Structural and Biochemical Studies of Ribonucleotide

STRUCTURAL AND BIOCHEMICAL STUDIES OF RIBONUCLEOTIDE

REDUCTASE INHIBITION BY dATP AND Sml1

SANATH RANJAN WIJERATHNA

Submitted in partial fulfillment of the requirement

For the degree of Doctor of Philosophy

Dissertation Advisor: Dr. Chris G. Dealwis, Ph.D.

Department of Pharmacology

CASE WESTERN RESERVE UNIVERSITY

August 2012

CASE WESTERN RESERVE UNIVERSITY

SCHOOL OF GRADUATE STUDIES

We hereby approve the thesis/dissertation of

Sanath Ranjan Wijerathna

Candidate for the Ph. D. degree*

Signed John J. Mieyal (Chair of the Committee)

Chris G. Dealwis

Thomas Radivoyevitch

Masaru Miyagi

Robert Bonomo

Date 6/13/2012

 We also certify that the written approval has been obtained by for any

proprietary material contained therein.

DEDICATION

This thesis is dedicated to my family for their love and encouragement throughout

the years

Table of Contents

Table of contents i

List of tables vi

Figures vii

Appendix x

Acknowledgement xi

List of Abbreviations xiv

Abstract 1

Chapter 1 Introduction and Background 3

1.1 An Overview of Ribonucleotide Reductase 3

1.2 Classification of RNRs 6

1.3 The Catalytic Mechanism 12

1.3.1. Radical Generation and Transport 13

1.3.2 Substrate Reduction 15

1.3.3 Regeneration of the Active Site 17

1.4 Regulation of Class Ia Ribonucleotide Reductase 18

1.4.1 Allosteric Regulation of Ribonucleotide Reductase 19

1.4.2 Transcriptional Regulation of Ribonucleotide Reductase 30

1.4.3 Regulation of RNR by Subunit Compartmentalization 36

1.4.4 Regulation of RNR by Small Protein Inhibitors 38

1.4.5 Regulation of RNR by Selective Protein Degradation 43

1.5 Inhibitors of Ribonucleotide Reductase 44

1.5.1 Translational Inhibitors 44

1.5.2 Inhibitors of the Large Subunit 44

1.5.3. Inhibitors of the Small Subunit of RNR 47

1.6 Summary, Rationale and Aims of the Current Study 48

Chapter 2 Materials and Methods 52

2.1 Protein Expression and Purification 52

2.1.1 Site Directed Mutagenesis 52

2.1.2 Preparation of the Peptide Affinity Column 52

2.1.3 Protein Concentration Determination 53

2.1.4 Expression & Purification of ScRR1 55

2.1.5 Expression & Purification of ScRR2●ScRR4 58

2.1.6 Expression & Purification of Sml1 60

2.1.7 Expression & Purification of HuRRM1 63

2.1.8 Expression & Purification of HuRRM2 65

2.2 Labeling of Sml1 65

2.3 Determining the Enzymatic Activities of RNRs 67

2.3.1 Iron Loading to the Small Subunit of RNR 67

2.3.2 Preparation of Boronate Columns 71

2.3.3 Preparation of Radioactive Stocks 71

2.3.4 Determining the Specific Activities of the Large &

Small Subunits of RNRs 72

2.3.5 Determining the Effect of dATP on Human &

Yeast Enzyme Activity 73

2.3.6 Determining the Mode of Inhibition of ScRR Activity

by Sml1 73

2.3.7 Determining the IC50 Value of HuRRM1 for Sml1 74

2.3.8 Enzyme Kinetics and Data Analysis 74

2.4 Size Exclusion Chromatography 75

2.4.1 Characterization of HuRRM1 Oligomers 76

2.4.2 Purification of the ScRR1 Hexamer 76

2.4.3 Purification of the SCRR1●dATP Holo Complex 77

2.4.4 Purification of the ScRR1● TTP● Sml1 Complex 77

2.4.5 Purification of the dATP Induced ScRR1

Hexamer●Sml1 Complex 78

2.5 Crystallization Techniques 79

2.6 Protein Foot Printing 79

2.6.1 Selection of Buffers 79

2.6.2 Proteolysis & Mass Spectrometric Analysis 80

2.6.3 Identification of the Sites of Modification 81

2.6.4 Calculation of Peptide Modification Rates 81

2.7 Chemical Cross-linking 82

2.8 Fluorescence Spectroscopy 83

iii

Chapter 3 Nucleotide Induced Oligomerization of Human

and Yeast Ribonucleotide Reductase 85

3.1 Introduction 85

3.2 Results 86

3.3.1 Purification of HRRM1 and HRRM2 86

3.2.2 Size Exclusion Chromatography of ScRR1 and HuRRM1 90

3.2.3 dATP-induced Oligomerization 92

3.2.4 The Effect of Subunit Oligomerization on Enzyme

Activity 95

3.2.5 The Structural Basis of RNR Oligomerization 97

3.2.6 Validation of the dATP-Induced Hexamer by

Site-Directed Mutagenesis 100

3.2.7 ATP Hexamers Differ from dATP Hexamers 107

3.2.8 Purification and EM Analysis of ScRR1●dATP Holo

Complex 109

3.3 Discussion 112

Chapter 4 Kinetic Mechanism of ScRR Inhibition by Sml1 116

4.1 Introduction 116

4.2 Results 117

4.21 Characterization of C14SS60C Sml1 117

4.2.2 Sml1 Binds to the ScRR1 Hexamer 119

4.2.3 Enzymatic Activity of ScRR1 with Sml1 123

4.2.4 Mode of Inhibition of ScRR1 by Sml1 is non-linear 126

4.2.5 Identification of Sml1 like Proteins and Implication of

