantage of extremely high protein yields, but also a simple means of purification. Cells from a 1.5 L fermentation were harvested, resuspended and incubated for 15 minutes in 10 mL of STE buffer (20% sucrose; 25 mM Tris-HCl, pH 8.0; 10 mM EDTA, 1 mg/mL lysozyme). The solution was then sonicated at 125 W for 2 minutes total, followed by centrifugation at 18000 rpm in an SS-34 rotor for 20 minutes. The pellet was retained and washed several times with 1.0 % Triton X-50 pET22-ets pET22-asn pET22-ser pFLAG-ser T7 lac Ntol A/do j e lB^ / jMGSH 1MGSH T7 lac T7 lac IMGSH Ets-domain [331-440] 1 cys - I -cys Ets-domain [331-440] Bam HI Ets-domain [331-440] —1 r-ser * term Ets-domain [331-440] —r~ ser Bgll pET22-cys pFLAG-cys Ets-domain [331-440] - r -ala -r— asn ompA j T T T V" ,teg::|KLCH Ets-domain [331-440] —1 Hnncflll 89" T— ser c I T T 1 Bam HI pET22-his -—^HBCZZ3 -J // /peiB'/^| MGHHHHHHCHM j Ets-domain [331-440] ' ' T7 /ac i i ala asn term Figure 2.7 : Signal Peptide Constructs — A series of first generation constructs shown in Panel A explored a variety of signal sequences (ompA, pelB), promoters (T7, tac) and Ets-112 kDa protein fragments (wild-type, C350A/C416N, synthetic gene C350N/C416S). None of these contructs resulted in secreted, cleaved Ets-1AN331 fusion protein; instead, inclusion bodies were produced an expression level of 30 mg/litre culture. A protocol was derived to purify the inclusion bodies and cleave them chemically at cysteine residues. In order for this scheme to work, endogenous cysteines in Ets-1AN331 were removed and a new cassette containing a cysteine at the signal peptide/Ets-lAN331 junction was incorporated (Panel B). Under denaturing conditions, the inclusion bodies were purified using nickel affinity chromatography facilitated by a cassette containing six histidines shown in Panel C. Solid triangles indicate potential cleavage sites by E. coli signal peptidase, enterokinase and NTCB chemical methods. Chapter 2 — Overexpression and Purification of ETS Domain Proteins 51 100 in 50 mM Tris-HCl, pH 8.0, followed by 5 M urea. Ets-1AN331 clearly had remarkable insolubility to resist the 5 M urea wash. Only sonication in 10 mL of 6 M Gu«HCl in 50 mM Tris-HCl, pH 8.0 would dissolve the protein. After centrifugation at 18000 rpm in SS-34 rotor for 30 minutes, the supernatant contained >95 % pure Ets-1AN331 fusion protein. The repeated sonications throughout the protocol resulted in the vast majority of the Ets-1AN331 fusion protein as disulfide-linked multimers. A large population of dimers in earlier preparations was confirmed by mass spectrometry. Following this observation, the denatured protein was reduced by incubation in the presence of 100 mM DTT at 37°C for 1 hour. Fusion protein was conviently precipitated by addition of 5 volumes of water and resuspended in 2 mL of 6 M Gu»HCl, 0.5 M Tris-HCl, pH 8.0. The excessive concentration of Tris proved necessary during further manipulations. Since the fusion protein was soluble only in denaturing solution, protease digestion was immediately discounted as a method to cleave the signal peptide. Instead, chemical methods were sought. Cyanogen bromide treatment was not an option since it would require replacement of two endogenous methionines (M384, M432) and substitution of a methionine at the signal peptide/Ets-1 junction. Ultimately, cysteine residues, rather methionine residues, were targeted as sites of chemical cleavage. Modifications to the fusion protein were performed as follows: Two endogenous cysteines present in the wild-type Ets-1 protein were conveniently "mutagenized" by replacing the wild-type gene with one of the double mutants or the synthetic gene. To specifically incorporate a cysteine at the junction between the signal peptide and Ets-1AN331, the oligonucleotide adaptors used in the original cloning of pET22-ets and pFLAG-ser were replaced. The resulting vectors, pET22-cys and pFLAG-cys are depicted in Panel B of Figure 2.7. Chapter 2 — Overexpression and Purification of ETS Domain Proteins 52 A pH 8.0, 25°C Cys-SH + NTCB Cys-S-CN + NTB pH9.0,37°C Cys-S-CN ->• 2-iminothiazolidine-4-carboxylate B N H ^ / CH - C — His329 — Met330...Asp440 Figure 2.8 : Upon reaction with NTCB (Panel A), the cyanylated cysteine residue spontaneousry cyclizes under alkaline conditions to form a stable 2 -iminorhiazolidine-4-carboxylic acid N-terminal cap (Panel B). The reagent, NTCB (2-nitro-5-thiocyanobenzoic acid), chemically cleaves cysteine residues through a two-step reaction (Jacobson et al., 1973; Degani and Patchornik, 1974). This reagent has been recently used in analyses of caldesmon (Sutherland and Walsh, 1989) and creatine kinase (Clottes et al, 1994). At neutral pH, cysteine, and serine/threonine to a lesser extent (Liao and Wadano, 1979), are initially cyanylated, liberating yellow NTB (2-nitrothiobenzoic acid). Once the cyanylation reaction is complete, the pH is raised to moderately alkaline conditions causing the thiocyanate to attack the amido group of cysteine. Cyclization results in cleavage of the polypeptide backbone, and yields a blocked N-terminal thiazolide (Figure 2.8). A free amine may be regenerated by treatment with Raney nickel (Otieno, 1978). However, this reaction was not considered since methionine side chains would be converted to 2-aminobutyrate. Considerably more NTCB was required to quantitatively cleave the fusion protein than the stoichiometric amount reported by Degani and Patchornik (1974). The following procedure continued from the inclusion body preparation at the stage where the Chapter 2 — Overexpression and Purification of ETS Domain Proteins 53 reduced fusion protein had been precipitated and resolubilized in 2 mL of Gu»HCl in 0.5 M Tris-HCl (pH 8.0): 100 mg of NTCB (Sigma, St. Louis MO) was dissolved in a 200 uL of dimethylsulfoxide and added to the fusion protein solution. The cyanylation reaction is spontaneous and rapid under these conditions. A high concentration of Tris buffer is necessary to buffer the protons being liberated from the reaction. NTCB reacts with the protein to form the NTB, 2-nitrothiobenzoic acid, which can be monitored at 412 nm. At this wavelength, NTB is very sensitive to changes in pH. If the pH drops below 4.0, the brillant yellow-red colour will disappear necessitating dropwise addition of a 10 M sodium hydroxide solution to bring the pH back up to 8.0. The reaction was allowed to proceed at 37°C for 20 minutes. To separate the derivatized protein from the remaining NTCB, the solution was applied to a PD-10 disposable desalting column (Pharmacia) that had been previously equilibrated with 6 M Gu»HCl, pH 8.0. The protein was eluted from the column with 3.5 mL of the same buffer. To initiate the cleavage reaction, the pH was adjusted to 9.0 with NaOH and allowed to incubate at 37°C for 8-10 hours. Finally, the cleaved N-blocked protein was purified by reverse-phase chromotography according to the method described in section 2.3.4. At least 15 mg/L culture of the 12 kDa fragment could be obtained from the inclusion body procedure, corresponding to a 2.0 mM NMR sample in 500 uL. Unfortunately, the N-blocked protein was very intolerant to renaturation, compared to the 12 kDa protein produced from the pCW expression series. A number of detergents (CHAPS, Tween-20, octylglucoside, Triton X-100), solvents (DMSO, ethanol, methanol, dioxane, trifluoroethanol), salt concentrations and pH's were assayed. However, no condition was found that could renature >5 % of the denatured material. Nevertheless, this exercise showed that inclusion body production can offer the advantages of high yields and a simplified purification for proteins that can be renatured. Chapter 2 — Overexpression and Purification of ETS Domain Proteins 54 2.3.7. Fusion Proteins Small protein fragments less than fifty residues and proteins whose expression is limited by their solubility or tolerance in the bacterium may benefit by expression as a fusion protein. Typical fusion partners are glutathione-S-transferase (Smith et al., 1992), maltose-binding protein (Maina et ai, 1988), protein-A (Dang, 1988) and thioredoxin (LaVallie et al., 1993). Extending the application further, thioredoxin-histidine tag fusion proteins attempt to provide both enhanced solubility and a simple means of purification (Wilkinson et al., 1995). Though no fusion protein constructs were designed for the Ets-1AN331, this success of this approach deserves mention. 2.3.8. Conclusions From the series of constructs designed to overexpress Ets-1AN331, the combination of the tac based promoter and Topp-2 cells was chosen based on its ability to support expression of milligram amounts of native protein in minimal media. Several double mutants in this series (pCW) were available but ultimately, the wild-type construct was chosen. Very little was known about the tertiary structure of the ETS domain at this time. Hence, there was a risk, although very low, that a double mutation may adversely affect the structure of the protein. Often, when a protein is not expressed well, a different fragment, or even a different Ets-1 family member may be considered. The minimal ETS domain from PU.l serves as an example. Unlike Ets-1, the minimal domain defined by the PU.l sequence is very soluble and is expressed well in E. coli (Pio et ai, 1995). However, PU.l only retains 39% sequence identity to Ets-1 and would not be suitable for addressing biological questions, beyond DNA binding, that were pertinent only to Ets-1 proteins. Two other protolytic fragments of Ets-1, an 18 kDa fragment (Ets-1AN280) and a 14 kDa fragment (Ets-1AN314), were also available for analysis but were considered to be too large for an initial structural study. In subsequent chapters, I will discuss the cloning, overexpression Chapter 2 — Overexpression and Purification of ETS Domain Proteins 55 and characterization of these deletion mutants because one deletion mutant, Ets-1AN280, possesses an additional activity capable attenuating ETS domain DNA binding. 2.4. Solution Attributes of Uncomplexed Ets-1AN331 2.4.1. Thiol Titrations Ellman's reagent (5,5'-dithiobis-2-nitrobenzoic acid; DTNB), is commonly used to determine the number and reactivity of free thiols in proteins (Ellman, 1959). Upon reaction with a free thiol, NTB production is conveniently monitored at 412 nm and quantitated from its £4i2nm °f 13700 M _ 1 at pH 7.0. An assay was performed by adding DTNB from a stock solution prepared in dimethylsulfoxide to samples of freshly reduced Ets-1AN331, and Ets-1AN331 in a 1:1 complex with the SCI (5'-GCCGGAAGTG-3') duplex (Nye et al, 1992). The DNTB:protein ratio was 50:1. Relative to a denatured control preparation, both thiols in the complexed and uncomplexed protein were reactive (data not presented). This confirms that native Ets-1AN331 does not contain a disulfide. Moreover, both cysteines appear to be solvent exposed. 2.4.2 Analytical Ultracentrifugation Although the ETS domain is believed to bind DNA as a monomer, uncomplexed fragments of Ets-1 may not necessarily be monomeric in solution due to unfavorable interactions between exposed hydrophobic or charged surfaces. The fact that aggregation occurred at concentrations > 1.0 mM protein and salt concentrations < 0.5 M NaCI, and the two thiols in the wild-type form were solvent exposed strongly suggested that multimeric forms of Ets-1AN331 in solution were plausible. Analysis of a 0.19 mg/mL solution of Ets-1AN331 (in 20 mM sodium phosphate, pH 6.3, 0.5 M NaCI, 1 mM DTT, 0.02 sodium azide) was carried out on a Beckman Model E instrument by Dr. L. Hicks at Alberta Peptide Institute (Edmonton, AB). Ets-1 AN331 did show signs of intermolecular association, with apparent molecular weights ranging from those expected for the monomeric form to the dimeric form. The average apparent Chapter 2 — Overexpression and Purification of ETS Domain Proteins 56 molecular weight was determined to be 16440 Da (the calculated molecular weight of Ets-1AN331 without an N-terminal methionine is 12967 Da). This deviation of several kDa from the true molecular weight is attributed to partial aggregation or association in solution. Ultracentrifugation was not performed on any Ets-1 AN331/DNA complexes. 2.5. Ets-1AN331 Has Sequence Specific D N A Binding Activity 2.5.1. Ultraviolet Crosslinking of Ets-l/DNA complexes The bandshift assay, or electrophoretic mobility shift assay, is widely used to detect protein-nucleic acid complexes. To perform a bandshift assay, a mixture of protein, DNA and radiolabeled DNA (also termed probe DNA) is preincubated and loaded on a polyacrylamide gel. The protein-DNA complexes and free probe DNA are resolved by their differences in electrophoretic mobilities. Ultraviolet (UV) crosslinking is excellent alternative to bandshift analysis. A l l components in a UV crosslinking experiment are preincubated and loaded onto a polyacrylamide gel like a bandshift assay with two important differences. First, UV-irradiation is used to covalently crosslink the complex together. Following UV-irraditation, the intermolecular events in the solution are frozen and therefore allow a wider range of chemical manipulations to be performed on the complexes. Second, complexes are resolved on denaturating polyacrylamide gel. A molecular weight of the complex can then be estimated from its relative electrophoretic mobility to protein standards. Upon UV-irradiation, highly reactive free radicals are produced at the nucleic acid bases, particularly thymine. Several outcomes are possible for the free radical: (i) it may form a covalent bond to a nearby protein residue; (ii) it may form a covalent bond with a nearby base (e.g. pyrimidine dimers); (Hi) or, it may simply may react with the solvent. Chapter 2 — Overexpression and Purification of ETS Domain Proteins 57 Once the covalent crosslink is formed, the complex can be electrophoretically resolved or digested and sequenced (Liu et al, 1994). Crosslinks are formed under any solution condition as long as there are no other UV-absorbing components, such as ATP, present. The crosslink formed is essentially zero-length between the protein and nucleic acid. Hence, the crosslink can supply a very powerful distance restraint for future modelling studies. However, due to the non-selectivity of the free-radical species and the numerous base-protein contacts, the location of the crosslink may be ambiguous. In addition, crosslinking generally occurs with low efficiency. Ambiguity and low efficiency, in turn, can be countered by specifically labelling the protein or nucleic acid with a photosensive reagent (Chen and Ehbright, 1993; Willis et al, 1993). The UV source can vary from a low-power germicidal lamp to a high-power laser. High-power UV-lasers offer one major enhancement over lamp irradiation (Williams and Konigsberg, 1991; Hockensmith et al, 1991). The pulse is of sufficiently short duration (5 ns) that diffusion effects are negated. In essence, the crosslinks reflect a molecular snapshot of the solution. Previously, Nye et al. (1992) used a selected-and-amplified binding assay to obtain a high-affinity consensus sequence for recombinant deletion mutants of Ets-1. Corresponding to the highest affinity sequence detected, SCI (kaiss = 3.8 x IO' 1 1 M), an oligonucleotide, 5'-GCCGGAAGTG-3', and its complement were synthesized on an Applied Biosysterns Model 391 instrument. The oligonucleotides were synthesized at a 40 nmol level, deprotected with ammonium hydroxide and dried. No further oligonucleotide purification was performed. Concentrations were calculated according to the sum of the individual nucleotide contributions at 260 nm. Equimolar amounts of each oligonucleotide were annealed by slow cooling from 80°C in 50 mM Tris-HCl, 5 mM MgCl 2 , 50 mM NaCI. A 5 pmol aliquot of the SCI duplex was 3 2 P labelled at the 5' end in a 50 uL reaction containing 5 nmol of 3 2Py-ATP and T4 polynucleotide kinase (Pharmacia). The radio-labelled SCI probe was purified from the free label with a disposable Sep-Pak C-18 column (Waters/Millipore). 58 NaCI concentration (mM) 100 150 200 250 300 350 400 450 500 1000 Figure 2.9 : The observed UV-laser induced crosslinking of Ets-1AN331 to a 10 bp 5'-radiolabelled SCI oligonucleotide duplex demonstrates that Ets-1AN331 is active for DNA binding. The oligonucleotide complex was unusually resistant to the effect of increasing ionic strength. From the autoradiogram of an SDS-PAGE gel above, it is apparent that the optimum salt concentration was 150 mM under the conditions assayed (20 mM sodium phosphate, pH 6.2, 2 u\M protein, 2 uM DNA, 10 uM poly(dI/dC)). Complexes could still be detected at 500 mM NaCI. The band corresponding the complex is denoted by an arrow; the free oligonucleotide is not shown in the figure. From the ratio of the integrated strongest peak (at 150 mM NaCI) at the free probe (not shown), a crosslinking efficiency of ca. 1 % was calculated. Chapter 2 — Overexpression and Purification of ETS Domain Proteins 59 A Quanta-Ray model GCR14S pulsed Nd:YAG laser (Spectra Physics) laser tuned to 266 nm and 60 mj power was used for crosslinking studies. A single 5 ns pulse of 6.4 mm in diameter was applied to samples containing Ets-1AN331 and SCI at a 1:1 ratio in a sample volume of 20 p.L. Al l irradiated mixtures contained 2 jiM protein, 2 uM SCI duplex, 10 uM poly(dI/dC), 20 mM sodium phosphate, pH 6.2, at a salt concentration varied from 100-1000 mM NaCI. A portion of the irradiated mixture was electrophoresed on a 15% denaturing polyacrylamide gel containing 0.1% SDS. Typically, at least a 36 hour period at -70°C was required to visualize bands on an autoradiogram. The ~ 1% efficiency of the direct crosslinking method described above was very poor thus precluding accurate quantitation of DNA binding affinity. However, this method revealed some interesting binding characteristics of the complex. Notably, it demonstrates that the expressed protein is active. Salt concentrations in excess of 0.5 M NaCI were required to eliminate binding (Figure 2.9). A similar resistance to ionic strength has been observed by Fisher et al. (1994) in BiaCore binding affinity studies and by Cuddeford (1994) in a more detailed UV-laser binding study than the one described here. The low dependence of binding on ionic strength suggests that electrostatic interactions play a minor role in DNA binding relative to interactions such as hydrogen bonding or hydrophobic interactions. In addition, no discernable differences in binding could be found between pH 5.5 - 8.0 (data not shown). In an NMR study, protein concentration, ionic strength, pH and temperature must all be optimized. During the course of the crosslinking study, it was apparent that the general properties of the Ets-1 fragment markedly changed upon DNA binding. Combinations of these parameters were easily investigated by incubating 20 JJ.L of concentrated protein, or protein/SCI complex, in tightly sealed PCR microtubes at 20°C, 25°C, 30°C and 32°C and 37°C. Whereas the uncomplexed protein was soluble only at a pH > 6.0, salt concentration > 0.5 M NaCI, temperature < 25 °C and protein concentration < 1.0 mM, the protein-DNA complex was stable under a wide range of conditions. In fact, the complex remained soluble above 30°C under low ionic strength conditions. Chapter 2 — Overexpression and Purification of ETS Domain Proteins 60 2.5.2. Protein / Nucleic Acid Titration Studies Initially, a series of two-dimensional homonuclear COSY, NOESY and TOCSY spectra were acquired for a 1.0 mM sample of the SCI duplex (in 20 mM sodium phosphate, pH 7.0, 50 mM NaCI, 25°C) with the aim of completing the proton assignment of the backbone-sugar and nucleotide bases. The strategy used was an alternating sugar/base NOE walk described in Wuthrich, (1986). However, the assignments were hampered by low signal-to-noise and an apparent "missing" base pair presumably due the symmetry of CCGG in the SCI sequence. Ultimately, the NMR spectrum of the duplex was assigned by Dr. J. Skalicky (Mcintosh laboratory, University of British Columbia) upon acquistion of a new series of *H and natural abundance 1 3 C spectra on a 5 mM sample of the SCI duplex. These assignments, in turn, were used to study the binding of DNA by ETS-1AN331 and Ets-1AN280. The UV-laser crosslinking experiments demonstrated that complexes could be made in a variety of salt concentrations. This knowledge facilitated the following DNA titration protocol: to a 0.15 mM sample of SCI oligonucleotide in 20 mM sodium phosphate, pH 6.2, 0.1 M NaCI, 1 mM DTT, 25 °C), successive aliquots of a concentrated solution of 1 5 N labelled 12 kDa Ets-1 at 0.5 M NaCI were added and 1 5 N decoupled, 1 H jump-return spectra were acquired in order to observe the imino protons of the duplexed DNA. From the titration shown in Figure 2.10, a complex was detected to be in the slow-exchange timescale shown by the appearance of a new set of guanine and thymine imino resonances as protein was titrated. Lineshapes are characteristically broadened due to the increase in molecular weight (the SCI duplex is ca. 7 kDa; the Ets-lDN331/SCI complex is ca. 20 kDa). An interesting peak at 12 ppm also appeared. This resonance was attributed to either a downfield shifted amide proton or H e l tryptophan indole proton of Ets-1AN331 since in similar runs where 1 5 N decoupling was not used, this resonance occurred as a doublet. Recently, the structure of a human Ets-1 ETS domain in complex was solved by NMR methods (Werner et al., 1995b) and this downfield shifted resonance at 12 ppm was confirmed to be W338 H e l . 61 Figure 2.10 : Protein titration of Ets-1AN331 with the SCI oligonucleotide duplex. (Panel A) The SCI duplex represents a high-affinity binding site for Ets-1 as determined by a selected-and-amplified binding assay (Nye et al, 1992). The invariant GGA core motif is boxed. (Panel B) Jump-return 1 H spectra of a 0.15 mM SCI duplex (20 mM sodium phosphate, pH 6.5, 0.15 M NaCI, 5 mM DTT) titrated with Ets-1AN331. The accompanying percentage indicates the ratio of protein to DNA. The appearance of a new set of peaks indicates that DNA binding is occurring at a slow-exchange rate. A peak at 12 ppm marked ets corresponds to the downfield-shifted W338 Hel resonance of Ets-1AN331 (Werner et al, 1995b). 6 1 ppm Chapter 2 — Overexpression and Purification of ETS Domain Proteins 63 A control oligonucleotide duplex was also tested for binding to Ets-1AN331. This duplex had a G->C mutation in the core GGA consensus binding sequence. During a similar protein titration experiment, no imino resonances were observed to change their respective chemical shifts or broaden (data not presented). Therefore, no binding of Ets-1AN331 to this mutant oligonucleotide duplex occurred. 2.5.3. Summary UV-laser crosslinking studies and NMR titrations both demonstrated that Ets-1AN331 is fully active for DNA binding. The NMR titrations demonstrated that DNA binding was specific as and occurred on a slow-exchange timescale consistent with tight binding. In terms of solution characteristics, a wider range of temperature and ionic strength was available to the protein-DNA complex than that of the free protein alone. However, the increased molecular weight precluded structural analysis of the complex. In the following section, circular dichroism and Fourier-transform infrared spectroscopy are employed to investigate the stability and secondary structural aspects of Ets-1AN331 alone, and in complex with DNA. 2.6. Secondary Structural Characteristics of Ets-1AN331 Circular dichroism (CD) and Fourier transform infrared (FT-IR) spectroscopy require small amounts of protein relative to NMR spectroscopy and offer a straightforward means of assessing both protein secondary structure and stability. 2.6.1. CD spectroscopy The circular dichroism effect observed for an amide chromophore in the far UV (190-250 nm) depends on its secondary structure (Johnson, 1988). For example, the presence of local minima at 208 nm and 222 nm is diagnostic of a-helices. Extended and B-sheet conformations also have characteristic spectra but not as pronounced as that observed for a-helices. Consequently, predictive algorithms designed to deconvolute a CD spectrum into secondary structural components tend to be more accurate for proteins Chapter 2 — Overexpression and Purification of ETS Domain Proteins 64 rich in cc-helices. Qualitative estimates of tertiary structure can be obtained from examining spectral contributions of aromatic chromophores in the near-UV spectrum (250-320 nm). Protein concentrations were determined from absorbance measurements in 6 M Gu«HCl, 50 mM Tris-HCl, pH 8.0 using a calculated extinction coefficient of 32000 M" 1cm'1 at 280 nm (Edelhoch, 1967; Gill and von Hippel, 1989). A spectrum of 15 uM Ets-1 AN331 (in 20 mM sodium phosphate, pH 6.4, 0.5 M NaCI, 25°C) in a 0.1 cm pathlength jacketed quartz cell was acquired with a Jasco J-730 CD spectropolarimeter. The spectrum was solvent subtracted and processed using Jasco software. From the spectrum, shown in Figure 2.11, a considerable degree of a-helical contribution is apparent. The proportion of a-helix was not estimated from the spectrum. Deconvolution methods used to estimate a-helical content rely on data that extends from 250 nm to 190 nm. Chloride ion absorption is considerable at 190-200 nm; therefore, CD spectra of uncomplexed Ets-1AN331 (in 0.5 M NaCI) could not be acquired in this range. A "CD melt" experiment, designed to measure the stability of Ets-1 AN331, was performed by monitoring ellipticity at 222 nm as the temperature was increased from 25°C to 70 °C with a temperature gradient of 1°C/minute. A sigmoidal curve suggested a cooperative unfolding with a T m of 46°C . This result demonstrates that the Ets-1 AN331 is a moderately stable, folded protein. Ets-1AN331, however, could not be reversibly renatured. The solvent, trifluoroethanol (TFE), is commonly used to induce a-helix formation in peptides and proteins and also to prevent aggregation (Lehrman et al, 1990). Addition of TFE from 5-20% (v/v) initially decreased the apparent ellipticity and shifted the character of the spectra to that of a random coil conformation. Further addition of TFE to a final concentration of 50% (v/v) induced a-helical formation to a level twice that of the 65 Chapter 2 — Overexpression and Purification of ETS Domain Proteins 66 native protein. Presumably, TFE addition had initially denatured the protein, then, upon further addition, induced the extended chain to adopt a non-native a-helical conformation. Similar thermal denaturation studies were carried out on the SCI duplex (monitored at 250 nm) and the Ets-1AN331:SC1 complex (monitored at 222 nm). The SCI duplex alone had a T m of 50°C. Upon complexation, the T m rose to 57°C indicating that binding stabilizes both protein and DNA against thermal unfolding (data not presented). 2.6.2. FT-IR Spectroscopy The stretching and bending vibrations found in the Amide I region of the FT-IR spectrum of a protein (1700-1620 cm - 1) are sensitive to both a-helical and (3-sheet secondary structures (Dong et al., 1990; Surewicz et al., 1993). Even though a high protein concentration is necessary to ensure adequate signal-to-noise in FT-IR studies, it is offset by the requirement of a very small sample volume. A solution of Ets-1AN331 in 20 mM sodium phosphate, pH 6.5, 10 mM DTT, 0.5 M NaCI was concentrated in a Centricon-10 device (Amicon) to 1.2 mM. Spectra of Ets-1 AN331 and an aliquot of similar buffer for reference were acquired on a Perkin-Elmer Model 2000 FTIR with a spectral range of 2200-1200 cm - 1 and a spectral resolution of 2 cm - 1. 1000 scans were signal averaged and processed to a final digital resolution of 0.2 cm"1/point. 6 uL of solution was sufficient to fill a calcium fluoride cell with a 6 urn pathlength. Post-processing of the spectra involved subtracting the spectral contribution of the buffer and water vapor followed by baseline flattening. The second derivative of the spectrum was calculated and its features qualitatively assessed (Figure 2.13). Undulations in the 1800-1700 cm - 1 region of the second derivative spectrum are representative of the noise. Several peaks were evident in the Amide I region and were confidently assigned to various secondary structural contributions based upon consultation of reference spectra Gl 1656.0 ( Amide I I 1 1 1 1800 1700 1600 1500 wavenumber (cm"1) Figure 2.12 : The FT-IR spectrum of EtslAN331 demonstrates the presence of both a-helix and 8-sheet conformations. In the second derivative (lower trace) of the Amide I region of the FTIR spectrum (upper trace), the peaks at (1687.2 cm"1,1679.5 cm"1), 1656.0 cm"1,1650.3 cm"1, and (1636.4 cm"1,1630.9 cm"1) arise from B-turn, a-helix, extended and B-sheet conformations, respectively. A second primarily a-helical contribution (1548.8 cm-1) lies in the Amide II region of the spectrum. The peak at 1517.1 cm"1 is characteristic of tyrosyl vibrations. Chapter 2 — Overexpression and Purification of ETS Domain Proteins 68 (Surewicz et al., 1993). The major peak in the spectrum at 1656.0 cm - 1 was due to a-helices in Ets-1 AN331 and corroborated the conclusions made from the CD study. Nearby, a second prominent peak at 1650.3 cm - 1 was indicative of extended conformations. Surprisingly, there were second derivative maxima at 1636.4 cm' 1 and 1630.9 cm - 1 that indicated a substantial population of residues in a B-sheet conformation. This B-sheet observation was further supported by peaks at 1687.2 cm - 1 and 1679.5 cm - 1 arising from B-turn conformations. In the Amide II region, a second oc-helical contribution was assigned at 1548.8 cm"1. The seven tyrosines in Ets-1AN331 produced a prominent peak at 1517.1 cm"1. 2.6.3. Conclusions The CD and FT-IR analyses of Ets-1AN331 together demonstrated that the ETS domain was largely oc-helical yet also contained a significant contribution from B-sheet secondary structure. The results of the CD thermal denaturation study showed that the protein was folded and moderately stable. The identification of a B-sheet component in the ETS domain was exciting as it provided hints of a novel secondary structure. In addition, studies of an Ets-1/high-affinity oligonucleotide complex showed that binding was specific and was not dominated by electrostatic interactions. A summary of data that was presented in this section is given as Table 2.3. In the next chapter, NMR spectroscopic methods are enlisted to precisely determine the secondary structure of Ets-1AN331. 69 Table 2.3 : Summary of the Properties of Ets-lAN33lt Technique Mass Spectrometric Analysis Edman Protein Sequencing Analytical Ultracentrifugation DNTB Thiol Titrations U V Laser Crosslinking Conclusions (1) The calculated isotope averaged molecular weight of Ets-1AN331 is 13098.98 Da. Without the N-terminal methionine, the calculated molecular weight is 12966.88 Da. (2) A molecular weight of 13097.5 Da is detenriined by electrospray mass spectrometric analysis on an HPLC purified sample of Ets-1 AN331. This results suggests that the N-terminal methionine is retained. (1) N-terminal sequence analysis identifies two species of Ets-1AN331 that are distinguished by the presence or absence of the N-terrninal methionine. Incomplete cleavage by the bacterial aminopeptidase is suspected. The relative amount of each species is variable and may depend on the type of growth media and the efficiency of each purification step. (1) Ets-1AN331 is primarily monomeric under high-salt and reducing conditions. (2) As the predicted molecular weight by this method (16 kDa) is higher that the observed molecular weight by mass spectrometry, suggesting the presence of minor quantities of soluble aggregates. (1) The two cysteines in wild-type Ets-1AN331, C350 and C416, are both as reactive to DNTB as an unfolded control. This suggests that both of these cysteines are solvent exposed. (1) Ets-1AN331 is active for monomeric sequence-specific DNA binding of a high-affinity SCI 10 bp oligonucleotide duplex. (2) Electrostatic interactions between Ets-1AN331 and DNA serve a minor role at the protein/DNA interface. Complexes are observed up to 0.5 M NaCI. D N A Imino Titration by N M R Circular Dichroism Spectroscopy (1) Ets-1AN331 forms a complex with the SCI oligonucleotide duplex at a 1:1 ratio that is in slow-exchange on the NMR chemical shift time scale (milliseconds). (1) Ets-1AN331 forms a folded protein domain with a T m =46°C. (2) Ets-1AN331 is significantly a-helical. Fourier Transform Infrared Spectroscopy (1) Ets-1AN331 is composed not only of a-helices but also of P-strands. tin some instances, N-terminal histidine tagged Ets-1AN331 was substituted. C h a p t e r 3 Assignment and Secondary Structure of the ETS Domain A discussion of basic NMR theory and the application of modern multidimensional heteronuclear experiments to assign protein structures is presented in this chapter. The CBCA(CO)NH experiment, commonly used to assist assignment of the protein backbone, is discussed by deconstructing it into a series of pulse modules. The amalgamated chemical shift, J coupling, NOE and amide hydrogen exchange data from over eighty experiments on nine samples produced a detailed secondary structure of the Ets-1 deletion mutant, Ets-1AN331. The secondary structure provided, enough insight to place the ETS domain in the winged helix-turn-helix family. 3.1. Theory and Technique 3.1.1. Introduction In the presence of a static magnetic field, B 0 , a single nucleus will precess about the field at an angular velocity, or Larmor frequency, of co that is proportional to its gyromagnetic ratio, y. B 0 is conventionally aligned along the z axis. co = y B 0 v = 2rcco [3.1] X H nuclei precess about the z axis at a frequency, v, of 500 MHz in the 11.7 T field produced by the electromagnet used throughout these studies (Equation 3.1). The population of spin 1/2 nuclei (!H, 1 3 C and 1 5 N) will be aligned with or against the magnetic field according to a Boltzmann equilibrium. A slight excess of spins in the low energy spin state produces a net magnetization that can be described by one vector, M Q , which, by convention, is parallel to the z axis. Application of RF pulses in the coil surrounding the NMR sample will produce an oscillating magnetic field, Bj, orthogonal 70 Chapter 3 — Secondary Structure of the ETS Domain 71 to the static field, B Q . In the classical description of NMR, a coordinate system, rotating at the frequency co of the nuclei in the B D field, is utilized. Therefore, the field B1 is stationary when applied at the resonance frequency co. In the rotating frame, the Bt field will exert a torque on the nuclei causing them to rotate around B i . The strength and duration of the pulse is adjusted such that the nuclei experience typically a 90° or 180° rotation around one of the x, -x, y, or -y axes. After a 90° x pulse, the M z (or M Q ) magnetization will be converted to M-y magnetization. As the magnetization in the xy plane precesses about the static field B Q , a free induction decay (FID) is recorded. The Fourier transform of the FID in the time domain converts it into an NMR spectrum in the frequency domain. The precise frequency, or chemical shift, of a given proton in the spectrum depends upon its local B Q field which may differ from the applied field due to factors such as electronic shielding. As the term "free induction decay" suggests, relaxation mechanisms will cause the nuclei to eventually return to Boltzmann equilibrium with M Q aligned along B 0 - The Bloch equations (Bloch, 1946) describe over time, f, what happens to the bulk magnetization, M , in terms of its component vectors, M x , My and M z . d M z /dt = -Y [ - M x B j sin(cot) - M y B a cos(cot) ] - ( M z - M Q ) / T I [3.2a] d M x /dt = y [ MyB 0 + M z B i sin(cot) ] - M x / T 2 [3.2b] d M y /dt = -y [ M X B 0 + M z B a cos(cot) ] - M y / T 2 [3.2c] The trigonometric terms present in the equation highlight the precession of the nuclei in the laboratory frame. The longitudinal or spin-lattice relaxation, Ti , describes an exponential return of of magnetization as nuclei emit energy to their surroundings and return to equilibrium. The transverse spin-spin or relaxation term, T 2 / describes an exponental loss of phase coherence as the nuclei precess about the xy plane. The Fourier transform of a cosine-modulated, exponentially decaying signal is a Lorentzian lineshape with a half-height width of (nT2)_1. Since T 2 is inversely proportional to the correlation time for molecular rotation or tumbling, larger proteins will produce broader Chapter 3 — Secondary Structure of the ETS Domain 72 lineshapes. This results in an upper protein molecular weight limit for solution NMR studies near 40 kDa. 3.1.2. The Building Blocks of a Modern Pulse Sequence In all pulse sequences, spin echoes are an essential tool for the selection of chemical shift or scalar coupling contributions (Hahn, 1950; Carr and Purcell, 1954). Figure 3.1 summarizes variations of the spin-echo for coupled spins. Spin echoes can refocus contributions of chemical shifts and couplings and can also eliminate the effects of magnetic field inhomogeneity. Pulsed field gradient technology, a central feature of magnetic resonance imaging, is exploited in several ways to improve sensitivity and minimize experimental artifacts (Figure 3.2). One of the foremost building blocks is the INEPT pulse sequence (Figure 3.1). The INEPT pulse sequence is used to transfer magnetization from one coupled nucleus to another in a frequency independent manner. This pulse sequence also accomplishes a selective inversion of the spin populations of the nuclei. Selective population inversion of two nuclei with differing magnetogyric ratios provides a means for enhancing the detection of the nucleus with the lower magnetogyric ratio. For example, if magnetization is passed from to a nucleus with a lower magnetogyric ratio such as 1 5 N (YH - 9.8 YN) and then detected on 1 5 N , the improvement in sensitivity corresponds to a factor of ( Y H / Y N ) / or 9.8-fold. Multidimensional experiments derived from the basic building blocks quickly become very long trains of pulses. During this period, T2 relaxation is greatly diminishing the signal to be acquired. To minimize the time required to apply all of the pulses, pulse sequence programmers strive to concatenate tasks wherever possible. Resolution also suffers in multidimensional experiments. During the experiment, spectral-widths are kept at a minimum to reduce the number of points that have to be acquired in each indirect dimension. Even with as few as 32 increments and a minimal phase cycling scheme, a four-dimensional experiment can still take eight days to 73 n I^ B uncoupled system 1 H 11 A A ! ACQ -chemical shift is refocused H n WWW ! •chemical shift is refocused coupled, homonuclear 1 H 11 A J A • ACQ coupling is not refocused coupled, heteronuclear 'H 1 5 N A ! ACQ •chemical shift is refocused coupling is refocused coupled, heteronuclear mJL ACQ 1 5 N : A • A : •chemical shift is not refocused coupling is refocused, "the hard 180" coupled, heteronuclear 1 H I ] A 1 1 5 N J A 1 ^ ! A C Q 'chemical shift is refocused coupling is not refocused INEPT 90>< I H n A 1 5 N ! A 90y A J L 1a£ ACQ 90x refocused INEPT I H I 1 A B A [ 1 A B A 1 5 N | A (a) (b) / A = (4J N H)- 1 at acquistion, 1 5 N magnetization is antiphase along the y-axis; [-2HzNy ]. The ratio of the intensities of the two members of the acquired doublet is: H Y N + Y H ^ Y H ] : [ ( Y N - Y H ^ Y H ) Once the antiphase-doublet has (c) (d) / decouple refocused, proton decoupling may be used to collapse it to a singet. Phase cycling eliminates the initial 1 5 N magnetization ! ACQ (e) H, H, . y -2HXNZ 2HzNy N x Polarization transfer results in a (yH / y N ) signal enhancement. Figure 3.1: A selection of basic experiments involving echoes. In the depicted pulse sequences, a unfilled rectangle denotes a 90° pulse along the x-axis and a filled rectangle denotes a 180" pulse along the x-axis. Product operator formalism is used to describe the state of the ^ and 1 5 N magnetization throughout the INEPT pulse sequence. 