Sml1 in Therapeutics 137

4.3 Discussion 141

Chapter 5 Structural and Biochemical Characterization of Sml1

and ScRR1 Interactions 145

5.1 Introduction 145

5.2 Results 145

5.2.1 Identification of Buffers for Foot Printing 145

5.2.2 Protein Foot Printing 147

5.2.3 Chemical Cross-linking and Mass Spectrometry 156

5.2.4 Characterization of N-terminal Mutants 163

5.2.5 Peptide Inhibition Assay and Mutant activity 167

5.2.6 Determining the Cryo-EM Structure of ScRR1

Hexamer●Sml1 Complex 170

5.3 Discussion 174

Chapter 6 Summary and Future Directions 180

6.1 Inhibition of RNR by dATP 180

6.2 Inhibition of Yeast RNR by Sml1 186

Appendix 194

Tables

Table 2.1 Primers used in this Study 54

Table 2.2 Components of the Ferrozine assay 70

Table 3.1 Specific Activities of Wild-type and Mutant HuRR 89

Table 4.1 Specific Activity of ScRR1 in the Presence of dATP and

Sml1 124

Table 4.2: Estimation of Kinetic Parameters for Different Models

of ScRR1 Inhibition by Sml1 133

Table 5.1 Modification Rates of the ScRR1 Peptides in the

Presence and Absence of Sml1 152

Table 5.2 Cross-linked ScRR1•Sml1 Peptides Identified

by Mass Spectrometry 160

Figures

Figure 1.1 Role of RNR in Ribonucleotide Reduction 5

Figure 1.2 Functionally important sites in Class Ia RNR 8

Figure 1.3 Radical propagation pathway of Class I RNR 14

Figure 1.4 Catalytic Mechanism of Class Ia RNR 16

Figure 1.5 ATP and dATP binding to the ATP cone 22

Figure 1.6 Specificity Site Regulation in Eukaryotes 28

Figure 1.7 Regulation of Yeast RNR 37

Figure 1.8 Regulation of Ribonucleotide Reductase by Sml1 41

Figure 2.1 Purification of ScRR1 57

Figure 2.2 Purification of ScRR2•ScRR4 59

Figure 2.3 Purification of Sml1 62

Figure 3.1 Purification of HuRRM1 Subunit 87

Figure 3.2 Purification of HuRRM2 Subunit 88

Figure 3.3 Standard Curve for the Determination of Molar

Masses (Mr) of RNR 91

Figure 3.4 SEC Analysis of HuRRM1 and ScRR1 Oligomers 93

Figure 3.5 Specific Activity Measurement of HuRRM1 and

ScRR1 in the Presence of dATP 96

Figure 3.6 A Photograph of ScRR1●dATP Hexamer Crystals 98

Figure 3.7 Hexameric RNR1 based on the Low Resolution X-Ray

Crystal Structure of the ScRR1 Hexamer 99

FIGURE 3.8 Ribbon Diagram of the Hexamer Interface of

vii

HuRRM1 in Model B 101

FIGURE 3.9 Purification of the HuRR1 Mutants andTheir Specific

Activities 102

Figure 3.10 SEC Analysis of HuRRM1 Mutants 104

Figure 3.11 Activity of D16R in the Presence of dATP 106

Figure 3.12 The ATP hexamer Interface is Different from

that of the dATP hexamer 108

Figure 3.13 Purification of the ScRR1●dATP Holo Complex 110

Figure 3.14 Electron Micrograph of ScRR Holo Complex 111

Figure 4.1 Purification of C14S S60C Sml1 118

Figure 4.2 SEC Analysis of ATP and dATP-induced ScRR1

Hexamer 120

Figure 4.3 SEC Analysis of ScRR1 Hexamer•Sml1 Interactions

Using C14S S60C Sml1 121

Figure 4.4 Effect of Sml1 on Hexamer Formation 122

Figure 4.5 Sml1 Inhibition of ATP and dATP –induced Holo

Enzymes 125

Figure 4.6 Mode of Inhibition of ScRR1 by Sml1 in the Presence

of dGTP and [3H]-CDP 127

Figure 4.7 Mode of Inhibition of ScRR1 by Sml1 in the Presence

of dGTP and [14C]-ADP 129

Figure 4.8 Models Describing [3H] CDP and [14C] ADP Reduction 136

viii

Figure 4.9 Sml1 orthologs in fungi and Sml1 and

human RNR interactions 139

Figure 5.1 Identification of Buffers for Protein Foot Printing 146

Figure 5.2 Purification of ScRR1•TTP•Sml1 Complex 148

Figure 5.3 Sequence Coverage of X-ray Exposed ScRR1 upon

Proteolysis 150

Figure 5.4 Protein Foot-printing of ScRR1 and the Sml1•ScRR1

Complex 154

Figure 5.5 Chemical Cross-linking of ScRR1 and Sml1 157

Figure 5.6 Mass Spectrometric Analysis of Cross-link Fragments 161

Figure 5.7 Characterization of the N-terminal Mutant Δ22 ScRR1 164

Figure 5.8 Binding of ScRR1 or its Peptides to C14S S60C

Sml1 using Fluorescence Spectroscopy 166

Figure 5.9 Specific Activity Measurements with Mutants 168

Figure 5.10 Purification of dATP-induced ScRR1 Hexamer•Sml1

Complex and GFP Sml1 172

Figure 5.11 Characterization of GFP-Sml1 and ScRR1-dimer

Interactions 173

Appendix

Appendix 1A MALS Analysis of dATP-induced HuRRM1 Hexamer 194

Appendix 2A SEC analysis of RNR Holo Complex with t-HuRRM1 and

HuRRM2 196

Appendix 3A MALS Analysis of D16R HuRRM1 Mutant 197

Appendix 4A Cryo-EM Analysis of dATP-induced ScRR1

Hexamer Sml1 Complex 198

Appendix 5A Class Averages of GFP-Sml1 bound ScRR1 dimer and

Hexamer 199

Appendix 6A Circular Dichroism (CD) Absorption Spectrum of

wild-type and Δ 22 ScRR1 200

ACKNOWLEDGEMENTS

I would like to thank my advisor Dr. Chris Dealwis for allowing me to pursue my graduate studies in his laboratory, first at the University of Tennessee and subsequently at

Case Western Reserve University. I am grateful for his support over the years. I greatly appreciate his commitment to using multidisciplinary approach to problem-solving. I would like to thank my committee members Dr. Robert Hettich, Dr. Elias Fernandez, Dr.

Barry Bruce and Dr. Todd Reynolds at the University of Tennessee for early guidance in my research project. I would like to thank Dr. Robert Hettich at Oak Ridge National

Laboratory for introducing me to mass spectrometry and also for his excellent mentorship. I would like to thank Jana Kiseler for helping me interpret the mass spectrometry data. I would also like to thank Dr. Zongli Li and Dr. Thomas Walz who were instrumental with our cryo-EM studies.

I would also like to extend my heartfelt gratitude to my current committee members, Dr. John Mieyal, Dr. Tomas Radivoyevitch and Dr. Masaru Miyagi. They were instrumental in critical assessment of the data and manuscript preparation and also in molding me into a confident scientist. Special thanks go to Dr. John Mieyal, who was also the chair of my committee for his guidance as I completed my thesis. I would also like to thank Dr. Tomas Radivoyevitch for his help with statistical programming and many insightful discussions of the work described in this thesis. Special thanks also go to

Dr. Anthony Berdis who provided me with helpful insight and advice on enzyme kinetic data interpretation and manuscript preparation. I would also like to thank Dr. Danny

Manor who provided me helpful advice on fluorescence spectroscopy. I would also like

xi to thank Camala Thompson and other non-academic staff of the Department of

Pharmacology for their timely support that made my research progress smoothly.

All the past and current members of the Dealwis laboratory have contributed immensely to the work presented in this thesis and I greatly appreciate their friendship and dedication towards achieving my goal. I would like to extend my sincere gratitude to

Dr. Tomoaki Uchiki for guidance and support as I learned mass spectrometry in the initial stages of my research project. Dr. Brad Bennett was instrumental in providing me mental support and intellectual stimulation on a regular basis during his stay in the lab.

Dr. Hai Xu was helpful in as I learned protein crystallization, data collection and data analysis in the lab and at the Advanced Photon Source at the Argonne National

Laboratory. I would like to thank Dr. Anna Gardberg for providing me helpful advice on my experiments. My fellow graduate student Jim Fairman not only provided me moral support but was also instrumental over the years for the successful completion and publishing of many aspects of the material presented in Chapter 3 of this thesis. I greatly appreciate Dr. Catherine Faber for proofreading my thesis on such short notice. I would also like to thank Beth Helmbrecht, John Lamaccia, Jay Prendergast, Shalini Jha, Ryo

Nakano, Grant Zimmerman and Andrew Zhang for their help with my research projects. I would also like to thank Dr. Faiz Mohammed, Dr. Prem Kaushal and Dr. Giri

Gokulrangan for their continuous encouragement and helpful discussions as I finished this thesis.

I would like to acknowledge my undergraduate mentors, Dr. Senarath Athauda and Dr. Cyril Amarasinghe who encouraged me to engage in research. Dr. Senarath

Athaudha allowed me to work in his lab as an undergraduate student and taught me many

xii different techniques in protein chemistry. I would also like to thank the late Dr. Cyril

Amarasinghe, who first woke an enthusiasm for science in me with his hours-long descriptions of early developments in molecular biology. I would also like to thank Dr.

Jay Wimalasena and Dr. Shamila Rajarathna for helping me gain admission at the

University of Tennessee to pursue my graduate studies.

I greatly appreciate my parents for their continued support of my education to date. I would also like to thank my sister and her husband for their continuous support during this journey. Even though I cannot list each and every name, I pay my deepest respects to all of the teachers who were intimately involved in my primary and secondary education.