9 0 ) c Application of the reverse gradient will rephase dephase... previously dephased rephase — I magnetization. If A is too Z I ^ long, diffusion will prevent I rephasing. 90* 180„ 1 H H A II A ! onward... T h e f i r s t g r a d i e n t dephases a perfect t n e S jg n a | j n t n e x y p | a n e . 180 I I Couplings and chemical -™ " shifts continue to evolve. A 180x pulse flips the sense of precession requiring application of a gradient of the same sign to rephase the wanted magnetization. Components arising from pulse imperfections in the 180x pulse are not rephased. Gradients do not have to be symmetrical about the 180x pulse. These gradients are also known as crusher gradients and replace the need to phase cycle the 180" pulse. 1 H 90y A J I A decouple one pulse refocused 9°x <5N fl 1 Sj i x I j ACQ INEPT * 1 Initial 1 5 N magnetization is tipped into the -y axis and dephased with a gradient akin to a homospoil pulse. The 90y . y phase cycle is no longer required to remove the initial 1 5 N signal. Figure 3.2: Gradients can be used to create an echo, to remove pulse imperfections and to serve as a homospoil. Chapter 3 — Secondary Structure of the ETS Domain 75 complete. While a two-dimensional experiment may be processed to a final proton resolution of 2 Hz/pt, a 3D experiment may only have a resolution of 10 Hz/pt. Though lower signal-to-noise, lower resolution and increased time requirements may seem overwhelmingly negative, these disadvantages often become minor before the prospect of reduced chemical shift degeneracy by dispersion in several dimensions. 3.1.3. The CBCA(CO)NH Experiment One of the great strides in the assignment of larger (70-300 residue) proteins is the use of three- and four-dimensional scalar correlation experiments to "walk" the protein backbone from residue to residue. One representative example is the CBCA(CO)NH experiment (Grzesiek and Bax, 1992). In this experiment, the amide proton and 1 5 N of a particular residue (i) become linked to both the C a and CP of residues (i) and (i-1). Many optimizations have since been made to the CBCA(CO)NH experiment that notably include pulsed field gradient technology and sensitivity enhancement techniques (Muhandiram and Kay, 1994). Summarized below, magnetization is transferred from the H a and H ^ protons through a path of backbone atoms ending at the amide proton. Chemical shift contributions are allowed to evolve during a>i, CO2 and CO3. This produces a three-dimensional spectrum with the axes represented by H N , N , and C a / ^ . This experiment is described in further detail in Figure 3.3. H a ( i . i ) / H P ( i-i) - > C a(i-i)/C P(i-l) - > C'(i-i) - > N ( i ) > H N ( i ) 7 6 (a) (b) ^ (c) (d) (e) (f) (g) (h) (i) -h 1 1 H 13 C«/P I I | v 2 j x b 13C« 13C' 1 5 N G WALTZ-16 : T - t , 1T-'I / 2UTJ 1 I SEDUCE-1 I 1 I I I I 1 1 'z • n-n n J L g1 g2 g3 g4 g5 g6 g7 I n-n n-n § g9 g10 g11g12 g13 g14 Figure 3.3: The CBCA(CO)NH experiment (Grzesiek & Bax, 1992; Muhandiram & Kay, 1994) incorporates nearly every optimization found in a modern pulse sequence. The ability of the sequence to discriminate carbonyl and C a / ' 3 shifts is particularly impressive, (a) A 90° pulse on 1 3 C followed by gradient gl (solid) destroys any carbon magnetization before the pulse sequence starts, (b) INEPT transfer of !H magnetization to ^C. This INEPT sequence utilizes "crusher" gradients (g2 and g3; Figure 3.2) and a gradient (g4; similarly for g7 and g8) to dephase any transverse magnetization (Hy) before the 13C 90° pulse, (c) H z C y antiphase magnetization is refocused by a 180° pulse flanked by crusher gradients, g5 and g6, and evolves under chemical shift only in period tj. A "hard" 180° pulse on 1 5 N and a 77 "soft" shaped off-resonance pulse on 13c (null on IK**) refocuses unwanted couplings that would otherwise evolve in tj. With three other nuclei, 1 3 C , 1 5 N and 1 H , to monitor in this constant-time period, Muhandiram and Kay (1994) switch on the proton decoupler at the optimal time when proton couplings have refocused. The echo-time, tb, has been calculated using the method of Kay et al. (1990). (d) After the constant-time period, the relevant magnetization is C^^^cosf^a/B)^]. During (d), magnetization is converted to C^^C'y anti-phase magnetization. C a / P and C resonances are separated by -100 ppm and can be treated as distinct nuclei. The sequence covered by (e) is an optimized version of a refocused INEPT followed by another INEPT to a third spin. In terms of product operators, magnetization follows the path: 2C a / p ,C' -> C" -> 2C z N y . Before the last 90° pulse, when C' Z N Z magnetization is present, a gradient (g8) is applied. During the second constant time period, 1 5 N magnetization evolves under chemical shift with a hard 180° pulse used to decouple C coupling and SEDUCE-1 decoupling to eliminate 2JNCCX • Immediately after SEDUCE-1 decoupling, at point (g), a gradient pair (g9/gl4) is applied to select only single-quantum 15N/!H coherence with g9/gl4 being satisfied by the equation, YNGztp = ±YHGztp where p is the coherence order, t is time, and G z is gradient strength. Finally, during (h) a sensitivity enhanced reverse-INEPT is run and in-phase proton magnetization H y , at point (i), is acquired. The final spectrum correlates C^P (coj) of residue (i-1) with 1 5 N (co2) and H N (co3) of residue (i). Chapter 3 — Secondary Structure of the ETS Domain 78 Phase-cycling is reduced to only four cycles per increment by the judicious use of pulsed field gradients and concatenated pulses. Superimposed on the States-TPPI quadrature detection scheme (States et al., 1982) is a gradient cycling scheme to retrieve both of the " N - " and "P-type" coherence pathways. Constant-time acquistion limits the acquistion to (SWi /Jcc) i n tl a n d ( S W 2 / J C N ) in *2— typically, only 32 points are collected in each dimension. Since T 2 relaxation is also constant during both acquistion periods, the chemical shift modulated cosine term is not multiplied by an exponential decay term. Mirror-image linear prediction (Zhu and Bax, 1992) performs well on this type of modulated data to extent the number of points up to a factor of two. Additional zero-filling and appropriate apodization ensures an acceptable final digital resolution for interpretation. 3.2. Experimental Methods 3.2.1. Expression of Isotopically Labelled Proteins Uniformly 15N-labelled Ets-1 AN331 protein was produced from Topp-2 cells harbouring plasmid pCW-ets in M9 minimal medium containing 1.0 g/L 99% 15NH4C1 or 1 5 (NH 4 ) 2 S0 4 and 1.0 g/L 99% 15N-labelled Isogro algal extract (Isotec; Miamisburg, OH). Uniformly 1 5 N - and 13C-enriched protein was produced from cells grown in M9 minimal medium containing 1.0 g/L 99% 1 5 NH 4 C1 and 2.0 g/L 99% 1 5N/ 1 3C-labelled algal extract. Addition of labelled algal extract dramatically improved the growth rate and overexpression levels. The amides of tyrosine, lysine, valine and leucine were selectively a-15N-labelled in a synthetic rich medium described by Mcintosh and Dahlquist (1990). Proteins were assayed for purity and activity according to the methods described in Chapter 2. Identical buffer conditions (20 mM sodium phosphate, pH 6.4,10 mM dithiothreitol, 0.01% sodium azide and 10% D 20 or 99% D 20, 20°C) were used for all samples. Chapter 3 — Secondary Structure of the ETS Domain 79 3.2.2. Instrumentation and Software Experiments were performed on a Varian Unity 500 NMR spectrometer (carrier frequency 499.9761 MHz in 1 H) equipped with three radio-frequency channels and a pulsed-field gradient accessory. The NMR data were processed and analysed using Felix v2.30 (Biosym Technologies; San Diego, CA), nmrPipe (Dr. F. Delagio; NIDDK, National Institutes of Health, Bethesda, MD) and PIPP/STAPP/CAPP software (Dr. D. Garrett, NIH, Bethesda, MD) accompanied by several processing and file-manipulation scripts that I wrote. Many of these scripts are now routinely used in the laboratory for protein assignment. One of these scripts, xprocpar, will read the experimental parameters from the Varian procpar file and write out a conversion and processing scripts appropriate to the dimensionality, the phasing required in the indirect dimensions and the carrier frequency of the nuclei involved. Before processing, experiments that incorporated sensitivity-enhanced sequences required grad_sort_nd software formatting (Dr. L. Kay, University of Toronto) to enable the sensitivity enhanced data to be separated in cosine and sine modulated terms. 3.2.3. NMR Experiments The following experiments were recorded with a 0.7 mM sample of uniformly 1 5 N -labelled Ets-1AN331: 2D HSQC (Kay et al., 1992) and HMQC-J (Kay and Bax, 1990) and 3D TOCSY-HSQC, NOESY-HSQC (Fesik and Zuiderweg, 1990) and HNHA (Vuister and Bax, 1993). To monitor the time course of hydrogen exchange, a series of HSQC spectra were acquired after transferring the sample into D2O sample buffer (pH* 6.5) by passage through a Sephadex G-25 spin-column. After complete exchange, 2D homonuclear DQF-COSY, clean TOCSY (65 ms MLEV-17 mixing period) and NOESY (100 ms mixing period) spectra were recorded on the sample. 1 H - 1 5 N HSQC spectra were also obtained from 0.2-0.4 mM samples of Ets-1AN331 selectively labelled with [a-15N]-tyrosine, -valine, -leucine, or -lysine. The following 3D experiments were recorded with a 0.7 mM sample of uniformly 13c/15N labelled Ets-1 AN331: HNCO (Muhandiram and Kay, 1994), HNCACB (Wittekind and Mueller, 1993), CBCA(CO)NH (Grzesiek and Bax, 1992), CBCACOHA Chapter 3 — Secondary Structure of the ETS Domain 80 (Kay, 1993), H ( C C O ) N H , C ( C O ) N H (Grzesiek et al., 1993), 2D- (H8)CB(CYC5)H8 and (Hp)CP(CyC5Ce)He experiments (Yamazaki et al, 1993), constant-time aromatic and aliphatic 1 3 C-HSQC and HCCH-TOCSY (Kay et al, 1993). The correlations involved in each of these experiments are illustrated in Figure 3.4. For experiments in which magnetization was detected on the amide H N , sensitivity-enhanced gradient pulse sequences were employed along with selective flip-back pulses to return water magnetization to equilibrium before acquistion of the NMR signal (Grzesiek and Bax, 1993; Muhandiram and Kay, 1994). The initial timings in the indirect dimensions were set to 1/(2* sw) resulting in first-order phase shifts of 180° and the inversion of aliased signals. Quadrature detection was accomplished in the indirect dimensions using the States-TPPI method (Marion et al, 1989). The biosynthetic pathway is elegantly exploited to provide a means of stereospecifically assigning leucine and valine methyl groups (Neri et al, 1989). Through pyruvate, acetolactate and a,p-dihydroxyvalerate intermediates, the carbon units are passed to leucine and valine. From a,(3-dihydroxyvalerate, the pro-R methyl group carbon is assembled as a two-carbon unit along with the C P (in valine) or CY (in leucine). The pro-S methyl is assembled onto the new amino acid as a one-carbon unit. Expression in a 90% 12C-glucose/10% 13C-glucose medium results in pro-R methyls and neighbouring C P or CY being simultaneously labelled with 1 3 C . In a 1 H, 1 3 C-HSQC experiment, pro-R methyls appear as doublets in the 1 3 C dimension whereas pro-S methyls appear as singlets. Patterns of 1 3 C / 1 2 C labelling also facilitate the identification of additional amino acid types such as alanine and isoleucine in the HSQC spectrum. 3.3. Results 3.3.1. Preliminary Spectra Ets-1AN331 is an amino-terminal deletion mutant of the murine Ets-1 proto-oncoprotein, containing, the ETS domain (Figure 2.2; residues 331-415; 110 residues) and the native carboxy terminus (residues 416-440). The latter 25 amino acids were found to be Chapter 3 — Secondary Structure of the ETS Domain 81 essential for solubility. Ets-1AN331 represents a minimally sized fragment of Ets-1 that is fully functional for DNA binding and amenable for structural studies by NMR methods. Previously discussed in Chapter 2, a significant effort was taken to optimize conditions for NMR analysis. Unfortunately, the sample concentration had be reduced to limit aggregration. However, even at 0.7 mM, the spectra looked favourable with significant *H and 1 3 C chemical shift dispersion (Figure 3.5). Figure 3.6 depicts a one-dimensional 1 3 C spectrum of ETS-1AN331. The corresponding 1 5 N experiment was not acquired. 3.3.2. Assignment of Ets-1AN331 The ! H - 1 5 N HSQC spectrum of uniformly 1 5 N enriched Ets-1AN331 is essentially a complete fingerprint of the protein, with 101 of the expected 105 amides detected (Figure 3.7). At lower pH favoring less proton chemical exchange with water, the H n groups of arginine would be visible at 80-85 ppm. Near complete assignments of the main chain 1H, 1 3 C and 1 5 N resonances of Ets-1 AN331 were obtained using a combination of three complementary approaches. First, the amide resonances of the tyrosine, valine, leucine and lysine residues, which represent 32 % of the protein sequence, were identified from 1H- 1 5N HSQC spectra of selectively labelled Ets-1 AN331 samples. These served as unambiguous reference points for the specific resonance assignments. Second, a combination of 3D 1 H / 1 3 C / 1 5 N scalar correlation experiments was used to derive intraresidue and sequential connectivities between the main chain atoms in a sample of uniformly 1 5N/ 1 3C-enriched Ets-1AN331 (Ikura et al, 1990). Third, in cases where degeneracies or insufficient signal-to-noise ratios precluded confident assignments using these through-bond, triple-resonance experiments, a main chain directed strategy was followed to interpret the 3D ! H / 1 5 N TOSCY-HSQC and NOESY-HSQC spectra of uniformly 15N-enriched Ets-1AN331 (Englander and Wand, 1987). Figure 3.8 illustrates the main chain assignment of the first helix of Ets-1AN331 using a combination of CBCA(CO)NH / HNCACB spectra and TOCSY-HSQC/NOESY-HSQC spectra. Consult Appendix 1 for the Ets-1AN331 chemical shift table. 82. H C H H C H H C H H C H CBCA(CO)NH H C H I H C H -u-i-m I I H H H C H I H<pH H O HNCO H C H H C H I :' H C H H C H ; - _jo_A. _ TOCSY-HSQC H H O I H O *fcA | 4 o NOESY-HSQC H C H H C H SI « I H<pH_ H ^ H -H—C—c H c—c-H i o i l O H(CCO)NH H C H I H C H H 1 o i l O C(CO)NH H H O H H O HCCH-TOCSY Figure 3.4: Three-dimensional heteronuclear experiments used to assign Ets-1 AN331. Shaded atoms are detected in the experiments. A square denotes through-bond connections and a circle denotes through-space connections. 83 8T 750.0 700.0 50.0 0.0 ppm Figure 3.6 : 1 3 C spectrum of Ets-1AN331. (ujdd) N g i 85 o CM LO CM O CO O CD O o co E Q_ Q_ O o o •7 ^ a; ~ CD 3 M O s 3 e o <n o cn .5 cn ^5 53 ai p u S b -2 °1 « ^ OH VO O a, o £ « CO 0 3 cn CD cn T3 3 cn cu >-. e ' S a; O O Pu gj ~ -t-> a; 0) - C 0) <S to S o s § cn ns o § £ 6 o o p . M l cn - 3 U C O cn X z to C C a; cc «u cn ai u C co C O cn ai « c 3 cn 60 O b 43 CO J-.2 •£ •B •+< 6 £P eg cn C. cn •3 cu C cj c 2 o pi (uidd) Hi n o o (0 o CO • c I o (0 (uidd) o e t o n O 0 4> u X 7 3 2 C ^ CT3 O -r CO m — > C/> o 5 ra C - C O " U c a — > 73 — o> z X c 0 a. 3 7 3 01 x; H > , 7 3 X l C I 1 to 2 I O i < £ u s m •c u 0- a , 5 = CD a> 2 u < nj C is S tu o> O a . c _ S c SO B '35 U c 3 a" Ol cn ,•- « S 5 a i 3 7 3 « C X I Oi J2 o e § 3 o> Ol o S 2 7 3 C U 7 3 C Ol a. CA ca U < V 7 3 O o i a. S cn o> > O c .S S 5 tn xi c ° to o c Ol e Ol Q . X 01 > Cn U 0 H U a CO X 01 x: § * 01 a" •£ U - c X . 01 bO oi § * 1M OI • f j X l cn cn c C u O '£ 'fj cn —i Ol >< t: X i o _ u 2 a x! xl T-5 C u '35 g a 3 2 cn •£ a. X CU OJ „ ^ Bl ^ X . CXI c tu .2 O u Z cn C 3 >" £ O 5 H X ! T J V to u 4 . 01 0 ; £. » o L- ^ 1 c ^ cn C C ° E 7 3 w o ^ c ' TO co u ai 5^ 3 c 60 Ol O ^ "I-e S -S -B- 2 A " » a ^ ca OL ? U « •6 < 2 « u 2 v ? y re X 2 is a i _ L c -c X cn 01 cj 0 0 C ns n 0 cn - J 2 5 72 ^ S c 3 (3 Ol 01 - J x; 1-— o x: M -ao > 3 cn o u cn 1 Q . U '= a cn cn . X 52 cu § •£ 0 S c p  o u LU O 2 ao c .1 7 3 z S —. o J- cn 3 QJ 1- o o <J - 01 . ± X Chapter 3 — Secondary Structure of the ETS Domain 87 3.3.3. Secondary Structure Analysis by NMR The secondary structure of Ets-1 AN331 was derived from four approaches: deviations from random coil chemical shifts, short-range NOE correlations, 3 J H N H O C scalar couplings and amide hydrogen exchange rates. Chemical Shift Analysis A strong correlation exists between secondary structure and the deviations of the H a , C a , CP and C chemical shifts of an amino acid in folded protein from its random coil values (Spera and Bax, 1991; Wishart et al, 1992; Wishart and Sykes, 1994). For example, residues in a-helices show downfield or positive C a and C secondary shifts; whereas those in B-sheets show upfield or negative H a and CP shift perturbations. The observed secondary shifts for Ets-1AN331 define four pronounced a-helical regions (HI through H4), as well as four short sequences with B-strand conformations (Si through S4). For a given helix of residues N , N l , N2 and N3, where N is the residue preceding the helix, a helix capping box is observed when the side chains of N and N3 make hydrogen bonds with the amides of N3 and N, respectively (Chakrabarrty et al, 1993). Asparagine, glutamine, aspartic acid and glutamic acid tend to occupy the N and N3 positions. Chemical shift patterns are diagnostic of helix capping boxes particularly for the capping residue which has pronounced upfield C a chemical shift, relative to its random coil value (Gronenborn and Clore, 1994). These chemical shift patterns also are observed in peptides (Lyu et al, 1993). Of the four helices observed in Ets-1AN331, helices HI and H2 exhibit the characteristic chemical shift pattern and the correct residue type at positions N and N3 (Figure 3.9). Though helix H3 has the correct chemical shift pattern, K388, at position N3, is not a suitable hydrogen bonding partner for the amide at position N . Helix H4 does not fit the prototype for a helix capping box since its CP is not downfield shifted at position N and also, T426 is incapable of making a hydrogen bond to the amide at postion N3. In helix H2, a proline is predicted to be the lead, or N-terminal bounded residue (Presta and Rose, 1988). A proline may serve as an acceptable lead residue since, in the trans configuration, it is constrained to a region of ((j),y) angle space 88 C a CB N N1 N2N3 © L W © H e l i x 1 C a CB N N1N2N3 N Y E K H e l i x 3 C a CB N N1N2N3 H e l i x 2 (5) P D ( ? ) V A R C a CB N N1N2N3 H e l i x 4 T P E E L H N N1N2N3 P r o t o t y p e f downfield shift ( A o b s > A r a n d o m c o N ) | upfield shift ( A o b s < A r a n d o m c o H ) Figure 3.9 : N-terminal Helix Capping Boxes — In a given helix with residues (N, N l , N2, N3), the N-cap residue takes on a very specific geometry (<t>=-94\ y=167°). This geometry occurs when the side chain of residue N can hydrogen bond to the amide of residue N3. The "box" is formed when the side chain of residue N3 correspondingly hydrogen bonds to the amide of residue N. In a capping box, the N residue is typically upfield shifted 3 ppm in C a and downfield shifted in CP- The C a chemical shift is the more reliable indicator of the two carbon chemical shifts. In the figure above, helices HI and H4 fit the prototype of a capping box. Helix H3, with lysine at position N3, and helix H4, with threonine in at position N , cannot make a hydrogen bond to their respective N and N3 residue partners. Side chain to side chain NOE data can be used to verify candidate capping boxes by this method. J-coupling, hydrogen exchange and NOE data predict helical bounds specified by dashed lines under the amino acid sequences. Chapter 3 — Secondary Structure of the ETS Domain 89 similar to an a-helix. Furthermore, in three of the ten helices of triose phosphate isomerase (1TIM; Banner et ai, 1975), a proline assumes the lead position. Sequential and Short-Range NOE Correlations Regular a-helical and B-sheet residues show characteristic NOE patterns between H N , H a and HP protons (Wuthrich, 1986). The NOE correlations between these main-chain protons in Ets-1AN331, obtained from 2D homonuclear NOESY and 3D heteronuclear 1 H / 1 5 N NOESY-HSQC spectra, are summarized in Figure 3.10. Evidence for four a-helical regions in Ets-1AN331 is provided by a contiguous series of H N i-H N j+i NOE interactions, combined with weak H a i - H N , + i NOE's and a limited number of observable H a i -H N ;+3 NOE's. The four stretches of residues in 3-strand confromations are identified by relatively strong H a;-HNi+iNOE's and the absence of sequential H N ; - H N i + i NOE's. More diagnostically, the patterns of cross-strand NOE interactions between the H a and H N protons of these residues provide unambiguous evidence that Ets-1AN331 contains a four-stranded anti-parallel (3-sheet (Figure 3.10) ^ J H N H O Scalar Coupling The 3jHNHa coupling constant is diagnostic of the conformation of a residue because of its dependence upon the <j) backbone dihedral angle. Regular helical regions are characterized by 3 J H N H O couplings on the order of 4 - 6 Hz, while peptides in extended conformations show couplings larger than 7 Hz (Wuthrich, 1986). Only 3 J H N H O values larger than ca. 8 Hz were measured confidently due to the broad amide 1 5 N linewidths (> 12 Hz) in the HMQC-T spectrum of Ets-1AN331 and the low signal-to-noise ratios in the H N H A spectrum (Figure 3.10). Consistent with the conclusions drawn from chemical shifts and NOE patterns, all of the residues assigned to the four helices showed couplings below this limit. The residues with 3 J H N H O couplings greater than 8 Hz were generally located near the ends of the helical regions and in 8-strands. Amide Hydrogen Exchange The rate of exchange of an amide H N with water depends upon hydrogen bonding and solvent accessibility. Thirty amides in Ets-1AN331 with exchange rates of less than Figure 3.10: Summary of the NMR evidence used to derive the secondary structure of Ets-1AN331. The locations of the four a-helices (H) and four B-strands (S) are marked by cylinders and arrows, respectively. A bar above the protein sequence highlights the 85 residue ETS domain, (i) HX: Amide hydrogen-deuterium exchange measured after transfer of the protein to D 2 0 buffer at pH* 6.5 and 20°C. Filled circles indicate slow amide exchange (t^/2 > 1000 min), half-filled circles indicate intermediate exchange ( 50 min < ti/2 < 1000 min) and open circles indicate that a rate could not be determined under these conditions (ti/2 < 50 min). (ii) H N - H 2 0 xpk: An asterisk denotes rapid amide hydrogen exchange (t^ /2 < 0.5 sec) or a NOE to long-lived bound water as evident by a strong H N - H 9 0 crosspeak in the NOESY-HSQC spectrum. (Hi) 3 J H N H C X : Amides with 3 JHN-HCX > 8 H Z are marked by the letter J . (iv) Unambiguous NOEs between H N and H a are labeled as d N N ( U + 1 ) , d a N ( i d a N ( i i + 2 ) , and d a N ( i i + 3 ) . The relative NOE strength (weak/medium/strong) is reflected by the bar thickness (tmix=100 msec), (v) The main-chain H a , C a , C p , and C chemical shifts are plotted as the difference from the corresponding random coil values (Wishart et al, 1992; Wishart and Sykes, 1994). 92 O) cu X cn - S 3 H is re CL. c re T3 cu T3 C re (-1 3 O 14-1 CU O CU t: £P TO C . re re S -a -C .S ^ 'cu 3 u, cu re -a a jL •O 5P O Oi •*-* Xl cn f—i re cu .— D , o cn i - 1 re i-; cn g co -a « 2 § 60 60 re ^ 4^ QJ C J cu 8 C 01 60 O I-I T3 60 cn C £ ° n >M O In - ° re c cu cn cu a. 0) l-l re s cu X CJ cn cn ai 3 c 5 £ -2 > cn Z. u, 0) o cn 2> 0 ^ ^ 01 5 s T3 in 01 oi LH 5 Ol cn — •81 . o CO O co £ < cu •22 g W re 60 c 01 -3 x C re C o X cn O c re M oi re 1-4 c 3 5 Sri 0) 0) X 0-0 cn ^ a cn >< _cy re cs 01 Q re K OH In ' r e X 0) 4-1 I* O 60 _u cn .a £ 6 c QJ . 3 -6 -a o .a s £ cn . 3 Chapter 3 — Secondary Structure of the ETS Domain 93 2.0xl0~4 s"1 were detected by recording a series of 1 H - 1 5 N HSQC spectra after transfer of the protein into D 2 O buffer at pH* 6.5. This corresponds to a protection of at least 1000-fold relative to an amide in an unstructured peptide (Englander and Poulsen, 1969). The amides with significantly retarded exchange rates were located in helices HI and H2, S-strands SI, S2 and S4 and a small section of the C-terminal sequence. This provides additional support for the hydrogen boding expected within these secondary structural elements. Although the amides in helices H3 and H4 and strand S3 were not measurably protected under these conditions, the first HSQC spectrum was recorded 1 h after transfer of the protein to D 2 0 , and thus only those amide protons with significantly retarded (>1000-fold) exhange kinetics were detected by this approach. Twenty-four amides in Ets-1AN331 with fast exchange rates, comparable to those of an unstructured, hydra ted polypeptide (ti/2 < 0.5s), were identified from strong crosspeaks to water in a 3D 1 H - 1 5 N NOESY-HSQC spectrum. These amides lie in loops and turns between the a-helices and (3-strands (Figures 3.10 and 3.11). No attempt was made to distinguish exchange cross-peaks from NOE's to long-lived bound waters at this stage of the analysis. Such an analysis has been performed, however, for a larger deletion mutant of Ets-1 (discussed in Chapter 5). Summary Analysis of the NMR chemical shift, NOE, J-coupling and amide hydrogen exchange data demonstrates that the first 85 residues of Ets-1AN331, corresponding to the ETS domain proper, have the secondary structure H1-S1-S2-H2-T-H3-S3-S4. The 25 residues beyond the the ETS domain contain a fourth helical region, H4. The NMR evidence for this secondary structure is discussed here. Helix HI contains L337 to T346, as evidenced by H N r H N i + 1 and H a i - H N i + 3 NOEs, the absence of 3jNHHa > 8 Hz, negative H a secondary shifts, and positive C a and C secondary shifts. Residues 339 - 345 also show strong protection from hydrogen-deuterium exchange, indicating that they constitute a stable hydrogen-bonded helix. Although sequential H ^ i - H ^ i NOEs and positive C°- secondary shifts suggest that the Chapter 3 — Secondary Structure of the ETS Domain 94 helix may continue beyond T346, the large 3jHNHa measured for D347 and the fast amide exchange of residues 348-352 indicate that the backbone of these residues may form an irregular or flexible structure. Helix HI precedes a B-hairpin comprised of strands SI (1354 - T357) and S2 (E362 -L365) linked by a four residue turn (Sibanda et ai, 1989). The B-sheet conformation of these amino acids is readily identified by strong H a ; - H N i + i NOEs, distinct positive H a and CP secondary shifts, slow amide hydrogen exchange, and most importantly, by cross-strand NOEs between residues in SI and S2. The anti-parallel alignment of the two strands is established from the H a - H a NOE correlations between 1354 and L365 and between W356 andF363 (Figure 3.11). Following the (3-hairpin, P368, initiates helix H2. This helix ends near residues 379 - 381, which appear to deviate from a standard helical conformation as indicated by a relative loss of protection from amide hydrogen exchange, strong H a i -H N i+i NOEs, and irregular patterns of secondary chemical shifts. Helix H3 is composed of Y386 - Y397, as identified by characteristic C a and C secondary shifts and H N ; - H N i + i and H ai-H Nj+3 NOEs. However, the protection from amide hydrogen exchange is less than what is observed for helices HI and H2. The residues between helices H2 and H3, which include K383-N385, are clearly non-helical, considering the large 3 J N H - H C C coupling measured for K383 and fast amide hydrogen exchange. The exact size of this region, which forms a turn between helices H2 and H3, remains ill-defined because the end of helix H2 is uncertain. A second B-hairpin, formed by strands S3 (1402 - T405) and S4 (V411 - F414) completes the secondary structure of the ETS domain. These two extended strands are defined primarily by pronounced downfield H a and CP secondary shifts and diagnostic cross-strand NOE connections. The amides in the five residue turn or loop linking S3 and S4 are not protected from hydrogen exchange and appear to be exposed to solvent, giving exchange or possibly NOE correlations to H 2 0 in a 1 H - 1 5 N NOESY-HSQC spectrum. Chapter 3 — Secondary Structure of the ETS Domain 95 Together, the two B-hairpins, SI - S2 and S3 - S4, form a four-stranded anti-parallel B-sheet with a distinct hydrophobic face (Figure 3.11). The alignment of the strands is (+1, +2x, -1) by the nomenclature of Richardson (1981) with strand S4 sandwiched between S2 and S3. In the region bounded by residues Y397 to V411, many H N a n d 1 5 N resonances were unusually weak or broad, thereby making assignments difficult to obtain. D398, R409 and Y410 could not be assigned, and combinations of triple resonance experiments and selective isotopic labelling yielded only partial assignments for K408 and V411. These residues include the region joining H3 and S3, strand S3 itself, and the turn or loop extending to S4. Line broadening due to slow conformational averaging of this region of Ets-1AN331 is the likely reason for the weak or missing resonances. The water magnetization is only minimally perturbed in the experiments employing pulsed field gradients and selective "flip-back" pulses, and thus rapid amide hydrogen exchange is not expected to cause significant line broadening under these experimental conditions. It is noteworthy that Vuister et al. (1994) reported similar behavior for the H N and 1 5 N resonances from the corresponding residues in Drosophila HSF. The final twenty-five amino acids in Ets-1AN331, although not part of the conserved ETS domain, also adopt a defined structure. With the exception of the extreme C-terminal residues D438, A439, and D440 which have unusually sharp resonances due to rapid conformational averaging, the H N and 1 5 N peaks from this portion of Ets-1AN331 are comparable in linewidth and intensity to those from the bulk of the ETS domain. Helix H4 spans E427 through M432 using the criteria of contiguous H N i -H N i+ i NOEs, and positive C a and C and negative H a secondary chemical shifts. The residues immediately preceding E427 also show H N i - H N , + i NOEs and protection from hydrogen exhange. However, the patterns of H a i - H N ; NOEs, backbone J couplings, and secondary chemical shifts preclude a confident assignment of the secondary structure for this region of the protein. Chapter 3 — Secondary Structure of the ETS Domain 96 3.4. The Winged Helix-Turn-Helix Motif NMR analysis demonstrates that the ETS domain is composed of three a-helices and a four-stranded anti-parallel B-sheet. The three helices (H1-H3) and the four P-strands (Sl-S4) are linearly arrayed in the order H1-S1-S2-H2-H3-S3-S4. Additional evidence supports this analysis: (i) Both a-helical and P-sheet secondary structural elements are observed through a combination of CD and FT-IR studies of Ets-1AN331 described in Chapter 2. (ii) Given all Ets protein sequences, the PHD algorithm (Rost and Sander, 1993) predicts a consensus secondary structure for the ETS domain closely resembling that determined by NMR methods (Chapter 1) (Hi) In the alignment of over twenty ETS domains, amino acid insertions or deletions occur only between the helices and strands, HI and SI, SI and S2, S2 and H2, H2 and H3. The secondary structure analysis also indicates a fourth helical region in the C-terminal sequence following the ETS domain of Ets-1AN331. Although not formally considered part of the ETS domain, there is limited primary sequence conservation in this region among a subset of Ets proteins (Klambt et al, 1993). This extension is also required to maintain the solubility of the murine Ets-1 ETS domain. Therefore, it is likely that this C-terminal extension is an important structural component of Ets-1. Several studies have in fact implicated this region in the regulation of Ets-1 DNA-binding (Hagman et al., 1992; Lim et al, 1992). The ordering of a-helices and p-strands of the Ets-1 ETS domain strongly resembles that of the DNA-binding domains of E. coli catabolite activator protein CAP, E. coli biotin repressor birR, hepatocyte nuclear factor HNF-3y, yeast heat shock factor HSF and the globular domain of histone H5 (Figure 3.11), among others. The DNA-binding domains of these five proteins also have similar tertiary structures (Brennan, 1993; Burley, 1994). Crystallographic analyses demonstrate that three helices in each of these domains pack to form a hydrophobic core which rests on an amphipathic P-sheet scaffold. DNA contacts are made through a helix-turn-helix motif (helices 2 and 3), the P-strands, the intervening loop regions, and the first helix. One particularly large loop within the P-Chapter 3 — Secondary Structure of the ETS Domain 97 sheet of HNF-3y led to the designation "winged-helix" to describe the structural motif of this protein (Clark et ai, 1993). The designation of the "winged helix-turn-helix" (wHTH) for the entire superfamily of DNA-binding proteins has been proposed for this group of DNA-binding proteins (Donaldson et al, 1994; Clubb et al, 1994). In this new, broader context, the "wing" would be the p-sheet region and any associated loops. This proposal distinguishes the wHTH superfamily from other HTH proteins, such as the homeodomain family, and highlights the role of the sheet region (with or without loops) in potential DNA contacts. Based on the close resemblance of their secondary structures, it was proposed that the ETS domain has a tertiary structure similar to that of the wHTH proteins, with helices H2 and H3 corresponding to a helix-turn-helix motif. The structural similarity of the wHTH proteins occurs in the absence of any pronounced sequence similarity. Thus, previous attempts to align the ETS domain with the wHTH motif were largely unsuccessful due to the lack of clues derived from only primary sequence information. In the next chapter, the tertiary structure is discussed in light of these predictions. 98 to s ts | ! t- t OS • E~ • 3 O • « o o w o CO Q >• >• >• CO X CM CM CO CO a o I I I E s a > E-IN • a, Q CO t! T < a o o u ui > a a > o >• s z a w a > > > td 2 > S CO 3 • J Z > u. a: ! T > 'a. O a z o 3 ot a z a E CD 3. Q Z X o > PC < < < a: 0) c o *H 5 co T J c o C J QJ cn QJ LH co QJ cn r*, CO QJ cn ^ 8 QJ 45 I a QJ J5 H T J QJ O l-i _ OH CO cn C QJ H-> O N O N l-i .O c QJ cu N ^ *3 ON T J c CO cn C g '5b QJ VH OH O O c 3 T J QJ N I 3 ^ 45 W C^ o 45 cn cn S § ? QJ * C < (5 'co U g 45 cn • -H 1-1 CO g CO i- to ffi .P c PH QJ T J l-l o T J cn H W QJ X cn ,f5 'co g o T J T J QJ c LH QJ H—> QJ T J QJ T j < Q QJ 45 3 u O 3 i n cn H-> CD cn C >^  2 cj 3 QJ c o T j c CO C O o \ O N LH c CO T J c o cj QJ cn QJ 45 H o 1-1 (0 l-l CO > CD 45 CD H-» CO T J 2 I .2 I -O <o O T J C CO T J c CO cn co to T J C 8 - • CJ C3 C -cn co 3 C co g T J QJ c .I3 1 « 45 I* QJ £ co cn QJ cn co QJ 45 >^ co O 45 2 *H 4 - DH QJ 15 cn, cn T J 55 CO L H Sj "H O N O N - . (5 3 O) , < s ~ QJ CD X T J f5 CO cj C CD 3 cr CD cn QJ X H T J QJ CJ 3 T J O LH H-« .s QJ LH QJ cn OH CO u cn CD cn C CN co QJ LH 3 60 CD CO cu C O cn O a x H cn >TH QJ HH 42 • £ cn QJ co • 3 cn CJ IH 3 co OJJ |_ —4 O 45 cn 3 -a cn CD cj 55 CD SH « CD O cn OH QJ ^ 4-H H-> 43 !5 ! H QJ cn i5 S 53 O M - i OH ^ QJ ^ N QJ g ^ 2 o £ 05 C h a p t e r 4 The Tertiary Structure of the ETS Domain Most of the experimental details of the NMR analysis of Ets-1AN331 have been covered in the previous chapter. Here, NOE experiments and methodology leading to the tertiary structure are described. The tertiary structure confirms the conclusion that the ETS domain is a winged helix-turn-helix family member. A fourth helix, H4, carboxy-terminal to the ETS domain and presumed to be disrupted in the v-Ets oncoprotein, is positioned near the first helix of the ETS domain. Because contacts are made to DNA in the vicinity of helix HI of the ETS domain, helix H4 may serve a role in the intramolecular repression of DNA binding. A structural description for DNA binding is presented in the context of biochemical and genetic studies reported for Ets-1. 4.1. Introduction From the secondary structure analysis, the ETS domain was shown to possess a winged helix-turn-helix (wHTH) motif. The wHTH motif is characterized by a helix-turn-helix motif built upon an anti-parallel (3-sheet. By homology, the wHTH family exhibited such diversity that most sequence comparison methods could not identify any relationship among what are now known members. However, closer manual inspection of the sequences of wHTH proteins such as CAP, HSF, HNF-3y and Histone H5 revealed a distinct preservation of residues that form the hydrophobic core of the wHTH motif. In this chapter, the NMR solution structure of the Ets-1AN331 deletion mutant serves as a focal point for a discussion of the structural conservation among the Ets proteins and the wHTH family. 