The work described in this thesis would not have been possible without the unconditional support and love of my wife and two children. I especially recognize my wife, Nirmala for her unwavering endurance, and love and the encouragement she has given me to accomplish this feat at last.

xiii

List of Abbreviations

A site Activity site

Ala Alanine

Arg Arginine

Asn Asparagine

ATP Adenosine triphosphate dATP Deoxyadenosine triphosphate

ASA Accessible Surface Are

Cys Cysteine

CD Circular Dichroism

C site Catalytic site cpm counts per minutes dNDP Deoxyriobonuclsoside 5’ diphosphates dNTP Deoxyriobonuclsoside 5’ triphosphates

Da Dalton

DTT Dithiothreitol

EDTA Ethylenediaminetetraacetic acid

E. coli Escherichia coli

EM Electron Microscopy

FA Fluorescence anisotropy

FQ Fluorescence quenching

FRET Fluorescence Energy Resonance Transfer

xiv

Gln Glutamine

Glu Glutamate

Gly Glycine

HEPES 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid

His Histidine

H. sapiens Homo sapiens

HuRRM2 small subunit of human RNR

HuRRM1 large subunit of human RNR

Ile Isoluecine

IPTG Isopropyl β-D-1-thiogalactopyranoside

LB Luria-Bertani media

Lys Lysine

KCl Potassium chloride kDa Kilodalton

Met Methionine mg/ml Milligrams per milliliter

MgCl2 Magnesium chloride

NaCl Sodium chloride

NDP Riobonuclsoside 5’ diphosphates

NTP Riobonuclsoside 5’ triphosphates

OD280 Optical density at 280 nanometers

OD600 Optical density at 600 nanometers

Phe Phenylalanine

PMSF Phenylmethylsulphonyl fluoride psi Pounds per square inch; unit of pressure

PF Protein foot prinitng

RNR Ribonucleotide Reductase

RNR1 Large subunit of ribonucleotide reductase

RNR2 Small subunit of ribonucleotide reductase

RNR3 Isoform of the large subunit of ribonucleotide reductase

found in Saccharomyces cerevisiae

RNR4 Isoform of the small subunit of ribonucleotide reductase

found in Saccharomyces cerevisiae

S. cerevisiae Saccharomyces cerevisiae

Ser Serine

S- site Specificity site

STORM Stochastic Optical Reconstruction Microscopy

TB Terrific Broth media

Thr Threonine

THI Three Helix Insert

TRIS Tris(hydroxymethyl)aminomethane

Trp Tryptophan

Tyr Tyrosine

TTP 2’-deoxythymidine 5’-triphosphate

Val Valine v/v Volume per volume

xvi w/v Weight per volume x g Gravity

xvii

Structural and Biochemical Studies of Ribonucleotide Reductase Inhibition

by dATP and Sml1

Abstract

SANATH RANJAN WIJERATHNA

Ribonucleotide reductase (RNR) catalyzes the conversion of ribonucleoside 5′- diphosphates (NDP) to deoxyribonucleoside 5′-diphosphates (dNDP). This enzyme maintains balanced and adequate deoxyribonucleoside triphosphate (dNTP) fluxes required for DNA replication and repair. All RNRs are subject to many different modes of regulation, principal among them, allosteric and transcriptional regulation. Of these mechanisms, the structural basis of dATP and ATP induced hexamer formation is less well understood. We hypothesized that hexamer formation is a prerequisite for dATP- induced inhibition of this enzyme. Structure determination of the dATP hexamer using

X-ray crystallography revealed two possible packing arrangements, which we called model A and B. Mutagenesis (D16R and H2E) and enzyme inhibition studies carried out in the presence of dATP produced results consistent with the packing arrangement seen in model B, where the dATP bound ATP binding cones interact in an anti-parallel manner forming the dATP hexamer. However, the ability of these mutants to form

1 hexamers in the presence of ATP suggests that the ATP-induced hexamer adopts an interface that differs from that of the dATP-induced hexamer. Consistent with published data, structure determination of a dATP induced RNR holo complex by cryo-EM revealed an α6β2 subunit arrangement. In Saccharomyces cerevisiae (Sc), a small protein inhibitor known as Sml1 inhibits RNR activity by binding to its large subunit (ScRR1).

Here we show that Sml1 binds to and inhibits both ATP-induced ScRR1 hexamers and dATP-induced ScRR1 hexamers. Based on kinetic studies, the mode of this inhibition of

ScRR by Sml1 appears to be S-linear I-parabolic non-competitive. The I-parabolic component of this model implies a cooperative mode of binding of Sml1 to both the

ScRR1 dimer and the ScRR1 hexamer. A database search revealed that Sml1 orthologs are absent in many fungi especially in those that cause life threatening infections in immunocompromised individuals. Our preliminary studies on the binding site of Sml1 on

ScRR1 suggest that Sml1 binds around the β/α barrel domain. Structural and mechanistic insights gained in the course of this work could pave the way toward the design of new

RNR-directed drugs that can be targeted against cancers and some fungal infections.

Chapter 1: Background and Introduction

1.1 An Overview of Ribonucleotide Reductase

The enzyme ribonucleotide reductase (RNR) is found in all cellular life forms and several species of viruses. All known RNRs catalyze the conversion of ribonucleoside 5’- di or tri phosphates to deoxyribonucleoside 5’-di or tri triphosphate providing the precursors essential for DNA replication and repair (Figure 1.1, p 5) [1, 2]. However, the uridine 5’-di or tri phosphate needs to be further converted to thymidine 5’- di or tri phosphate before incorporation into DNA [3]. RNRs use intricate allosteric regulatory mechanisms to maintain the adequate and balanced pools of dNTPs required for DNA replication and repair. An excess or deficiency of cellular dNTPs leads to genomic instability, which ultimately results in carcinogenesis or cell death [3-5]. Due to its central role in dNDP/dNTP synthesis, over the years RNR has become an attractive target for anti-cancer [6], antimicrobial [7-9] and antiviral therapy [10, 11].

The discovery of RNR dates back to 1960. Enzyme extracts obtained from bacterial, mammalian and avian sources and experiments conducted in vivo in different cell types indicated that deoxyribosyl compounds are derived by reducing corresponding ribosyl compounds [12-17]. These earlier experiments also identified ATP, Mg2+ and a reducing agent as required for the reduction of ribosyl compounds[14, 18, 19]. The enzyme system that converts ribonucleoside diphosphates (NDPs) to their corresponding deoxyribonucleoside diphosphates (dNDPs) was discovered by Reichard et al. and was named ribonucleotide reductase (RNR)[20].

The enzyme system found in the E. coli extract catalyzed the reduction of all four

NDPs, a process requiring two different protein fractions which were named B1(i.e. α subunit) and B2 (i.e. β subunit) in addition to thioredoxin [19, 20]. Different nucleoside triphosphates affected the rate of enzyme reaction to varying extents, and of these nucleotides ATP had a stimulatory effect on CDP reduction while dATP had an inhibitory effect [1, 21]. Interestingly, the catalytic activity of the β subunit of E. coli

RNR was dependent on non-heme iron [22, 23]. The spectrum of the β subunit showed an unusual absorption at 410nm that later led to the discovery of a free radical that is intimately connected with the enzyme’s activity[24]. Site-directed mutagenesis studies located this organic free radical on tyrosine 122 of the small subunit [25]. The metallocofactor assembly in the small β subunit provides the basis of classification of RNR into three main classes [2, 26].

This chapter will describe many aspects of RNR biology in depth, beginning with an overview of the classification of RNRs into three main classes based on their metallocofactor assembly. A description of the catalytic mechanism of RNRs, with reference to free radical generation, substrate reduction and active site regeneration, will follow.

RNRs have multiple regulatory mechanisms, some of which are organism specific. This chapter will provide a detailed account of allosteric and small protein mediated inhibition of RNRs, and include a concise description of RNR inhibitors used in clinical practice.

The rationale for attaining the overall objective of this thesis, specifically a deeper understanding of the molecular basis of the inhibitory mechanisms of RNR, will conclude this introduction.

Figure 1.1 Role of RNR in Ribonucleotide Reduction

Substrate reduction involves three main steps: 1) radical generation 2) substrate reduction

3) active site regeneration. Radical generation is shown for the three classes of RNR. The free radical is transferred to the catalytic site, generating a transient thiyl (S•) radical which is required for initiation of substrate reduction at the 2’OH (red) group. Class I and

Class II enzymes depend on two active site cysteines for the reducing equivalents required for substrate reduction. Class III enzymes oxidize formate to CO2 to obtain reducing equivalents. The figure was taken from [27] and modified.

1.2 Classification of RNRs

Functional RNRs are hetero-oligomeric complexes during substrate reduction in many organisms[2]. Most RNRs have a large (α) and a small (β) subunit. The large subunit carries out substrate reduction while the small subunit generates and stores the free radical required for catalysis. Three classes of RNR have been described based on their metal cofactors, which generate the free radicals essential for catalysis [26]. Class I enzymes generate a tyrosyl free radical in a diiron-oxygen cluster located in the β subunit [28]. Class II enzymes directly generate a deoxyadenosyl free radical using cobalt-containing cobalamin as the cofactor. Class III enzymes utilize a 4Fe-4S type iron sulfur cluster coupled to S-adenosylmethionine as their metal cofactor to generate a glycyl free radical [3]. In all three classes of RNR, these free radicals are ultimately delivered to a conserved active site cysteine residue to generate the thiyl radical (S•).

Class I enzymes:

Class I enzymes are expressed in almost all eukaryotes, in some prokaryotes, and in certain viruses [29]. The functional forms of eukaryotic Class I RNR have an active subunit composition of α2β2 and/or αnβm. The reduction of the NDP substrate occurs in the large α subunit, which also has two allosteric sites (the S-site and the A-site) in addition to the catalytic site (the C-site) (Figure 1.2, p 8) [30, 31]. The specificity site

(the S-site) determines which cognate substrate is reduced at the C-site, while binding of

ATP or dATP to the activity site (the A-site) determines the overall activity of the enzyme [21, 32-34]. Class I enzymes are further divided into three sub classes based on the organization of the RNR genes, the subunit topology, and the metal cluster assembly

6 of the β subunit [2, 35, 36].

In RNR Classes Ia and Ib, generation of the free radical (Y) in the β subunit is oxygen dependent. Therefore, these enzymes are functional under aerobic conditions.