99 Chapter 4 — Tertiary Structure of the ETS Domain 100 4.2. The N M R Approach to Structure Determination 4.2.1. Restraints A restraint is an experimentally determined distance range between two atoms or an angle range between two intersecting planes of atoms. It is the basic experimental unit that defines an NMR structure. On its own, one restraint is not very powerful. However, a sufficient number of restraints can collectively define a tertiary structure with atomic precision. During a structure calculation, each restraint is compared to the current structure. If a given distance or angle is out of the bounds specified by the restraint, an energy penalty is calculated. It is the goal of the structure calculation program to minimize the total value of the energy penalties. An energy penalty is calculated according to a function that often mimics the atomic property involved. For example, a bond that is allowed to stretch within a specified range follows a quadratic potential function for a spring. A square-well function is commonly adapted for many restraints as it allows a parameter to vary yet still be tightly confined within a given range. Structural attributes of proteins such as peptide bond lengths, co torsion angles, and aromatic ring planarity are nearly invariant and do not need to be explicitly defined in a structural calculation. Rather, these attributes are considered as "constraints" in a "force field" that is specific to the biomolecule. Variable weighting is assigned to experimentally determined restraints and molecular constraints to provide structural variability at the secondary and tertiary structural levels while maintaining structural invariability at the primary structural level. Hydrogen bonds are a special case as they are treated as distance restraints and not in terms of electrostatics. In addition to distance and angular restraint classes, hydrogen bonding, direct J coupling, and chemical shift information can be incorporated into a structure calculation. Representative NMR experiments and the parameters that they measure are listed in Table 4.1. Of the restraint classes, distance restraints from NOESY experiments are the most plentiful and therefore contribute the most information towards obtaining a 202 Table 4.1: General Classes of NMR Restraints Class NMR Experiment* Distance pH^HJ^D-NOESY [^N^H^HJ-SD-NOESY FQWHJ-SD-NOESY FC^N^H^HBD-NOESY Dihedral Angle (<|> angle) HMQC-J (<t> angle) HNHA* ft angle) HNHBt Chemical Shift chemical shift assignments* Hydrogen Bonds Amide hydrogen exchange NMR Property The Nuclear Overhauser Effect (NOE) is a through-space dipole-dipole interaction that has a distance dependency that varies as (1 /d A B ) 6 . The upper limit of an NOE restraints is ca. 5.0 A . The lower limit of an NOE restraint is generally set at the van der Waals repulsion limit (1.8 A). 2-3 bond couplings between protons measured in Hz may be translated into torsion angle information according a Karplus relationship. Chemical shifts, particularly those of H N , N, C a and CP, are sensitive to the secondary structure. Information is obtained by calculating and comparing trends among the small ( < 1 ppm) chemical shift deviations of these nuclei relative to their random coil values. Hydrogen bonds in cc-helices and cross-strand hydrogen bonds in p-sheets are identified according to their resistance to chemical exhange with the solvent (typically D 2 0) * Only a few representative experiments are listed. Please consult Oschkinat et al. (1994) for a complete listing. t These experimental approaches were not used in the structural calculation of Ets-1AN331 Chapter 4 — Tertiary Structure of the ETS Domain 102 tertiary structure. Theoretically, a complete set of torsion angles is sufficient to describe a structure. However, this approach is precluded by the limited precision that a torsion angle restraint can be described and the limited number of torsion angle restraints that can be obtained experimentally. A more detailed description of each of these restraint classes will be discussed as they are encountered in the upcoming experimental sections. An NMR "structure" is a stereochemically acceptable solution to the distance, torsion angle and chemical shift restraints. Since every possible restraint cannot be experimentally obtained, no single solution exactly defines the structure. An ensemble of twenty or more solutions is therefore calculated as a representative sampling of all possible solutions. The degree of similarity among the structures of the ensemble is expressed as the root-mean-square (RMS) deviation of the atomic coordinates. As a guide, a "fourth-generation" high-resolution NMR structure described by at least 15 restraints per residue will typically have respective backbone and side-chain RMS deviations in the order of 0.4 A and 1.0 A (Clore and Gronenborn, 1994). 4.2.2. Structure Generation An NMR structural determination consists of three general phases of restraint assignment, structure generation and structure assessment (Figure 4.1). This procedure is iterative due to the presence of chemical shift degeneracy in all NMR spectra. Chemical shift degeneracy translates directly into ambiguity in restraint assignment (Figure 4.2). Therefore, at early stages of a structural analysis, only the most trustworthy, or unambiguous data is included in calculations. Unambiguous data generally includes a-helix, P-strand and selected long-range NOE restraints. Ideally, sufficient unambiguous data will be available to identify a tertiary structure that can reduce a set of related ambiguous restraints to one obvious choice. This technique, termed "bootstrapping", requires considerable manual intervention (Meadows et al., 1992). Bootstrapping can be made more efficient through the use of short scripts and programs 103 NMRdraw PIPP/STAPP X-PLOR lnsight-ll QUANTA PROCHECK Process ing • data conversion • automated scripting xprocpar xprocpar <-pipe> NOE Ass ignment • data preparation NOE assignment • data sorting data statistics make_siad siftPCK count_noe Structure Generation - importing restraint data - making a starting template - simulated annealing refinement pck2xplor, notes2xplor strip2backbone Assessment • choosing an ensemble - visualization - E v d w , solvent accessibility planarity, chirality topEnergy, noViolations superimpose Figure 4.1 : A flowchart outlining the steps taken to generate the Ets-1AN331 structure. Commercial software used is indicated on the left of the chart. Software that I wrote is indicated on the right of the chart. 104 #1 R 0 #2 #3 R O R 0 #4 R O In this hypothetical example, the Ha chemical shifts of residues #2 and #3 are degenerate and similarly, the Ha chemical shifts of residues #53 and #54 are degenerate. An NOE is detected between one of #2 and #3 and one of #53 and #54. The true NOE is between #2 and #54 in this example. It is assumed that the Ca shifts of the four residues in this example are not degenerate. - N - C - C - N - C - C - N - C - C - N - C - C i i ' A IA I i H H H ( H ) H ( H ) H H H H H - C - C - N - C - C - N - C - C - N 1 I I  I I  I O R O R O R #54 #53 #52 or #54 In this hypothetical 2D NOESY experiment, the low dimensionality combined with the degeneracy gives four ambiguous NOE assignments for this crosspeak. In the absence of 3D or 4D NOESY spectra, the experimentalist would discard this crosspeak or choose one of the four possible assignments based upon clues from the structure or corroborating assignments from nearby residues. #2 or #3 In this hypothetical 4D-[i3C, 1 3C, 1H, 1H]-NOESY experiment, the intersecting plane of the Ca resonances of #2 and #54 are depicted. At the Ha shifts of #2 and #54, a crosspeak is evident. At four dimensions, the NOE assignment is now unambiguous. Though not shown in this figure, the crosspeak symmetry of a 3D-[ 1 3 C, 1 H, 1 H]-NOESY experiment can also lead the experimentalist to the same unambiguous conclusion. — •#54 Figure 4.2 : Chemical shift degeneracy is a source of ambiguous distance restraint information. The use of 1 3 C or 1 5 N to increase the dimensionality of the NOESY experiment can resolve degeneracy in many cases. However, 3D and 4D NMR of Ets-1AN331 was limited in signal-to-noise due to the low concentration of the sample. Therefore, many ambiguous NOE's were not assigned. Chapter 4 — Tertiary Structure of the ETS Domain 105 such as those listed in Figure 4.1 to eliminate repetitive tasks. At each point requiring user intervention, one must be aware of subjectivity. Subjectivity will naturally take a less prominent role if the basis dataset is large enough to lead the process to only one general solution. Recently, a method has been described by Nilges (1995) that allows ambiguous data to be included in structural calculations. Thus, many potentially incorrect decisions are prevented. Many approaches to structure generation and assessment are available and reflect the ongoing evolution of the NMR technique towards one unified approach. In the following sections, I will restrict my discussion to the software packages that were used to generate and assess the structure of Ets-1AN331. Distance geometry (DG) is a common method for generating a starting structure. The DG algorithm chooses exact distance restraints at random from the experimental distance bounds and successively fits them to a mathematical solution while striving to maintain normal bond lengths throughout the molecule. Following distance geometry, the result resembles more of a mathematical solution to the restraint data rather than a structure as many torsion angles and bond lengths are sub-optimal and therefore, must be corrected. To accomplish this, a simulated annealing protcol (Nilges et al, 1988) is followed. Simulated annealing (SA) is a term to describe a minimization / molecular dynamics schedule where van der Waals interactions, bond length, dihedrals and planarity contributions are factored in at different rates while maintaining as many user defined NOE and dihedral restraints as possible. Each factor contributes an energy term to a target function that the system is constantly trying to minimize. In combination, the hybrid DG/SA approach allows a considerable amount of conformational space to be sampled, thus avoiding problems of local minima. Alternatively, an ab initio approach may be taken. This method begins by randomizing all atoms in a 20x20x20 A box and then uses a modified SA scheme to assemble the structure. The computational requirements of full DG/SA and ab initio SA are roughly Chapter 4 — Tertiary Structure of the ETS Domain 106 equivalent. The exclusion of all atoms except C a in calculation of the target function substantially decreases the time to generate DG and ab initio SA solutions (Nilges, 1995). 4.3. Experimental Methods 4.3.1. Protein Samples An E. coli expression system described in Chapter 2 was used to produce uniformly and selectively 1 3 C , 1 5 N and 2 H labelled Ets-1 AN331. Selectively ring deuterated proteins were prepared from synthetic media containing 100 mg/L of two of L-8i, £2, £,1,2, ^2-[2H.5]tryptophan, L-8i,e2,^-2H.5]phenylalanine, or L-5i Ei^-pH^tyrosine (Cambridge Isotope Laboratories and Isotec). Fractionally 13C-labelled Ets-1AN331 was prepared using minimal media containing 10% (0.3 g/L) 13C6-glucose and 90% (2.7 g/L) unenriched glucose (Neri et al, 1989). All samples were 0.6 - 0.8 mM protein in 20 mM sodium phosphate (pH 6.5), 500 mM sodium chloride, 0.01% sodium azide, 10 mM DTT, in 10% or 99% D 2 0 . 4.3.2. NMR Experiments The resonances from side chain aliphatic 1 H and 1 3 C nuclei were identified using HCCH-TOCSY experiments with 8 msec and 16 msec DIPSI-3 mixing periods on a sample of uniformly 1 3 C / 1 5 N labelled protein in H 2 O (Kay et al, 1993). Mainchain assignments were previously described in Chapter 3. These assignments were verified with sensitivity-enhanced versions of the H(CCO)NH and C(CO)NH experiments using 8 msec DIPSI-3 mixing periods (Grzesiek et al, 1993; Muhandiram and Kay, 1994). The correlations involved in these experiments are shown in Figure 3.4. The methyl groups of leucine and valine were assigned stereospecifically from gradient ! H - 1 3 C HSQC spectra of 10% fractionally 1 3 C labelled protein (Neri et al, 1989; Wider and Wuthrich, 1993). Aromatic ring proton resonances were assigned using 2D DQF-COSY, TOCSY and NOESY experiments recorded on Ets-1AN331 in 99% D 2 0 in which two of tryptophan, tyrosine and phenylalanine were selectively deuterated (Mcintosh et al.,1990). Aromatic ring carbon resonances were identified from constant time 1 H - 1 3 C Chapter 4 — Tertiary Structure of the ETS Domain 107 HSQC, ( H P J C P C C Y C ^ H 5 and (HP)CP(CYC 5CE)He experiments (Santoro and King, 1992; Vuister and Bax, 1992; Yamazaki et al., 1993). NOE distance restraints were measured from sensitivity-enhanced 3D 1 5 N-NOESY-HSQC (125 msec mixing period) and simultaneous 1 5 N , 1 3C-NOESY-HSQC (50 and 150 msec) experiments (Zhang et ai, 1994; Pascal et al, 1994). In addition, many critical long distance restraints involving aromatic protons were obtained using 2D 1H- 1H-NOESY spectra (125 msec) recorded with samples of Ets-1 AN331 in 99% D2O with selectively deuterated aromatic rings. The high-resolution and high signal-to-noise of these spectra also provided an avenue for aromatic spin-system identification. 3 J H N H O C couplings were measured with the HMQC-T experiment (Kay and Bax, 1990) using software provided by Dr. L.E. Kay (Univ. Toronto). 4.3.3. Structural Restraints With a large body of evidence suggesting that the ETS domain was a member of the wHTH family, the iterative bootstrapping approach was deemed to be a reasonable protocol to follow providing that many consistency checks were made throughout the course of the structure determination. From sequence alignments and inspection of the available wHTH family structures, a group of residues were identified as being invariant contributors to the hydrophobic core. Throughout the assigment process, these residues were used as guides for assignments but not to the extent that NOEs were intentionally sought out that were consisent with the wHTH motif. As it would be ultimately shown in the final structure, the anti-parallel (3-sheet serves as an excellent platform to build the hydrophobic core. Consistent with its role in maintaining the integrity of the protein, many unambigous NOEs were obtained for residues which contributed to the B-sheet. NOE distance restraints were assigned and integrated in the PIPP software package (Dr. D. Garrett, NTH) and categorized into weak (1.8-5.0 A), medium (1.8-3.5 A) and strong (1.8-2.9 A) classes according to an r 1 / 6 relationship. A strong crosspeak intensity (<2.9 A) was calibrated by measuring several H a-HP NOE's, H N[ - H N ; + i NOE's in a -helices and Chapter 4 — Tertiary Structure of the ETS Domain 108 H 5 - H e NOE's in aromatic rings. A correction of 0.5 A was added to the upper bounds of restraints involving methyl groups. Prochiral HP groups, as well as the aromatic ring protons of phenylalanine, were treated as pseudoatoms. The programs pcklxplor and noteslxplor, written by this author, facilitated the conversion of restraint datafiles between commercial software packages and are available via the internet at <http://otter.biochem.ubc.ca/www/nmrtools.html>. Based on 3 JHNHC< couplings, § torsion angles were restrained to three regimes following a quadratic potential energy function: <7 Hz, -60 ± 40°, (8 Hz < J <9 Hz), -120 ± 60°, >9 Hz, -140 ± 40°. Main chain hydrogen bonding restraints were included for residues showing protection from amide hydrogen exchange as well as dihedral angles and NOE patterns diagnostic of a-helical or B-sheet secondary structures. A loose restraint of <J) = -94° ± 40° was assigned to the N residue position of helices whose C a and CP chemical shift profiles matched the prototype for an N-terminal capping boxes (Figure 3.9). No medium or long-range restraints were observed for the four N-terminal residues 331-334, suggesting a disordered conformation in solution. Therefore, these residues were omitted from the structure calculations. Long distance NOE distance restraints that characterized the tertiary structure of Ets-1AN331 were comprised of aromatic/aromatic, aromatic/methyl and methyl/methyl NOE's. A summary of the numbers of distance and angle restraints used in the calculation of the structure of Ets-1AN331 are listed in Table 4.2. 4.3.4. Structure Calculations Although both DG/SA and the ab initio approaches produced similar results, an ensemble of Ets-1AN331 structures was generated most directly by the following method: (i) The Ets-1 sequence was aligned to the K. lactis HSF sequence using regions of amino acid identity and secondary structures as cues (Harrison et al, 1994). (ii) For each residue in Ets-1 with a counterpart in HSF, N , C a , CP and C coordinates were extracted from the HSF structure. (Hi) Al l other side chain atoms and loop regions were given default coordinates of 9999.0 A. (iv) This file, which now resembled output from a sub-embedded distance geometry protocol, was submitted to standard restrained SA and Table 4.2 : Restraint Summary for Ets-1 AN331 Distance Restraints intra-residue 229 inter-residue short-range (1 < I i -j I < 4) 383 inter-residue long-range ( I i -j I > 4) 123 hydrogen-bond 29 764 Dihedral Angle Restraints backbone, § angle 77 sidechain, % angle 0* 77 * No experiments were performed to obtain side chain x angle information. Chapter 4 — Tertiary Structure of the ETS Domain 110 refinement protocols (Nilges et al, 1988) written for the X-PLOR (Brunger, 1988) software package. A total of fifty simulated annealing structures were calculated using a list of restraints listed in Table 4.2. Specifics of the SA and refinement protocols may be found in Table 4.3. The top twenty structures that had no NOE violations > 0.5 A, no dihedral angle violations > 5°, and a low target energy were selected for refinement. An average structure was calculated and subjected to restrained regularization according to a low temperature refinement protocol also outlined in Table 4.3. 4.3.5. Structural Assessment All of the twenty structures in the final ensemble had an acceptably low target energy values with good stereochemistry (Table 4.4). The average structure was further assessed with QUANTA (Molecular Simulations; Burlington, MA) and PROCHECK (Laskowski et al., 1993). The average structure and the ensemble of twenty calculated structures are designated respectively as 2 ETC and 1ETD in the Brookhaven Protein Data Bank. The secondary structure of Ets-1AN331 did not change from the earlier prediction using information from chemical shifts, 3 J H N H C C coupling constants, and patterns of interresidue short-range NOEs. The angular order parameters, S§ and Sy, were calculated from the ensemble of the twenty refined structures according to Hyberts et al. (1992) using software generously provided by Dr. G. Wagner (Harvard Medical School). If a given torsion angle in each of the structures of the ensemble is invariant, that angle is described by an order parameter of 1.0. In contrast, if the angle is completely unrestricted, it will be described by an order parameter of 0.0. Interhelical angles were measured using the interhx module of the ribbons suite of programs (Carson, 1991). Interhelical angles are defined such that 0° is parallel and 180° is antiparallel. A comprehensive comparison of interhelical angles and RMS differences between HTH proteins may be found in Vuister et al. (1994b). Ml Table 4.3 : Ets-1AN331 structure generarion.and refinement protocol* Search Cool Refinel Refine2 Regularize General: Temperature (K) 2000 2000 1000 500 300 Steps 3000 3000 2000 1000 1200 Masses (a.m.u.) 100 100 100 100 — Energy Constants: Kbonds (kcal /(mol A 2 )) 100 -> 1000 1000 1000 1000 1000 Kans,es (kcal /(mol rad2)) 0.05 r> 5 200 200 200 200 Kplamr (kcal/(mol rad2)) 50 -> 1000 1000 1000 1000 1000 Krepe, (kcal /(mol A 2 )) 0.50 -> 0.90 0.90 -> 0.75 0.90 -> 0.75 0.90 -> 0.75 0.75 (kcal/(mol A 2 )) 20 -> 0.003 0.003 -> 4 0.003 -> 4 0.003 -> 4 1 K N O £ (kcal /(mol A 2 ) ) 50 50 50 50 50 tAll protocols are adapted from the simulated annealing and refinement protocols described in (Nilges, 1988). Chapter 4 — Tertiary Structure of the ETS Domain 112 4.4. Structural Features The Ets-1AN331 is an N-terminal deletion mutant of Ets-1 which contains the 85 residue ETS domain and a 25 residue extension that ends at the native C-terminus of Ets-1. The mean minimized structure of Ets-1AN331 was calculated by averaging the coordinates of the individual structures, aligned to each ther using the backbone atoms of residues 336-377, 386-396, 401-405 and 411-435, followed by restrained regularization. The best fit superimposition of the ensemble of twenty SA structures determined for Ets-1AN331 is presented in Figure 4.3. Excluding disordered sequences, the ensemble of Ets-1AN331 structures has an average RMS deviation of 1.65 A versus the mean minimized structure for all backbone (N, C a , C and O) atom and has an average RMS deviation of 1.34 A for backbone atoms in helices and strands. This degree of precision is directly attributable to the limited solubility of the Ets-1 fragment (< 0.8 mM), the necessity to use high ionic strength conditions (0.5 M NaCI) and the effects of internal motions on a millisecond timescale. Specifics of the structure, including the number of *H assignments, the number NOE restraints per residue, the RMS deviations of the C a atom positions about the mean minimized structure and the angular order parameters are shown in Figure 4.4. With respect to the average structure, the majority of backbone residues lie within the favored regions of the Ramachandaran plot shown in Figure 4.5. Ets-1AN331 is composed of four a-helices (HI 337-345, H2 369-377, H3 386-396, and H4 428-434) packed against a four-stranded anti-parallel B-sheet (SI 354-357, S2 362-365, S3 402-405, S4 411-414). This is in agreement with the secondary structure with a minor difference in the definition of the C-terminal boundary of helix H2. (consult Chapter 3). The three helices (H1-H3) and the four P-strands in the ETS domain are well defined, with low RMS deviations and angular order parameters approaching unity. The C-terminal boundary of helix H3 is uncertain because the signal-to-noise from residues 396-399 is poor due to conformational averaging. Residues 428-434 in the C-terminal extension form the fourth helix, H4. The RMS deviations of the residues in this helix versus the average structure are larger than observed with the first three helices (Figure 4.4C), yet independently, backbone atoms of residues 428-434 superimpose with a RMS 113 Figure 4.3 : Ets-1AN331 contains a winged helix-turn-helix motif and a helical C-terminal extension. The stereoview shows a superimposition of 20 simulated annealing structures of Ets-1AN331 aligned against the average, rrunimized structure (thick bar) on the backbone atoms of residues 335-377, 386-396, 401-405, and 411-415. For clarity, only the average structure is shown for residues 416-436. The N-terminal 4 residues (331-334) are disordered and were not included in the structure calculations. 111-Table 4.4 : Structural Statistics of the 20 Ets-1AN331 Structures X-PLOR Energies (kcallmol) Etotal ^•bond ^angle ^•improper ^•vdw ^•dihedrals* <SA>" 167.9 ± 6.5 9.94 ± 0.28 64.5 ± 0.2 11.9 + 0.3 56.8 ± 2.9 23.5 ± 2.8 1.36 ± 0.02 <SA>° 168.8 9.18 60.7 12.0 65.3 21.2 0.40 RMS deviations from ideal geometry: bonds(A) angles (*) impropersC) 0.0024 ± 0.0000 0.3606 ± 0.0004 0.2825 ± 0.0041 0.0023 0.3501 0.2805 RMS deviations from experimental restraints:' NOE's (A) 0.0242 ± 0.0016 dihedrals (°) 0.5268 ± 0.0015 0.0231 0.2930 Atomic RMS differences: <SA> versus <SA> <SA> versus <SA> with no loops <SA> versus <SA> only a-helices or fi-strands backbone atoms 2.37 A 1.65 A 1.34 A all heavy atoms 3.97 A 3.54 A 3.48 A a<SA> is the average of the ensemble of the 20 final structures. b<5A> is the set of coordinates obtained by averaging <SA> following a least squares superimposition of the backbone atoms for residues in a-helices or P-sheets. C A force constant of 50 kcal/mol/A2 was used for NOE restraints. rfA force constant of 200 kcal/(mol/rad2) was used for dihedral angle restraints. eBackbone atoms include N, C°, O, C . Figure 4.4 : Summary of NMR restraints and structural parameters for Ets-1 AN331. (A) The number of restraints depends on the extent of spectral assignments for a given residue. Solid bars indicate the maximum number of proton assignments expected per residue, while grey bars indicate the actual number of assignments made. Methylene and symmetrically related aromatic protons are counted only once. (B) Total number of intraresidue (open), short-range (grey), and long-range (solid) NOE restraints per residue used to determine the final set of Ets-1AN331 structures. All restraints are counted once for each hydrogen involved. (C) The RMS deviation of the backbone (grey) and all heavy atoms (solid) for the ensemble of 20 structures of Ets-1AN331 aligned against the average, minimized structure. The locations of the four a-helices and four B-strands are indicated by the thick and thin horizontal bars, respectively. (D, E) The angular order parameters S<J) and Svj/ for the (j) and \\r main chain dihedral angles, respectively, observed in the ensemble of protein structures. The N-terminal residues 331-334 were omitted from the structure calculations. 227 Phi (degrees) Figure 4.5 : Ramachandaran plot of the Ets-1AN331 mean minimized structure generated by PROCHECK (Laskowski et ah, 1988). The degrees of shading indicate the most favored, allowed and generously allowed regions of phi/psi space from an analysis of 118 high-resolution crystal structures. Glycine residues are represented as triangles, all other non-proline residues are represented as squares. The loop between helix HI and strand SI (D347-S352) is poorly defined in the structure. Chapter 4 — Tertiary Structure of the ETS Domain 118 deviation of 0.66 A. The angular order parameters of these residues also approach unity. Therefore, the conformation of helix H4 is defined locally, while the position of this helix relative to the remainder of Ets-1AN331 is not precisely established. This is in part due to extensive chemical shift degeneracy in the region of T425-V435. In contrast, low angular order parameters and high RMS deviations for residues in the turns or loops linking HI/SI, H2/S3, H3/S3 and S3/S4, as well as the termini of the protein, show that these regions are disordered in the ensemble of calculated structures due to a limited number of restraints or conformational flexibility. 4.5. The Winged Helix-Turn-Helix Motif The global fold of Ets-1AN331, with three a-helices packed on a four-stranded anti-parallel B-sheet, confirms the placement of Ets-1 in the wHTH superfamily of DNA binding proteins. The designation, wHTH, highlights the role of the B-sheet and wing-like loops in the structure of the binding domain as well as in possible protein-DNA or protein-protein interactions. At the same time, it distinguishes this superfamily from others such as homeodomains that only use a helix-turn-helix DNA binding motif (Clark et al, 1993; Brennan et al, 1993; Clubb et al, 1994). Figure 4.6 presents ribbon diagrams of the DNA binding domains of CAP, HSF and HNF-3y along with that of Ets-1. The common fold of three a-helices on an anti-parallel B-sheet is evident. This structural similarity was not readily recognized until recently due the absence of any significant sequence similarity among the various wHTH proteins. However, in light of the tertiary structures of these proteins, a pronounced pattern of aliphatic and aromatic residues contributing to their respective hydrophobic cores can be detected. For example, L337, W338 and L341 (HI), W375 (H2), and L389 and L393 (H3) greatly influence the relative placement of the three a-helices in the Ets-1 ETS domain. K. lactis HSF contains the analogous residues L202, W203 and V206 (HI), F232 (H2), and F249 and L253 (H3) (Harrison et al., 1994). Important residues that establish the interior hydrophobic face of the amphipathic B-sheet in Ets-1 include 1354 and W356 Figure 4.6 : Comparison of the wHTH DNA binding domains and C-terminal extensions of (a) murine Ets-1, (b) CAP (Schultz et al, 1991), (c) Drosophila HSF (Vuister et al, 1994b), and (d) HNF-3y (Clark et al, 1994). The a-helices are red and B-strands are blue. Each protein is oriented with helices HI and H3 in the plane of the page. The helix-turn-helix motif is formed by helices H2 and H3, and the recognition helix (H3; vertical) binds or is proposed to bind in the major groove of duplex DNA. A distinctive B-hairpin loop or wing (green) makes additional flanking sugar or phosphate contacts in the crystal or NMR structures of the DNA complexes of CAP, HNF-3y and the Fli-1 ETS domain. The analogous extended loop in HSF does not appear to be involved directly in DNA binding to the heat shock response elements (Vuister et al, 1994b; Hubl et al, 1994). Sequences C-terminal to the DNA binding domains of Drosophila HSF and HNF-3y adopt a position similar to that of helix H4 in Ets-1AN331 (yellow). In HNF-3y, this region forms a pronounced second wing (green) and provides additional phosphate and specific minor-groove base contacts. A function has not been described for the analogous sequence in HSF. In Ets-1, evidence exists for intramolecular association of helix H4 with sequences N-terminal to the ETS domain, resulting in inhibition of DNA binding (Petersen et al, 1995; Jonsen et al, 1996). The figure was drawn using Molscript (Kraulis, 1991). 120 Chapter 4 — Tertiary Structure of the ETS Domain 121 (SI), F363 and L365 (S2), 1402 (S3), and Y412 and F414 (S4) with counterparts in HSF being 1215 and W217 (SI), 1224 and V226 (S2), W258 (S3) and W277 and F279 (S4). Although the wHTH proteins share a common fold, Figure 4.6 also illustrates the considerable variation that exists in terms of interhelical angles, helix, strand and turn or loop lengths, and the numbers of B-strands. In comparison to interhelical angles summarized previously for several wHTH proteins (Vuister et al., 1994b; Harrison et al, 1994), the Ets-1 ETS domain shows helical crossing angles of 141° for H1/H2, 108° for H1/H3, and 109° for H2/H3. The four residue turn found in the prokaryotic HTH motif, such as CAP, is replaced with eight residues in Ets-1 and HNF-3y. The wHTH proteins also differ in the number of strands forming the P-sheet (Table 4.4). In all cases, at least two p-strands follow the HTH motif, providing a loop or wing for possible DNA contacts. A third helix precedes the HTH motif, except in the case |j.-transposase, which forms a distinct class of wHTH proteins by having this helix located after the P-sheet (Clubb et al, 1994). These structural variations undoubtedly reflect the biological diversities exhibited by member of the wHTH family. For example, wHTH proteins function as monomers, dimers or even trimers to specifically, or in the case of histone H5, to nonspecifically bind DNA (Table 4.5). 4.6. Homology Among ETS Domain Sequences The ets gene family has over twenty members from a variety of metazoan species (Lautenberger et al, 1992; Degnan et al, 1993). Relative to Ets-1, these family members show sequence identity within their ETS domains ranging from 96% in the case of Ets-2 to 39% in PU.l . The diagnostic features of the ETS domain include a pattern of conserved hydrophobic residues, highlighted by three essentially invariant tryptophans, and a C-terminal region rich in basic amino acids (Wasylyk et al, 1993; Janknecht and Nordheim, 1993). To investigate the structural basis of this sequence conservation, fifteen ETS domains were aligned (Figure 4.7A) and scored according to an evolutionary mutation matrix at each residue position (Henikoff and Henikoff, 1992). When color IZ2_ T3 0 Xj E u c 01 01 E ra Z S 3 u 2 5 ^ ro ro ro u, u. Ui u«X X X X I—I CO 1% ro Pi ^ & & & „ x 2 2 2 i—i o" " c-4 co^" *—I i—I i—« T—t t—I 01 %C 2 <u 60 <" » 1 c cn 3 1 cn 01 01 6 o t-l a. o o X a. cu < Ui o c o c o S CO CO S o 3 X 0 "3b in 1 at c o cn cn Oi Oi 01 s o c o 6 'CO cn u O CO I D 2 ai 6 o> T3 01 0) e T3 o u • a. u g o x — *. cn — i i io — o cn CX a . 3 S CU 0) X 3. X X c o a. •»3 £ 01 6 o c o s cn cn cn cn ai ai OJ ai 01 01 01 see o o o c c c o o o S £ 6 3 2 X a. < =5 r x X _>, e s s . CC (3 jn cn cn W W W J? c" c £ co ffl ' E S c 3 3 £ x x 2 Q w w EC in in ~ 1—• a -» Ox 1 . ^ CO O ON <3 "S <-< C - ro ~ «i j = 1 1 ^ <u S '5> " JS 3 in (J « 2> | , ON ON 1 - 4 CO "3 2? t* ON ON <U 1-1 1 - 1 Ui « ^ o • « 6 •a | i o « "C 2°.!£ r- ON ti ^- 2 ro ON ^ C ON :• ro w a C< - T — co ->-> c " £ o sf «-3 ON . « ro '-' 3 C > > o "a „ Q *a r^ s , , cu ' — ' c J « ON L O j-< i-l ON 1 ' "U Q ON ON oi "S Ul 3 N , — , X j NO . v 3 i—• Tt< x ; ^ CD ON m oi u C 01 Ul 01 MM 01 cu SI Chapter 4 — Tertiary Structure of the ETS Domain 123 coded on the structure of the Ets-1 ETS domain (Figure 4.7B), a hydrophobic core invoving highly conserved residues from all three helices and four (B-strands is readily apparent. Notably, extensive packing contacts are made by the theee tryptophans (W338, W356, W375), as well as F340, L341, and L345 in helix HI, 1354 in strand SI, F363 in strand S2, Y386 and L389 in helix H3, and Y412 and F414 in sheet S4. These residues are essential to maintain the structural integrity of the ETS domain. In contrast, the regions of the highest sequence variability, including amino acid insertions or deletions, map to the loops or turns connecting the a-helices and B-sheets. The conserved basic amino acids of the ETS domain are found within the helix-turn-helix motif (helices H2 and H3), the final two strands of the B-sheet, and the loop or wing connecting these strands (Figure 4.7A,B). As discussed below, these postively charged side chains contribute to the DNA binding interface of the ETS domain. The pronounced conservation of residues in the recognition helix H3 correlates well with the conserved 5'-GGA(A/T)-3' core sequence found in all regulatory elements bound by Ets proteins. -To maintain solublilty at NMR concentrations, a moderately high (0.5 M NaCI) salt concentration was present in all Ets-1AN331 samples. This high salt requirement is relieved upon complexation with a high-affinity 10 bp oligonucleotide suggesting that unfavorable electrostatic interactions may be an important factor affecting solubility. Indeed, when an electrostatic profile is calculated and overlaid upon the accessible molecular surface of Ets-1AN331, a dipolar character is readily apparent (Figure 4.8). A patch of basic residues is found at the helix H2/H3 turn and throughout H3 itself. An patch of acidic residues is found on the opposite end of the molecule near helix HI and the C-terminal extension. Such an arrangement may be important in assisting the ETS domain to orient itself with the nucleic acid. Upon complexation, the basic patch is essentially neutralized by the nucleic acid creating a complex with an overall acidic profile. An expressed minimal DNA binding fragment of the wHTH family member, LexA, also requires moderately high salt (0.3 M NaCI) to maintain solubility in the I2T H1 S1 S2 H2 H3 S3 S4 mu-Etsl av-Ets2 v-Ets hu-GABP hu-Erg2 h u - F l i l dr-Ets6 mu-PEA3 hu-Elkl hu-SAPl hu-Elf1 dr-E7 4A dr-Ets4 dr-Yan hu-PU.1 IQ|WQ|^ LHXBTDKS rQtWQniLiilTJKS IQlWQPBLSUiTDKS I Q l w Q | IQ|WQ|| IWQI |TDKD fSDSS ISDSA DSS SDDPT §REQG IQKPQ IQDKA TCPKY; |LKffi|QDREY CPRF IKESASPQV NGTA Rli|vi)^KCiQfflNDRNQKYSDL IRBYQJHLDSIRSGDMK DSf CQSFISW7G DGWElKLS CQS.-ISW:-G DGWE|KLA CQSFf:SWTG DGWE|KLS ARDclsfvG DEGEfeKLN NSSckptfEG TNGEKKMT NAScfrfEG TNGE|KMT G QSG: RG; SRDGG: SND Gi •QREKGI: REKG IDRSKGI CRDTG' KDKG' 3PDSVARPMQXRK NKPXMNYEKLSRGLHYYYLIKXI IriKTA GKRYV^ RJfV DPDEVAR RVJGRRK MX?KMMXEX1SRGI,8'/TYDKNI IHKTS GKRYVfRfV DPDEVARR#tlKRK NK PKMDSHKLS RG LH YX Y DK.W : i 05TA GKRYV^ Ritv QPELVAQK$f< DPDEVAR: "' DPDEVAR: DPDEVAR: EPEEVAR: DAEEVAR: QAEEVAR: DSKAVSR: DSKAVSR. DSVRVAK: DPAGLAKUJI QFSSKHXEALAHRW3IQ8GNR fYfYDGDMICKVQ GKRFVffKfV " """"DKNIMlfVH GKRYAfk|D DKNIMTfVH GKRYA3§KgD DKNIMTKVK GKRYA1 EKGIMQ§VA GERY KNIIRlfvS GQKE IIKRVN GQKF QRGILAKVE GQRLV Y|YQRGILAKVD GQRLV 'DKLS§Sr|Q§YKKGIMKlTERSQRL' DKMS|ALlYSfYRVNILR|vQ GERHC QK i^Alj|NSGKTGEVT#VK KKL' Figure 4.7 : (This page) A representative population of fifteen ETS domains were aligned in order of decending sequence homology to murine Ets-1 (mu-Ets-1). Invariant side chains are shaded and the secondary structure of Ets-1AN331 indicated by boxes (a-helices) and arrow (P-strands). (Opposite page) The sequences were scored per residue relative to the murine Ets-1 sequence according to a BLOSUM62 evolutionary mutation matrix (Henikoff and Henikoff, 1992) The sequence conservation is presented as a white (most conserved) to blue (least conserved) colour gradient on residues 336-415 of the average, minimized structure of murine Ets-1. The figure was generated using the program, Insightll (Biosym/Molecular Simulations) Chapter 4 — Tertiary Structure of the ETS Domain 125 126 Figure 4.8 : An examination of the electrostatic profile of Ets-1AN331 reveals a highly dipolar character which may contribute to the poor solubility of the protein at NMR concentrations. Accordingly, 0.5 M NaCI was present in all samples to minimize electrostatic interactions. In the above figure, red denotes electronegative surfaces and blue denotes electropositive surfaces. In the overlayed backbone worm, residues which make contacts to DNA are coloured green. Much of the electropositive surface coincides with the DNA binding surface. This figure was generated by the program, GRASP (Nicholls et ai, 1993). Chapter 4 — Tertiary Structure of the ETS Domain 127 uncomplexed form (Fogh et al, 1994). An electrostatic profile of LexA repressor reveals a similar dipolar character with a basic patch near the recognition helix and an acidic patch on the opposite face of the molecule. It is interesting to note that the ETS domains of PU.l and Fli-1 do not suffer from poor solubility in the uncomplexed form (Pio et ai, 1995; Liang et al, 1994a,1994b; Dr. C. Arrowsmith, University of Toronto; Dr. T. Handel, University of California at Berkeley, personal communications). From an examination of the PU.l sequence, a similar acidic patch is predicted to be absent. This suggests that the dipolar character of the Ets-1 ETS domain contributes significantly to its poor solubility. Furthermore, in contrast to the ETS domain of Ets-1, the ETS domain of PU.l is situated in the N-terminal portion of the native protein. 