The Class Ic enzymes do not generate Y but the synthesis of the active metal cluster in the β subunit requires oxygen [28].

Class Ia: Class Ia enzymes are found in almost all eukaryotic organisms, in prokaryotes, and in viruses that infect eukaryotes [2, 29]. The α subunit is encoded by the nrdA locus and the β subunit is encoded by the nrdB locus [37]. The β subunit possesses the diferric tyrosyl radical (FeIIIFeIII-Y.) required for nucleotide reduction [38]. The catalytic mechanism of Class Ia enzymes requires the transport of this free radical over a 30- 35 Å distance to the catalytic site [30, 39]. At the end of each catalytic cycle, active site cysteines become oxidized and their regeneration in Class Ia enzymes depends on thioredoxin or glutaredoxin based systems, which ultimately receives its reducing equivalents from NADPH [40]. The quaternary structures of this class of enzymes are complex and these enzymes undergo nucleotide dependent oligomerization. For this reason, different subunit compositions have been proposed where αn (β2)m (n= 2, 4, 6 : m= 1,2, 3)[41-46]. The binding of either ATP or dATP to the ATP binding cone at the N- terminus of this class of enzymes results in oligomerization of the α subunit. These oligomers then bind the corresponding β subunits and yield α4β4 oligomers in E. coli and

α2β2 and α6(β)m oligomers in eukaryotes [41, 46, 47].

Figure 1.2 Functionally important sites in Class Ia RNR.

X-ray crystal structure of the large subunit of yeast RNR (PDB ID 2CVS). Nucleotides in the S-site, C-site and A-sites are shown as solid objects. ScRR2•ScRR4 peptides define the P-site. Loop2 (white) and loop 1 (magenta) are close to the dimer interface.

Class Ib: Class Ib enzymes are found only among prokaryotes. These organisms include many notable human pathogens such as Mycobacterium tuberculosis [48], Bacillus anthracis, Staphylococcus aureus [49], and Salmonella typhimurium [50] . The Class Ib enzymes have a subunit composition of α2β2. In many organisms, the genes that encode subunits α and β are clustered with two other genes to form an operon nrdHIEF. The genes, nrdE and nrdF, code for α and β subunits, respectively. The functions of nrdH and nrdI have recently been elucidated. NrdH is a disulfide oxidoreductase with a characteristic CXXC motif and is thought to function as an electron donor to NrdE [51].

NrdI is a flavodoxin-like protein and is thought to be involved in the generation and maintenance of the active MnIIIMnIII-Y• cluster of NrdF [52]. In addition, Class Ib enzymes lack the N-terminal ATP binding cone [53].

Class Ic: Class Ic enzymes were initially described in Chlamydia species and later found in archea and eubacteria [54]. Compared to the other Class I enzymes, the α subunit of

Chlamydia spp. has an extended N-terminus, which is required for dATP mediated inhibition [36]. The β subunit has a metal cluster assembly quite different from those of

Class Ia and Class Ib. Structural and paramagnetic resonance studies have shown that the

β subunit lacks a tyrosine radical [54]. In Chlamydia species, this metallo-cluster has been identified as MnIVFeIII [55]. The conserved Y residue found in Class Ia and Ib is replaced by F127. The MnIVFeIII site seems to serve the function of Y• [56]. Apart from

Chlamidiae, several other human pathogens like M. Tuberculosis and Tropheryma whipplei also possess Class Ic β subunits [57]. The tyrosyl radical of RNR appears to be a key target of the antiproliferative effect of nitric oxide (NO). These alternative metal

9 clusters in Class Ic, and the lack of a radical, may be an adaptation to NO mediated toxicity [58] .

Class II:

Class II enzymes are found among prokaryotes and some lower order eukaryotes [29].

Class II RNRs are composed of an α subunit encoded by the nrdJ gene [2]. The α subunit has only the C-site and the S-site [59]. Many Class II RNRs also lack the A- site and therefore cannot be inhibited by dATP or activated by ATP or ATP. The free radical generation in Class II RNR is oxygen independent. The metallo-cofactor named 5- deoxyadenosylcobalamin binds to the α subunit and undergoes homolytic cleavage to yield the 5’-deoxyadenosyl radical [60, 61]. The radical then propagates across a distance of 6 Å via a conserved set of Tyr residues to the active site cysteine to generate the transient thiyl radical [62]. Unlike other classes of RNR, some Class II RNRs also reduce ribonucleoside 5’-triphosphates [59]. Several forms of quaternary structural arrangement of the α subunit have been reported. In most cases, the α subunit functions mainly as a monomer (α) or dimer (α2) [63, 64]. Recently, two crystal structures of Class II RNR have been solved [63, 65]. In the Lactobacilus leichmannii structure, the α subunit is a monomer. Unlike other Class II RNRs, the S-site is formed in the monomer itself by the insertion of additional 130 amino acid residues between two helices of the four helical bundle. The other crystal structure is from Thermotoga maritima and shows the α subunit in the dimeric form.