4.7. D N A Binding by the ETS Domain The ability to bind DNA is prerequisite for the transcriptional activity of Ets-1. Thus, characterization of the minimal DNA binding domain presents a starting point to begin studying the biology of the Ets family. .. To probe for major conformational changes, a series of CD spectra of Ets-1AN331 were collected in the presence and absence of a 10 bp high affinity SCI oligonucleotide duplex (consult Chapter 2 for a description of uncomplexed Ets-1AN331). No major differences could be discerned between the uncomplexed and complexed forms suggesting that the secondary structure content of the protein did not change (refer also to Petersen et al, 1995). Biochemical data, on its own, can provide a significant level of understanding towards the DNA binding characteristics of the wHTH family. One particularly important issue that biochemical data can address is the question of whether or not a common fold also specifies a common mode of DNA binding. Figure 4.9 summarizes the published DNA Chapter 4 — Tertiary Structure of the ETS Domain 128 footprinting information gathered from methylation/ethylation interference, hydroxyl radical protection and DNase I protection experiments for CAP, HNF-3y and Ets-1. Clearly, the synthesis of data points demonstrates a preference for binding one face of the DNA duplex and the dominance of the major groove as the source of specific contacts. A DNase I hypersensitive site in the minor groove opposite to the core recognition site is a common characteristic found between HNF-3y and Ets-1 and hints that the two monomeric proteins may bind DNA in a similar manner. From the HNF-3y crystal structure, the DNA in the vicinity of the recognition site is twisted open slightly resulting a wider major groove and a very shallow kink (Clark et al., 1994). A short oligonucleotide was used to crystallize HNF-3y; perhaps, with a larger duplex the kink may have been more pronounced. Such perturbations in the duplex could be sufficient to provide a DNase I hypersensitive site. By invoking a similar mode of binding for the ETS domain and combining the chemical shift perturbation data of the Fli-1 structural study, potential contacts would be made from Helix HI, Helix H3 and the wing following a 5'-GGA-3' direction. Though not presented in Figure 4.9, dimeric RTP (Bussiere et al., 1995) follows this model with contacts to DNA made helix 1, helix 3 and a large DNA binding wing. Fine structure changes are expected in the ETS domain upon complexation, particularly in the vicinity of the recognition helix, H3. To identify the protein-DNA interface, a strategy involving titration of selectively labelled proteins was undertaken. A 1 5 N -tyrosine label was chosen since there are four tyrosines in helix H3 which are expected to shift upon complexation. 1 H, 1 5 N-HSQC spectra were acquired of Ets-1AN331 and 1:1 [15N-Tyr]Ets-lAN331 / SCI duplex at a concentration of 0.2 mM. At this concentration, the spectra were noisy but were still quite interpretable. The results are presented in Figure 4.10. The most perturbed residues were Y386 and Y397. These two residues occupy the N - and C-terminal positions of the recognition helix and it is not surprising that their resonances are perturbed upon complexation. Residue Y410 occupies a position in the wing of the ETS domain. In the complexed form, one amide resonance from Y410 disappeared due to a combination of the increased molecular weight of Ets-129 cu > < o o •5* c "5 E o •a to UJ VJ CO I u_ Z X 'A c/1 u> a. < o z a X 3J cn < C 'co 6 0 T J cn E—1 Li-eu X T J c CO c?T 1 P H 2 X < cj cn C O o cj CO C O cj < z Q ON cu >. 3 C cn CO S T J £ cn cu t—i OH W CU o C o c cu T J CU a, CU T J CU X cn 3 O bO J D "<3 c co cu I-I CO C 'co o T J cn w CU T J c CO OL, C O 0 3 rr S 2 c cu cu cu X ) T J CU & CU cn X O .2P & O (0 3 X> c o cj cu cn CO C M cn cn cn T J 3 CO C cu cu X bO 3 cu T J c CO t—I X X "CU cn cu 3 T J CO C cu cu X CO T J >-CO I [-L, 2 X LcT as O N T J CU C X o cn CO 5 CO T J —: co « C J cu c C CO B •-3 CO T J c CO cu CO X> O cn to cO T J P- , < u cj CO •3 3 ^ 'S OH S c .2 c co cu X ! cj O CN . . O N C O O N O N T - H O N > - ' cu 3 cj cj cn s "S O « M - t — v <V O N 3 7 - 1 co •t; c N > cu CSH T J c co X c co cn 3 O CJ M _ L bo -g X O CN O N O N CU 2 B J30 X < St S2 S3 S4 116.0 120.0 124.0 £ Q. Q. M28.0 H3 39^7 Figure 4.10 : Pane/ A, 1 5N-HSQC spectra were acquired on a 15N-tyrosine labelled Ets-1AN331 sample at 0.2 mM in the absence and presence of the 10 bp SCI duplex. For clarity, the positions of the amide crosspeaks are denoted by circles. After complexation at a 1:1 ratio, the chemical shifts of several tyrosyl amides were perturbed and are denoted as triangles. An arrow denotes the direction and magnitude of the amide chemical shift perturbation upon DNA binding. Assignments of the bound state were assisted by the published amide chemical shift data of Werner et al. (1995b). The absolute value of the amide HN chemical shift differences is plotted against residue number in Panel B. Y386, situated at the N-terminus of helix H3. is the most perturbed upon complexation with DNA. The amide resonance corresponding Y410 in the complex is believed to be extremely broadened and therefore not visible. By repeating this experiment with other selective labels, the specific regions of the ETS domain that interact with the DNA can be determined. Chapter 4 — Tertiary Structure of the ETS Domain 131 1AN331 in complex and conformational variation on a micro- to millisecond time scale. Y412 and Y414 in strand S4 participate in the hydrophobic core of the ETS domain and thus make good probes for any major conformational changes. Negligible chemical shift differences were observed for these residues suggesting that a major conformational reorganization did not occur upon DNA binding. Finally, Y424, in the C-terminal extension region outside the ETS domain, was also insensitive to the effects of the DNA binding. After submission of a manuscript describing this work (Donaldson et al, 1996), a solution structure of a complex of human Ets-1 ETS domain with a 17 bp high-affinity binding site was reported (Werner et al, 1995b). A significant amount of intramolecular and intermolecular (protein-DNA) data allowed a high resolution structure to be calculated. A second ETS domain structure from Fli-1 in complex with DNA was published in the same year that this study was completed (Liang et al, 1994b). However, the Fli-1 ETS domain is not described at a similar level of detail as the human Ets-1 ETS domain. No intramolecular distance restraints are included in the Fli-1 ETS domain structure calculations. Therefore, the DNA duplex is not present in the structure. The conformation of the Fli-1 ETS domain, however, does reflect the complexed state. In the following discussion of DNA binding, the human Ets-1 ETS domain, for now, will be exclusively discussed. Afterwards, features of the Fli-1 ETS domain will be compared and contrasted to the murine and human Ets-1 ETS domains. The sequences of human Ets-1 and murine Ets-1AN331 are 100 % homologous in the ETS domain. Unike Ets-1AN331, the sequence boundaries chosen for the expressed polypeptide were not determined by homology; rather, a Xgtll expression library of human Ets-1 gene fragments was screened for activity (Jorcyk et al, 1993). The specific protein fragment chosen for optimal binding affinity and solubility corresponds to the equivalent murine deletion mutant, Ets-1AN324AC416. The protein fragments are Chapter 4 — Tertiary Structure of the ETS Domain 132 schematically represented in Figure 4.11 along with the Fli-1 ETS domain, for comparison. The DNA duplex chosen was longer yet similar in sequence; (Werner et al.: 5'-TCGAGCCGGAAGTTCGA; SCI: 5'-GCCGGAAGTG-3'). 324 440 ] hu-Ets-1 mu-Ets-1 Fli-1 Figure 4.11 : The solubility of human Ets-1 protein fragment was improved by including residues N-terminal, rather than C-terminal to the ETS domain (solid). Solubility and spectral quality of the human Ets-1 protein fragment was further improved upon formation of a complex with a 17 bp oligonucleotide duplex. The NMR structure of the human Ets-1 ETS domain in complex with DNA is described by Werner et al. (1995b). A second NMR solution structure of the Fli-1 ETS domain in complex with DNA has been reported (Liang et al., 1994b). Solubility of the Fli-1 ETS domain is also significantly higher than the murine Ets-1 deletion mutant, Ets-1AN331, described in this study (as high as 3.8 mM; Liang et ai, 1994a,b) The secondary and tertiary structures of human Ets-1 fragment are similar to Ets-1AN331 implying that a major, global conformational change in the ETS domain does not result from binding DNA. As expected, the recognition helix H3 confers much of binding through contacts to the GGAA core sequence. The helix is situated in the major groove in an orientation similar to HNF-3y (Figure 4.9). For many bZip, homeodomain and other DNA binding proteins that utilize a helix to bind the major groove, the side chains involved in DNA contacts may be considered in terms of the probe helix model (Suzuki, 1994). From a probe helix, the majority of DNA Chapter 4 — Tertiary Structure of the ETS Domain 133 contacts arise from residues at positions 1, 4, 5 and 8, relative to the N-terminal residue of the helix. In the absence of structural, mutational and biochemical data, a set of empirical rules may be consulted. These rules specify a scheme where a window of four residues is scanned across a protein's sequence at the 1, 4, 5, and 8 positions and scored according to how well those four residues can make putative contacts to a particular DNA sequence (assumed to be localized to the major groove). The scoring system itself is derived from a study of the numbers and identities of all amino acids involved in specific DNA contacts as observed in the available high resolution structures of complexes (Suzuki, 1994). From the contact summary in Figure 4.12, the best probe helix would include residues K388, R391, G392 and Y395. Indeed, this group of residues provides most of the observed contacts. However, given only these residues in the absence of further structural information, it is not possible to predict unambiguously the mode of ETS domain DNA binding. It is interesting to note that in the original study, the scoring algorithm of Suzuki (1994) predicted the basic region of helix H2 to be the best candidate for a probe helix. The structure described by Werner et al. (1995b) in Figure 4.13 demonstrates how the ETS domain can bind DNA in such a specific, high-affinity manner. A considerable binding surface is available with contacts not only made by the recognition helix (H3) but also by residues of helix H2, the HTH turn and helix HI. A comparison of the free structure of Ets-1AN331 with the bound structure of the human Ets-1 ETS domain reveals that there are no global conformational changes in the ETS domain due to DNA binding. However, the protein-DNA interaction is facilitated by the DNA, as evidenced by a slightly unwound major groove and a prominent 60° kink. Helix HI binds the minor groove like other wHTH proteins but in an unprecedented manner for an wHTH family member. At the N-terminus of Helix HI, the indole ring of W338 intercalates between two GC basepairs situated 5' to the GGA core sequence. This interaction is assisted by L337 and the backbone amide of L341. The resonance at 12 ppm which appears in Ets-1 AN331/SCI protein-DNA titration (Figure 2.10) is due to the 134 A 4 T M • I J" T M 1 O O O A O A I i C o - V 1 • I I c o O G Figure 4.12 : DNA binding by the ETS domain. The three panels are based on data described in Werner et al. (1995). In the most general sense, the orientation of the ETS domain in complex is similar to that of a homeodomain protein. In Panel A, an approximate orientation of the ETS domain recognition helix H3 in an unfolded major groove encompassing the core 5-GGAA-3' motif is shown. Unfilled and filled circles represent the hydrogen bond acceptor and donors for each base pair. A boxed 'M' represents the 5-methyl group of thymine. The recognition helix makes the majority of contacts to the DNA, yet several very important contributions are made by residues in helix H2 (shaded residues in Panel B, opposite page) and by helix HI. The numbering (1,4,5,8) shown below the primary sequence highlights the residues predicted to make specificity contacts according to the probe helix model (Suzuki, 1994). The probe helix model is not entirely accurate at describing ETS domain DNA binding since DNA distortions are involved. In Panel C, a contact summary is shown. The indole ring of W338 intercalates between two guanine bases in the minor groove, 5' to the GGAA core site (shaded sugar rings). The 10 bp sequence shown corresponds to the SCI high-affinity site previously used in the binding studies of Ets-1AN331. A 60° bend and slight degree of unwinding is necessary to accommodate all of the observed contacts. A K388T mutant will alternatively bind a GGAT core site (Bosselut et al, 1992). The relevant bases and K388 are flagged by an asterisk. Note that K338 does not directly contact the relevant basepair (A7/T14). A reorganization of the hydrogen bonding network therefore must occur to fulfill this alternate core site. l3fT B H2 H3 DPDEVARI#|GKRK^PKMNYE^^JR|^|ltdgfe^ 4 5 236 Figure 4.13 : The structure of the human Ets-1 ETS domain by Werner et al. (1995b) demonstrates that the "wing" between the (3-strands nearest the recognition helix (H3) does not bind DNA. Intercalation of W338 drives the formation of a 60° kink in the DNA duplex. Other residues in the turn of the HTH motif (K379, N380) and the recognition helix (R391, Y396) that make important DNA contacts are shown. A 237 1.0 £ 0.5 CO 0.0 -! X -0-5 -1.0 L B 340 350 360 370 380 residue number 390 400 410 B Figure 4.14 : Chemical shift perturbation analysis of the uncomplexed murine Ets-1 ETS domain and the complexed human Ets-1 ETS domain. A 100% sequence identity between the two Ets-1 ETS domains facilitates the analysis. (Panel A) The H N chemical shifts of the uncomplexed murine ETS domain (Ets-1AN331) was subtracted from their corresponding H N chemical shifts in the complexed human Ets-1 ETS domain (Werner et al, 1995b) and plotted against residue number. (Panel B) The absolute value of the chemical shift differences is coloured onto the Ets-1AN331 structure as increasing shades of green. A chemical shift difference could not be calculated for the following residues: P368, P392, S390, D398, A399, K409, Y410and V411. Chapter 4 — Tertiary Structure of the ETS Domain 138 highly downfield shifted W338 H e l proton (see Figure 2.9 regarding the peak marked ets). Presumably, the downfield shift arises from the rich electronic environment that the aromatic DNA bases can provide. Throughout the 10 bp footprint, conventional Watson-Crick base pairing is maintained. Werner et al. (1995b) also demonstrated that the wing (S3-S4 loop) neither contributes DNA contacts nor even approaches the DNA. Chemical shift perturbations observed upon Ets-1AN331-DNA binding and those observed in Fli-1 upon DNA binding therefore must reflect an indirect structural change (Liang et al., 1994b). In contrast, the wing of the wHTH members, E. coli CAP and HNF-3y, plays a role in DNA binding (Figure 4.6). The role of the wing in Drosophila and K. lactis HSF is currently unknown, although, it is suspected to make DNA contacts (Harrison et al., 1994; Damberger et al, 1994; Vuister et al., 1994). An alternative function of the wing as a mediator of protein-protein interactions may be possible for the ETS domain. In the human tumor necrosis factor promoter region, a PEA3/Ets-1 consensus binding sequence is found directly juxtaposed to an AP-1 sequence consenus binding sequence (Kramer et al., 1995). In addition, the ETS domain has also be shown to interact with N F - K B (John et al, 1995). The structures of the uncomplexed murine ETS domain and the complexed human ETS domain offer a "before" and "after" view of DNA binding. As the murine and human ETS domains demonstrate 100% sequence identity, an opportunity arises to examine what residues experience changes in their immediate environment by chemical shift perturbation analysis. This analysis extends what was initiated by the Ets-1AN331 1 5 N -tyrosine selective labelling. Chemical shift differences are summarized in Figure 4.14. with interpretation aided by colour-coding the chemical shift differences onto the structure of Ets-1AN331. As expected from the analysis, the recognition helix (H3) experienced significant amide chemical shift changes throughout its length from Y386 to Y396. Intercalation by W338 represents the most significant chemical shift perturbation upon DNA binding. The amide proton of L337 also experiences significant chemical shift Chapter 4 — Tertiary Structure of the ETS Domain 139 perturbation as a result of its vicinity to the aromatic ring of W338 and the bases of the DNA. The remainder of helix HI is not significantly perturbed. From the series of perturbations in strand S2 that culminate in a large perturbation experienced by the amide of W356, a degree of plasticity in the P-sheet scaffold appears to be necessary to accommodate a structural conversion to the uncomplexed to the complexed state. The events that occur in the vicinity of strand S2 may be associated with the perturbations found throughout helix H2. Similarly, perturbations found throughout strand S3 likely follow the events that occur nearby at helix H3. In summary, chemical shift perturbation analysis offers a means of obtaining specific data regarding structural transitions without the prerequisite of a high resolution structure. From the extent of the chemical shift perturbations, it is evident that communication of the DNA binding event extends beyond the recognition helix. Potential recipients of a conformationally mediated signal include regulatory modules of Ets-1 and other cellular factors. An adjacent regulatory module that attenuates the DNA binding affinity of the murine Ets-1 ETS domain will be discussed in Chapter 5. In the postulated mechanism of this intramolecular regulation, structural communication throughout the ETS domain is a central feature. 4.8. The ETS Domain of Fli-1 In addition to the structures of murine and human Ets-1, a third NMR solution structure of an ETS domain from human Fli-1 has been described (Liang et al.,1994a, 1994b). Fli-1 shows a 69% sequence identity with the ETS domains of human and murine Ets-1. Reflective of this sequence simlarity, the structure of the Fli-1 ETS domain is topologically similar to the Ets-1 ETS domain. The smaller Fli-1 fragment lacks the helical C-terminal extension found in Ets-1AN331 and was characterized as a complex with a 16 basepair DNA duplex. Therefore, no major changes of secondary structure or global fold arise in the Ets-1 ETS domain due to DNA binding or the presence of the additional C-terminal helix. A electrostatic profile of the Fli-1 ETS domain reveals the Chapter 4 — Tertiary Structure of the ETS Domain 140 absence of an acidic patch observed on Ets-1 AN331. In accordance with the prediction that electrostatic interactions markedly affect the solubility of the ETS domain (Figure 4.8), the Fli-1 ETS domain demonstrates an exceptional level of solubility (3.8 mM; Liang et ai, 1994a,b). The resolution, expressed as a backbone RMS deviation with respect to the ensemble of calculated structures, of the Fli-1 ETS domain structure is similar to that of Ets-1 AN331 (Fli-1: 0.98 A, Ets-1: 1.34 A). In contrast to human Ets-1, the corresponding DNA component of the Fli-1 ETS domain complex was not reported due to the limited availability of chemical shift assignments and structural restraints. However, amide backbone chemical shift perturbations relating the uncomplexed and complexed structures were reported. The chemical shift perturbation analysis of Fli-1 confirms the majority of DNA contacts that were observed in the human Ets l /DNA complex by Werner et al. (1995b). Specifically, chemical shift perturbations were observed throughout the recognition helix (H3) and in the N-terminal region of helix HI. The chemical shift perturbations observed in helix HI support the conformational changes that must occur to achieve intercalation of the tryptophan indole ring (W338 in Ets-1AN331). Chemical shift perturbation were also observed in the "wing-like" loop between strands S3 and S4 leading Liang et al. (1994b) to propose a model of DNA binding in which the wing makes specific contacts to the DNA. 4.9. A Comparision of Three ETS Domains A comparison on the interhelical angles observed among the murine Ets-1AN331, human Ets-1 and Fli-1 ETS domains is listed in Table 4.6 and presented in Figure 4.15. At the level of secondary structure, the three ETS domains are identical. At a level of sequence conservation observed between the three ETS domain, a much higher degree of tertiary structural similarity is expected (Pastore and Lesk, 1991), especially between the complexed ETS domains of human Ets-1 and Fli-1. A significant degree of similarity may Figure 4.15 : The ETS domains of uncomplexed murine Ets-1 (Ets-1AN331), complexed human Ets-1, and complexed Fli-1 were superimposed according to the backbone (N, C a , C , O) atoms of the three a-helices (Helix HI 337-346, Helix H2 369-383, Helix H3 386-396). Interhelical angles are listed in Table 4.5. The turn of the HTH motif is shown for all three proteins. However, for clarity, only the loop and p-sheet regions of murine Ets-1 AN331 are shown. 242 Table 4.6 : Summary of ETS Domain Interhelical Anglesa Proteinb Hl/H2anglec H1/H3 angle H2/H3 angle murine Ets-1 141° 108° 109° (uncomplexed) human Ets-1 165° 97° 72° (complexed) human Fli-1 112° 93° 113° (complexed) flInterhelical angles were calculated with the interhx module of the ribbons software package (Carson, 1991). ''References: murine Ets-1 (this study, and Donaldson et al, 1996); human Ets-l/DNA complex (Werner et ai, 1995b); human Fl i - l /DNA complex (Liang et al., 1994b). cInterhelical angles are defined such that 0° is parallel and 180° is antiparallel Chapter 4 — Tertiary Structure of the ETS Domain 143 in indeed exist but is masked by the limited precision at which Ets-1AN331 and Fli-1 ETS domain structures can be described. The position of a helix in an NMR derived structure is dependent upon the number of long-range side chain-side chain distrance restraints that tether it to the rest of the molecule. Even with many long distance restraints, a variation in angles similar to a propeller or see-saw motion can be observed if a considerable number of N - and C-terminal restraints cannot be obtained. From the ensemble of structures presented in Figure 4.3, the location of helix H2 in Ets-1AN331 is not precisely defined due to the conformationally unrestrained turn that follows it. Similarly, unrestrained regions occur following helices HI and H3. Thus, observed differences in interhelical angles among the three ETS domains are likely a result of both the limited structural resolution and effects of DNA binding. As predicted by the chemical shift perturbation analyses of Ets-1AN331/human Ets-1 and the uncomplexed/complexed forms of the Fli-1 ETS domain, it is very probable that bona fide structural differences exist between the uncomplexed and complexed forms of the ETS domain. In LexA repressor, a 20° shift in helix H2 has been observed upon complexation (Fogh et ai, 1995). The structural changes accompanying DNA binding may be conceptually discussed in terms of known modes of domain and subdomain movement as described by Gerstein et al. (1994). In this work, the authors aimed to identify common structural changes by examining structures of unliganded and liganded proteins. From their study, Gerstein et al. (1994) identified shearing and hinge-like motions as two general classes of domain and subdomain movement. In such motions, the secondary structures tend to remain unaffected as the (<|>,v|/) space that defines a secondary structure, especially an a-helix, is quite limited. Recalling the CD spectroscopic study of the free and bound murine Ets-1AN331 ETS domain (Chapter 2 and Petersen et al, 1995), the secondary structure content did not change. Chapter 4 — Tertiary Structure of the ETS Domain 144 Shearing motions involve an interface of a-helices or (3-sheets. The extent of shearing is determined by the extent of interdigitation of the relevant side chains (Perutz, 1989; Gerstein et ai, 1994). As helix H2 demonstrates a wide range of interhelical angles in the three ETS domain structures, its residue profile may be particularly suited for shearing and an associated hinge-like motion. The short chain residues, V371 and A372, define the N-terminal interface of helix H2 and slightly interdigitate with residues L365 and V412 in the P-sheet. W375 consitutes the "pivot" of helix H2 and is not completely buried in the structure. The C-terminus of helix H2 degenerates into a highly basic, mobile loop involved in DNA binding and nuclear targetting. In sum, the N-terminal residues of helix H2 have the potential to mediate a shearing motion while the i l l -defined loop has the potential to coordinate a hinge motion. Shearing motions resulting in up to a 1.5 A shift and a up to a 20° rotation have been observed in the helices of tryptophan repressor protein HTH motif (Lawson et al, 1988). A hinge-like motion is often ascribed to P-strand secondary structures. P-strands tend to move in pairs to maintain hydrogen bonding. In two regions of the ETS domain, there is potential for a hinge motion. Leaving helix HI, strands SI and S2 may form a hinge that assists helix H2 movement. Since L345 H 5 2 maintains its distinction as the most upfield shifted proton resonance in both Ets-1AN331 and complexed human Ets-1 (-0.64 ppm and -0.57 ppm, respectively), the electronic environment defined by helix HI, strand SI and strand S2 has likely not been significantly altered. Likewise, the undefined loop after helix H3 that spans residues D398 to N400 may also serve to mediate a hinge-like motion of strands S3 and S4. Marked chemical shift perturbations are observed in strands SI and S3, the outer-most strands of the anti-parallel P-sheet, that support a hinge-like motion. From chemical shift perturbation analysis and a direct structural comparison of the free (Ets-1AN331) and bound (human Ets-1 and Fli-1) forms, a conformational change occurs throughout the entire ETS domain upon DNA binding. As will be discussed in the following chapter, communication of the DNA binding event to an adjacent module will play a major role in the modulation of the DNA binding affinity of murine Ets-1. C h a p t e r 5 Intramolecular Regulation of D N A Binding With the DNA binding domain so central to the overall function of a transcription factor, it is not surprising that many mechanisms exist to influence its activity. In the Ets family, the combined effects of phosphorylation state, protein-protein interactions, alternative splicing and promoter architecture all contribute to a complex regulatory program. This chapter discusses the role of autoinhibition of DNA binding, a regulatory mechanism that has been identified in Ets-1 and five other Ets proteins. Specifically, I present data which shows that the inhibitory module is intimately associated with the ETS domain and is composed of four helices. The helices of the 'inhibition domain' experience a rich range of molecular motions and local stabilities as determined from NMR relaxation and hydrogen exchange experiments. These data are discussed in light of a potential mechanism of inhibition. 5.1. Introduction The reductionist approach relies upon the dissection of modules, particularly DNA binding domains, to simplify structural study of transcription factors. One interesting observation that has recurred throughout many studies is that the DNA binding activity or more precisely, the binding affinity, is greater when the domain has been segregated from the full-length protein. DNA binding domains which are susceptible to this phenomenon of intramolecular inhibition include p53, N F K B , TATA-box binding protein (TBP), Islet-1 (Isl-1), and a wHTH family member, HSF (Hupp et al, 1992; Grimm and Bauerle, 1993; Lieberman et al, 1991; Sanchez-Garcia et al, 1993; Rabidran et al, 1993). Similar observations have also been reported for members of the Ets family: Sap-1, Elk-1, Erp, Net, Ets-1 and Ets-2 deletion mutants all have a higher DNA binding affinity than the native protein (Hagman and Grosschedl, 1992; Lim et al, 1992; Nye et al, 1992; Wasylyk et al, 1992; Dalton and Treisman, 1994; Giovane et al, 1994; Janknecht et al, 1994; Lopez et al, 1994). Given the variance in the positioning of the ETS domain within 145 Chapter 5 — Intramolecular Regulation of DNA Binding 146 this group of Ets proteins, more than one type of autoinhibition may be occurring. Hence, for the remainder of this chapter, I will restrict my discussion to Ets-1. Ets-1 has been shown to contain inhibitory sequences both amino and carboxy terminal to the ETS domain. Deletion of either region results in as much as a 10 to 20 fold increase in affinity for DNA (Hagman and Grosschedl, 1992; Lim et ai, 1992; Nye et al, 1992; Wasylyk et al, 1992; Fisher et al, 1994; Petersen et al, 1995). Several natural variants of Ets-1, differing in either the amino or carboxy sequences flanking the ETS domain, also show enhanced DNA binding. An alternatively spliced isoform of Ets-1, p42, lacking exon VII, binds to DNA with high affinity (Wasylyk et al., 1992; Fisher et al., 1994; Figure 1.2). This exon encodes the entire amino inhibitory sequence. The oncogenic viral form of Ets-1, v-Ets, contains a 16 amino acid substitution within the carboxy terminal inhibitory sequence that also results in increased DNA binding affinity (Leprince et al., 1983,1994, 1993; Nunn et al, 1983,1989; Hagman and Grosschedl, 1992; Lim et al, 1992; Hahn and Wasylyk, 1994). 5.2. The Inhibition Module of Ets-1 Previously, members of the Graves laboratory performed limited proteolysis of full-length murine Ets-1 to identify a number of protein fragments of varying DNA-binding affinity. The results of Petersen et al. (1995) and Jonsen et al. (1996) are shown in Figure 5.1. Ets-1AN331 was initially chosen for structural studies since it represented the minimal DNA binding domain (Chapter 2). Ets-1AN331 contains a wHTH DNA binding motif (Liang et al, 1994; Donaldson et al, 1994,1996) and an additional a-helix in the C-terminal flanking sequence. Since deletions or mutations in the C-terminal region abolish the inhibitory activity, this region is believed to interact with the N-terminal inhibitory region and the ETS domain. The structure of Ets-1AN331 alludes to such an association as the C-terminal inhibitory region rests very close to the N-terminus of the ETS domain. Chapter 5 — Intramolecular Regulation of DNA Binding 147 Ets-1 AN 170 AN280 AN301| A N 3 2 2 | AN331 AN280AC428 Figure 5.1: Deletion mutants that were designed to delineate the boundary of the ETS domain also uncover an inhibitory activity (Petersen et al., 1995; Jonsen et al, 1996). An asterisk denotes Ets-1 fragments which have been overexpressed for NMR studies. The Graves laboratory supplied me with a second deletion mutant, Ets-1AN280, containing the full inhibition domain, for structural analysis. The gene fragment had been previously inserted into a T7-based expression vector. Expression of this construct (pAED3-18k) in rich or M9 media supplemented with algal extract was more robust than Ets-1AN331, ca. 8 mg/L culture. I attempted to improve the expression further by subcloning the gene fragment into the pCW series, a tac promoter based vector which was used for Ets-1AN331 overxpression (Chaper 2). This construct, termed pCW-18k, did not appreciably express Ets-1AN280 in E. coli DH5a, TBI or Topp2 strains; therefore pAED3-18k was utilized. Purification of Ets-1 AN280 protein was straightforward. Upon lysis, the crude extract was applied to a similar Mono S column used for Ets-1 AN331 purification and eluted with a gradient of 150-600 mM NaCI. Ets-1AN280 eluted at 350 mM NaCI. Preliminary i H ^ N - H S Q C spectra demonstrated that Ets-1AN280 was a folded protein (Figure 5.2). As expected for a 162 residue, 18kDa polypeptide, the linewidths of Ets-148 1 5N Chemical S h i f t (ppm) o o o o LD O CM LD CM O 6 i3 . ! t: • ^ .4*. i ~ 1  o _, °* ci M « w 1 O o a c o o 4-1 - H CO o • H (D CO CU 1—1 X —' 13 C? >> 8 .a .5 to Si g c i s S 0 o .9 cn CU CN 0) CO T 3 T 3 £ S= O * a ^ -IT S 58 CD CO ffi g a, 53 ~ T3 O X. co •** S i cn -CO C CO 0--a & ii < i2 1—1 CO c" g „ too O u O o a a t3 g, £ c U 1 on a H H J H C 52 ^ -2 2 * a 8 i-C co Z ex i n o <—1 -M '- | ui CN ^ 0) cn 13 .S co 5 o .a 8 Io ^ cn C I H >H O u T3 co cn 3 * J (0 53 cu co C •S '5b H - » CO CU (0 "~' O H 3 O CU vfcl e: "cS •S C T3 .SP CU cn -•-> co O C J C Chapter 5 — Intramolecular Regulation of DNA Binding 149 1AN280 were broad. Like Ets-1AN331, Ets-1AN280 was only soluble at low concentrations by NMR standards (0.6-0.8 mM) and high salt conditions (0.4-0.5 M NaCI). Resonances characteristic of Ets-1 AN331, including the tryptophan H e l indole protons, Q339 H N and F363 H N also observed in Ets-1AN280. Collectively, these similar resonances suggested that the structure of the ETS domain was similar in Ets-1AN280 and Ets-1 AN331. However, the presence of many resonances in the "middle" of the HSQC spectrum (ca. 120 ppm in 1 5 N and 8.0 ppm in *H) indicated that a significant amount of random coil or extended conformations was also present in the Ets-1AN280 N-terminal inhibition region. CD thermal denaturation experiments demonstrated that the N-terminal inhibitory region was stabilizing to the protein as a whole. At moderate salt concentrations (0.15 -0.25 M NaCI) and at 8 pM protein, Ets-1AN280 and Ets-1AN331 exhibited sharp transitions in their respective thermal denaturation curves at 54°C and 45°C when monitored at 222 nm (Consult Chapter 2 for a CD melt profile of Ets-1AN331). These results imply that the N-terminal inhibitory region is cooperatively folded with the ETS domain to form a single structural domain. Assignment and secondary structure analysis of Ets-1AN280 was performed by J. J. Skalicky as described in Skalicky et al. (1996). This work identified two additional helices in the N-terminal inhibitory region and as expected, a considerable stretch of residues with random coil conformations. Specifically, the random coil conformations were identified by consistent random coil chemical shifts, few interresidue NOE's and strong heteronuclear 1H{ 1 5N} NOE's. Compared to Ets-1AN331, the bounds of the secondary structural elements in the ETS domain and C-terminal region were essentially invariant. Residues in intermediate exchange (the wing: G408-V411, and the H3-S3 loop: D398-N400) which were absent in Ets-1AN331 spectra were similarly absent in Ets-1AN280 spectra. Figure 5.2 summarizes the secondary structure assignments of Skalicky et al. (1996) and Chapter 5 — Intramolecular Regulation of DNA Binding 150 what is known from other Ets-1 deletion studies (Wasylyk et ai, 1992; Hahn and Wasylyk, 1994). Taken together, the inhibition domain begins at residue 301, coincident with the N-terminus of the first inhibition helix, HI1. Regarding the topology of the inhibition domain relative to the ETS domain, Wasylyk et al. (1992) noted that an anti-helix H2 monoclonal antibody did not bind well to inhibited Ets-1 deletion mutants. This observation suggests that some portion of the inhibitory domain must be in the vicinity of helix H2. As discussed in Chapter 4, the C-terminal inhibitory region (H4) lies near the N -terminus of the ETS domain. This led to the suggestion that the N - and C-terminal inhibitory regions directly contacted each other to inhibit DNA binding by the intervening ETS domain. This model was extended by chemical shift perturbation analysis (presented in Skalicky et al, 1996). As chemical shifts of a nucleus are highly sensitive to its conformation and immediate electronic environment, the differences between corresponding backbone amide 1 5 N and *H resonances of Ets-1AN280 and Ets-1AN331 were calculated. Several large differences of I AS I = 0.25 ppm in H N and 1.0 ppm in 1 5 N were noted. These differences are well above the "noise" expected from the comparison of two samples at slightly different pH's and temperatures (Ets-1 AN331: pH 6.45, 20°C; Ets-1AN280: pH 7.2, 28°C). The absolute shift differences were tabulated and mapped onto the Ets-1AN331 structure. When the largest perturbations are considered as a whole, a plausible interaction surface for the inhibition domain is apparent (Figure 5.4). This surface includes helix HI of the ETS domain, and residues immediately preceeding helix H4. A limited number of residues in helix H2 and helix H3 and the fi-shed are also perturbed. As these residues contribute to the hydrophobic core of the ETS domain, it is likely that these changes arise indirectly due to contact of the N-terminal inhibitory region with helices HI and H4. If the structure of the inhibited ETS domain was very different from the uninhibited form, many more perturbations relative to Ets-1AN331 would have been observed. The patch of chemical shift perturbations is on the opposite face as the recognition helix and implies that the inhibition domain cannot 151 •x X 3 3 c" •£ T J < c • •—^  Oil CN T - I ON ^ : 2 2 <3 o — . cn cn W cC I i CO T J C o Z! CJ c & c * cu X H i - i i cn W C § co T J ^ 3 73 OJ 13 3 o cn — , CU • 3 ~ 12 g . S > T J CU ! * o -*-> «.s 3 co O 3 - 3 co C L 3 ™ T J J£> cu 2 > C N « ON ~ 2 <u cn cu T J 3 6 3 M T J c cu C L cu g Ol X c j cn "co C "-2 . 