Class III

Class III enzymes are thought to be the most ancient RNR enzymes in evolutionary terms

[26]. They are found among some bacteriophages and in strict or facultative anaerobic bacteria [29]. The nrdD gene encodes the large α subunit and the nrdG gene encodes the small β subunit [66]. The sequence identity between Class I and Class III RNR is poor compared to the sequence identity between Class II and Class III RNR. The α subunit of a T4 bacteriophage is the only structure currently available for this class and it shows the

10-stranded β/α barrel containing the C-site is conserved [57]. The α subunit has two allosteric sites and a catalytic site. The C-site of the Class III enzymes differs from other classes in that it has only two cysteine residues. Site–directed mutagenesis studies have shown two cysteine residues (Cys 290, Cys 79) involved in the reaction mechanism [59].

Cys 290 is required for the generation of thiyl radical (S•) [67]. Class III RNRs use a thiol independent reduction system during catalysis, hence they lack the equivalent of

Cys 462 (E coli. numbering). Instead, Class III enzymes depend on the oxidation of formate to CO2, which is thought to occur via Cys 79, in place of cysteine oxidation [68,

69]. The 4Fe-4S iron sulfur cluster reduces S-adenosylmethionine (SAM) in the β subunit to generate the 5’-deoxyadenosyl radical, which in turn generates a stable glycyl radical on the large subunit [70, 71]. The transfer of the 5’-deoxyadenosyl free radical from the β subunit is also dependent on the canonical CX2C –CX2C motifs present in the C-terminus of the α subunit [72]. The structure of the nrdD subunit of the T4 bacteriophage RNR shows that the glycyl radical is located 5.2 Å away from the active site of the Class III enzymes [57]. Once generated, the glycyl radical is transferred to the C-site of the enzyme to generate the S• that is required for catalysis [67].

1.3 The Catalytic Mechanism

The sequence identity among the three classes of RNR is less than 10% [59].

Their catalytic cores, however, show remarkable similarity, which indicates a common reaction mechanism for substrate reduction (Figure 1.1, p 5). The conserved catalytic domain among all three classes of RNR contains the 10 stranded α/β barrel with the

‘RNR finger loop’ that harbors the thiyl free radical [30, 31, 46, 63, 65, 73]. The key residues required for substrate reduction are structurally conserved between Class I and

Class II enzymes. These residues include the two redox active cysteines, the general acid/base catalyzing glutamic acid and the asparagines that are its hydrogen bonding partners [31, 62]. The reaction mechanism of Class I enzymes is well studied in E. coli and will be presented briefly below [74, 75].

During each catalytic cycle, the Y● free radical is transported to the active site of the α subunit where it participates in catalysis. During this process the redox active cysteine pair at the active site undergoes oxidation, to form a disulfide bond. Prior to the next turn-over cycle, the oxidized cysteines must be regenerated. A brief description of each of these processes follows.

1.3.1 Radical Generation & Transport

Initiation of the catalytic cycle requires formation of the holoenzyme complex and binding of the cognate effector and substrate pair to the S-site and C-site. Once formed, the Y● needs to be transferred from the β subunit to C439 of the α subunit some 35 Å, away, generating the thiyl radical (Figure 1.1, p 5) [30, 76, 77]. The direct electron transfer (ET) between the active site of the β subunit and the thiyl (S●) radical in the active site of the α subunit requires the participation of radical intermediates of several amino acids. A model for free radical initiation and transfer was first proposed by Uhlin and Eklund based on a docking model generated from the shape complementarity of the structure of E. Coli RNR1(α) and RNR2(β) subunits[30, 39]. The proposed pathway included 10 residues between Y122 of the β subunit and C439 of the α subunit; a simplified view is shown in Figure 1.3. According to this model, Y122 is reduced by

W48 of the β subunit to generate W48H+●[27]. The W48● subsequently transferred to

Y356 at the C terminal tail of the β subunit with help of D237, which acts as proton sink

[78, 79]. Finally the radical is transferred across the α/β interface to two adjacent tyrosines, Y731 and Y730 before generating the S● radical in the active site (Figure 1.3 p, 14) [80].

Figure 1.3 Radical propagation pathway of Class I RNR

The pathway is well studied in E. coli. Location of the two Fe atoms (brown) is also indicated. Amino acid residues involved in transfer of tyrosyl free radicals from the

RNR2 (PDB ID1MXR) subunit to the substrate binding site at RNR1 (PDB ID 4R1R) is indicated (red dashed line).

1.3.2 Substrate Reduction

This is followed by the abstraction of the 3’-hydrogen atom from the ribose sugar by the thiyl radical (C439), generating the 3’-carbon radical (Figure 1.4, p 16) (Step 1)

[81]. The formation of the 3’-nucleotide radical facilitates both the protonation of the 2’- hydroxyl group of the ribose ring by one of the catalytically active redox pair, (C225) and the deprotonation of the 3’-OH by the glutamate (E441). Subsequently, the 3’-nucleotide radical isomerizes to the 2’-nucleotide radical with the concomitant loss of a H2O molecule (Step 2). The second cysteine (C462) at the catalytic site then delivers the reducing equivalents to the 2’-nucleotide radical, leading to the generation of 3’- ketodeoxynucleotide and the disulfide radical anion (Step 3). The free radical on the anion is then transferred back to the 3’-carbon of the deoxyribose sugar, and the E441 now acts as a general base and protonates the 3’-ketodeoxynucleotide radical, yielding the 3’-hydroxynucleotide radical (Step 4) [75, 82]. The free radical is then transferred back to the original free radical bearing cysteine (C439) and thus regenerating the thiyl radical and the 2’-deoxyribonucleoside diphosphate (Step 5). In Class I RNRs, the free radical is transferred back to the  subunit. During this process, the redox cysteine pair undergoes oxidation and must be reduced before the next catalytic cycle [83]. The latter is achieved by the reducing equivalents carried by the α subunit’s C-terminus and the

NADPH-thioredoxin /glutaredoxin based reductase system (Step 6) [84, 85].