3 eg * «J b N j / 2 t> >^  a S > N H -cn *•> co cn c s co 3 60 " g T J c X cn co 3 co ? T J C o & s . co cn T J . s £ 3 co O OJ i-t 2 3 co • s cn T ! to *a3 T J T J 3 CO O N O - O N g « CO _b0 t3 g 5 < £ " "cO cn ^ C J 3 cu g o 3 X X 2 cu X CU > CO X ON 2 T J 3 CO 2 CO t-CU X bo X CO X < 1 T J 3 X o CN 2 T J 3 CO [>. co 2 LO 2 cn cu 3 ON 60 ON 5 CJ 3 i i cn >> d cO T J 3 O c j CU cn > N X T J CU « i 5 "o3 cn cu i-i cu - t l CO X 3 o '5b cu I i CO cn cu H - * CO CJ T J 3 u, CO X >> CU bo j-» X bO 3 H 4-, W O cn cn g 3 3 5 T J — i CU — I i CO 4 1 CU cn -gj r T .5 ON Q T J CU i i CU 1 3 3 N O O N O N I l l CS g s C 3 o g > ^ i S X N I i T J O O O >*-£ SJ LO X cn •3J cu cu " 3 co 2 2 oo 2 •2 w JJ 3 cu co 6 -g g TJ C ^ f^.S O N C L f H co co ^ g ^ « O cn oo ^ •g <N 3 5 cn O 3 3 T J g ^ ' CN bo o 2 x 2 x; "2 cO CO S 3 ^ CO X 3H CJ 3 cu 3 c r cu cn < 2 Q o C H U O M 01 cn a, - 3 -S cu cu — S-L cj -3 cu O " 52 43 3 c C ai S cu cu cu Ii hn i i 3 g 3 O T J CU T J cu 3^ cu cn '% " 3 cu ca £ ai i—1 - 5 o cu £ CU £ .9 3 c j 3 i i »-i I i o i - 0 0 g <N 3 t i g *-2 g " x: T J ' 3 co bo .tJ S CO (U i i _ 3 "43 U X) H 2 -g _cn m .g W cu cj 3 CO CO bo 3 cu cj 3 cu T J T J QJ > I i CU X Figure 5.4 : Stereoview mapping the spectral perturbations due to the presence of the amino terminal inhibitory sequence on residues 1335 to D440 in the three-dimensional solution structure of Ets-1AN331 (Donaldson et al, 1996). This high affinity deletion mutant of Ets-1 is composed of the ETS domain and the carboxy terminal inhibitory region, which includes helix H4. The absolute values of the differences in the amide 1 5 N chemical shifts between Ets-1AN331 and Ets-1AN280 are presented on a scale from white (no change; IA5I = 0.0 ppm) to green (I A51 max = 3.4 ppm). The shift perturbations indicate that the amino terminal inhibitory region packs on the surface formed by helix HI of the ETS domain and helix H4 of the carboxy terminal inhibitory region. The figure was generated using the program GRASP (Nicholls et al, 1991). Three significant aspects are shown by this perturbation diagram: (i) The inhibition module make contacts to the N - and C- termini of the ETS domain consistent with deletion mutant analysis, (ii) Helix HI is a link between the ETS domain and the inhibition module. (Hi) The HTH of the ETS domain is essentially unpeturbed by the inhibition module. This implied that the inhibition module must function primarily by an allosteric mechanism. Chapter 5 — Intramolecular Regulation of DNA Binding 153 function by simply occluding protein-DNA contacts. Therefore, Ets-1 autoinhibition must be considered in terms of an allosteric mechanism. DNA titrations with a 10 bp high-affinity SCI oligonucleotide duplex were performed to assess the contribution of the inhibition module to the DNA binding surface. Jump-return one-dimensional imino proton spectra acquired at a number of DNA:protein ratios up to 1.2:1 were identical to the imino spectra observed for Ets-1AN331 (Figure 2.10). Furthermore, similar DNase I footprints were observed for Ets-1 AN280 and Ets-1AN331 (T Petersen and B Graves, personal communication). Together, these data reinforce the findings of the chemical shift perturbation analysis that the inhibition module does not directly block the DNA binding surface or change the number of DNA contacts. 5.3. Structural Modelling of the Inhibitory Module In the absence of any NOE restraint data to precisely position the helices of the inhibition module of Ets-1 AN280, an initial model of repressed Ets-1 was designed that packs the N - and C-terminal inhibitory sequences onto helix HI of the ETS domain as an anti-parallel helical bundle. The helices were packed and aligned according to chemical shift predictions, common patterns of helical alignment and their marked amphipathic profiles shown in Figure 5.5. The structure of Ets-1AN280 was not pursued due to the presence of 20 N-terminal residues in a random coil conformation that made the spectra too difficult to interpret. A three-dimensional model of the inhibited Ets-1 fragment was generated in the Insightll/Homology environment (Biosym Technologies / Molecular Simulations, Inc.; San Diego CA). In the first phase, the two inhibition helices, HI1 and HI2, and helix H4 were modelled as "perfect" a-helices ((]) =-57°) and aligned to a representative four-helix bundle, 1FLX (Quinn et al, 1990). In the second phase, the helical bundle was docked to the ETS domain structure and then residues were "mutagenized" to correspond to the Ets-1 sequence. Lastly, the loop regions were modelled. The entire assembly was 111 ^ a? £ O —! X cn _0 0) ^ <N X -S» O CO £ 1 .S 1/5 tin U J Chapter 5 — Intramolecular Regulation of DNA Binding 155 subjected to several rounds of short time scale (1 ps) unrestrained molecular dynamics at 300 K and minimizations in the X-PLOR 3.1 environment. This model, depicted in Figure 5.6, is consistent with the chemical shift perturbation data, monoclonal antibody binding data, and DNA titration data. Circular dichroism spectroscopy revealed a unexpected result when the complexes of Ets-1AN331 and Ets-1AN280 with DNA were compared (Petersen et ai, 1995; Donaldson, unpublished data). Upon DNA binding, Ets-1 AN331 exhibited a relatively minor change in the magnitude and shape of its amide CD spectrum suggesting that the secondary structural changes between uncomplexed and complexed forms of the ETS domain were minor. However, a marked reduction in molar ellipticity at 222 nm was observed when Ets-1 AN280 was complexed with DNA. This change in ellipticity was predicted to reflect a loss of ca. 9-11 a-helical residues. To probe the site of this apparent instability in the inhibition domain, Petersen et al. (1995) subjected the uncomplexed and complexed forms of Ets-1AN280 to partial proteolysis by trypsin and chymotrypsin. Almost all of the potential cleavage sites between residues 304-311 (helix HI1) were found to be susceptible to proteolytic attack only in the complexed form. Combined with the Ets-1AN280 model and the chemical shift perturbation data, these data lead to a model of Ets-1 autoinhibition in which DNA binding is linked to a conformational change in the ETS domain, resulting in the disruption of the inhibitory module and the local unfolding of helix HI1 (Figure 5.7). In a complementary study, Jonsen et al. (1995) noted enhanced proteolysis in the N-terminal region of the inhibition module that was independent of DNA binding when the C-terminal helix was disrupted by deletion (Ets-1 AN280AC428). This observation further supports the intimate association between the N - and C-terminal inhibitory sequences. In the following sections, I will describe structural and dynamical aspects of an optimized construct of the inhibited ETS domain with the goal of refining the model and mechanism of autoinhibition. Figure 5.6 : A model of Ets-1AN280. The inhibition domain was modelled as a helical bundle and grafted onto the structure of Ets-1 AN331. In this illustration, only residues 302-436 are shown. (Colour legend) Blue = P-strand; red = ETS domain a-helices H I , H2 and H3; red = inhibition module helices HI1, HI2 and H4. This figure was created by M O L S C R I P T v l . 4 (Kraulis, 1991) and R A S T E R 3 D (Merritt and Murphy , 1994). 257 Ets-1AN331 High Affinity Ets-1AC428 High Affinity Figure 5.7 : A model for intramolecular inhibition of Ets-1 DNA binding. Helices HI1, HI2, HI, and H4 are indicated by cylinders and the remainder of the ETS domain by a shaded oval. Possible DNA contacts include the helix(H2)-turn-helix(H3) motif as well as the N-terminus of HI (not shown for clarity). Residues 1 to 279 of Ets-1AC428 are represented by the hatched box. A jagged line indicates sequences that adopt an extended conformation. In solution, Ets-1AN280, or native Ets-1, exists predominantly in a low affinity form in which helices HI1, HI2, and H4 in the amino and carboxy terminal inhibitory regions, respectively, are packed with helix HI of the ETS domain to form a 4-helix bundle-like structure. DNA binding by Ets-1AN280 is accompanied by a conformational change involving the unfolding of HI1 (denoted by the changes in the shape of the oval and the shading of cylinders) (Petersen et al, 1995). Refolding of HI1 restores the structure of the inhibitory module and leads to disassocation from DNA. Ets-1AN331, which lacks the amino terminal inhibitory region, and Ets-1AC428, which lacks the final 12 residues of the carboxy terminal inhibitory region, exist constitutively in this altered conformation and thus bind DNA with high affinity (Jonsen et al, 1996). Chapter 5 — Intramolecular Regulation of DNA Binding 158 5.4. A Minimal Autoinhibited Ets-1 Polypeptide, Ets-1AN301 5.4.1. Overexpression and Characterization With twenty residues of random coil appended to the N-terminus of Ets-1AN280, the quality of the spectra of this protein were not sufficient for further structural analysis. From a consideration of all deletion mutants (Figure 5.3), the N-terminal boundary of the inhibition module was predicted to begin at residue 301, coincident with the N-terminal boundary of the first inhibitory helix. Therefore, a new deletion mutant of Ets-1, Ets-1AN301, was expected to fully retain the inhibitory activity. This deletion mutant was also predicted to be a more suitable candidate for NMR studies as linewidths should be narrower as a result of the lower molecular weight (18 kDa vs. 16 kDa). More importantly, deletion of twenty random coil residues would relieve much of the amide resonance degeneracy allowing interpretation of higher dimensional experiments such as an 1 5N-HSQC-NOESY and 15N-HSQC-TOCSY. Clearly, the process of identifying an optimal deletion mutant in the reductionist approach to structural characterization can be iterative one. A new plasmid, pET22-16kDa, spanning residues 301 to 440 of Ets-1 was constructed as follows: (i) The gene fragment encoding residues 301-440 was amplified from plasmid pET3-18k using the polymerase chain reaction (PCR). Primers were chosen to create Ndel and Hindlll sites flanking the gene and to create BamHI sites at the extreme 5' and 3' termini. The PCR fragment was prepared with BamHI, subcloned into pKS -Bluescript (Stratagene, La Jolla CA) and subsequently grafted into the synthetic Ets-1 genes of plasmids pET22-ser and pCW-ser via Ndel and EcoRI sites (originally described in Figure 2.4). The constructs were therefore hybrids of the murine and synthetic codon biased genes (Figure 5.8). The expressed hybrid protein, Ets-1AN301, contains a point-mutation, C416S. From the structures of the Fli-1, human and murine ETS domains, C416 did not serve a structural function and thus could be safely mutated. As Taql polymerase was used during the cloning procedure, the identity of the gene fragment was verified by dideoxy-sequencing. Chapter 5 — Intramolecular Regulation of DNA Binding 159 pCW-16k 1 tac | — Ndel term [— cys ser pET22-16k j T7 \ lac > — | term |— cys ser Figure 5.8 : Constructs designed to produce a minimally sized, autoinhibited Ets-1. Dark grey shading denotes the portion of the gene derived from the E. coli codon-biased synthetic ets-1 gene. These constructs express a 16kDa protein fragment spanning residues 301-440 in the native Ets-1 sequence. The two constructs were expressed in a variety of E. coli strains, summarized in Table 5.1. A vector/strain combination of pET22-16k/BL21::DE3(pLysS) resulted in the highest expression of Ets-1AN301 as judged from crude extract profiles on Coomassie Blue stained SDS-polyacrylamide gels. However, even in this best case, expression was low at approximately 1 mg/L bacterial culture. Plasmid selection of pET22-16k and pLysS were respectively maintained by 75 |a,g/mL carbenicillin and 34 ug/mL chloramphenicol in all media. Addition of algal extract to 0.1% (w/v) supported bacterial growth and expression in M9 media, a prerequiste for isotopic labelling. Expression of Ets-1AN301 by pET22-16k/BL21::DE3(pLysS) was improved by culturing the bacteria at 30°C, rather than 37°C. A lower temperature is believed to improve expression by enhancing the stability of Ets-1 AN301 and thus its resistance to bacterial cytoplasmic proteases. Rifampicin is a potent and selective inhibitor of bacterial RNA polymerase. Since T7 RNA polymerase is unaffected by rifampicin, this drug offered a means to enrich Ets-1AN301 transcripts in the pET22-16k/BL21::DE3(pLysS) expression system. Addition of Chapter 5 — Intramolecular Regulation of DNA Binding 160 Table 5.1: Plasmid/Strain Combinations Assayed for Ets-lAN301fl Plasmid Topp-2 DH5a BL21-DE3 BL21::DE3(pLysS) pCW-16kb < 1 mg/L « 1 mg/L N / A N / A pET22-16kc N / A N / A l m g / L 2.5 mg/Ld aAll cultures were incubated at 30°C and induced with 1 mM IPTG. ''Transcription in pCW-16k is driven by a tac promoter cTranscription in pET22-16k is driven by a T7 promoter ^Rifampicin was added at 1.5 hours post-induction to enhance expression rifampicin according to the following protocol resulted in increased expression of Ets-1AN301: (i) T7 polymerase expression in BL21::DE3(pLysS) cells was induced upon addition of IPTG to 1 mM when the ODgoo of the culture reached 1.2. (ii) After 1.5 hours of incubation, rifampicin was added from a stock solution at 25 mg/mL in methanol to a concentration of 50 pg/mL in the culture medium. (Hi) Cells were incubated for an additional 2.5 hours before harvesting and processing according to methods described in Chapter 2. Ets-1AN301 was purified to approximately 90% homogeneity by cation exchange chromatography. Given the expense of preparing isotopically labelled proteins, this level of purity was sufficient for NMR studies. Ets-1AN301 that was destined for further analysis by mass spectrometry or protein sequencing was rechromatographed on a reverse-phase C18 matrix to >99% homogeneity. As assayed by mass spectrometry, the observed mass of Ets-1AN301 was similar within 1.5 Da error to the calculated mass of Ets-1AN301 with a missing N-terminal methionine (mass spectrometry was performed by Dr. W. Wakerchuk, National Research Council, Ottawa ON). Ten cycles of Edman Chapter 5 — Intramolecular Regulation of DNA Binding 161 sequencing performed by Dr. J. Moore (Alberta Peptide Institute, Edmonton AB) verified that the N-terminal methionine was indeed absent. Unlike the other Ets-1 deletion mutants, Ets-1AN301 could be quantatively refolded by rapidly diluting 8 M urea solubilized protein with a 20-fold excess of buffer. However, like the other Ets-1 deletion mutants, Ets-1AN301 exhibited a similar limited solubility. The T m of Ets-1AN301 is 56°C (at 0.25 M NaCI and pH 6.5). As discussed previously, the T m of Ets-1AN331 is 46°C at that of Ets-1AN280 is 54°C under similar conditions. This observation suggested that all of the structural elements that contributed to the improved T m in Ets-1AN280 were still present in Ets-1AN301. Therefore, from this viewpoint, Ets-1 AN301 represents the minimally sized, autoinhibited protein fragment of Ets-1. 5.4.2. Backbone and Side chain Assignment of Ets-1AN301 The i H ^ N - H S Q C spectra of Ets-1AN301 were identical to spectra of Ets-1AN280 and thus provided compelling evidence that both proteins possessed a similar structure. The inhibitory activity was later confirmed by quantitative bandshift analysis (B. Graves, unpublished data). The difference in the quality of the two 1 H, 1 5 N-HSQC spectra is striking (Compare Figure 5.2 to the 1 H, 1 5 N-HSQC spectrum of Ets-1AN301 in Figure 5.9). A temperature of 28°C was chosen as a compromise to maintain solubility and to sharpen linewidths. The removal of the N-terminal random-coil residues of Ets-1AN280 in turn removed the cluster of resonances in the middle of the Ets-1AN301 spectrum. As heteronuclear experiments designed to assign the protein backbone are based largely on amide resonances, the improved spectral quality of Ets-1AN301 was advantageous. The backbone amide chemical shifts of Ets-1AN301 (0.7 mM in 20 mM sodium phosphate, pH 6.5; 5 mM dithiothreitol; 0.5 M NaCI, 0.02% sodium azide) were assigned primarily from a 1 5N-NOESY-HSQC spectrum. This was assisted by previous assignments of Ets-1AN280 and Ets-1AN331. No 15N-selective labels were used. 1 3 C a 262 (wdd) WMS | B 0 ! L U 9 L | 0 N S I . O CN 1— T -T— T~ _1 I I i_ CD 00 o CN CM CM CN ID CM CO CM O CO CM CO o o @ u (o) ® © o Q O O <3 o o E Q_ CL O a O o o * © o © 'O o-0 © 0 O O O h to CO 03 O x: O X cu c o S .2 « s < is cu g W I H O o ^ o •S 8 6 cu Z. 3 S3 6 £ Ja O o cu Ol u8 .S £ I o HI S 00 CL) "5 CN C/N MH cn C X , M cu U i 3 6 0 cn a « cn cu C3 O c 3 s, t5 2 1 1 i—I a -£ o "K c c > «» 6 0 CU 6 a . P ,M B X -C cu 2 £ in cn H CO co •• a\ .a i n T-, m cu IH " S .SP § c cu <n to B eg a § •« $ CM la tn co ' ? * H E-H V * H 6 0 C I D 3 cu n5 > H V H 3 In 6 0 cn P H Chapter 5 — Intramolecular Regulation of DNA Binding 163 and 1 3 CP chemical shifts were obtained from an HNCACB spectrum. Normally, this experiment would have also been used for backbone assignments; however, very few interresidue correlations were evident. The 1 5N-TOCSY-HSQC spectrum also proved to be of little value towards assignment of the side chains of this largely helical protein. C(CO)-TOCSY and H(CCO)-TOCSY experiments were acquired to assist side chain assignment but again, no useful data was obtained. Alternatively, a 24 ms mixing time HCCH-TOCSY experiment was acquired (these experiments are described in Chapter 3). Long chain aliphatic residues, such as leucine and isoleucine, were assigned by what I term, "the crossroads approach". In this approach, all peaks in the strip specified by a known C a or CP are tabulated. Next, all the HCCH-TOCSY crosspeaks specified by unknown methyl CY or C^ resonances are tablulated (the methyl shifts were previously obtained from a high resolution 1H, 1 3C-HSQC). To complete the side chain assignments, the shift patterns are consectutively matched up until all the methyls have been accounted for. In Ets-1 AN301, there were four isoleucine residues, seven valine residues and fifteen leucine residues which made assignment daunting, to say the least. This method was helped in part by prior knowledge of the Ets-1AN331 chemical shifts, of which a majority could be ported to the Ets-1AN301 database. Selective deuterium labels were extensively used in the manner of Ets-1 AN331 to assign the aromatic rings of phenylalanine, tyrosine and tryptophan. Surprisingly, even with samples as low in concentration as 0.4 mM, the quality of the !H-TOCSY and ^H-NOESY data was very good since experiments were two-dimensional and thus could be acquired at high-resolution. Coupled with a constant-time 1 H, 1 3 C-HSQC experiment, all possible assignments in 1 H , 1 3 C and 1 5 N were made for the four tryptophan rings. A majority of assignments were made for the four phenylalanines in Ets-1AN301. As with Ets-1AN331, the chemical shifts of F363H5, H e and H^ were degenerate. In Ets-1AN301, there are nine tyrosine residues which complicated the assignment process. Y307 and Y329 were believed either to be in close proximity to each other or degenerate in every respect as the ring and HP resonances in a 1H, 1H-NOESY experiment (in 99% D 2 O ) could Chapter 5 — Intramolecular Regulation of DNA Binding 164 not be unambiguously assigned. Y396 and Y410, located in the N-terminus of helix H3 and in the wing, respectively, suffered from low signal-to-noise as these residues were believed to be in slow conformational exchange. The proximity and chemical shift degeneracy of Y395, Y396 and Y397 further compounded the problems with tyrosine assignment using NOESY spectra. Low signal-to-noise and many missing correlations severely limited the utility of data obtained from (HB)CB(C)C5)H5 and (Hp)CB(C-iC5Ce)He experiments. HMQC-J and HNHB spectra were collected to obtain torsion angle information. The HMQC-J spectrum was examined in a qualitative sense to identify loop and P-sheet regions owing to their large 3 JHNHOC coupling constants. The HNHB experiment is designed to assist stereospecific assignments of HP resonances and to provide qualitative X 1 restraints (Archer et al., 1991). However, about 80% of the data were discarded, again, due to poor signal-to-noise. In helix HI2 of Ets-1AN301, there is a stretch of alanines ( 3 2 4 P A A A L A 3 2 7 ) that is difficult to assign and to accumulate NOE restraint data on due to the degeneracy of the alanine methyl groups. There is also considerable degeneracy in the H a protons due to the a-helical secondary structure. To allievate some of these difficulties, a sample was prepared in which the methyl groups of alanine where selectively labelled in 1 3 C . Conventional 1 H, 1 3 C-HSQC spectra could be now acquired at high-resolution without the interference of leucine, isoleucine and valine methyl groups. Indeed, initial spectra aquired at a sweep-width of 600 Hz in 1 3 C resolved all nine methyls groups in Ets-1AN301 (data not presented). From here, 3D-TOCSY-HSQC and 3D-NOESY-HSQC spectra were acquired to faciliate assignment and obtain restraint information. The only modification made to the pulse sequences involved the delay in the sensitivity enhanced nSTEPT sequence from 1/2JCH to 1/3JCH (Muhandiram and Kay, 1994). Unfortunately, the signal-to-noise was not sufficient to obtain any useful NOE or correlation data, despite the improved signal and relaxation characteristics afforded by observing methyl groups. Isotopic dilution of the 1 3 C in the 13C-alanine studies is likely a source of the low Chapter 5 — Intramolecular Regulation of DNA Binding 165 signal-to-noise in the acquired spectra. Unfortunately, no alanine auxotrophs are available to assess the effects of isotopic dilution. In summary, although deletion of the 20 amino acids in random coil conformations that were in present in Ets-1 AN280 improved the iH^N-HSQC spectra of Ets-1 AN301, the same problems of limited solubility, chemical shift degeneracy plagued this study. In the end, many of the side chain resonances were assigned, at least to a similar extent as Ets-1AN301. However, without any torsion angle or stereospecific assignments, there is no possibility of calculating a medium to high resolution structure of Ets-1AN301. 5.4.3. NMR Restraint Analysis and Topology of Ets-1AN301 Distance information was obtained from 3D- 1 5N-NOESY-HSQC and 3 D - 1 3 C / 1 5 N -NOESY-HSQC spectra acquired with 1 3 C / 1 5 N labelled Ets-1AN301, respectively, under similar buffer conditions used for assignment in the previous section. Given the incomplete assignment of the Ets-1AN301 spectra, NOE crosspeak assignment and accounting was adjusted to take advantage of the ambiguous NOE approach of Nilges (1995). To facilitate entry of ambiguous information, a manual method of tabulation was adopted. The data itself, with the suffix .notes, was organized according to the example shown in Figure 5.10. The .notes file was prepared and translated to ambiguous assignments in the X-PLOR 3.1 format with a small C-program, noteslxplor. To date, 230 long-range (I i-j I > 5) restraints have been accumulated in the ETS domain primarly from a manual vector-by-vector analysis of the 3D- 1 3C/ 1 5N-NOESY-HSQC spectrum. The number of long-range restraints are therefore roughly twice the number of restraints used to define the Ets-1AN331 structure. However, very few long-range restraints have been assigned to the inhibition module and may be reflective of its loose association with the ETS domain. Structures of Ets-1AN301 were calculated according to a similar SA protocol used for Ets-1AN331 except the starting structure was an elongated polypeptide with "pre-formed" secondary structure elements (consult Chapter 4 for the X-PLOR 3.1 protocols). A # a sample .notes f i l e c a l l e d "testme.notes" ! W123 HEI w 1 23 1 A23 4 HB# . m 5 67 2 R456 HG2 Q567 HA s 6 32 3 R456 HG2 Q567 HA 1654 HG2# m 1 43 7> might be R4 5 6 HG1, uncomment f o r now m 1 67 @ W123 HD1 (was assigned i n another spectrum) ! G124 HA1 w 4 56 1 T125 HG2# B notes2xplor - f i l e testme.notes \ -weak 4 0 2 2 1 0 \ -medium 3 0 1 2 0 5 \ -strong 2 7 0 9 0 2 \ > x p l o r . t b l c ; g e n e r a t e d b y n o t e s 2 x p l o r f o r f i l e : t e s t m e n o t e s ! W123 a s s i g n ( r e s i d 234 a n d name HB# ) r e s i d 123 a n d name H E i 4 . 0 2 2 1 0 a s s i g n ( r e s i d 456 a n d name HG2 ) o r ( r e s i d 5 67 a n d name HA ) r e s i d 123 a n d name H E I 3 . 5 1 7 0 5 a s s i g n ( r e s i d 456 a n d name HG2 ) o r ( r e s i d 654 a n d name HG2#) r e s i d r e s i d 567 123 a n d name a n d name HA H E I o r 2 . 7 0 9 0 2 ! G124 a s s i g n ( r e s i d 125 a n d name HG2#) r e s i d 124 a n d name HA1 4 . 0 2 2 1 0 Figure 5.10 : A simple .notes file format to facilitate the entry of ambiguous NOE restraints. (Panel A) An exclamation mark begins a new restraint entry, in this case for W123 HEI. The fields of the following input lines include: (i) restraints as strong, medium or weak; (ii) a proton chemical shift; (Hi) the number of ambiguous NOE's; (iv) restraints or a comment. (Panel B) A sample command line entry to convert the .notes file into an X-PLOR 3.1 restraint file. (Panel C) The annotated X-PLOR 3.1 restraint file. Chapter 5 — Intramolecular Regulation of DNA Binding 167 Although the inhibition module contributes a sizable degree of stabilization to the protein as a whole from the thermal denaturation studies, it is somewhat paradoxical that like an onion, the inhibition domain can be "peeled off" the ETS domain by deletion. F304, Y307 and V309, at the N-terminus of HI1, together display a large set of hydrophobic contacts for packing with other inhibition helices and the ETS domain itself. Preliminary restraint data suggests the ring of Y307 is in close contact with the ring of Y329 at the C-terminal position of helix HI2. Similarly, an extensive set of contacts is observed between W338 in helix HI and residues of helix H4 and possibly helix HI1. The end of helix HI1 is most likely to be tethered to other regions of the inhibition module and possibly the ETS domain via side chain-side chain contacts from L314. Indeed, preliminary long-range contacts are observed between L314 and a loop of hydrophobic residues from L421 to Y424. To locate the N-terminus of helix HI2, the side chain of L326 must be focused on. A preliminary set of assignments arising from the L326 H^ methyls groups places L326 near 1354 in the pocket formed by C-terminus of helix HI, the N -terrninus of strand SI, the C-terminus of strand S2 and the N-terminus of helix H2. These NOE's restraints would place helix HI2 over helix HI obscuring the HI/SI loop and a cysteine C350 contained therein from the solvent. Together, these observation suggest that the current model of the inhibition domain is correct, in a general topological sense. Given that these preliminary restraints are correct, the model depicted in Figure 5.6 should be adjusted by pulling the helices HI1/HI2 closer to the C-terminus of HI. 5.4.4 Paramagnetic Spin-Relaxation Since Ets-1AN301 is a C416S mutant, the structural role of the unique cysteine, C350, could be probed without complication. With the aim of collecting long range distance restraints to identify the position of the N-terminal inhibitory helices, an experiment was designed to derivatize C350 with a TEMPO (2,2,6,6-tetramethylpiperidine-l-oxyl, radical) spin label. Placing a free electron in a protein has the effect of obliterating any Chapter 5 — Intramolecular Regulation of DNA Binding 168 proton signal which are less than -10 A away. This phenomenon occurs because a free electron has a magnetic moment which is several orders of magnitudes greater than a proton, thus causing nearby nuclei to relax very quickly. Spin labelling has been previously useful in investigations into the subunit arrangement of ATP synthase (Girvin and Fillingame, 1995; Wilkens et al., 1995). 0 Me^ / M e / \ / \ Me N 1 Me 1 cr Figure 5.11: The paramagnetic relaxation reagent, (2-Iodoacetamido)-4-TEMPO. The reagent, iodoacetamido-TEMPO (Figure 5.11), was chosen over a maleimido analog due to its low reactivity with histidine and lysine (reviewed in Kosen, 1989). A 3-fold molar excess of iodoacetamido-TEMPO (Sigma, St. Louis MO) was added directly to an unfolded, freshly reduced sample of Ets-1AN301 in 2.5 mL of 6 M Gu»HCl at pH 8.0 and incubated in darkness at room temperature for 30 minutes. Previous attempts to derivatize native samples were unsuccessful due to a combination of poor reactivity of the spin label with the protected cysteine, long incubation times, and the inherent instability of the derivatized product. After incubation, the 2.5 mL of denatured protein was desalted over a Pharmacia PD-10 disposable column and eluted in 3.5 mL in the Gu«HCl buffer with 5 mM DTT to quench unreacted spin-label. The solution was diluted into 30 mL of refolding buffer (20 mM sodium phosphate, pH 6.5, 0.5 M NaCI, 0.02% sodium azide) and concentrated to 0.5 mL. Subsequently, the sample (at ca. 0.3 mM) was dialysed against refolding buffer containing 10% D 2 O . Chapter 5 — Intramolecular Regulation of DNA Binding 169 Figure 5.12 : Paramagnetic spin-relaxation of Ets-1AN301. The derivatized C350 in the H l / S l loop is coloured yellow. Amides resonances residing in the ETS domain and the C-terminal inhibitory region with unambiguously missing ^-H-^N resonances are coloured green. Amides residing in the N-terminal inhibitory region with unambiguously missing resonances are coloured magneta. Amides that are coloured white are either unperturbed by the TEMPO derivatization or are unknown. That is, due to spectral shifts upon derivatization, many amide from HI1, HI2 could not be unambiguously identified. This figure was generated using GRASP (Nicholls et al, 1993). 1 H, 1 5 N-HSQC spectra were acquired before and after derivatization. Unfortunately, the TEMPO-derivative perturbed the chemical shifts of Ets-1AN301 significantly, thus precluding a complete, unambiguous analysis of the relaxation effects of the spin label on the protein. Thus, data in Figure 5.11 must be viewed only in the "positive" sense. That is, coloured amides are close to the label, whereas white amides are unknown due to chemical shift degeneracy and incomplete assignments. Analysis of the two HSQC spectra confirmed the disappearence of C350 and 19 additional amides after spin labelling. In the ETS domain, missing amides were located in the H1/S2 loop where C350 resides, and portion of helix H2. Missing resonances from amides of L418 and L422 in the S4/H4 loop are plausible since this region is near C350. Most significantly, amides Chapter 5 — Intramolecular Regulation of DNA Binding 170 in two regions spanning residues 309-312 in helix HI1 and 324-327 in helix HI2 were rapidly relaxed by the spin label. In Figure 5.12, amides with missing resonances have been highlighted on a model of inhibited Ets-1. This model was generated to represent one possible packing scheme for the inhibition domain. Based on the spin-labelling data, the model appears to be correct in the assumption that the N-terminal inhibitory helices are packed in an anti-parallel manner. The model needs a certain degree of readjustment, however. The N-terminal inhibition helices, HI1 and HI2, should be brought closer to the HI/SI loop and residues in the S4/H4 loop should be remodeled to reflect their proximity to C350. The two cysteines in Ets-1AN331 (C350 in the ETS domain proper and C416 just outside the ETS domain) have been proposed to participate in redox regulation in vivo based upon the observation that human Ets-1 had a reduced DNA binding affinity when it was isolated from extracts of cells pretreated with serum containing 150 uM H2O2 (Wasylyk and Wasylyk, 1993). To support their redox hypothesis, the investigators reported that DNA binding could similarly be reduced in vitro by mutation of C350 to aspartic acid, alanine or serine or by modification with the reagents N-ethylmaleimide or diamide. The two murine Ets-1AN331 point mutants expressed in this study, C350N and C350A, contradict their mutational data since both mutants bound DNA as well as the wild-type protein (B. Graves and J. Petersen, unpublished data). Furthermore, no gross structural changes were detected in the 1 H, 1 5 N-HSQC spectra of corresponding regions in the various Ets-1 proteins studied. It is unknown how well the murine Ets-1 C350 point mutants are tolerated in the inhibited or full-length murine Ets-1. Upon TEMPO derivatization of Ets-1AN301, several ! H - 1 5 N amide resonances in the HSQC spectrum appeared significantly broadened and experienced chemical shift perturbations relative to an untreated control spectrum. Though Ets-1AN301 was clearly folded in both spectra, some destabilization may have occurred in the derivatized protein to accommodiate the bulky TEMPO adduct. Chapter 5 — Intramolecular Regulation of DNA Binding 171 5.5. Dynamics of Ets-1AN301 5.5.1. Introduction The large variance in linewidths observed throughout all Ets-1 spectra suggests that a wide range of motions are being sampled. Such conformational sampling is anticipated to achieve a tight DNA complex. The inhibitory region is also a good candidate to experience an interesting range of motions as it must be capable of locally unfolding upon DNA binding (Figure 5.7; Petersen et al, 1995). In this section, I will investigate the dynamics of the inhibitory domain and ETS domain through amide relaxation and amide hydrogen exchange studies of Ets-1 AN301. 5.5.2. Relaxation Theory All NMR experiments essentially occur in two parts, a series of RF pulses perturb a population of spins from equilibrium then a signal is measured which is dependent upon how the population redistributes itself over time. The redistribution process is the phenomenon of relaxation. As I briefly discussed in Chapter 3, relaxation in general can be described by two interrelated rates, Ti and T2. Ti describes an enthalpic change — an excited population of spins returns to equilibrium by exchanging energy with its surroundings. T2 describes an entropic change — an excited population of spins exchanges spin energy to create a loss of phase coherence that is manifested by a loss of the observable signal (Homans, 1992). Two fundamental rules are followed in either Ti or T2 relaxation processes, (i) The interaction that causes the relaxation must work directly on the spins and (ii) there must be a time dependency on the nature of the interaction. The material presented in this section is based upon a number of excellent discussions from Kay et al. (1989), Farrow et al. (1994), Barchi et al, (1994) and Mandel et al. (1995) and general overviews of the subject from Derome (1987), Homans (1992), Clore and Gronenborn (1993), Wagner (1993) and Peng and Wagner (1993). The mathematical roots of relaxation are found in the study of random processes to describe the behaviour of molecules tumbing in solution. The autocorrelation function, Chapter 5 — Intramolecular Regulation of DNA Binding 172 G(t), represents molecular motion in terms of how different the position of a given molecule is between one observation and another observation shortly after given by a time, t. Motion, in this case, refers to global diffusional and rotational processes associated with the molecule, as well as internal motions. Typically, the shape of G(t) is a decaying exponential — over longer time intervals, the relative orientation of a given molecule will be less similar. A time constant, xc, termed the correlation time, describes how fast the molecule is tumbling or more precisely how fast it is reorienting itself. Appended onto G(t) below is the function A(t) which describes the nature of the molecular motion. G(t) = A(t) • exp (-t / xc ) [5.1] The autocorrelation function is more meaningful when it is expressed in terms of frequency, rather than time. Fourier transformation of G(t), the autocorrelation function gives J(co), the spectral density function. The spectral density reflects the strength of a given frequency component at modifying the molecular motion. J(co) = ( 2 xc 11 + co^c2) • A(co) log (co) The shape of J(co) is Lorentzian. In practical terms,;as T c becomes slower than co, the function quickly maximizes at a value of 2xc»A(co).Thus, random processes act as a broadband source of frequencies up to (l/x c ) H Z . At frequencies faster than co, the function quickly approaches zero; in other words, the contribution of very fast frequency components relative to the tumbling time are negligable. A nucleus can favorably exchange energy with its surroundings when it experiences motions at its Larmor frequency, co. This, in essence, is Ti relaxation. From the representative plot of J(oo), it can Chapter 5 — Intramolecular Regulation of DNA Binding 173 be seen that when ic ~ co, the value of J(oo) changes dramatically. Experimental conditions such as field strength, temperature and molecular weight affect xc and therefore can be adjusted to improve the overall relaxation of the system being studied. Slow motions, such as those associated with a loop or hinge region, are responsible for additional contributions to T2 relaxation according to the following rationale. For a given slow motion, there will be a z-magnetization component that will be aligned with or against the applied magnetic field. Since all like spins will not be experiencing the same total magnetic field, the energy levels associated with the ensemble of like spins will be not be similar. In the rotating frame of reference, the relevant isochromat that describes the ensemble will begin to blur and lead to loss of signal. Since the loss of signal occurs in the transverse (i.e. the xy) plane, T2 relaxation is often called transverse relaxation. T2 relaxation may be further modified by chemical exchange processes. In dynamics studies of large proteins, the amide 1 5 N offers several advantages as a molecular probe: (i) Multidimensional experiments can be designed to exploit the chemical shift dispersion of 1 5 N and 1 H . (ii) The amide 1 5 N can provide direct information on the dynamics of the polypeptide backbone. (Hi) The amide is a two-spin system which simplifies the mathematics, (iv) The amide ^ N - 1 ! ! bond is essentially an invariant distance. Therefore, 1 5 N relaxation is dominated by the directly bonded 1 H and not the structure of the protein. Ti and T2 relaxation are described as follows in terms of the spectral density function from Kay et al. (1989): (1 / T i ) = d2[ J(coH-coN) + 3J(coN) + 6J(coH +CCN)] + c2 J(coN) [5.3] (1 / T2) = 0.5d2[ 4T(0) + J(COH-COn) + 3J(coN) + 6J(coH +CON ) +6J(coH) ] + (l/6)c2[4J(0) + 3J(coN)] [5.4] Chapter 5 — Intramolecular Regulation of DNA Binding 174 In the above equations, there are two constants, d and c, that represent dipolar and chemical shift anistropy contributions. d2=0.1[yH7hj(2Tt<l/rHN3>)]2 and c2 = (2/15)[YHBO(AC)] 2 . Dipole-dipole interactions arise when mutually tumbing nuclei affect each other through space. Chemical shift anisotropy (CSA) does not play a large role in the relaxation of quickly tumbing molecules since on a relative N M R time scale, the orientation of the shielding tensor (Aa) describing the ^ N - 1 ! - ! bond averages out. If the tumbling is slow enough, however, equivalence cannot be assumed. In these terms are parameters that reflect the dependence of Ti and T2 on bond lengths ( r H N) and the magnetogyric ratios (y) of the two nuclei involved. In addition to Ti and T2, the heteronuclear 1H{ 1 5N}-NOE is a sensitive indicator of backbone dynamics. Expressed in terms of the spectral density function, a quantitive description of the motions may be obtained (after Kay et ah, 1989): NOE = l + (YH/YN)d2[6J(coH + coN) - J(coH - <oN)] • T : [5.5] Ets-1AN301 A 11 10 9 8 7 A plot of the 1 H{ 1 5 N)NOE, and T 2 against x c, the overall correlation time for motion. This graph is reproduced from the calculations of Kay et al. (1989) using a 1 5 N frequency of 50.65 MHz, an order parameter, S2, of 1.0 indicating complete isotropic motion, and a N-H bond length of 1.02 A . Ets-1AN301 has a measured correlation time of 9.79 ns, indicated by the arrow in the figure. Since two different nuclei are being considered in the 1 H{ 1 5 N}-NOE, there is also a term describing the polarization enhancement, (YH/YN)- Since Y N has an opposite sign in relation to Y H / the 1 H { 1 5 N ( - N O E in the accompanying plot has an opposite sign in relation to a ^ H ^ H NOE. In addition, the accompanying plot demonstrates that the Chapter 5 — Intramolecular Regulation of DNA Binding 175 1H{ 1 5N}-NOE is most sensitive to flucuations in motion around T C . For fast motions, Ti equals T2, while for slow motions, T2 relaxation is dominant. In a qualitative sense, the 1H{ 1 5N}-NOE experiment is very useful since it allows random coil regions and very unrestrained regions to be identified by simply comparing the sign of the ^H, 1 5 N crosspeaks. At a 1 5 N frequency of 50.65 MHz used in this study, Kay et al. (1989) have calculated that the 1H{ 1 5N}-NOE can vary from -3.6 for CDNT; c « 1 to 0.82 for ( O N ^ C » 1- As shown in the plot, Ets-1AN301 has a calculated C0NT cof 3.1, placing it in the slow-motion limit portion of the graph. Therefore, in the absence of fast motion contributions, the observed 1H{ 1 5N}-NOE will be maximal and positive. However, for nuclei experiencing unrestrained fast motions, such as those found at the extreme N -and C-termini of a protein, the observed 1H{ 1 5N}-NOE will be near zero, or perhaps negative. The Tj, T2 and the heteronuclear 1H{ 1 5N}-NOE are measured in experiments detected via sensitivity enhanced gradient 1 H, 1 5 N-HSQC spectra. To extract relevant motional data, the spectral density function T(co) must be expressed in terms of a maximum of three variables since only three independent measurements (Ti, T2, 1H{ 1 5N}-NOE) have been made. According to the model-free approach described by Lipari and Szabo (1992a,b), the spectral density function is expressed by (i) an order parameter, S2, which reflects the amount of spatial freedom a ^ISPH vector has and (ii) a correlation time, x m . When internal motions are present, the total correlation time, T, is reconfigured to include the overall tumbling time, x m , and the effective correlation time for the internal motion, x e, where (1/f) = (1/Trn) + (lAe)- K 1 0) is expressed in the model-free formalism as: J(co) = Shm I (1 + co 2xm 2) + ( 1 - S 2 )T / ( 1 + ffl2T2 ) [5.6] The relaxation rate T2 is governed by a combination of dipolar interactions and chemical shift anisotropy. If T 2 cannot be accurately described by a combination of dipole-dipole and CSA contributions, an additional term reflecting chemical exchange (R e x) 1 S Chapter 5 — Intramolecular Regulation of DNA Binding 176 included. Thus, (1/T2) = (1/T 2 DD) + ( V T 2 C S A ) + R e x . In cases, where the inclusion of T E and R e x terms is not sufficient to accurately model the T i , T 2 and NOE data, an alternative spectral density function may be used (Clore et al, 1990). This alternative function accounts for a motion that is faster than the correlation time by an order of magnitude or less (Equation 5.7). J ( C 0 ) = S2xm / ( 1 + C 0 2 T M 2 ) + ( Sf2- S 2 )T / ( 1 + C 0 2 T 2 ) [ 5 . 7 ] Equation 5.9 incorporates two modifications to the standard model-free model. First, since a slow motion, x s, must be represented, (1/x) - ( l /x m ) + (1/T s). The correlation time for the faster motion is assumed to not contribute to the relaxation. Second, the order parameter, S2, must be represented by a product of fast and slow components, s2 = sf2.ss2. 5.5.3. Methods The pulse sequences used throughout the relaxation study were provided by the laboratory of Prof. L.E. Kay (U. Toronto) as described in Farrow et al. (1994) and similarly by Skelton et al. (1993). Al l pulse sequences employed gradient coherence selection and sensitivity enhancement, and were designed to ensure minimal water excitation and to eliminate cross-relaxation. Spectra were acquired at 28°C on one 0.7 mM sample of Ets-1AN301 in 20 mM sodium phosphate, pH 6.5, 0.5 M sodium chloride, 5 mM dithiothreitol, 0.02% sodium azide, 10% D 20. Since Ets-1AN301 has a tendency to aggregate above 0.7 mM, the sample was allowed to "cure" in the magnet for one day. The curing process was important since any precipitation during the data collection would superimpose an artificial decay profile over the Ti and T 2 exponential decays which were to be measured. Ti data points were collected out of order at 11, 56,122, 200, 333, 522, 788 and 1065 ms with one duplicate time point to facilitate error estimates. T 2 data points were collected in a similar manner using 16, 33, 50, 67, 84, 100 and 117 ms CPMG pulse trains. Heteronuclear NOE spectra were acquired with and without 3 seconds of 1 H saturation and a total recycle delay of 5 seconds. Chapter 5 — Intramolecular Regulation of DNA Binding 177 All spectra were processed in the NMRpipe (F. Delaglio, Nat'I Inst's. Health) environment with Lorentz-to-Gaussian apodization. Peak intensities were measure in semi-automated manner with STAPP (D. Garrett, Nat'l Inst's. Health). Ti and T 2 data points were fit to a two-parameter decaying exponential by conjugate gradient minimization using software provided by the Kay laboratory (U. Toronto). The goodness of fit was assessed by comparing %2 against a 95% confidence threshold. Typical data are shown in Figure 5.13. Note that due to the phase cycling scheme used in the TI measurements, the observed signal decays to zero. This reduces the observed signal-to-noise but alleviates the need to fit the data with a three-parameter expression. The error in duplicate Ti and T 2 measurements was 3.3% and 1.6%, respectively. The average values of Ti and T 2 for Ets-1AN301 were 599 ms ± 7.5 ms and 93.9 ms ± 4.2 ms, respectively. Peak intensity ratios between the 1H{ 1 5N}-NOE spectra collected with and without proton saturation were tabulated. The uncertainty of the NOE values (O"NOE) were estimated by incorporating the RMS noise (a) of the two spectra (calculated within the nmrDraw display software) into the following expression: The global correlation time, x m , of Ets-1AN301 was computed from residues (I335-V415) in the ETS domain. This subset was chosen because the structure of Ets-1AN331 showed that the ETS domain represented an autonomous, globular unit likely to have one distinct x m . Ti / T 2 ratios were calculated to be 7.241 ± 1.158 for the ETS domain and 7.183 ± 4.540 for all residues. Residues were selected from the ETS domain whose Ti / T 2 ratio was within 1.0 a of the mean value. A grid search on a two-time scale spectral density function (Equation 5.7) using (Ti,T2) and (Ti, T 2 , NOE) data from the 1.0 a ETS domain subset produced xm estimates of 9.79 ns ± 0.44 ns and 9.78 ns ± 0.50 ns, respectively. NOE = (I s a t - Iunsat ) / Wsat [5.8] ONOE= [ ( 0"sat / Isat ) 2 + ( 0"unsat/ W a t ) 2 ] 1 / 2 [5.9] 178 Figure 5.13 : Two-parameter monoexponential curve fitting of the T 1 and T 2 data. T a data points of residues Y424 • , D440 • and E370 • and T 2 data points of L314 • , E370 A and V411 • are shown. Chapter 5 — Intramolecular Regulation of DNA Binding 179 A total of 106 amides were measured out of a possible 139 amides in Ets-1AN301. The remaining amides represented those which were missing due to intermediate exchange or were in unresolvable clusters. Of the 106 amides, a total of 92 amides successfully met all of the criteria for one of the five dynamical models summarized in Table 5.2. The criteria used for model selection proceeded from the simplest to the most complex model as follows: (i) A model was discarded if all the parameters could not be fit to within a 95% confidence threshold, (ii) A model was discarded if any of the parameter errors exceeded the parameter values. (Hi) If a unique model had not yet been identified, the model with the lowest overall x2 value was selected, (iv) If two models remained with the same overall x2 value, the model with the fewest parameters was selected. 5.5.4. Results General Comments Graphs of the 1 5 N Tj, T 2 , 1 H{ 1 5 N}NOE data and of the various model fitting parameters are depicted in Figures 5.14 and 5.15. Of the 92 amides measured, 27 could be fit to the simplest model (S2 and xm). Overall, for the amides which could be measured, Ets-1AN301 has consistently high S2 values (1.0 < S2< 0.8) with the exception of one region at the end of helix HI. On their own, S 2 values can only comment on the angle, 0, around which the 1 5 N - 1 H bond vector reorients. As S 2 decreases from 0.95 to 0.80, 0 increases from 10.5° to 22° (Torchia et al, 1994). In the absence of any other motions, an order parameter of 0.80 can also be accommodated by small changes around the mean 1 5 N - H bond length of 1.05 A (Torchia et ai, 1994). Inspection of the similarly consistent Ti values suggests that as an ensemble, Ets-1AN301 behaves as one discrete unit. The observed 9°C increase in T m , relative to Ets-1AN331, from the circular dichroism studies further supports this claim that the inhibition domain has more qualities of being an integral unit that it does of being an appendage to the ETS domain. x m , the overall correlation time, was determined to be 9.79 ns. At present, there is no analytical ultracentrifugation data or detailed structural data to 280 Table 5.2 : Summary of the model-free dynamical parameters Model Description 2 simplest model; assumes isotropic motion of the N-H vector with the rest of the molecule S2,xe a fast motion affects x m ; (1 A) = (1 A m ) + (1 A e ) S^Rgx the amide experiences chemical exchange broadening S 2 / x e ,R e x fast motions and chemical exchange affect relaxation Amotions occur at two time scales differing by at least one S2,Sf2/xs order of magnitude. The faster component is assumed to contribute negligably to the overall relaxation. References: Lipari and Szabo (1982a,b); tClore et al. (1990a,b) 181 1.00 --0.75 1.2 -1.0 -0.8 -* — ' 0.6 T— H 0.4 | 0.2 i 0.0 1 o CO o IN CO o o o co in co co co Figure 5.14 : 1 5 N Tv T 2 and ^{^NJ-NOE plots for Ets-1 AN301. Amide resonance overlap, missing amide resonances due to exchange on an intermediate timescale and proline residues contribute to the missing entries in the plots. 310 320 330 340 350 360 370 380 390 400 410 420 430 440 14 H ' CO 310 320 330 340 350 360 370 380 390 400 410 420 430 440 0.1 I i i ' T 111 i i i 111 i i I T l ii i i i iiiii 1 i T l T T l II 1 1 III 1 1 111 i 1 i 1 1 1 1 1 i i I i i I ' lTi i f 1 1 1 1 ITl 1 <T 1 iTl | 1 iT l i 1 1 i i T . il 1 1 1 1 1 i 1 1 1 1 1 1 1 1 | II 1 1 1 1 1 II l 1 ITl 1 i 1 1 1 1 310 320 330 340 350 360 370 380 390 400 410 420 430 440 0.5 • • ! i l l 1 T 'I I i i n i i i | i i i i i i i i i | TTTTTTTTTTTT "ITI n II 111 II 310 320 330 340 350 360 370 380 390 400 410 420 430 440 residue number Figure 5.15 : Model free parameters for Ets-1AN301 Chapter 5 — Intramolecular Regulation of DNA Binding 183 indicate any contribution of rotational anisotropy to x m . The correlation time is slightly longer than that reported for other proteins in this molecular weight range, suggesting possible aggregation of the protein. In the following sections, I will discuss the dynamical aspects of Ets-1AN301 in terms of several local regions of interest. Amino and Carboxy Termini From structural studies of Ets-1 AN331, the last four residues (P437 to D440) appear to be unrestrained in solution. This is clearly the case, as the residues near the C-terminus have long 1 5 N Ti and T 2 relaxation times, negative 1H{ 1 5N}-NOE values and low S 2 values. The final six residues of Ets-1AN301, which comprise part of helix H4, were fit to a two order parameter model with a higher frequency component in the order of 1 ns. In the structure of Ets-1AN331, the extreme C-terminus is also disordered. Ets-1AN301 is a deletion mutant especially designed to begin at the first occurrence of secondary structure (helix HI1). Only G302 has an extremely low order parameter and a negative NOE value and a higher than average T 2 indicating fast, loosely constrained motion at the helical N-terminus. The ETS Domain The ETS domain is composed of three helices and a four-stranded anti-parallel (3-sheet. Underlying its role as a structural scaffold, the simplest spectral density function was sufficient to describe the dynamics of most of the p-sheet. A few residues, spanning 1400-H403 in strand S3 do have chemical exchange and sub-nanosecond to nanosecond motions attributed to them. Perhaps a degree of conformational sampling is required to allow the ETS domain to form a stable complex with DNA. The v-Ets oncoprotein has a 1401V point mutation which alters its in vivo activity relative to wild-type (Leprince et ai, 1994). Furthermore, a H403D point mutation confers a temperature-sensitive phenotype to the E26 virus and to Elk-1 constructs (Golay et al, 1992; Janknecht et al, 1994). It is Chapter 5 — Intramolecular Regulation of DNA Binding 184 therefore tempting to conclude that mutations in this region may not only be distruptive to the hydrophobic packing of the ETS domain but also disruptive to the dynamic character of this region. A wide variety of motions describe the loop regions of the ETS domain. As such, the HI/SI loop and S1/S2 loops could not be adequately described by any of the five spectral density functions. The complex dynamics observed for these two loops may arise from a form of conformational crosstalk between the ETS domain and the nearby inhibitory module. The S2/H2 loop appears to be quite restrained highlighting the many important contacts it makes to the ETS domain hydrophobic core. The H2/H3 loop, which forms the "turn" of the helix-turn-helix and is also a putative nuclear localization signal, has high S 2 values indicating a degree of local order. However, this turn is blurred in the superimposition of the ensemble of Ets-1AN331 structures. Therefore, this apparent "motion" described by the ensemble of structures is due to the lack of NOE distance restraints in that region. Chemical exhange terms (Rex) as high as 12 sec 1 for K379 in the H2/H3 loop suggest that it interacts with the solvent. The H3/S3 loop and S3/S4 wing remains very dynamic in the inhibited ETS domain, undergoing conformational exchange in the microsecond to millisecond range. The model free parameters of helix HI and helix H3 (the recognition helix) are described by S 2 values of ca. 0.90 and fast internal motions (ie) in the order of 50 - 150 ps. An S 2 values of ca. 0.90 and a x m paramenter was sufficient to model the relaxation characteristics of helix H2. Several hydrophobic contacts are made between helices HI and H3 and several more indirectly via the (3-sheet. The similarity of the complexed human Ets-1AN320AC415 structure of Werner et al. (1995b) and Ets-1AN331 uncomplexed structure suggest that only minor conformational changes occur. Can helices HI and H3 also be further linked in a dynamic sense? Since helix HI is also a component of the inhibition domain, is structural flexibility necessary to propagate the signal to the inhibition domain from helix H3 upon sensing the minor conformational changes induced by DNA binding? To fully answer these questions, the dynamical Chapter 5 — Intramolecular Regulation of DNA Binding 185 profile of complexed Ets-1 AN301 will have to be investigated. The Inhibition Module A rich set of motions is evident throughout the inhibition module. This survey divides the inhibition module into six regions: (i) the first inhibition helix HI1 (G302-K311), (ii) a flexible, extended region (D312-P323), (Hi) the second inhibition helix HI2 (A324-T330), (iv) a flexible loop connecting HI2 to helix HI of the ETS domain(G331-I335), (v) helix HI of the ETS domain and (vi) helix H4 of the C-terminal inhibitory region. Helix HI1 of the inhibition module serves the primary role in the proposed mechanism of Ets-1 intramolecular inhibition (Figure 5.6). It is this helix that is observed to unfold when the ETS domain binds DNA (Petersen et al, 1995). To accomplish the necessary unfolding and refolding, helix HI1 would likely need to sample many conformations. In terms of energy, the barrier separating the folded and unfolded state must be low. Examination of the order parameters for helix HI1 in uncomplexed Ets-1AN301 suggests that helix HI1 is a bona fide helix — a view supported by the extensive short and medium range sequential H«-H N and H N - H N NOE's observed in NOESY-HSQC spectra. However, significant chemical exchange and fast motions are also observed throughout the helix suggesting that some conformational sampling is indeed occurring. A region spanning D312-K318 is described by low order parameter values and fast motions of 1.0-1.5 ns. Together, these observations suggest that this region is very flexible and assumes an extended conformation. Consistent with slow T 2 relaxation (100-150 ms vs. an average of 80 ms), the ! H - 1 5 N lineshapes throughout this region are sharper. As Ti values are consistent with those of the entire molecule, this region is still associated with Ets-1AN301. This is in contrast to the extreme C-terminus (P437-D440) which must undergo very unrestricted motion. The prolines (P319 and P323) in this loop may damper motions enough to place the remainder of this region in an exchange regime that precludes observation. Chapter 5 — Intramolecular Regulation of DNA Binding 186 Helix HI2 is an a-helix, being so abundant in alanine residues. As discussed earlier in this chapter, attempts were made to gain distance restraint information from NOE spectra of 1 3 C labelled alanine methyls but failed due to poor signal-to-noise. Poor signal-to-noise, in fact, is observed throughout the entire helix suggesting that this helix may have peculiar relaxation properties. However, the Ti , T 2, 1H{ 1 5N}-NOE data for helix HI2 was not significantly different than any other region of Ets-1AN301. Consequently, helix HI2 is described in terms of the model-free parameters by high S 2 values and no fast motions relative to the correlation time, xc, of Ets-1 AN301. The prevailing explanation for the poor spectral quality of helix HI2 is therefore a global slow motion on the microsecond timescale that cannot be identified by the model free approach. In terms of subdomain motions, as was previously discussed in Chapter 4 for the ETS domain, the extent that the five alanines in helix HI2 can interdigitate into the structure is relatively low. Thus, a translating, global motion for HI2 is plausible (Gerstein et al, 1994). Chemical exchange and fast motions both contribute to 15N-relaxation in the HI2/H1 loop. Specifically, chemical exchange is extreme at 1335, at ca. 10 s_1. Fast motions relative to the correlation time of Ets-1 AN301 range from 1.5 ns (S332, G333) down to 100 ps (G331,1335). The dynamic nature of this loop is further reinforced by a paucity of long-range distance NOE restraints. Helix HI is common to both the ETS domain and the inhibition module. This a-helix is described by very fast motions of ca. 100 ps. throughout its length. In the previous section (, a dynamical connection between helix HI and helix H3 was considered as both helices share the same fast global motions. Extending this idea further, these motions may be a consequence of the coupling of helix HI to the inhibition module. In future studies, it would be interesting to obtain dynamical information on Ets-1AN331 to discern the effects of the ETS domain and the inhibition module on the dynamics of this important helix. Chapter 5 — Intramolecular Regulation of DNA Binding 187 Virtually no secondary motions are attributed to residues V415-K436 suggesting that this region may serve as a scaffold or interface for inhibition helices HI1 and HI2 and the ETS domain helix HI. Few intermolecular contacts between helix H4 and other helices of the inhibition module and the ETS domain are observed. Few contacts combined with the relatively short length of this helix (ca. 6 residues) are likely contributors to instability in this region, but not to the extent taht was observed for helix HI1. This instability of helix H4 is manifested by lower order parameters and fast 1.0 ns motions extending from D434-D440. To summarize, the inhibition domain is described by many interesting motions which are a likely prerequisite to its function. These motions, in turn, explain the inability to define a unique structure of Ets-1AN301. Of the various secondary structural components of the inhibition module, helix HI1 is the most interesting as it is predicted to unfold upon DNA binding by the ETS domain. In the following section, I will discuss the results of experiments designed to probe the stability of helix HI1 by its susceptibility to amide hydrogen exchange. 5.6. A Hydrogen Exchange Study of Ets-1AN301 5.6.1. Introduction The resistance of hydrogen bonds in proteins to hydrogen exchange is a general indicator of stability (Englander and Kallenbach, 1984). Hydrogen exchange may be acid or base catalysed; the direct base catalysed reaction is shown in Figure 5.16 (Connelly et al, 1993; after Eigen, 1964). \ r A CONH +OH- JCONH OH-===CON- H O H j = = = CON" + H OH Figure 5.16 : Base catalysed proton abstraction. K D is a diffusion limited rate constant (10~10 M^sec"1). is the disassociation constant of the amide proton. Chapter 5 — Intramolecular Regulation of DNA Binding 188 The rate, kp r o t , describes the rate at which an amide in a protein exchanges with the solvent. Three main parameters influence k p r o t : (0 the secondary and teritary structure of the protein which establishes hydrogen bonding to the amide, (ii) the primary structure of the protein which exerts a local, nearest neighbour effect, and (Hi) the solvent pH, temperature and ionic strength. A "protection factor", P, is often quoted to assess the contribution of secondary and teritary structure to kp r ot- Thus, P = k r c / k p r o t [5.10] Here, k r c is the random coil exchange rate for an amide within a given polypeptide sequence, previously characterized by Bai et al. (1993). The intrinsic random coil amide exchange rate , k r c , is comprised of the individual acid-catalysed, base-catalysed and water-catalysed exchange rates and has been measured for each nearest neighbour pair of amino acids: k r c = kH+[H+] - koH-[OH"] + Kwater [5.11] As such, amide hydrogen exchange is directly related to pH. k'prot = kpro f 10 (pH'-pH) [5.12] In practical NMR terms, a solution pH of > 7 is generally avoided to limit the effects of amide hydrogen exchange. However, the introduction of pulsed field gradient technology has largely allieviated the detrimental effects of exchange by ensuring that water remains unexcited throughout the experiment. "Free" amide exchange rates are dependent upon primary sequence; hence, nearest neighbours account for a wide range of rates extending from 0.1 to 30 sec"1 at pH 6.5 (Molday et ai, 1972; Connelly et al, 1993; Bai et al, 1993). When embedded within a protein structure, the lifetime of an amide proton can range anywhere from less than one Chapter 5 — Intramolecular Regulation of DNA Binding 189 seconds to years (Figure 5.17). A stretch of five amides (L341-L345) in helix HI of the ETS domain were identified as very slow exchangers (no appreciable change in peak intensity after two days) in the previously discussed amide hydrogen/deuterium exhange experiments (Figure 3.10; Donaldson et ai, 1994; Liang et al, 1994; Skalicky et al, 1996). Twenty-five other amides with sufficient protection such that their amide exchange rates could be measured directly by 1 H, 1 5 N-HSQC spectra recorded after transfer to D 2 O . A different approach used in this section allows measurement of hydrogen exchange rates in the range of 0.1 to 20 s"1 by extracting the information from NOE and ROE chemical exchange experiments (Spera et al., 1991; Grzesiek and Bax, 1993). Chemical Exchange Rate (in s~1) 1 0 " 5 1 0 " 4 1 0 " 3 1 0 " 2 1 C T 1 1 1 0 1 1 0 2 1 0 3 • M ^ f D2O exchange f HX in this study / ...by 1D methods ...unprotected amide Figure 5.17 : Amide proton exchange can occur over a wide range of rates. 2D-HSQC D^O-transfer methods are effective at measuring the kinetics of slow amide hydrogen exchange. Stop-flow ID experiments extend this range. An approach originally described by Grzesiek and Bax (1993) is used in this study to detect faster amide exchange rates. The range of predicted unprotected amide exchange rates of Ets-1AN301 at pH 6.5 is also shown (Baiefa/., 1993). 5.6.2. Theory and Experimental Methods: NOESY and ROESY experiments measure the true NOE between the coupled 1 H and 1 5 N of the amide as well as chemical exchange processes including amide exchange. The acquisition of both NOESY and ROESY spectra allows the dipole-dipole (NOE) and amide hydrogen exchange information to be separated based on the different effects of dipolar and exchange terms in the two experiments. Chapter 5 — Intramolecular Regulation of DNA Binding 190 The decay in longitudnal (dM z/dt) or transverse magnetization (dMy/dt) is affected by by amide hydrogen exchange embodied in the rates kR and k;\[ (Grzesiek and Bax, 1993). The rates, kR and k^, represent the rate of exchange between the amide and water in ROESY experiments (Equation 5.13) and NOESY experiments (Equation 5.14). The ROESY experiment represents the simpler case as p 2 equals T i p , the relaxation rate of protons in the rotating frame. The NOESY case must take into consideration that pi is a sum of the amide proton spin-flip rate (Ti z z) and the Ti relaxation rate of the 1 5 N nucleus. The quantities, M z ° and M z w a t e r , represent the z-magnetization of the water at thermal equilibrium and during the experiment, respectively. kR can have different signs depending on whether amide hydrogen exchange or dipole-dipole interactions are dominant. kj\j is always positive, leading to the two extreme situtations below: kR = kN — for exclusive amide hydrogen exchange [5.15] kR - -2k]sj — for exclusive dipole-dipole interactions [5.16] The rates, dM z /d t and dMy/dt, were derived from experiments provided by Prof. L.E. Kay (U. of Toronto) on one uniformly 15N-labelled Ets-1AN301 sample used earlier for the relaxation series. Individual amide intensities were measured in using a semi-manual method and fit to a decaying monoexponential function according to the method described in Section 5.4.2. Delays of 0, 20, 40, 60, 90 and 150 ms were used to derive dM z /d t and delays of 2,5,10,15, 20 and 35 ms were used to derive dM y /dt . Equilibrium and experimentally available water magnetization ratios for the ensuing dMy/dt = p 2 M y - k R (M y - M y w a t e r ) [5.13] dM z /d t = pi(M z - M z°) - kN(Mz- M z w a t e r ) [5.14] Chapter 5 — Intramolecular Regulation of DNA Binding 191 NOE experiment was determined to be 64.4% (parallel) and 77.9% (anti-parallel), respectively. Similarly, values of 65.9% (parallel) and 74..2% (anti-parallel) for the ROE experiment were determined. NOE and ROE pulse sequences resemble one 2D block of 3D ROESY-HSQC experiment. Mixing times of 65 ms and 35 ms were chosen, respectively, for the NOE and ROE experiments. Selective water pulses were used to minimize the number of H a resonances which reside near the water resonance at 4.72 ppm from being excited. Grzesiek and Bax (1993) used purge pulses and gradients on a uniformly 1 5 N / 1 3 C labelled sample to eliminate H " contributions. However, a 1 5 N / 1 3 C labelled sample was not available so a long ROESY mixing period was alternatively chosen. H a contributions are attenuated because they have a much shorter Tip than the mixing time (Grzesiek and Bax, 1993). The pulse sequences were designed such that the two experiments needed to calculate the NOE or ROE were interleaved into one experiment. £NOE and c^ ROE were calculated and normalized according to Equations 5.17 and 5.18. M z + denotes that water is parallel to the amide proton magnetization. Programs were written (calculateHX_noe and calculateHX_roe.c) to determine k>j and kR according to the relationship derived by Grzesiek and Bax (1993): c^OE= ( M z + - M z - ) / ( M z + + M z - ) CROE = ( My + - My" ) / ( My + + My" ) [5.17] [5.18] k N = 2CNOE.(p 1 +kN) { Mz(0) + [exp((p1+kN)t)-l]MZQ} [5.19] [f+tKNOE) - f-(l+CNOE) + 2^0E].[exp((p 1+k N)t)-l].M z° 2CROE.(P2+kR)-My(0) [5.20] [f+(l-c^OE) - f-(l+CROE)Hexp((p2+kR)t)-l].My< o Chapter 5 — Intramolecular Regulation of DNA Binding 192 5.6.3 Results and Discussion A table of kR and kjsj values is presented in Appendix A.2. As exchange rates increase exponentially with respect to the change in pH, the kR and kjsr exchange rates are reported at both pH 6.5 and pH 7.5 to aid identification of amides which experience fast exchange in the regime of 0.1-20 s"1. kR , the rate of amide magnetization exchange with water is presented graphically in Figure 5.18. A rapidly exchanging hydroxyl proton can mimic the large, negative ROE expected for the interaction of an amide with bound water (Otting et al, 1991; Grzesiek and Bax, 1993). This is likely the case for residues S352, D357, G358, D359, E387, L418 (near D417) and E428. The most striking result of this study is that significant rates of amide hydrogen exchange were observed for the region spanning residues G302-D317. This region includes inhibition helix HI1 and six residues that adopt an extended structure. Recalling the 1 5 N relaxation study, helix HI1 is described by high order parameters, appreciable amide exchange rates and a fast motion (0.1-1.0 ns) component. Though kN and kR values were relatively low at pH 6.5, significant exchange was observed at pH 7.5. Thus, compared to the other helices in Ets-1 AN301, helix HI1 has a significantly higher amide exchange rate and a rather limited stability. As discussed in an upcoming section, a mutagenesis study would be warranted to determine the contributions of the sequence and intermolecular contacts to the stability of helix HI1. Table 5.3 lists the protection factors for regions of Ets-1AN301 experiencing fast amide hydrogen exchange. A region of six residues spanning K311-K318 has all the qualities of being highly unrestrained. From the earlier structural study, both sequential H N i -H N i+i and H a j -H N;+i were observed combined with no apparent intermolecular long-range amide NOE's. The relaxation study corroborated this structural assignment with low order parameters and higher-than-average T 2 relaxation rates. Protection factors were observed to be near or less than 1.0, conclusively demonstrating that the amides 2 9 3 ETS domain S1 S2 H2 O Q Q •>~ C\| CO CO CO CO HX residue number Figure 5.18 : Summary of Ets-1AN301 fast amide hydrogen exchange. The method of Grzesiek and Bax (1993) used in this section to measure fast amide hydrogen exchange rates is sensitive to rates from 1-20 sec1. In this figure, k R, the first-order rate constant for magnetization exchange during the ROESY mixing period, is plotted versus residue number at pH 6.5 (hollow vertical bars) and pH 7.5 (solid vertical bars). Missing entries are due to proline residues, amides in intermediate exchange, amides that could not be measured due to resonance overlap, and amides that have a negative value of k R . At the inset marked HX, Hollow squares denote amides that exchange very slowly as identified from amide hydrogen-deuterium exchange experiments (Chapter 3). Filled squares denote amides that experience exclusive fast hydrogen exchange; fulfilled by the criteria: (kR > 0), (0.9 k N < k R < 1.1 k N ) and (1 sec"1 < k R < 20 sec1). These residues map to a dynamic extended region following inhibition helix HI1, a loop region connecting helices HI2 and Helix HI (of the ETS domain), and the turn of the HTH motif (between helices H2 and H3). Protection factors these fast exchanging amides are listed in Table 5.3. Secondary structural elements of Ets-1AN301 are depicted above the graph for reference (horizontal bars, a-helices; arrows, B-strands). 194 Table 5.3 : Regions of Ets-1AN301 with significant* amide hydrogen exchange at pH 6.5 Location Residue k^lsec1]* kR [sec-1] pb Helix HI1 G302 17.0 ±1.7 13.8 ±0.6 1.2 ±0.1 Loop HI1/HI2 R311 5.5 ±0.5 1.7 ±0.2 3.2 ± 0.5 D313 4.7 ± 0.5 4.9 ± 0.2 1.0 ±0.1 L314 1.2 ± 0.1 1.9 ±0.1 0.6 ±0.1 N315 13.0 ± 1.3 5.2 ± 0.5 2:5 ± 0.3 K316 13.0 ±1.3 7.6 ± 0.4 1.7 ±0.2 D317 4.7 ± 1.3 4.6 ± 0.2 1.0 ±0.3 Helix HI2 Y329 5.5 ± 0.5 0.8 ± 0.2 6.9 ± 1.8 T331 6.6 ± 0.7 1.7 ±0.2 3.9 ±0.6 S332 20.0 ± 2.0 5.2 ± 0.3 3.8 ± 0.4 Loop HI2/H1 G333 24.0 ± 0.5 0.8 ± 0.2 30.0 ± 8.1 1335 25.0 ± 0.7 1.7 ±0.2 14.7 ±1.8 Q336 20.0 ± 2.0 5.2 ± 0.3 3.8 ±0.4 Turn HTH (H2/H3) M384 8.9 ±0.9 2.4 ±0.2 3.7 ±0.5 Helix H3 Y386 7.7 ± 0.8 4.0 + 1.4 1.9 ±0.7 Y395 6.2 ± 0.6 2.7 ±0.4 2.2 ± 0.4 Wing HI2/H1 V411 1.5 ± 0.2 6.4 ± 0.6 0.2 ± 0.0 *Significant amide hydrogen exchange follows the criteria outlined in Figure 5.18. (1.0 sec-1 < k R < 20 sec"1) and (0.9 k N < k R < 1.1 k N). flThe value, k r c, is the calculated hydrogen exchange rate of an amide in a random coil conformation based solely on primary structure effects according to Bai et al. (1993). Errors are estimated at ± 10%. ^The protection factor, P, equals k r c divided by the observed chemical exchange rate, k R. Chapter 5 — Intramolecular Regulation of DNA Binding 195 comprising this region are completely unprotected from hydrogen exchange with the solvent. Residues 1335, Q336 and L337, which are located immediately N-terminal to helix HI of the ETS domain, also demonstrate protection factors near unity (Figure 5.18). Of these three residues, a R e x value was only.assessed to 1335 (10 sec"1; Figure 5.15). As these residues must undergo a conformational change upon DNA binding to allow L337 and W338 to contact DNA, a degree of structural plasticity, as demonstrated by the amide hydrogen exchange and the relaxation experiments, is reasonable. 5.7. Characterization of the Ets-1AN301 / D N A complex A mechanism of autoinhibition cannot be constructed without knowledge of the structure of the complexed form. To this end, a complex of uniformly 15N-labelled Ets-1AN301 and a 16 bp oligonucletide duplex (5'-GCCAAGCCGGAAGTGT-3') was studied. The GGAA core binding site is asymmetric in this oligonucleotide based on DNA footprinting studies (B. Graves, unpublished data). To establish the feasibility of characterizing the Ets-1 AN301/DNA complex, a series of solvent conditions were assayed ranging from pH 6.0-7.0 and 50-250 mM sodium chloride (at 1.5 mM protein/DNA in 20 mM sodium phosphate, 1 mM DTT, 1 mM EDTA, 0.02% sodium azide). The DNA:protein ratio in solution was slightly > 1.0 as determined by examining the number and relative intensity of the DNA imino resonances by ID jump-return spectra. Given that the complex is in excess of 27 kDa, the linewidths were characterically broad in 1 H, 1 5 N-HSQC spectra collected at 28°C. The linewidths markedly improved at temperatures greater than 35°C but unfortunately at the expense of stability. Several 1 5N-NOESY-HSQC spectra were collected at 35°C and 40°C for assignment purposes; however, none of the spectra demonstrated a sufficient signal-to-noise level to permit sequential assignment. Chapter 5 — Intramolecular Regulation of DNA Binding 196 A large number of amide resonances shifted significantly in the 1 H, 1 5 N-HSQC spectrum of Ets-1AN301 upon complexation as was originally observed in selective labelling experiments of Ets-1AN331 (Figure 4.14). The W338 indole H e l resonance is notable, shifting to nearly 12 ppm in *H to reflect its intercalation in the DNA (Werner et al., 1995b). More significantly, however, was the appearance of many resonances in the center of the HSQC spectrum (ca. 120 ppm in 1 5 N and 8.0 ppm in 1 H). These are attributed to extended conformations that were not found in the uncomplexed 1 5 N -HSQC spectrum. A count of the amide, amino and indole H e l resonances in the 1 5 N -HSQC spectrum of the complex suggested that approximately twenty resonances were absent due to intermediate exchange effects. In contrast, Werner et al. (1995b) reported virtually complete assignments for the Ets-1AN320AC415/DNA complex. These missing resonances in Ets-1AN301 were not due to hydrogen exchange since water flip-back and pulsed field gradients in this HSQC pulse sequence strive to minimize water excitation. In addition, these resonances were also missing at pH 6.0, again, where chemical exchange is minimized (not shown). Therefore, the poor signal-to-noise and the missing amide resonances associated with Ets-1AN301 are due to a combination of the considerable molecular weight of the Ets-lAN301 complex, a lower working concentration (0.6 mM for Ets-1AN301 vs. 1.1 mM for Ets-1AN320AC415) and the proposed the helical unfolding of the N-terminal inhibition helix, HI1. To discern resonances of amides in extended conformations from those resonances of structured amides which happen to reside in the middle of the spectrum, a heteronuclear 1 H( 1 5 N}-NOE experiment was performed. The heteronuclear NOE spectrum demonstrated that several amides in the central cluster of resonances were highly mobile, and thus, in a flexible conformation (Figure 5.19). Since these resonances were not found in the ! H , 1 5 N -HSQC spectrum of uncomplexed Ets-1AN301, it is tempting to speculate that these resonances correspond to an unfolded helix HI1. 12 11 10 B 2 9 7 " W338HE1 - 110 (aliased) • 0 0 o 0 0 0 0 - / . •' 0 - 115 0 0 c '•' Oooo ^ • 0 Of ~. - 120 0 On 0 5 , - 125 o'» ' o 0 0 0 0 - 130 1) 0 • in h 110 h 115 A439 4.V 'A 5* Q_ h 120 Q_ D438 / $ \ , D440 K436 / h 125 h 130 12 11 10 9 8 7 1 H P P M Figure 5.19 : 1H{ 1 5N}-NOE study of the Ets-1 AN301 / D N A complex. (Panel A) Spectrum of the complex acquired without 1 H saturation (Iunsat)- (Panel B) Difference spectrum of the complex (Iunsat - Isat)- Amide signals in the spectrum shown in Panel B, correspond to 1 5 N which are experiencing fast motions relative to x m / the overall molecular correlation time. These inlcude the five C-terminal residues of Ets-1AN301 (K436 - D440, residue 437 is proline). A boxed region denotes a cluster of ca. 10 amides that have random-coil chemical shifts and also experience fast motions. These resonances are not present in uncomplexed Ets-1 AN301. The boxed resonances may represent the amides of inhibition helix HI1, which unfolds upon DNA binding. An asterisk denotes resonances belonging to a low molecular weight protein impurity in the Ets-1AN301/DNA sample. Chapter 5 — Intramolecular Regulation of DNA Binding 198 5.8. A Possible Mechanism for Inhibition The elucidation of any structural mechanism is a complicated task requiring high-resolution structural information assisted by the knowledge of functionally important residues. CD, proteolysis, and preliminary NMR data on the inhibited complex all suggest that a region is unfolding upon complexation (Figure 5.7; Petersen et al., 1995). Helix HI1 is favored to unfold since it is very flexible and is strongly affected by hydrogen exchange in a pH dependent manner. In the absence of structural data from NMR or crystallographic sources of the inhibited, bound state, the identity of the unfolded residues will remain uncertain. However, proteolysis studies have identified at least some unfolding in helix HI1 (Petersen et al, 1995). The structures of the ETS domains of human Ets-1 by Werner et al. (1995b) and of Fli-1 by Liang et al. (1994b) confirm that major conformational changes in the ETS domain do not occur upon DNA binding. In this study, the ETS domain is shown by 1 5 N relaxation and amide hydrogen exchange to be relatively stable, with some fast motions in helices HI and the recognition helix, H3. In contrast, the inhibition module is very dynamic, undergoing motions ranging from 50 ps to milliseconds. Together, these structural insights form a framework to explore a mechanism. The proposed mechanism that follows relies on the following assumptions: (i) the topology of the inhibition domain is described by an anti-parallel helical bundle and inhibition helix HI1 unfolds upon DNA binding; (ii) both the inhibited and the uninhibited forms bind DNA at nearly the same rate; (Hi) the rate of conversion between the bound, inhibited complex and the fully-active complex (Ets^DNA -> Ets*-DNA in Figure 5.20) is fast and concerted with the events occurring in the ETS domain; (iv) A 20-fold reduction in affinity is observed from quantitative bandshift assays (Petersen et al., 1995). This inhibitory effect is due to a decrease in off-rate driven by the refolding of helix HI1 and relief of other stresses imposed by the inhibition domain on the complex. Chapter 5 — Intramolecular Regulation of DNA Binding 199 (Figures 5.20, 5.21). (v) The conversion from an unfolded helix HI1 to a folded helix HI1 is likely concerted process with disassociation of the protein from the DNA. Ets + DNA Ets*+ DNA Ets—DNA Ets*—DNA Ets = [Ets-1 AN301] Ets* = [Ets-1 AN331] or de-repressed [Ets-1 AN301] Figure 520 : Upon formation of the complex (Ets-DNA), helix HI1 unfolds to produce the fully active, high-affinity form (Ets*). Since Ets-1AN331 does not possess the inhibitory activity, complex formation in this case may be considered as a two state equilibrium between Ets* + DNA and the Ets*-DNA complex. The on-rates of both the inhibited and uninhibited forms are faster than diffusion (M. Jonsen and B. Graves, unpublished data). This suggests that the ETS domain binds non-specifically to DNA and slides along to its promoter (Figure 5.22). a \ b * — *^  c ^— non-specific ETS binding site Figure 5.22 : The basic patch on the ETS domain can allow it to quickly orient itself on the DNA and slide to an adjacent Ets binding site, (a) seeking DNA by electrostatic interactions (b) non-specific binding (c) recognition of a 9 bp ETS binding site followed by unfolding of helix HI1 upon binding. The DNA binding surface specified by the HTH motif is basic and could therefore promote an interaction with the predominantly acidic DNA duplex. An electrostatic surface of Ets-1 AN331 is illustrated in Figure 4.8. 200 b uncomplexed complexed reaction coordinate Figure 5.21 : (a) In the proposed mechanism of inhibition, both the inhibited (Ets-1AN301) and uninhibited (Ets-1 AN331) forms non-specifically bind DNA at the same rate, limited only by diffusion. The starting free energies of the uninhibited and inhibited forms have been arbitrarily set to equal (b) The rate-limiting step is the association of either form with DNA. Upon DNA binding, the unstable inhibition domain unfolds to form the high-affinity, fully bound state. The dynamic nature of helix HI1 reduces the energy barrier associated with the unfolding transition and thus reflects only a 2-fold difference in the formation of the bound state (c) The bound state for both proteins are energetically favorable; however, the unfolding of the inhibitory module destabilizes the bound state of Ets-1AN301 by increasing the energy barrier for the reverse dissociation reaction. This leads to a 10-fold increase in the disassociation rate. Chapter 5 — Intramolecular Regulation of DNA Binding 201 In this proposed allosteric model, conceptually organized into steps, the first specific interaction between the ETS domain and an ETS binding site likely begins with the seating of the recognition helix (H3) into the major groove. Side chains of R391, G392 and R394 hydrogen bond with key bases (Figure 5.23). N380 in the turn of the HTH motif also makes a contact at this point. For reference, DNA contacts and the structure of the human Ets-1 ETS domain are provided in Figure 4.12 and Figure 4.13, respectively. The fast motions observed through helix H3 may be indicative of its ability to accommodate the numerous protein-DNA interactions. Water molecules are likely expelled at this point from the protein-DNA interface. Figure 5.23 : The first specific contacts made by the ETS domain to DNA likely occur at the major groove. Key residues are shown. The placement of a-helices (tubes) and p-strands (arrows) in this illustration approximate their respective positions in the structure. The precise location of the inhibition helices depicted in later figures is not known. The next step in the proposed binding of Ets-1A331 focuses on the rearrangement of the hydrophobic core at the helix H3 / helix HI interface (Figure 5.24). Rearrangement is necessary to expose the side chain of W338 so it may intercalate between two bases 5' to the GGAA core binding site (Figure 5.25). Residues L337, W375 and Y396 also make DNA contacts implying further conformational changes occur. In the vicinity, L341,1401 and 1402 mediate the conformational change. Structural perturbations also occur in the DNA duplex, resulting in a 60° kink (Werner et al.,1995b). Chapter 5 — Intramolecular Regulation of DNA Binding 202 Figure 5.24 : A side chain rearrangement begins at the helix H1/H3 interface. The residues of strand S3, helix H3 and helix HI are not the only ones to participate in the proposed conformational change. Nearby, several residues including Y307 in helix HI1 and Y329 in helix HI2 of the inhibition domain are also coupled to L337 and W338. Side chains from L429 and L433 in helix H4 of the inhibition module may also participate in the conformational change (Figure 5.26). Figure 5.26 : New associations must be sought to accommodate the conformational changes that have occurred at H1/H3. The inhibition domain is incapable of accommodating the conformational changes that have occurred. Helix HI1, which is unstable even in the unbound form, unfolds as a 203 TD CD X CD T— CL o CO E Z o o < c 1 </> TD LU 3 J5 f-c CD X JD CL i i in § c CD X _CD • CL S E LU O c » CO -rj £ £ 1c c 'c =3 2 0 4 QJ X ! QJ o T J C / J H W QJ cj CO M l CO QJ X, 33 QJ X CO 4-1 w .5 CO CO CO pq 4 - . -33 to ca S T J 3 3 TJ QJ CO O O H X -CJ QJ XI CO 33 QJ X d 33 M3 QJ x; TJ -QJ CO -9 TJ x; CJ QJ -3 ^ s CO 4-» »>*» I S QJ -p3 O H .5 QJ CO r x P S 8 ca X w c o oo CO CO o bJD .g j U "o TJ c QJ x; o ™ca x; .33 W C 3 QJ > QJ X O N O N o o 4-> c o QJ CJ g x u T J < o QJ o 4-> .g QJ 4-> 13 CJ I i QJ ~* CO COX ! r-T CO QJ S | 3 33 co CO ^ i I C 2 I o 33 X 3 QJ cu U X Tj 4 1 co . 3 cu ^ s -O U co ca s s QJ 3 ^ c3 T J £ 2 X) C QJ •B a .5 1/1 QJ CO •£ 2 * ^ 4-> co _C D H T J £ QJ O cj 33 4 - ^ > rt ^ X> S3 + I 33 .3 ^ 4-» X T J H CJ £ CO QJ QJ ^ X X) X QJ j2 ~ x; 33 " CO CO QJ co gj QJ S g *-* O H 2 CJ U ^ IB « CJ CJ .52 x; t« T J £ « ^ 4 i c QJ toO £ .2 ..—I T J % -S cj co QJ co £ T J . S P QJ CO N CO X3 ™ - & X > QJ 6b a OJC -3 U o 2 § ^ J l M l o "Xi O CO j_. _g < T i ^ 33 co C QJ QJ X l QJ > C3 x; QJ 4 i _QJ 6 o u QJ 4 1 O » 5 l T - H ^ CO 4 1 Z T J ~ 4 1 QJ +3 x; c "33 QJ C QJ •33 co • S X l ca 33 "2 ^ £ « ob > T J i l g o S QJ S _ *i 33 CCj QJ T—I •rj x; a ' 53 x > x\ Six CO o ^ x; n a g o 'o QJ CS -S3 xT OJ d "5 w % QJ CO s O H 3 C3 gj HH CO 3H <- QJ c > . - i u 3 1 v T J 5312 K g X 4 1 § « o ni CJ ° * i _ g C CO •a C .52 -2 .y -33 i t ! co T J 33 CJ CJ co y .s =3 Cfl QJ 4 1 cc C 3 ^ QJ |3 £ * ^ CO f-4 1 X( CO QJ M l WO o too , i 33 C QJ c ctf 4 1 >1 o O H w .2 -S -5 £ ° CM g QJ 33 QJ f i QJ X ° § CO r l c«  -5 o 6 c o QJ J i 33 4 1 CJ 33 i i x; CJ LD LO QJ 60 X QJ d 33 4 1 CJ 33 Si UH co g 4 1 o I i o-X W ,X< Qj CJ x; x. ^ M l QJ O TJ g 0 o - T3 C CJ O X •j- ,CJ X O H 1 I .33 U QJ CO M9 •** o QJ x: oo O CJ Tj cn g qj •'-< S y co X i co E O ^ O H ^ - V (Cj I—I •a co o . t J cu l i I 33; QJ co 33 4 1 U 3 a • CJ ^ ^ cj O X O H £ 133 ! > , O *3rt XITJ 5 CJ O 33 cu TJ C O 'co S 33 33 ca ns c- u c3^ > ^ H 4 1 QJ "w too O ca O H CU QJ > ca .33 'C CJ ca X ! > «« C cu QJ 5> co CJ x< pL, 2 ^ hrt " CJ E .52 -Q M l Q >^  o . ca bo^ S g ^ oo O ^ .33 N g X) ca O ^ TJ \g <C '2 X l Z •S 2 Q TJ •£ 33 C MH CO ca o co co TJ QJ C ' ' o CN "43 ffi 43 r W CO C3 333 O ^ W QJ CN u CO CJ O C O S 3 v 2 '-^  oo x; ca g §^  s > T j CJ M i r -o x: .5 Chapter 5 — Intramolecular Regulation of DNA Binding 205 result (Figure 5.27). Helix H4 and and helix HI2 likely maintain a similar position in the complex. The proposed position of helix HI2 in uncomplexed Ets-1AN301 is similar to what was observed by Werner et al. (1995b) for the residues N-terminal to the ETS domain in the complexed form of the human Ets-1 ETS domain (Figure 4.13). Figure 5.27: Helix HI1 of the inhibition domain unfolds. From this proposed model, it is evident that the DNA binding involves much more than the formation of base specific contacts by the recognition helix, H3. A concerted structural event must occur involving not only the recognition helix, but also intercalation of W338, DNA bending and finally unfolding of the inhibition helix HI1. At each point in this mechanism, there is potential for regulation. However, the opportunity for regulation also offers the opportunity for disregulation. The substitution of 16 residues in the C-terminal portion of the inhibition module by v-Ets is presumed to disrupt helix H4 and thereby derepress the ETS domain (Donaldson et al, 1994; Petersen et al., 1995; Donaldson et al, 1996; Skalicky et al, 1996). Mutational analysis of Ets-1AN301 may provide further clues as to which residues are important in mediating the inhibitory effect. Some of the following mutations listed in Chapter 5 — Intramolecular Regulation of DNA Binding 206 Table 5.4 are suggested to identify the hydrophobic contacts are important. Other mutations are designed with the aim of stabilizing the inhibitory domain in order to possibly potentiate the inhibitory effect. It should be noted at this point that helical folding and unfolding as an event itself is not specific to the Ets-1 inhibition module. For example, a helical folding event occurs in the C-terminal 22 amino acids of MATa2 upon formation of a ternary complex with MATal (Stark and Johnson, 1994; Phillips et ai, 1994; Li et al, 1995). In basic leucine zipper (bZip) proteins of which Fos, Jun, and GCN4 belong, the recognition helix exists in an unstable, "nascent" form prior to DNA binding (Weiss et al, 1990). Recently, helical unfolding was identified as a critical event in DNA binding by the restriction enzyme, BamHI (Newman et al., 1995). Upon binding DNA as a homodimer, the C-terminal helix of one monomer unfolds to make additional base specific contacts with the DNA. Furthermore, the unfolding event introduces assymmetry into the BamHI/DNA complex. Aside from its ability to attenuate the DNA binding affinity of Ets-1 and Ets-2, the inhibition module may also encourage or prevent associations as a result of unfolding (Petersen et ai, 1995). The alternatively spliced, inhibition module deficient isoform of Ets-1 is incapable of interacting with AP-1 at the polyoma virus enhancer; thus, an investigation of possible protein-protein interactions between Ets-1 and AP-1 is certainly warranted (Karin, 1994; Figure 1.4). Other Ets proteins are modulated by autoinhibition of the DNA binding activity, though it is unclear if a similar mechanism is invoked. Given that the location of the other inhibitory activities are variable and that there is no significant homology among the inhibitory regions and the Ets-1 inhibition domain, other mechanisms of autoinhibition likely exist. Along this line of study, it would be interesting to determine if the inhibitory activity is maintained in Ets-1 if the ETS domain of PU.l (39% identity) was substituted. As discussed in Chapter 1, the Ets protein, Elk-1 and serum response factor (SRF) 207 Table 5.4 : Proposed mutations to probe aspects of inhibition domain function Mutation F304A F304D R309D D310K L314Y L314V L314K V320E I320E Location Effect No NOEs have been observed to F304 to date. Is this residue Helix HI1 important? Is the bulky aromatic ring important? Is a hydrophobe important for function? Helix HI1 is quite amphipathic. What is the effect of Helix HQ altering the charge profile? Or adding the potential of forming a salt bridge? By stabilizing helix HI1, will inhibition increase? Currently, several tentative restraints have been accumulated Helix HI1 from the ETS domain to L314. Is there a necessity for a hydrophobe here? L314V is a functional substitution in Ets-2 HI1/HI2 Only a tentative assignment is available for V320 located in a mobile loop spanning P319 to P323. Are these residues important? A327E/G331K A325P Y329M G331A E343A E343Q Helix HI2 Helix HI Helix HI2 is a bona fide helix which suffers from a global motion that is slow on the NMR timescale. The lack of restraints in this helix to the inhibition domain and the ETS domain suggests that it may be loosely associated with the protein. Is a helix important? Or just the contacts made by residues residing at the termini? Would a salt bridge stabilize the helix? Would a charged residue be destabilizing to an aspect of HI2 packing? The E343 sidechain would seem to be destabilizing if the inhibition domain assumes a four anti-parallel helix bundle configuration. What is the effect of mutating E343 to glutamine, found in other ETS domain? Is the charge required there? H430F H430D H430A Helix H4 H430 has a peculiar upheld chemical shift and many NOEs to it. Is it a crucial residue in mediating H1/H4/HI1 interactions? Chapter 5 — Intramolecular Regulation of DNA Binding 208 cooperate through protein-protein interactions to. activate transcription at the c-fos promoter. Autoinhibition as well as phosphorylation state are important factors in the regulation of the Elk-1 /SRF ternary complex. Undoubtedly, this system will be revealing some exciting structural insights in the years to come. 5.9. Conclusions Throughout this dissertation, I have employed both a reductionist and constructionist approach to study the structural aspects of DNA binding by the ETS domain. The Ets-1 ETS domain, as defined by homology, was a poor candidate for a structural study either in the uncomplexed or complexed form due to its limited solubility. Thus, in the case of Ets-1, a purely reductionist approach to the study of the ETS domain would have failed. In contrast, structural studies of the Fli-1 and PU.l ETS domains have succeeded as both protein fragments are very soluble (Liang et al, 1994,1995; Pio et ai, 1995). Together, the Ets-1, Fli-1 and PU.l studies have shown that even for very related molecules, structural studies can still be very independent tasks. To obtain a minimally sized fragment of murine Ets-1, deletion mutagenesis and limited proteolysis experiments were performed (Nye et al., 1992; Jonsen et al., 1996). Alternatively, expression library screening was performed to delineate the boundary of human Ets-1 for structural studies (Jorcyk et al, 1993). Only by pursuing larger fragments of Ets-1, was the structure ultimately solved (Werner et ai, 1995b; Donaldson et ai, 1996). The combined results of many laboratories beginning with the work of Lim et al. (1992) lead to the identification of a bipartite inhibitory activity that flanked the amino- and carboxy-terminal regions of the Ets-1 ETS domain. Again, in early structural studies of the inhbited Ets-1 ETS domain, a purely reductionist approach did not entirely succeed. In order to structurally pursue the autoinhibited form of Ets-1, a larger deletion mutant, Ets-1AN280, had to be characterized at the level of secondary structure (Skalicky et al., 1996). Though this deletion mutant was not feasible for further structural studies, it did form the basis for an early model of Ets-1 autoinhibition and an optimized deletion Chapter 5 — Intramolecular Regulation of DNA Binding 209 mutant, Ets-1AN301 (Petersen et ai, 1995). The deletion mutant, Ets-1AN301, as shown by the results covered in this chapter has provided many valuable insights into the structure and dynamics of the autoinhibited ETS domain. Though it was not covered in this dissertation, the reductionist approach has been successful at initiating structural studies of the Pointed domain. The Pointed domain is a regulatory domain of unknown function found in a subset of Ets proteins, including Ets-1. It is the hope of the author that ultimately, a full structural understanding of the Ets family will be obtained — even if it has to be accomplished piece-by-piece. Bibliography Alexander W, and Dunn A. 1995. Structure and transcription of the genomic locus encoding murine c-Mpl, a receptor for thrombopoetin. Oncogene 10: 795-803. Anton I, and Frampton J. 1988. Tryptophans in myb proteins. Nature 336: 719. Archer S, Ikura M , Torchia D, and Bax A. 1991. An alternative 3D NMR technique for correlation backbone 1 5 N with sidechain HP resonances in larger proteins. /. Magn. Reson. 95: 636-641. Bachman B. 1972. Pedigrees of some mutant strains of E. coli K12. Bacterial. Rev. 36: 525-557. Bae S, Takahashi E, Zhang Y, Ogawa E, Shigesada K, Namba Y, Satake M, and Ito Y. 1995. Cloning, mapping and expression of PEBP2 alpha C, a third gene encoding the mammalian Runt domain. Gene 159: 245-248. Bai Y, Milne J, Mayne L, and Englander S. 1993. Primary structure effects on peptide group hydrogen exchange. Proteins 17: 75-86. Bachmair A, Finely D, and Varshavsky A. 1986. In vivo half-life of a protein is a function of its amino-terminal residue. Science 234:179-186. Banner D, Bloomer A, Petsko G, Phillips D, Pogson C, Wilson I, Corran P, Furth A, Milman J, Offord R, Priddle J, and Waley S. 1975. Structure of chicken muscle triose phosphate isomerase determined crystallographically at 2.5 angstroms resolution using amino acid sequence data. Nature 255: 609. Barchi J, Grasberger B, Gronenborn A, and Clore G. 1994. Investigation of the backbone dynamics of the IgG-binding domain of streptococcal protein G by heteronuclear two-dimensional 1 H - 1 5 N nuclear magnetic resonance spectroscopy. Prot. Sci. 3: 15-21. Bellacosa A, Datta K, Bear S, Patriotis C, Lazo P, Copeland N , Jenkins N , and Tsichlis P. 1994. Effects of provirus integration in the Tpl-1/Ets-1 locus in Moloney murine leukemia virus-induced rat T-cell lymphomas: levels of expression, polyadenylation, transcriptional initiation, and differential splicing of the Ets-1 rnRNA. /. Virol. 68: 2320-2330. Ben-David Y, Giddens E, Letwin K, and Bernstein A. 1991. Erythroleukemia induction by Friend murine leukemia virus: Insertional activation of a new member of the ets gene family, Fli-1, closely linked to c-ets-1. Genes. Dev. 5: 908-918. Bodenhausen G, and Ruben D. 1980. Natural abundance nitrogen-15 NMR by enhanced heteronuclear spectroscopy. Chem. Phys. Lett. 69:185-189. 210 Bibliography 211 Bories J-C, Willerford D, Grevin D, Davidson L, Camus A, Martin P, Stehelin D, and Alt F. 1995. Increased T-cell apoptosis and terminal B-cell differentiation induced by inactivation of the Ets-1 proto-oncogene. Nature.377: 635-638. Brennan R. 1993. The winged-helix DNA-binding motif: Another helix-turn-helix takeoff. Cell 74: 773-776. Brunner D, Ducker K, Oellers N , Hafen E, Scholtz H, and Klambt C. 1994. The ETS domain protein pointed-P2 is a target of MAP kinase in the sevenless signal transduction pathway. Nature 370: 386-389. Brunger A. 1992. X-PLOR 3.1: A System for X-Ray Crystallography and NMR. New Haven: Yale University Press. Buratowshi S. 1994. The basics of basal transcription by RNA polymerase II. Cell 77: 1-3. Burley S. 1994. DNA-binding motifs from eukaryotic transcription factors. Curr. Opin. Str. Biol. 4: 3-11. Bussiere D, Bastia D, and White S.o 1995. Crystal structure of the replication terminator protein from B. subtilis at 2.6 A. Cell 80: 651-660. Buttice G, and Kurkinen M. 1993. A polyomavirus enhancer A-binding protein-3 site and Ets-2 protein have a major role in the 12-0-tetradecanoylphorbol-13-acetate response of the human stromelysin gene. /. Biol. Chem. 268: 7196-7204. Cano E, and Mahadevan L. 1995. Parallel signal processing among mammalian MAPKs. Trends Biochem. Sci. 20:117. Carlsson P, Waterman M , and Jones, K. 1993. The LEF-1 /TCF-lcc HMG protein contains a context-dependent activation domain that induces the TCRa enhancer in T cells. Genes Dev. 7: 2418-2430. Carson M . 1991. Ribbons 2.0 Manual. University of Alabama at Birmingham Press. Chakrabartty A, Doig A, and Baldwin R. 1993. Helix capping propensities in peptides parallel those in proteins. Proc. Natl. Acad. Sci. USA. 90:11332-11336. Chen Y, and Ehbright R. 1993. Phenyl-azide photocrosslinking analysis of Cro-DNA interaction. /. Mol. Biol. 230: 453-460. Chi T, Lieberman P, Ellwood K, and Carey M. 1995. A general mechanism for transcriptional synergy by eukaryotic activators. Nature 377: 254-257. Chien C, Bartel P, Sternglanz R, and Fields S. 1991. The two-hybrid system: A method to identify and clone genes for proteins that interact with a protein of interest. Proc. Nat 7 . Acad. Sci. USA. 88: 9578-9592. Clore G, Szabo A, Bax A, Kay L, Driscoll P, and Gronenborn A. 1990. Deviations from the simple two parameter model free appoach to the interpretation of 1 5 N nuclear magnetic relaxation of proteins. /. Am. Chem. Soc. 112: 4989. Bibliography 212 Clottes E, Couthon F, Denoroy L, and Vial C. 1994. Creatine kinase compactness and thiol accessibility during sodium dodecyl sulfate denaturation estimated by resonance energy transfer and 2-nitro-5-thiocyanobenzoic acid cleavage. Biochim. Biophys. Acta. 1209: 171-176. Clubb R, Omichinski J, Savilahti H, Mizuuchi K, Gronenborn A, and Clore G. 1994. A novel class of winged helix-turn-helix protein: the DNA-binding domain of Mu transposase. Structure 2: 1041-1048. Connelly G, Bai Y, Jeng M-F, and Englander SW. 1993. Isotope effects in peptide group hydrogen exchange. Proteins. 17: 87-92. Cuddeford L. 1994. Characterization of the binding of the ETS domain to DNA by UV-laser crosslinking. Honours Thesis, University of British Columbia. Dalton S, Marais R, Wynne J, and Treisman R. 1993. Isolation and characterization of SRF acessory proteins. Philos. Trans. R. Soc. London Biol. 340: 325-332. Dalton S, and Treisman R. 1994. Characterization of SAP-1, a protein recruited by serum response factor to the c-fos serum response element. Cell 76: 411. Dang C. 1988. Detection and use of recombinant staphlococcal protein: A fusion protein to localize nucleic-acid-binding domains of proteins. Anal. Biochem. 174: 313-317. Davis R. 1993. The mitogen-activated protein kinase signal transduction pathway. /. Biol. Chem. 268:14553-14556. DeBoer H, Comstock L, and Vasser M. 1983. The tac promoter: A functional hybrid derived from the trp and lac promoters. Proc. Natl. Acad. Sci. USA. 80: 21-25. Degani Y, and Patchornik A. 1974. Cyanylation of sulfhydryl groups by 2-nitro-5-thiocyanobenzoic acid. High-yield modification and cleavage of peptides at cysteine residues. Biochemistry. 13:1-11. Degnan B, Degnan S, Naganuma T, and Morse D. 1993. The ets multigene family is conserved throughout the Metazoa. Nucl. Acids Res. 21: 3479-3484. Deng T, and Karin M. 1994. c-Fos transcriptional activity stimulated by H-Ras-activated protein kinase distinct from JNK and ERK. Nature. 371:171-175. Derome A. 1987. Modern NMR Techniques for Chemistry Research. Oxford: Pergamon Press. Dickson B. 1995. Nuclear factors in sevenless signalling. Trends Genet. 11:106-111. Donaldson L. 1992. Overexpression, purification and characterization of the transactivation domain of Vmw65 from herpes simplex virus type 1. MSc. thesis, McMaster University. Bibliography 213 Donaldson L, and Capone J. 1992. Purification and characterization of the carboxyl-terminal transactivation domain of Vmw65 from herpes simplex virus type 1. /. Biol. Chem. 267:1411-1414. Donaldson L, Petersen J, Graves B, and Mcintosh L. 1994. Secondary structure of the ETS domain places murine Ets-1 in the superfamily of winged-helix-turn-helix DNA-binding proteins. Biochemistry. 33:13509-13516. Donaldson L, Petersen J, Graves B, and Mcintosh L. 1996. Solution structure of the ETS doman from murine Ets-1: A winged helix-turn-helix DNA binding motif. EMBO J. in press. Dong A, Huang P, and Caughey W. 1990. Protein secondary structures in water from second-derivative Amide I infrared spectra. Biochemistry. 29: 3303-3308. Dudek H, Tatravahi R, Rao V, Reddy E, and Reddy E. 1992. Myb and Ets protein cooperate in transcriptional activation of the mim-1 promoter. Proc. Natl. Acad. Sci USA. 89: 1291-1295. Edelhoch H. 1967. Spectroscopic determination of tryptophan and tyrosine in proteins. Biochemistry 6:1948-1954. Eigen M . 1964. Proton transfer, acid-base catalysis, and enzymatic hydrolysis. Angew. Chem. Int. Ed. Engl. 3:1-19. Ellman G. 1959. Anal. Biochem. Biophys. 82: 70. Englander S, and Poulsen A. 1969. Biopolymers. 7: 379-393. Englander S, and Kallenbach N . 1984. Hydrogen exchange and structural dynamics of proteins and nucleic-acids. Q. Rev. Biophys. 16:521-655. Ezaz-Nikpay K, Uchino K, Lerner R, and Verdine G. 1994. Construction of an overproduction vector containing the novel srp (sterically repressed) promoter. Prot. Sci. 3:132-138. Farrow N , Muhandiram R, Singer A, Pascal S, Kay C, Gish G, Shoelson S, Pawson T, Forman-Kay J, and Kay L. 1994. Backbone dynamics of a free and a phosphopeptide-complexed src homology 2 domain studied by 1 5 N NMR relaxation. Biochemistry 33: 5984-6003. Fesik SW, and Zuiderweg E. 1988. Heteronuclear three-dimensional NMR spectroscopy: A strategy for the simplification of homonuclear two-dimensional NMR spectra. /. Magn. Reson. 78: 588-593. Fesik SW, Gampe RJ, Holzman TF, Egan DA, Edalji R, Luly JR, Simmer R, Helfrich R, Kishore V, and Rich DH. 1990. Isotope-edited NMR of cyclosporin A bound to cyclophilin: evidence for a trans 9,10 amide bond. Science 250:1406-9. Bibliography 214 Fesik SW, Zuiderweg ER, Olejniczak ET, and Gampe RJ. 1990. NMR methods for determining the structures of enzyme/inhibitor complexes as an aid in drug design. Biochemical Pharmacology 40:161-7. Fisher R, Fivash M, Casas-Finet J, Erickson J, Kondoh A, Bladen S, Fisher C, Watson D, and Papas T. 1994. Real-time DNA binding measurements of the ETS1 recombinant oncoproteins reveal significant kinetic differences between the p42 and p51 isoforms. Prot. Sci. 3: 257-266. Fleischmann L, Holtzclaw L, Russell J, Mavrothalassitis G, and Fisher R. 1995. Ets-1 in astrocytes: expression and transmitter-evoked phosphorylation. Mol. Cell Biol. 15: 925-931. Fogh R, Ottleben G, Ruterjans H, Schnarr M, Boelens R, and Kaptein R. 1994. Solution structure of the LexA repressor DNA binding domain determined by *H NMR spectroscopy. EMBO J. 13: 3936-3944. Frankel A, and Kim P. 1991. Modular structure of transcription factors: Implications for gene regulation. Cell 65: 717-719. Freund J, Vertesy L, Koller K, Wolber V, Heintz D, and Kalbitzer H. 1995. Complete nuclear magnetic resonance assignments and structural characterization of a fusion protein of the alpha-amylase inhibitor tendamistat with the activation domain of the human immunodeficiency virus type 1 Tat protein. /. Mol. Biol. 250: 672-688. Gegner J, Graham D, Roth A, and Dahlquist F. 1992. Assembly of an MCP receptor, CheW, and kinase CheA complex in the bacterial chemotaxis signal transduction pathway. Cell. 70: 975-982. Gegonne A, Bosselut R, Bailly R, and Ghysdael J. 1993. Synergistic activation of the HTLV-1 ets-responsive region by transcription factors Etsl and Spl. EMBO J. 12: 1169-1178. Gerstein M , Lesk A M , and Chothia C. 1994. Structural mechanisms for domain movements in proteins. Biochemistry 331: 6739-6749. Giese K, Kingsley C, Kirshner J, and Grosschedl R. 1995. Assembly and function of a TCRa enhancer complex is dependent on LEF-1 induced DNA bending and multiple protein-protein interactions. Genes Devel. 9: 995-1008. Gill S, and vonHippel P. 1989. Calculation of protein extinction coefficients from amino acid sequence data. Anal. Biochem. 182: 319-326. Gille H, Strahl T, and Shaw P. 1995. Activation of ternary complex factor Elk-1 by stress activated protein kinases. Curr. Biol. 5: 1191-1200. Giovane A, Pintzas A, Maira S, Sobieszczuk P, and Wasylyk B. 1994. Net, a new ets transcription factor that is activated by Ras. Genes Dev. 84:1502-1513. Bibliography 215 Girvin M, and Fillingame R. 1995. Determination of local protein structure by spin label difference 2D NMR: The region neighbouring Asp61 of subunit c of the F 1 F 0 ATP synthase. Biochemistry 34: 1635-1645. Golay T, Introna M, and Graf T. 1988. A single point mutation in the v-Ets oncogene affects both erythroid and myelomonocytic cell differentiation. Cell 55:1147-1158. Goldberg Y, Treier M, Ghysdael J, and Bohmann D. 1994. Repression of AP-l-stimulated transcription by c-Ets-1. /. Biol. Chem. 269:16566-16573. Golub T, Barker G, Lovett M, and Gilliland D. 1994. Fusion of PDGF receptor beta to a novel ets-like gene, tel, in chronic myelomonocytic leukemia with t(5;12) chromosomal translocation. Cell 77: 307-316. Grimm S, and Bauerle P. 1993. The inducible transcription factor NF-kappa B: Structure-function relationship of its protein subunits. Biochem J. 290: 297. Gronenborn A, and Clore G. 1994. Identification of N-terminal helix capping boxes by means of 1 3 C chemical shifts. /. Biomol. NMR. 4: 455-458. Gross G. 1989. High level gene expression in Escherichia coli. Chimicaoggi. 3: 21-29. Grzeisiek S, and Bax A. 1992. Correlating backbone amide and side chain resonances in larger proteins by multiple relayed triple resonance NMR. /. Am. Chem. Soc. 114: 6291-6293. Grzesiek S, and Bax A. 1993. Measurement of amide proton exchange rates and NOEs with water in 1 3C/ 1 5N-enriched calcineurin B.J. Biomol. NMR. 3: 627-638. Grzesiek S, Anglister J, and Bax A. 1993. Correlation of backbone amide and aliphatic side-chain resonances in 1 3 C / 1 5 N proteins by isotropic mixing of 1 3 C magnetization. /. Magn. Resort. 101:114-119. Giintert P, Qian Y, Otting G, Muller M , Gehring W, and Wuthrich K. 1991. Structure determination of the Antp (C39S) homeodomain from nuclear magnetic resonance data in solution using a novel strategy for the structure calculation with the programs DIANA, CALIBA, HABAS and GLOMSA. /. Mol. Biol. 217: 531-40. Gunther C, Nye J, Bryner R, and Graves B. 1990. Sequence-specific DNA binding of the proto-oncoprotein Ets-1 defines a transcriptional activator sequence within the long terminal repeat of the Moloney murine sarcoma virus. Genes. Dev. 4: 667-679. Hagemeier C, Bannister A, Cook A, and Kouzarides T. 1993. The activation domain of transcription factor PU. l binds the retinoblastoma (RB) protein and the transcription factor TFIID in vitro: RB shows sequence similarity to TFIID and TFIIB. Proc. Natl. Acad. Sci. USA. 90:1580-1584. Hagman J, and Grosschedl R. 1992. An inhibitory carboxyl-terminal domain in Ets-1 and Ets-2 mediates differential bindng of ETS family factors to promoter sequences of the mb-1 gene. Proc. Natl. Acad. Sci. USA. 89: 8889-8893. Bibliography 216 Hahn E. 1958. Spin echoes. Phys. Rev. 60: 580-594. Hahn S, and Wasylyk B. 1994. The oncoprotein v-Ets in less selective in DNA binding than c-Ets-1 due to the C-terminal sequence change. Oncogene 9: 2499-2512. Hansen A, Petros A, Mazar A, Pederson T, Reuter A, and Fesik S. 1992. A practical method for uniform isotopic labelling of recombinant proteins in mammalian cells. Biochemistry 31:12713-12718. Harrison S. 1991. A structural taxonomy of DNA-binding proteins. Nature. 353: 715-719. Harrison C, Bohm A, and Nelson H. 1994. Crystal structure of the DNA binding domain of the heat shock transcription factor. Science. 263: 224-226. Henikoff S, and Henikoff J. 1992. Amino acid substitution matrices from protein blocks. Proc. Natl. Acad. Sci. USA. 89:10915-10919. Higashino F, Yoshida K, Noumi T, Seiki M, and Fujinaga K. 1995. Ets-related protein E1A-F can activate three different matrix metalloproteinase gene promoters. Oncogene 10:1461-1463. Hill C, Marais R, John S, Wynne J, Dalton S, and Treisman R. 1993. Functional analysis of a growth-factor responsive transcription factor complex. Cell 73: 395-406. Hipskind R, Rao V, Mueller C, Reddy E, and Nordheim A. 1991. Ets like protein Elk-1 is homologous to the c-fos regulatory factor p62. Nature 354: 531-534. Hockensmith J, Kubazek W, Vorachek W, Evertsz E, and vonHippel P. 1991. Laser crosslinking of protein-nucleic acid complexes. Meth. Enzymol. 208: 211-235. Homans S. 1992. A Dictionary of Concepts in NMR. Oxford: Clarendon Press. Hori R, and Carey M . 1994. The role of activators in asssembly of RNA polymerase II transcription complexes. Curr. Opin. Genet. Dev. 4: 236-244. Hupp T, Meek D, Midgely C, and Lane D. 1992. Cell 71: 875. Ikura M, Kay L, and Bax A. 1990. A novel approach for sequential assignment of *H, 1 3 C and 1 ^ N spectra of proteins: Heteronuclear triple-resonance three-dimensional NMR spectroscopy. Biochemistry 29: 4659. Jacobson G, Schaffer M, Stark G, and Vanaman T. 1973. Specific chemical cleavage in high yield at the amino peptide bonds of cysteine and cystine residues. /. Biol. Chem. 248: 6583-6591. Janknecht R, Zinck R, Ernst W, and Nordheim A. 1994. Functional dissection of the transcription factor Elk-1. Oncogene 9: 1273-1278. John S, Reeves R, Lin J, Child R, Leiden J, Thompson C, and Leonard W. 1995. Regulation of cell-type-specific interleukin-2 receptor alpha-chain gene expression: Bibliography 217 potential role of physical interactions between Elf-1, HMG-I(Y), NF-kappa B family proteins. Mol. Cell. Biol. 15:1786-1796. Johnson W. 1988. Secondary structure of proteins through circular dichroism spectroscopy. Ann. Rev. Biophys. Biophys. Chem. 17:145-166. Jonsen M , Petersen J, and Graves B. 1995. Intramolecular interactions mediate negative regulation of Ets-1 DNA binding, submitted. Jorcyk C, Watson D, Mavrothalassis G, and Papas T. 1993. The human ETS1 gene: genomic structure, promoter characterization and alternative splicing. Oncogene 6: 523-532. Kagoshima H, Shigesada K, Satake M, Ito Y, Miyoshi H, Ohki M, Pepling M , and Gergen P. 1993. The runt domain identifies a new family of heteromeric transcriptional regulators. Trends. Genet. 9: 338-341. Karim F, Urness L, Thummel C, Klemsz M, McKercher S, Celada A, Beveren CV, Maki R, Gunther C, Nye J, and Graves B. 1990. The ETS-domain: A new DNA-binding motif that recognizes a purine-rich core DNA sequence. Genes. Dev. 4:1451-1453. Karin M , and Hunter T. 1995. Transcriptional control by protein phosphorylation: signal transmission from the cell surface to the nucleus. Curr. Biol. 5: 747-755. Kaufmann E, Muller D, and Knochel W. 1995. DNA recognition site of Xenopus winged helix proteins. /. Mol. Biol. 248: 239-254. Kay L, Torchia D, 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. Kay L, Ikura M, Tschudin R, and Bax A. 1990. Three-dimensional triple resonance NMR spectroscopy of isotopically enriched proteins. /. Magn. Reson. 89: 496. Kay L, and Bax A. 1990. New methods for the measurement of N H - C a H coupling constants in 1 5 N labelled proteins. /. Magn. Reson. 86:110-126. Kay L, Keifer P, and Saarinen T. 1992. /. Am. Chem. Soc. 114:10663-10665. Kay L. 1993. Pulsed-Field Gradient-Enhanced Three-Dimensional NMR Experiment for Correlating " C ^ , 1 3 C , and 1 H ° Chemical Shifts in Uniformly 1 3 C labelled Proteins Dissolved in H 20. /. Am. Chem. Soc. 115: 2055-2057. Kay L, Xu G-Y, Singer A, Muhandiram D, and Forman-Kay J. 1993. A gradient-enhanced HCCH-TOCSY experiment for recording side-chains 1 H and 1 3 C correlations in H 2 0 samples of proteins. /. Magn. Res. 101: 333-337. Kharrat A, Macias M, Gibson T, Nilges M, and Pastore A. 1995. Structure of the dsRNA binding domain of E. coli RNase III. EMBO J. 14: 3572-3584. Bibliography 218 Kidd I, and Emery V. 1993. The use of baculoviruses as expression vectors. Appl. Biochem. Biotechnol. 42: 137-159. Klambt C. 1993. The Drosophila gene pointed encodes two ETS-like proteins which are involved in the development of the midline glial cell. Development 117:163-176. Klemsz M, McKercher S, Celada A, Beveren CV, and Maki R. 1990. The macrophage and B cell-specific transcription factor PU.l is related to the ets oncogene. Cell 61: 113-124. Kodadek T. 1995. From carpet bombing to cruise missiles: the second-order mechanisms used by transcription factors to ensure specific DNA binding in vivo. Curr. Biol. 267-278: Koizumi S, Fisher R, Fujiwara S, Jorcyk C, Bhat N , Seth A, and Papas T. 1990. Isoforms of the human ets-1 protein: generation by alternative splicing and differential phosphorylation. Oncogene. 5: 675-681. Kosen P. 1989. Spin labelling of proteins. Meth. Enzymol. 177: 86-118. Kramer B, Wiegmann K, and Kronke M. 1995. Regulation of the human TNF promoter by the transcription factor Ets. /. Biol. Chem. 270: 6577-6583. Kraulis P. 1991. MOLSCRIPT: a program to produce both detailed and schematic plots of protein structures. /. Appl. Cryst. 24: 946-950. LaMarco K, Thompson C, Byers B, Walton E, and McKnight S. 1991. Identification of Ets-and Notch-Related Subunits in GA Binding Protein. Science 253: 789-792. Landsman D, and Wolffe A. 1995. Common sequence and structural features in the heat-shock factor and Ets families of DNA-binding domains. Trends Biochem. Sci. 20: 225-226. Laskowski R, MacArthur M, Moss D, and Thornton J. 1988. PROCHECK: a program to check the stereochemical quality of protein structures. /. Appl. Cryst. 26: 283-291. Laudet V, Niel C, Duterque-Coquillaud M, Leprince D, and Stehelin D. 1993. Evolution of the Ets gene family. Biochem. Biophys. Res. Comm. 190: 8-14. Lautenberger J, Burdett L, Gunnell M, Qi S, Watson D, O'Brien S, and Papas T. 1992. Genomic dispersal of the ets gene family during metazoan evolution. Oncogene. 7: 1713-1719. LaVallie E, DiBlasio E, Kovacic S, Grant K, Schendel P, and McKoy J. 1993. A thioredoxin gene fusion expression system that circumvents inclusion body formation in the E. coli cytoplasm. Biotechnology 11:187-193. Lehrman S, Tuls J, and Lund M. 1990. Peptide alpha-helicity in aqueous trifluoroethanol: Correlations with predicted alpha-helicity and the secondary structure of the corresponding regions of bovine growth hormone. Biochemistry 29: 5590-5596. Bibliography 219 Lemarchandel V, Ghysdael J, Mignotte V, Rahuel C, and Romeo P. 1993. GATA and Ets cis-acting sequences mediate megakaryocyte-specific expression. Mol. Cell. Biol. 13: 668-676. LeMaster D, and Richards F. 1985. 1 H - 1 5 N heteronuclear NMR studies of Escherichia coli thioredoxin in samples labelled by residue type. Biochemistry 24: 7263-7268. Leprince D, Gegonne J, de Taisne A, Schneeberger A, Lagrou C, and Stehelin D. 1983. A putative second cell-derived oncogene of the avian leukaemia retrovirus E26. Nature 306: 395-397. Leprince D, Crepieux P, and Stehelin D. 1992. C-ets-1 DNA binding to the PEA3 motif is differentially inhibited by all the mutations found in v-ets. Oncogene 7: 9-17. Li T, Stark M , Johnson A, and Wolberger C. 1995. Crystal structure of the MATal /MATa2 homeodomain heterdimer bound to DNA. Science 270: 262-269. Liang H, Olejniczak E, Mao X, Nettesheim D, Yu L, Thompson C, and Fesik S. 1994a. The secondary structure of the ets domain of human Fli-1 resembles that of the helix-turn-helix DNA-binding motif of the Escherichia coli catabolite gene activator protein. Proc. Natl. Acad. Sci. USA. 91:11655-11659. Liang H, Mao X, Olenjniczak E, Nettesheim D, Yu L, Meadows R, Thompson C, and Fesik S. 1994b. Solution structure of the ets domain of Fli-1 when bound to DNA. Nature Str. Biol. 1: 871-876. Liao T, and Wadano A. 1979. Inactivation of DNase by 2-nitro-5-thiocyanobenzoic acid. II. Serine and threonine are the sites of reaction on the DNase molecule. /. Biol. Chem. 240: 9602-9607. Lieberman P, Schmidt M , Kao C, and Berk A. 1991. Two distinct domain in the yeast transcription factor IID and evidence for a TATA box-induced conformation change. Mol. Cell. Biol. 11: 63. Lim F, Kraut N , Frampton J, and Graf T. 1992. DNA binding by c-Ets-1, but not v-Ets, is repressed by an intramolecular mechanism. EMBO J. 11: 643-652. Lipari G, and Szabo A. 1982. Model-free approach to the interpretation of nuclear magnetic resonance relaxation in macromolecules. 2: Analysis of experimental results. /. Am. Chem. Soc. 104: 4559-4570. Lipari G, and Szabo A. 1982. Model-free approach to the interpretation of nuclear magnetic resonance relaxation in macromolecules. 1: Theory and range of validity. /. Am. Chem. Soc. 104: 4546-4559. Liu J, Fan Q, Sodeoka M , Lane W, and Verdine G. 1994. DNA binding by an amino acid in the C-terminal half of the Rel homology region. Chemistry and Biology 1: 47-57. Lopez M , Oettgen P, Akbarali Y, Dendorfer U, and Libermann T. 1994. ERP, a new member of the ets transcription factor/oncoprotein family: cloning, Bibliography 220 characterization, and differential expression during B-lymphocyte development. Mol. Cell. Biol. 14: 3292-3309. Love J, Li X, Case D, Giese K, Grosschedl R, and Wright P. 1995. Structural basis for DNA bending by the architectural transcription factor LEF-1. Nature 376: 791-794. Lyu P, Wemmer D, Zhou H, Pinker R, and Kallenbach N . 1993. Capping interactios in isolated a-helices: Position-dependent substitution effects and structure of a serine-capped peptide helix. Biochemistry.32: 421-425. Macleod K, Leprince D, and Stehelin D. 1992. The ets gene family. Trends. Biochem. Sci. 17: Maina C, Riggs P, Grandea A, Slatko B, Moran L, Tagliamonte J, McReynolds L, and Guan C. 1988. An Escherichia coli vector to express and purify foreign proteins by fusion and separation from maltose-binding protein. Gene 74: 365-373. Majerus M , Bibollet-Ruche F, Telliez J, Wasylyk B, and Bailleul B. 1992. Serum, AP-1 and Ets-1 stimulate the human Ets-1 promoter. Nucl. Acids Res. 20: 2699-2703. Mandel A, Akke M , and Palmer A. 1995. Backbone dynamics of Escherichia coli ribonuclease HI: Correlations with structure and function in an active enzyme. /. Mol. Biol. 246:144-163. Mao X, Miesfeldt S, Yang H, Leiden J, and Thompson C. 1994. The FLI-1 and chimeric EWS-FLI-1 oncoproteins display similar DNA binding specificities. /. Biol. Chem. 269: 18216-18222. Marion D, Driscoll P, Kay L, Wingfield P, Bax A, Gronenborn A, and Clore G. 1989. Overcoming the overlap problem in the assignment of 1H NMR spectra of larger proteins by use of three-dimensional heteronuclear 1H-15N Hartmann-Hahn-multiple quantum coherence and nuclear Overhauser-multiple quantum coherence spectroscopy: Application to interleukin 1 beta. Biochemistry 28: 6150. Mcintosh LP, and Dahlquist FW. 1990. Biosynthetic incorporation of 1 5 N and 1 3 C for assignment and interpretation of nuclear magnetic resonance spectra of proteins. Q Rev. Biophys 23:1-38. Meadows R, Olejniczak E, and Fesik S. 1994. A computer-based protocol for semiautomated assignments and 3D structure determination of proteins. /. Biomol. NMR. 4: 79-96. Merritt E, and Murphy M. 1994. Raster3D Version 2.0 - A program for photorealistic molecular graphics. Acta. Cryst. D50: 869-873. Mitchell P, and Tjian R. 1989. Transcriptional regulation in mammalian cells by sequence-specific DNA binding proteins. Science 245: 371-378. Moreau-Gachelin F, Tavitian A, and Tambourin P. 1988. Spi-1 is a putative oncogene in virally induced murine erythroleukaemias. Nature 331: 277-280. Bibliography 221 Moreau-Gachelin F. 1994. Spi-l/PU.l: an oncogene of the Ets family. Biochim. Biophys. Acta. 1198: 149-163. Muhandiram D, and Kay L. 1994. Gradient-enhanced triple-resonance three-dimensional NMR experiments with improved sensitivity. /. Mag. Res. Series B 103: 203-216. Muthusamy S, Barton K, and Leiden J. 1995. Defective activation and survival of T cells lacking the Ets-1 transcription factor. Nature 377: 639-642. Natesan S, and Gilman M. 1993. DNA bending and orientation-dependent function of YY1 in the c-fos promoter. Genes. Dev. 7: 2497-2509. Nelsen B, Gang T, Erman B, Gregoire J, Maki R, Graves B, and Sen R. 1993. Regulation of lymphoid-specific immunoglobulin (J. heavy chain gene enhancer by ETS domain proteins. Science 261: 82-86. Neri D, Szyperski T, Otting G, Senn H, and Wuthrich K. 1989. Stereospecific nuclear magnetic resonance assignments of the methyl groups of valine and leucine in the DNA-binding domain of the 434 repressor by biosynthetically directed fractional 1 3 C labelling. Biochemistry 28: 751-7516. Newman M , Strzelecka T, Dorner L, Schildkraut I, A Aggarwal. 1995. Structure of BamHI endonuclease bound to DNA: Partial folding and unfolding on DNA binding. Science 269: 656-663. Nicholls A, Bharadwaj R, and Honig B. 1993. GRASP — Graphical representation and analysis of surface properties. Biophys. J. 64:166. Nikolov D, Chen H, Halay E, Usheva A, Histake K, Lee D, Roeder R, and Burley S. 1995. Crystal structure of a TFIIB-TBP-TATA element ternary complex. Nature 377: 119-128. Nilges M , Gronenborn AM, Briinger AT, and Clore GM. 1988. Determination of three-dimensional structures of proteins by simulated annealing with interproton distance restraints. Application to crambin, potato carboxypeptidase inhibitor and barley serine proteinase inhibitor 2. Protein Engineering 2: 27-38. Nilges M . 1995. Calculation of protein structures with ambiguous distance restraints. Automated assignment of ambiguous NOE crosspeaks and disulphide connectivities. /. Mol. Biol. 245: 645-660. Nunn M , Seeburg P, Moscovici C, and Duesberg P. 1983. Tripartite structure of the avian erythroblastosis virus E26 transforming gene. Nature 306: 391-397. Nunn M , and Hunter T. 1989. The ets sequence is required for induction of erythroblastosis in chickens by avian retrovirus E26. /. Virol. 63: 398-402. Nye J, Petersen J, Gunther C, Jonsen M, and Graves B. 1992. Interaction of murine Ets-1 with GGA-Binding sites establishes the ETS domain as a new DNA-binding motif. Genes. Dev. 6: 975-990. Bibliography 222 O'Neill E, Rebay I, Tjian R, and Rubin G. 1994. The activities of two Ets-related transcription factors required for Drosophila eye development are modulated by the Ras/MAPK pathway. Cell 78: 137-147. Ogata K, Hojo H, Aimoto S, Nakai T, Nakamura H, Sarai A, Ishii S, and Nishimura Y. 1992. Solution structure of a DNA-binding unit of Myb: A helix-turn-helix related motif with conserved tryptophans forming a hydrophobic core. Proc. Natl. Acad. Sci. USA. 89: 6428-6432. Ogata K, Morikawa S, Nakamura H, Hojo H, Yoshimura S, Zhang R, Aimoto S, Ametani Y, Hirata Z, Sarai A, Ishii S, and Nishimura Y. 1995. Comparison of the free and DNA-complexed forms of the DNA-binding domain from c-Myb. Nature Str. Biol. 2: 309-319. Oschkinat H, Mueller T, and Dieckmann T. 1994. Protein structure determination with three- and four-dimensional NMR spectroscopy. Angew. Chem. Int. Ed. Engl. 33: 277-293. Otieno S. 1978. Generation of a free alpha-amino group by Raney nickel after 2-nitro-5-thiocyanobenzoic acid cleavage at cysteine residues: application to automated sequencing. Biochemistry 17: 5468-5474. Otting G, Liepinsh E, and Wuthrich K. 1991. Protein hydration in aqueous solution. Nature 254: 974-980. Padmanabhan S, Marqusee S, Ridgeway T, Laue T, and Baldwin R. 1990. Relative helix-forming tendencies on nonpolar amino acids. Nature 344: 268-270. Pascal S, Muhandiram D, Yamazaki T, Kay J, and Kay L. 1994. Simultaneous acquisition of 1 5 N and 1 3 C edited NOE spectra of proteins dissolved in H 2 O . / . Magn. Reson. 103:197-201. Pastore A, and Lesk A M . 1991. Brave new proteins: What evolution reveals about protein structure. Curr. Opin. Biotechnol. 2: 592-598. Pelligrini L, Tan S, and Richmond T. 1995. Structure of serum response factor core bound to DNA. Nature 376: 490-497. Peng J, and Wagner G. 1994. Investigation of protein motions via relaxation measurements. Meth. Enzymol. 239: 563-597. Perutz M. 1989. Q. Rev. Biophys. 22:139. Petersen J, Skalicky J, Donaldson L, Mcintosh L, Alber T, and Graves B. 1995. Modulation of transcription factor Ets-1 DNA binding: DNA-induced unfolding of an a-helix. Science 269:1866-1869. Phillips C, Stark M , Johnson A, and Dahlquist F. 1994. Heterodimerization of yeast homeodomain transcriptional regulators a2 and al induces an interfacial helix in a2. Biochemistry 33: 9294-9302. Bibliography 223 Pingoud V, Zinck R, Hipskind R, Janknecht R, and Nordheim A. 1994. Heterogeneity of ternary complex factors in HeLa cell nuclear extracts. /. Biol. Chem. 269: 23310-23317. Pio F, Ni C-Z, Mitchell R, Knight J, McKercher S, Klemsz M, Lombardo A, Maki R, and Ely K. 1995. Co-crystallization of an ETS domain (PU.l) in complex with DNA. /. Biol. Chem. 270: 24258-24263. Pongubala J, Beveren CV, Nagulapalli S, Klemsz M, McKercher S, Maki R, and Atchison M. 1993. Effect of PU.l activation on interaction with NF-EM5 and transcriptional activation. Science 259:1622-1625. Power B, Ivancic N , Harley V, Webster R, Kortt A, Irving R, and Hudson P. 1992. High-level temperature-induced synthesis of an antibody VH-domain in Escherichia coli using the pelB secretion signal. Gene 113: 95-99. Presta L, and Rose G. 1988. Helix signals in proteins. Science 240:1632-1641. Price M, Rogers A, and Treisman R. 1995. Comparative analysis of the ternary complex factors Elk-1, SAP-la and SAP-2 (ERP/NET). EMBO ]. 14: 2589-2601. Quinn T, Richardson D, and Richardson J. 1990. De novo design, expression, and characterization of FELIX: A four-helix bundle protein of native-like sequence. Science 249: 884. Rabidran S, Haroun R, Clos J, Wisniewski J, and Wu C. 1993. Regulation of heat shock factor trimer formation: role of a conserved leucine zipper. Science 259: 230-234. Radke K, Beug H, Kornfeld S, and Graf T. 1982. Transformation of both erythroid and myeloid cells by E26, an avian leukemia virus that contains the myb gene Cell. 31: 643-653. Reddy E, Rao V, and Papas T. 1987. The erg gene: a human gene related to the ets oncogene. Proc. Natl. Acad. Sci. USA. 84: 6131-6135. Reddy E, and Rao V. 1988. Structure, expression and alternative splicing of the human c-ets-1 proto-oncogene. Oncogene Res. 3: 239-246. Redfield C, Smith L, Boyd J, Lawrence G, Edwards R, Smith R, and Dobson C. 1991. Secondary structure and topology of human interleukin 4 in solution. Biochemistry 30:11029-11035. Roberts SG, Choy B, Walker SS, Lin Y-S, and Green M. 1995. A role for activator-mediat ed TFIIB recruitment in diverse aspects of transcription regulation. Curr. Biol. 5: 508-516. Richardson J. 1981. Adv. Prot. Chem. 34:167-339. Bibliography 224 Robinson M, Lilley R, Little S, Emtage J, Yarranton G, Stephens P, Millican A, Eaton M , and Humphreys G. 1984. Codon usage can affect efficiency of translation of genes in Escherichia coli. Nucl. Acids Res. 12: 6663-6671. Rost B, and Sander C. 1993. Prediction of protein structure at better than 70% accuracy. /. Mol. Biol. 232: 584-599. Rost B, and Sander C. 1994. Combining evolutionary information and neural networks to predict protein secondary structure. Proteins 19: 55-72. Sadowski I, Ma J, Treizenberg S, and Ptashne M. 1988. GAL4-VP16 is an unusually potent transcriptional activator. Nature 335: 563-564. Sailer M, Helms G, Henkel T, Niemczura W, Stiles M, and Vederas J. 1993. 1 5 N and Re-labeled media from Anabaena sp. for universal isotopic labelling of bacteriocins: NMR resonance assignments of leucocin A from Leuconostoc gelidum and nisin A from Lactococcus lactis. Biochemistry. 32: 310-318. Sanchez-Garcia I, and Rabbitts T. 1993. Redox regulation of in vitro DNA-binding activity by the homeodomain of the Isl-1 protein. /. Mol. Biol. 231: 945-949. Santoro J, and King G. 1992. A constant-time 2D overbodenhausen experiment for inverse correlation of isotopically enriched species. /. Magn. Reson. 97: 202-207. Scholz H, Deatrick J, Klaes A, and Klambt C. 1993. Genetic dissection of pointed, a Drosophila gene encoding two ETS-related proteins. Genetics 135: 455-468. Seth A, and Papas T. 1993. The c-ets-1 proto-oncogene has oncogenic activity and is positively regulated. Oncogene 5:1761-1767. Sgouras D, Athanasiou MA, Beal G, Fisher R, Blair D, and Mavrothalassitis G. 1995. ERF: an ETS domain protein with strong transcriptional repressor activity, can suppress ets-associated tumorigenesis and is regulated by phosphorylation during cell cycle and mitogenic stimulation. EMBO /. 14: 4781-4793. Shanblatt S, and Revzin A. 1987. Interactions of the catabolite activator protein (CAP) at the galactose and lactose promoters of Escherichia coli probed by hydroxyl radical footprinting. The second CAP molecule which binds at gal and the one CAP at lac may act to stimulate transcription in the same way. /. Biol. Chem. 262: 11422-11427. Sheridan P, Sheline C, Cannon K, Voz M, Pazin M, Kadonaga J, and Jones K. 1995. Activation of the HIV-1 enhancer by the LEF-1 HMG protein on nucleosome-assembled DNA in vitro. Genes. Dev. 9: 2090-2104. Shokat K. 1995. Tyrosine kinases: Modular signalling enzymes with tunable specificities. Curr. Biol. 2: 509-514. Shore P, Bisset L, Lakey J, Waltho J, Virden R, and Sharrocks A. 1995. Characterization of the Elk-1 DNA-binding domain. /. Biol. Chem. 270: 5805-5811. Bibliography 225 Sibanda B, and Thornton J. 1985. (3-hairpin families in globular proteins. Nature 316:170-174. Sibanda B, Blundell T, and Thornton J. 1989. Conformation of (3-hairpins in protein structures: A systematic classification with applications to modelling by homology, electron density fitting and protein engineering. /. Mol. Biol. 206: 759-777. Skalicky J, Donaldson L, Petersen J, Graves B, and Mcintosh L. 1996. Structural coupling of the inhibitory domain flanking the ETS domain of murine Ets-1. Prot. Sci. in press. Skelton N , III AP, Akke M, Kordel J, Ranee M, and Chazin W. 1993. Practical aspects of two-dimensional proton-detected 1 5 N spin relaxation measurements. /. Magn. Reson. 102: 253-264. Smith D, and Johnson K. 1988. Single-step purification of polypeptides expressed in Escherichia coli as fusion with glutathione S-transferase. Gene 67: 31-40. Sorensen P, Lessnick S, Lopez-Terrada D, Liu X, Triche T, and Denny C. 1994. A second Ewing's sarcoma translocation, t(21;22), fuses the EWS gene to another ETS-family transcription factor, ERG. Nature Genet. 6:146-151. Spera S, Ikura M , and Bax A. 1991. Measurement of the exchange rates of rapidly exchanging amide protons: Application to the study of calmodulin and its complex with a myosin light chain kinase fragment. /. Biomol NMR. 1:155-165. Stark M , and Johnson A. 1994. Interaction between two homeodomain proteins is specified by a short C-terminal tail. Nature 371: 429-432. States D, Haberkorn R, and Ruben D. 1982. A two-dimensional nuclear Overhauser experiment with pure absorption phase in four quadrants. /. Magn. Reson. 48: 286-292. Steitz T. 1990. Structural studies of protein-nucleic acid interactions: the sources of sequence specific binding. Q. Rev. Biophys. 23: 205-280. Studier F, and Moffatt B. 1986. Use of Bacteriophage T7 RNA polymerase to direct selective high-level expression of cloned genes. /. Mol. Biol. 189:113-130. Sun W, Graves B, and Speck N. 1995. Transactivation of the moloney murine leukemia virus and T-cell receptor (3-chain enhancers by cbf and ets requires intact binding sites for both proteins. /. Virol. 69: 4941-4949. Surewicz W, Mantsch H, and Chapman D. 1993. Determination of protein secondary structure by fourier transform infrared spectroscopy: A critical assessment. Biochemistry 32: 389-394. Sutherland C, and Walsh M. 1989. Phosphorylation of caldesmon prevents its interaction with smooth muscle myosin. /. Biol. Chem. 264: 578-583. Bibliography 226 Suzuki M . 1993. Common features in DNA recognition helices of eukaryotic transcription factors. EMBO J. 12: 3221-3226. Suzuki M. 1994. A framework for the DNA-protein recognition code of the probe helix in transcription factors: The chemical and stereochemical rules. Curr. Biol. 2: 317-326. Swindells M. 1995. Identification of a common fold in the replication terminator protein suggests a possible mode for DNA binding. Trends Biochem. Sci. 20: 300-302. Thompson C, Brown T, and McKnight S. 1991. Convergence of Ets- and Notch-related structural motifs in a heteromeric DNA binding Complex. Science 253: 762-766. Treier M , Bohmann D, and Mlodzik, M. 1995. JUN cooperates with the ETS domain protein Pointed to induce photoreceptor R7 fate in the Drosophila eye. Cell 83:753-760. Wagner G. 1993. NMR relaxation and protein mobility. Curr. Opin. Str. Biol. 3: 748-754. Walter J, Dever C, and Biggin M . 1994. Two homeodomain proteins bind with similar specificity to a wide range of DNA sites in Drosophila embroys. Genes Dev. 24: 1678-1692. Wang Y-C, Petryniak B, Thompson C, Kaelin W, and Leiden J. 1993. Regulation of the Ets-related transcription factor Elf-1 binding to the retinoblastoma protein. Science 260:1330-1335. Wasylyk B, Wasylyk C, Flores P, Begue A, Leprince D, and Stehelin D. 1990. The c-ets proto-oncogenes encode trasnscription factors that cooperate with c-Fos and c-Jun for transcriptional activation. Nature 346:191-193. Wasylyk C, Kerckaert J, and Wasylyk B. 1992. A novel modulator domain of Ets transcription factors. Genes Dev. 6: 965-974. Wasylyk C, and Wasylyk B. 1993. Oncogenic conversion of Ets affects redox regulation in vivo and in vitro. Nucl. Acids Res. 21: 523-529. Wasylyk B, Hahn S, and Giovane A. 1993. The Ets family of transcription factors. Eur. J. Biochem. 211: 7-18. Watson D, McWilliams M, Lapis P, Lautenberger J, Schweinfest C, and Papas T. 1988. Mammalian ets-1 and ets-2 genes encode highly conserved proteins. Proc. Natl. Acad. Sci. USA. 85: 7862-7866. Weiss M , Ellenberger T, Wobbe C, Lee J, Harrison S, Struhl K. 1990. Folding transition in the DNA binding domain of GCN4 on specific binding to DNA. Nature 347: 575-578. Werner M, Huth J, Gronenborn A, and Clore G. 1995a. Molecular basis of human 46X,Y sex reversal revealed from the three-dimensional solution structure of the human SRY-DNA complex. Cell 81: 705-714. Bibliography 227 Werner M, Clore GM, Fisher CL, Fisher RJ, Trinh L, Shiloach J, and Gronenborn A M . 1995b. The solution structure of the human ETS1-DNA complex reveals a novel mode of binding and true side chain intercalation. Cell 83: 761-771. Wharton R, Brown E, and Ptashne M. 1984. Substituting an a-helix switches the sequence-specific DNA interactions of a repressor. Ce//38: 361-369. Wharton R, and Ptashne M . 1985. Changing the binding specificity of a repressor by redesigning a helix. Nature 316: 601-605. White C, Kempi N , and Komives E. 1994. Expression of highly disulfide-bonded proteins in Pichia pastoris. Structure 2:1003-1005. Whitmarsh A, Shore P, Sharrocks A, and Davis R. 1995. Integration of MAP kinase signal transduction pathways at the serum response element. Science 269: 403-407. Wider G, and Wuthrich K. 1993. A simple experimental scheme using pulsed field gradients for coherence-pathway rejection and solvent suppression in phase-sensitive heteronuclear correlation spectra. /. Magn. Reson. 102: 239-241. Wilkens S, Dahlquist F, Mcintosh L, Donaldson L, and Capaldi. R. 1995. Structural features of the 8 subunit of the Escherichia coli ATP synthase determined by NMR spectroscopy. Nature Str. Biol. 2: 961-967. Wilkinson D, Ma N , Haught C, and Harrison R. 1995. Purfication by immobilized metal affinity chromatography of human atrial natriuretic peptdide expressed in a novel thioredoxin fusion protein. Biotechnol. Prog. 11: 265-269. Williams K, and Konigsberg W. 1991. Identification of amino acid residues at interface of protein-nucleic acid complexes by photochemical crosslinking. Meth. Enzymol. 208: 516-539. Willis M , Hicke B, Uhlenbeck O, Cech T, and Koch T. 1993. Photocrosslinking of 5-iodouracil-substituted RNA to DNA and proteins. Science 262:1255-1257. Wishart D, Richards F, and Sykes B. 1992. The chemical shift index: A fast and simple method for the assignment of protein secondary structure through NMR spectroscopy. Biochemistry 31:1647-1651. Wishart D, and Sykes B. 1994. /. Biomol. NMR. 4:171-180. Wittekind M, and Mueller L. 1993. HNCACB, a high-sensitivity 3D NMR experiment to correlate amide-proton and nitrogen resonances with the alpha- and beta-carbon resonances in proteins. /. Magn. Reson. 101: 201. Woods D, Ghysadel J, and Owen M. 1992. Identification of nucleotide preferences in DNA sequence recognised specifically by c-Ets-1 protein. Nucl. Acids. Res. 20: 699-704. Wotton D, Ghysdael J, Wang S, Speck N , and Owen M. 1994. Cooperative binding of Ets-1 and core binding factor to DNA. Mol. Cell. Biol. 14: 840-850. Bibliography 228 Wuthrich K. 1986. NMR of Proteins and Nucleic Acids. Wiley:New York. Xin J, Cowie A, Lachance P, and Hassell J. 1992. Molecular cloning and characterization of PEA3, a new member of the Ets oncogene family that is differentially expressed in mouse embryonic cells. Genes. Dev. 6: 481-496. Yamazaki T, Yoshida M , and Nagayama K. 1993. Complete assignments of magnetic resonance of ribonuclease H from Escherichia coli by double- and triple-resonance 2D and 3D NMR spectroscopies. Biochemistry 32: 5656-5669. Zhang L, Lemarchandel V, Romeo P-H, Ben-David Y, Greer P, and Bernstein A. 1993. The Fli-1 proto-oncogene, involved in erythroleukemia and Ewing's sarcoma, encodes a transcriptional activator with DNA-binding specificities distinct from other Ets family members. Oncogene 8:1621-1630. Zhu G, and Bax A. 1992. Improved linear prediction of damped NMR signals using modified "forward-backward" linear prediction. /. Magn. Reson. 100: 202-207. Appendix Table A . l : Ets-1AN331 Chemical Shift Assignments Table A.2 : Hydrogen Exchange Results for Ets-1AN301 Table A.3: Materials and Methods Quick Reference 229 Table A.l: Main chain 'H, 1 3C, and 1 5 N resonance assignments for murine Ets-1 AN331 at pH 6.45 and 20°C.a Residue HN C« (H«) CP(HP) C* 0^ 330)* 34.1 (3.84) 29.0 (3.84 , 2.20) 178.2. G331 115.1 (8.44) 45.5 (4.42) 45.5 (4.80 , 4.04) 174.5 S332 117.1 (8.45) 54.6 (3.96) 59.0 (4.63) 174.5 G333 112.7 (8.49) 44.8 (4.80, 4.20) P334 63.2 (4.55) 32.4 (2.32, 1.91) 177.2 1335 124.1 (8.29) 61.4 (4.31) 39.8 (1.98) 174.0 Q336 125.0 (8.23) 54.0 (4.92) 31.8 (1.41 , 1.58) 177.9 L337 126.9 (9.35) 58.8 (3.01) 40.1 (2.07 , 1.26) 178.0 W338 115.2 (7.58) 60.2 (4.00) 27.5 (3.28 , 2.31) 175.9 Q339 121.1 (5.66) 58.0 (4.15) 28.1 (1.84) 178.6 F340 124.1 (8.07) 61.0 (4.23) 40.1 (3.13 , 2.92) 176.4 L341 118.4 (8.18) 58.2 (3.51) 40.4 (1.91 , 1.86) 178.2 L342 118.4 (7.16) 57.9 (3.91) 41.9 (1.62, 1.62) 178.3 E343 122.4 (7.82) 59.8 (3.80) 29.3 (2.42 , 2.10) 179.6 L344 119.0 (7.55) 57.7 (4.11) 42.2 (1.78 , 1.26) 180.0 L345 119.1 (7.98) 56.9 (3.76) 42.4 (1.53 , 1.63) 178.6 T346 110.0 (7.55) 61.2 (4.65) 64.1 (3.93) 179.3 D347 124.2 (7.15) 52.2 (4.95) 42.7 (3.34 , 2.63) 176.4 K348 126.8 (9.00) 59.5 (4.04) 32.8 (1.96, 1.61) 178.3 S349 115.6 (8.73) 61.2 (4.80) 53.1 (4.26 , 3.90) 176.4 C350 121.8 (8.14) 59.6 (4.42) 27.8 (3.32 , 2.48) 174.5 Q351 119.6 (7.14) 58.2 (4.40) 28.2 (2.15) 176.8 S352 114.2 (8.29) 52.1 (3.92) 60.5 (4.26) 174.7 F353 117.2 (7.49) 56.2 (5.13) 41.4 (3.35 , 3.22) 174.6 1354 124.6 (7.71) 60.7 (5.19) 39.9 (1.89) 178.6 S355 118.7 (8.50) 56.9 (4.84) 57.6 (4.02 , 3.97) 173.9 W356 125.4 (8.92) 56.9 (5.37) 29.9 (4.01 , 2.98) 178.0 T357 113.6 (8.73) 63.0 (4.38) 65.0 (4.52) 176.8 G358 111.5 (8.94) 44.7 (4.28, 2.99) 179.6 D359 123.6 (7.71) 51.9 (4.91) 41.4 (2.87 , 2.45) 175.6 G360 114.7 (8.80) 47.4 (3.72, 3.21) 175.0 W361 129.0 (9.32) 54.4 (5.79) 30.2 (4.35 , 3.77) 176.4 E362 125.0 (8.43) 56.7 (5.45) 31.7 (2.38) 175.9 F363 129.0 (10.33) 56.7 (5.85) 43.8 (2.92 , 2.78) 176.8 K364 120.5 (9.67) 53.7 (5.48) 37.5 (1.63 , 1.75) 175.2 L365 125.4 (8.86) 53.5 (4.97) 40.9 (1.28 , 1.15) 174.6 Residue N C« (Ha) CP (HP) C S366 122.3 (7.57) 52.3 (3.81) 61.1 (4.33) 175.2 D367 118.5 (8.03) 51.3 (5.07) 41.8 (3.25 , 2.66) 173.3 P368 65.3 (3.91) 32.3 (1.52 , 1.44) 177.9 D369 117.5 (7.84) 57.7 (4.49) 40.4 (2.84 , 2.75) 178.6 E370 124.8 (7.61) 58.0 (4.06) 28.8 (2.01 , 2.01) 178.5 V371 119.9 (7.59) 68.0 (3.46) 31.3 (2.28 , 2.29) 177.4 A372 121.1 (8.23) 55.8(4.13) 17.8 (1.46) 179.3 R373 120.5 (8.42) 59.9 (4.08) 30.8 (2.01 , 1.89) 180.5 R374 120.5 (8.35) 55.6 (4.67) 40.2 (2.94 , 2.75) 177.1 W375 126.3 (8.75) 58.4 (4.14) 30.7 (3.13 , 3.27) 176.8 G376 107.3 (8.70) 48.2 (3.92, 3.28) 178.3 K377 122.5 (7.78) 58.9 (4.07) 32.3 (1.93) 179.1 R378 121.8 (7.68) 56.7 (4.08) 29.6 (1.79 , 1.64) 177.2 K379 116.3 (7.52) 52.7 (3.94) 30.9 (1.15 , 0.36) 175.2 N380 118.9 (7.49) 54.3 (4.27) 36.9 (2.69 , 3.15) 173.9 K381 119.6 (8.70) 52.8 (4.77) 34.3 (3.17 , 2.99) 173.7 P382 64.2 (4.32) 32.5 (2.35 , 1.96) 178.0 K383 118.8 (8.64) 55.5 (4.41) 30.6 (2.02 , 1.85) 176.6 M384 123.0 (7.86) 55.9 (4.33) 29.9 (1.97 , 2.20) 174.0 N385 122.9 (7.42) 51.6 (4.86) 40.1 (3.09 , 3.40) 175.2 Y386 122.3 (9.63) 62.5 (4.22) 37.7 (3.11 , 3.39) 176.3 E387 124.6 (8.56) 60.7 (3.80) 28.8 (2.14 , 2.37) 179.6 K388 121.2 (8.38) 59.8 (4.03) 33.9 (2.10, 1.73) 180.1 L389 126.7 (8.57) 59.4 (4.08) 42.8 (1.59 , 1.59) 178.6 S390 116.1 (8.97) 62.4 (3.89) (3.67 , 3.67) 177.1 R391 124.8 (7.65) 59.9 (4.05) 30.1 (1.99 , 3.71) 179.1 G392 110.5 (7.90) 46.9 (3.92, 3.71) 176.0 L393 122.3 (8.03) 58.4 (4.12) 41.4 (1.63 , 1.63) 178.4 R394 119.6 (7.70) 58.5 (3.02) 29.8 178.7 Y395 121.2 (7.33) 60.7 (4.28) 38.4 (3.10, 2.82) 176.4 Y396 114.7 (7.52) 5 9 . 4 (4.43) 38.8 (3.32 , 3.32) 178.0 Y397 124.1 (8.13) 58.9 (4.23) 38.8 (3.11 , 2.93) D398 K399 118.8 (8.57) 54.8 32.9 175.4 N400 116.9 (8.32) 54.9 (4.16) 37.1 (3.03,2.82) 173.4 1401 115.2 (7.68) 63.2 (4.09) 40.4 (1.46) 175.8 Residue N C« (Ha) CP(HP) c 1402 119.7 (6.93) 59.7 (5.08) 43.9 (1.55) 173.9 H403 124.5 (9.09) 53.0 (5.13) 34.4 (3.18 , 2.99) 173.7 K404 126.7 (8.47) 56.3 (4.25) 34.1 (1.60, 1.60) 176.3 T405 127.5 (8.10) 63.8 (4.11) 68.9 (3.60 , 3.60) 182.0 A406 107.4 (8.77) 53.4 (4.15) 18.6 (1.42) 178.4 G407 110.9 (8.73) 45.7 (4.60, 4.14) 174.5 K408 119.5 (7.09) 54.0 (4.70) 35.7 R409 Y410 V411 61.8 (4.91) 33.8 (1.51) 174.6 Y412 128.8 (8.75) 56.1 (4.87) 43.7 (3.41 , 2.46) 180.3 R413 119.1 (9.24) 55.3 (5.30) 36.3 177.2 F414 126.0 (8.30) 60.0 (5.25) 40.4 (3.43 , 3.22) 176.5 V415 117.0 (8.26) 61.6 (4.53) 31.6 (2.23) 175.4 C416 118.1 (7.36) 54.9 (4.88) 29.3 (-3.17 , 2.96) 173.2 D417 120.6 (8.60) 53.5 (4.62) 39.2 (3.01 , 2.71) 176.9 L418 127.3 (7.42) 56.4 (3.84) 40.7 (0.05 , 0.36) 178.0 Q419 122.7 (8.62) 59.8 (4.48) 28.1 (2.40 , 2.26) 179.6 S420 115.8 (7.81) 51.7 (3.82) 61.3 (4.22) 175.6 L421 122.5 (7.26) 57.5 (4.22) 43.4 (1.58) 178.3 L422 116.4 (8.57) 55.4 (4.33) 43.5 (1.76, 1.38) 174.8 G423 107.4 (7.93) 45.5 (3.86) 175.2 Y424 117.4 (6.90) 56.2 (5.13) 43.0 (3.43 , 2.43) 175.0 T425 111.5 (8.53) 60.1 (4.60) 62.8 (3.93) 173.3 P426 65.5 (3.01) 30.1 (1.20, 0.85) 177.1 E427 115.3 (8.26) 60.5 (3.76) 28.6 (1.96, 1.85) 180.2 E428 121.4 (7.64) 58.8 (3.99) 30.6 (2.45 , 2.28) 179.2 L429 123.3 (8.07) 57.9 (3.84) 41.6 (1.54, 1.34) 178.8 H430 118.4 (8.77) 57.7 (4.20) 30.6 (3.22 , 3.08) 177.9 A431 121.4 (7.60) 54.8 (4.32) 18.1 (1.54) 180.6 M432 119.7 (7.78) 58.4 (4.21) 33.1 (2.21) 177.5 L433 119.5 (7.72) 54.7 (4.35) 43.0 (1.81 , 1.54) 175.6 D434 120.3 (7.82) 55.2 (4.49) 40.2 (3.00 , 2.68) 175.4 V435 121.0 (8.13) 63.2 (3.97) 32.9 (1.95) 176.0 K436 P437 129.9 (8.49) 53.6 (4.76) 63.3 (4.45) 32.9 (1.86, 1.79) 174.1 32.3 (2.34, 1.99) 176.7 Residue N C a ( Ha) CP (HP) C D438 122.7 (8.47) 54.6 (4.60) 41.4 (2.74 , 2.66) 175.7 A439 126.4 (8.23) 52.4 (4.41) 19.8 (1.43) 176.5 D440 127.0 (8.01) 56.2 (4.41) 42.4 (2.71 , 2.61) 181.0 a The 'H chemical shift(s) associated with each heteronucleus is given in parentheses. HP are not stereospecifically assigned. The 'H and 1 3C shifts are reported in ppm relative to sodium 2,2-dimethyl-2-silapentane-5-sulfonate (DSS) at 0.0 ppm, and the l 5 N shifts in ppm relative to 2.9 M 15NH4C1 in 1 M HC1 at 24.93 ppm. b The analysis of an HPLC-purified sample of Ets-1 AN331 by mass spectrometry and Edman sequencing indicated that the protein lacked a terminal methionyl residue. However, weak signals from the H«, C a , CP and C of the N-terminal Met330 and the amide 1 5 N and H N of Gly 331 were observed in the NMR spectra of Ets-1 AN331 prepared using standard FPLC chromatography (Chapter 2). This N-terminal heterogeneity, that results from incomplete processing of Ets-1 AN331 by methionine aminopeptidase during overexpression, was not observed by sequencing a portion of an HPLC-purified sample. 234 Table A.2 : Fractional NOE and ROE difference intensities (CjNOE / CROE), amide proton spin flip rates (pj), rotating frame relaxation rates (p2) and amide-water magnetization exchange rates (kN and k R) in Ets-1AN301 at pH 6.5 and pH 7.5. pH 6.5 pH 7.5 pH 6.5 pH7.5 Res.it rj^OE CROE fROE Pl+kN pj+kR p2+kR kN* pH6.5 kR pH7.5 kR 302 1 . 16 1 .00 20 . 2 33 .5 15 .0 (7 .6) 13 .8 (0 .6 ) 303 0 . 17 0 .08 1 .45 9 .4 36 .0 55 . 8 3 .3 (1 .4) 1 .1 (0 .2) 14 . 1 (1 • 4) 304 0 . 19 0 . 15 .2 . 69 15 . 3 54 .8 84 . 8 4 . 4 (2 • 8) 1 .4 (0 .3) 11 .3 (2 .4) 305 0 .06 0 . 05 1 00 12 . 5 55 . 6 68 . 5 1 .4 (0 .9) 0 .5 (0 .3 ) 7 . 0 (1 .0) 306 0 . 03 - 0 . 04 7 . 8 42 5 0 .7 (0 .4) - 0 5 (0 .9 ) 307 0 . 00 0 11 0 25 10 3 55 3 59 . 5 0 .0 (0 .0) 1 0 (0 .2) 2 .2 (0 .3 ) 308 0 11 43 . 1 1 . 6 (0 .6 ) 309 0 .02 0 13 0 58 8 9 46 4 54 9 0 4 (0 3) 1 5 (0 6) 6 3 (0 .8 ) 310 0 10 0 03 1 00 8 7 38 4 59 6 1 9 (0 9) 0 4 (0 2) 9 0 (0 9) 311 0 11 0 15 11 1 65 1 3 0 (1 3) 1 7 (0 2) 313 0 24 0 33 3 62 9 1 31 8 73 0 4 2 (1 6) 4 9 (0 2) 19 5 (3 2) 314 0 10 0 12 1 00 7 8 34 5 45 3 2 1 (0 8) 1 9 (0 1) 12 2 (0 7) 315- 0 30 0 37 6 63 10 6 33 5 113 4 5 2 (2 2) 5 2 (0 2) 11 7 (4 8) 316 0 55 0 65 1 98 14 3 45 1 54 9 8 9 (3 9) 7 6 (0 4) 16 8 (1 7) 317 0 28 0 35 2 28 11 7 38 3 67 7' 5 2 (2 5) 4 6 (0 2) 15 1 (2 0) 324 0 03 - 0 11 9 7 28 6 0 6 (0 5) -1 9 (2 5) 325 0 00 0 13 10 6 53 6 0 1 (0 3) 1 2 (1 3) 326 0 03 0 02 11 2 52 1 0 7 (0 4) 0 2 (0 3) 327 0 01 - 0 07 0 18 9 4 45 3 61 8 0 1 (0 1) - 0 7 (0 6) 1 5 (0 4) 328 0 05 0 02 0 36 10 0 46 7 47 1 1 0 (0 6) 0 2 (0 2) 4 1 (0 3) 329 0 08 0 11 0 32 12 6 51 7 55 2 1 8 (1 2) 0 8 (0 2) 2 8 (0 3) 331 0 15 0 17 1 28 9 1 45 2 57 9 2 8 (1 3) 1 7 (0 2) 11 2 (1 1) 332 0 29 0 46 6 76 12 0 43 3 170 2 5 4 (2 7) 5 2 (0 3) 2 2 (2 3) 333 0 19 0 29 9 6 33 4 3 5 (1 6) 3 8 (0 2) 335 0 07 0 37 2 35 12 9 56 8 84 6 1 6 (1 1) 2 9 (0 4) 10 1 (2 0) 336 0 04 0 26 0 22 13 4 65 6 73 2 1 1 (0 7) 1 6 (0 4) 1 2 (0 3) 337 0 00 0 61 19 0 67 8 0 0 (0 1) 3 7 (0 7) 338 0 06 - 0 07 - 0 03 10 4 49 0 53 5 1 2 (0 7) - 0 6 (0 2) - 0 3 (0 1) 339 0 03 0 03 - 0 01 13 2 66 6 71 9 0 7 (0 5) 0 2 (0 3) - 0 1 (0 2) 340 0 10 - 0 08 0 04 8 6 43 0 46 6 1 9 (0 9) - 0 9 (0 1) 0 5 (0 1) 341 0 01 0 06 - 0 05 14 0 65 9 66 6 0 3 (0 2) 0 3 (0 3) - 0 3 (0 3) 342 0 02 - 0 03 0 03 11 8 61 9 59 0 0 5 (0 4) - 0 2 (0 3) 0 3 (0 2) 343 0 06 - 0 06 - 0 06 12 3 58 4 62 2 1 4 (0 9) - 0 4 (0 2) - 0 4 (0 2) 344 0 04 0 01 0 01 13 9 68 3 72 1 1 0 (0 7) 0 0 (0 3) 0 1 (0 2) 346 0 02 - 0 03 1 00 11 7 55 3 57 1 0 5 (0 3) - 0 2 (0 3) 7 5 (0 9) 347 0 05 - 0 17 - 0 17 10 0 47 3 48 1 1 8 (0 2) 348 0 16 - 0 15 0 17 11 1 52 7 74 7 3 2 (1 7) -1 4 (0 3) 1 0 (0 2) 349 0 17 - 0 19 - 0 14 10 2 48 0 50 4 3 2 (1 6) -2 0 (0 2) -1 5 (0 2) 350 0 05 - 0 07 0 07 11 6 45 2 55 9 1 1 (0 7) - 0 8 (0 3) 0 6 (0 2) 351 0 03 - 0 01 - 0 11 11 0 53 1 57 5 1 0 (0 1) 352 0 00 0 04 2 55 16 6 54 9 55 2 0 0 (0 1) 0 3 (0 1) 22 5 (2 3) 353 0 01 0 07 - 0 04 9 4 50 6 50 8 0 2 (0 1) 0 6 (0 2) - 0 4 (0 1) 354 0 02 0 10 0 15 9 2 52 0 47 7 0 4 (0 2) 0 9 (0 2) 1 5 (0 2) 355 0 01 0 17 12 3 47 3 0 2 (0 2) 1 4 (0 2) 356 0 15 56 6 1 3 (0 2) 357 0 11 - 0 56 - 0 46 16 2 70 9 68 9 2 9 (2 0) - 3 7 (0 8) -3 2 (0 5) 358 0 16 - 0 87 -1 00 15 7 73 5 71 4 3 9 (2 6) - 5 1 (1 0) - 6 5 (1 0) 359 0 18 - 0 26 - 0 26 10 6 46 1 43 3 3 5 (1 8) -2 7 (0 2) - 3 1 (0 2) 360 0 14 - 0 06 0 11 10 4 41 8 54 2 2 7 (1 5) - 0 6 (0 2) 1 1 (0 2) HI1 HI2 HI SI 235 pH 6.5 pH7.5 pH6.5 pH7.5 pH6.5 Res.U QJOE (ROE £ROE Pi+kN pj+fcR p2+kR kNt kR pH7.5 kR 361 0 07 -0 12 0 02 13 3 54 6 65 9 1 6 (1 1) -1 0 (0 3) 0 1 (0 2) 362 0 19 -0 28 -0 30 11 7 54 8 56 3 3 8 (2 1) -2 5 (0 3) -2 7 (0 3) 363 0 02 0 13 0 01 14 8 56 3 47 5 0 6 (0 4) 1 0 (0 4) 0 1 (0 4) 364 0 02 -0 07 -0 10 13 0 60 0 56 7 0 6 (0 4) -0 5 (0 4) -0 8 (0 3) 365 0 05 0 00 -0 08 11 9 71 0 68 9 1 1 (0 7) 0 0 (0 4) -0 5 (0 3) 366 0 00 0 11 0 10 11 5 54 4 52 0 0 0 (0 3) 0 9 (0 3) 0 9 (0 2) 367 0 00 0 11 0 09 9 7 52 7 45 6 0 0 (0 5) 0 9 (0 2) 0 9 (0 1) 369 0 01 0 07 0 05 9 0 46 2 44 6 0 1 (0 1) 0 6 (0 1) 0 5 (0 1) 370 0 02 0 04 0 05 11 1 56 1 56 8 0 4 (0 3) 0 3 (0 2) 0 4 (0 1) 371 0 04 0 16 0 15 12 2 67 8 66 2 0 9 (0 6) 0 9 (0 3) 0 9 (0 2) 372 0 03 0 10 0 09 10 5 54 8 55 8 0 6 (0. 4) 0 7 (0 2) 0 8 (0 2) 373 0 05 0 12 0 06 10 1 54 0 51 6 1 1 (0 7) 1 0 (0 2) 0 8 (0 2) 374 0 04 0 11 0 03 12 9 62 9 62 6 1 0 (0 7) 0 7 (0 3) 0 2 (0 2) 375 0 00 0 09 -0 01 13 0 56 6 61 8. 0 0 (0 1) 0 7 (0 3) -0 1 (0 2) 376 0 03 -0 29 0 04 13 4 62 8 53 1 0 6 (0 4) -1 9 (0 4) 0 4 (0 2) 377 0 01 -0 01 0 04 10 0 37 7 47 0 0 2 (0 1) -0 2 (0 4) 0 4 (0 2) 379 0 03 0 06 0 47 12 8 54 9 50 2 0 8 (0 5) 0 5 (0 7) 3 8 (0 3) 380 0 00 0 34 0 70 9 3 48 9 63 9 0 0 (0 3) 3 3 (0 4) 5 2 (0 6) 381 0 00 0 02 0 05 9 4 39 4 50 2 0 0 (0 .2) 0 2 (0 5) 0 5 (0 .2) 384 0 06 0 21 0 64 10 3 40 5 56 5 1 3 (0 7) 2 4 (0 2) 5 7 (0 5) 385 0 00 0 39 0 70 11 6 49 3 54 8 0 0 (0 1) 3 7 (0 5) 6 6 (0 6) 386 0 49 0 73 18 9 64 7 9 2 (6 4) 4 0 (1 4) 387 0 08 -0 88 0 51 9 3 50 3 57 0 1 6 (0 8) -9 4 (0 8) 4 8 (0 4) 388 0 01 0 02 0 08 11 0 54 9 59 6 0 3 (0 2) 0 2 (0 2) 0 7 (0 2) 389 0 00 0 00 0 17 11 4 67 9 66 8 0 0 (0 1) 0 0 (0 3) 1 2 (0 3) 390 0 07 -0 15 0 04 10 9 50 6 56 0 1 6 (1 0) -1 4 (0 3) 0 4 (0 2) 391 0 00 0 09 0 12 9 9 77 9 52 2 0 0 (0 1) 0 5 (0 1) 1 4 (0 2) 392 0 03 0 01 0 16 11 4 47 3 . 58 6 0 8 (0 5) 0 1 (0 2) 1 3 (0 2) 393 0 04 0 04 0 16 9 5 53 5 67 3 0 8 (0 4) 0 4 (0- 3) 1 1 (0 2) 394 0 03 0 04 0 10 10 5 54 7 57 8 0 6 (0 4) 0 3 (0 2) 0 9 (0 2) 395 0 11 0 27 2 31 11 3 52 2 62 2 2 4 (1 3) 2 7 (0 4) 14 6 (1 8) 396 0 00 0 10 0 .17 9 7 47 5 50 2 0 0 (0 3) 0 9 (0 3) 1 5 (0 2) 397 0 22 ' 46 8 2 3 (0 4) 400 0 07 0 34 1 00 10 9 55 '1 73 0' 1 5 (0 9) 3 1 (0 5) 5 9 (0 8) 401 0 00 -0 06 0 00 12 9 62 8 57 0 0 0 (0 2) -0 4 (0 5) 0 0 (0 3) 402 0 13 -0 11 0 14 7 7 32 2 39 9 2 4 (0 9) -1 6 (0 1) 2 0 (0 1) 403 -0 34 55 4 -3 3 (0 3) 404 0 28 0 95 15 0 80 2 5 9 (3 5) 4 5 (0 8) 405 0 00 0 11 0 66 13 3 63 4 74 1 0 0 (0 1) 0 1 (0 0) 4 1 (0 6) 406 0 00 0 68 39 05 13 5 59 5 113 4 0 0 (0 1) 6 1 (0 6) 30 6 (16 8) 411 0 60 0 64 20. 6 54 1 11 6 (7 2) 6 4 (0 6) 412 0 11 -0 21 -0 17 12 4 49 0 58 2 2 6 (1 6) -2 1 (0 4) -1 4 (0 2) 413 0 03 -0 07 -0 02 11 1 57 4 57 8 0 6 (0 4) -0 5 (0 2) -0 1 (0 2) 414 0 38 65 4 2 3 (0 4) 416 0 04 0 15 0 37 10 7 48 6 58 7 0 8 (0 5) 1 3 (0 2) 3 2 (0 3) 418 0 11 -0 26 -0 29 12 1 47 8 52 6 2 3 (1 5) -2 4 (0 5) -3 2 (0 5) 419 0 07 -0 06 0 24 10 4 56 0 67 0 1 5 (0 8) -0 5 (0 3) 2 0 (0 4) 420 0 12 0 10 0 89 6 5 46 3 60 5 2 2 (0 8) 1 1 (0 2) 7 0 (0 7) 421 0 00 0 04 -0 02 11 2 48 0 53 7 0 0 (0 1) 0 4 (0 2) -0 2 (0 2) 422 0 00 0 05 0 04 16 2 74 4 66 6 0 0 (0 4) 0 3 (0 9) 0 3 (0 7) 423 0 01 0 08 0 11 10 9 54 8 63 1 0 1 (0 1) 0 1 (0 0) 0 8 (0 2) 424 0 03 0 04 0 02 9 2 43 6 42 7 0 6 (0 4) 0 4 (0 1) 0 3 (0 1) 425 0 00 0 01 0 00 • 11 5 54 0 58 1 0 0 (0 1) 0 1 (0 3) 0 0 (0 2) 236 pH 6.5 pH 7.5 pH 6.5 pH7.5 pH 6.5 pH7.5 Res. # (NOE (ROE (ROE pj+kN pj+kR p2+kR kNt kR kR 428 0 10 - 0 25 - 0 18 11 2 58 4 60 9 2 2 (1 2) -2 2 (0 3) -1 5 (0 2) 430 0 04 0 03 0 01 10 9 50 2 53 5 0 9 (0 5) 0 3 (0 2) 0 1 (0 1) 431 0 00 0 10 0 23 10 6 53 9 60 2 0 0 (0 1) 0 8 (0 2) 1 7 (0 2) 432 0 00 0 12 0 12 9 7 53 4 53 3 0 0 (0 0) 1 1 (0 2) 1 1 (0 1) 433 0 00 0 14 0 T_5 11 9 57 2 55 4 0 0 (0 2) 1 2 (0 2) 1 4 (0 2) 434 0 00 0 23 0 28 8 6 44 2 44 9 0 0 (0 5) 2 4 (0 2) 3 2 (0 2) 435 0 00 0 05 0 12 8 8 44 2 44 6 0 1 (0 1) 0 6 (0 1) 1 4 (0 1) 436 0 08 0 01 0 36 7 2 33 1 41 9 1 6 (0 6) 0 2 (0 1) 4 8 (0 2) 438 0 10 0 19 0 83 5 0 22 5 35 2 1 7 (0 4) 3 8 (0 1) 12 0 (0 5) 439 0 13 1 13 27 1 29 4 2 5 (0 1) 19 2 (0 8) 440 0 05 0 06 0 10 0 9 33 5 6 8 0 9 (0 0) 1 0 (0 0) 2 9 (0 1) H4 Errors associated with k N and kR are shown in brackets. Sidebars indicate a-helices and arrows indicate p-strands. A jagged line identifies a region in the inhibition module which displayed significant amide hydrogen exchange. 237 Table A.3 : Materials and Methods Quick Referenced Cloning of Ets-1 gene fragments Ets-1AN331 Chapter 2.3 Ets-1AN301 Chapter 5.4 Expression and Characterization Growth conditions Section 2.3.1, Section 3.3.1,Section 5.4.1 Thiol titrations Section 2.4.1 Analytical centrifugation Section 2.4.2 Ultraviolet Laser Crosslinking Section 2.5.1 CD spectroscopy Section 2.6.1, Section 5.4.1 FT-IR spectroscopy Section 2.6.2 NMR Methods Assignment Section 3.3.2, Section 5.4.2 Restraint analysis Section 2.4.1, Section 4.3.1, Section 5.4.3 Structure generation Section 4.3.4, Section 4.3.5, Section 5.4.3 Structure analysis Section 4.3.5, Section 5.4.3 Paramagnetic spin-relaxation Section 5.4.4 Dynamics Section 5.5.3 Amide hydrogen exchange Section 3.2.3, Section 5.6.2 ^Please also refer to the Table of Contents 


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