Figure 1.4: Catalytic Mechanism of Class Ia RNR

1.3.3. Regeneration of the Active Site

During catalysis two active site cysteines, Cys225 and Cys462 form a disulfide bridge. In Class I enzymes, the electrons required to regenerate them are provided by structurally conserved CXXC motifs (Cys754 and Cys759) present in their flexible C – terminal tails [84, 85, 87, 88] (Figure 1.1, p 5). Genetic experiments with the large subunit of S. cerevisiae RNR have demonstrated that the C-terminus of one subunit acts in trans to regenerate the active site of its neighboring monomer [88]. The reduction of active site disulphides results in oxidation of the CXXC motif and its re-reduction requires external electron donors such as thioredoxin (Trx) or glutaredoxin (Grx) [89-91].

Class II enzymes, like Class I enzymes, use a redox active pair of cysteines at their C- terminal tail to reduce the active site [62]. In contrast, Class III RNRs lack one of the corresponding cysteines in the redox active pair in the catalytic site. Instead, Class III enzymes oxidize formate to carbon dioxide as the primary electron donor [92]. However,

Class III enzymes use a Trx system to keep the conserved C-terminal cysteines in the reduced form required for the generation of glycyl radicals [72].

External electron donors such as Trx and Grx cannot directly access the narrow active site cleft and thus require the C-terminal tails of the large subunit RNRs. Sequence alignment of C-terminal tails of Class I and Class II RNRs from different organisms shows conserved cysteine residues indicating the involvement of external electron donors. Both Trx and Grx were originally discovered as dithiol electron donors of E.Coli

RNR [18, 19, 93]. Trx reduction requires NADPH and thioredoxin reductase (TrxR)

17 while Grx is first reduced via glutathione (GSH) and the oxidized GSH is later reduced by glutathione reductase (GR) and NADPH [94, 95]. Many isoforms of Trx and Grx exist and at least one form of redoxin is essential for cell viability [96]. In E coli, Grx1 functions as the most efficient dithiol redoxin for RNR [97], and it and Trx1 are the two main hydrogen donors for E.coli RNR [98]. In contrast to bacterial RNR, for which the phase of the cell cycle does not matter, mammalian RNRs seem to distinguish between

Trx1 and Grx1 systems, favoring Trx1 as an S-phase electron donor and Grx1 during

DNA repair and mitochondrial DNA synthesis, despite the fact that both redoxin systems have similar catalytic efficiency. Furthermore, the Grx system seems to use a glutathionyaltion based mechanism during C- terminal regeneration of mammalian RNR

[91].

1.4 Regulation of Class Ia Ribonucleotide Reductase

All known RNRs are subject to multiple modes of regulation. Allosteric regulation at the protein level and transcriptional regulation at the nrd gene level seem to be conserved across many species. Some RNRs are also subject to additional different modes of regulation: (1) regulation by binding of small protein inhibitors, (2) regulation by subunit compartmentalization, (3) regulation of mRNA stability, (4) regulated protein degradation and (5) regulation of cofactor assembly and free radical generation. Most of these mechanisms are well studied in E. coli, yeast and mammalian cell culture systems.

This section will describe RNR regulation principally with reference to these organisms, placing special emphasis on allosteric regulatory mechanisms and inhibition of RNR in budding yeast by the small protein inhibitor known as Sml1.

1.4.1. Allosteric Regulation of Ribonucleotide Reductase

Two different mechanisms bring about allosteric regulation of RNRs: substrate specificity regulation and overall activity regulation. Substrate specificity regulation is important for maintaining a balanced dNTP pool whereas overall activity regulation determines the dNTP pool size [33]. Binding of ATP/dNTPs to the S-site regulates substrates specificity. Binding of ATP or dATP to the A-site regulates the overall activity of the enzyme. The initial model of allosteric regulation was derived by characterization of E. coli RNR and studies with RNR purified from calf thymus [99, 100]. Although these models were based on in vitro studies, subsequent analysis of dNTP pools in cultured mammalian cells treated with exogenous deoxyribonucleosides [101-104] or with inhibitors of DNA synthesis [105, 106] provided supportive evidence for the existence of two different allosteric sites [107].

1.4.1.1 Activity Site Regulation

Initial evidence that ATP acts as an activator and dATP as an inhibitor originated from pioneering studies carried out with purified E. coli extracts [32]. Based on their studies, Reichard et al. proposed that enzyme activation and inhibition was a result of

ATP and dATP binding to one or more allosteric sites. Radiolabeled nucleotide binding experiments carried out later by Reichard et al. showed that ATP and dATP bound a low affinity binding site when compared to other dNTPs. No detectable binding of dGTP or

TTP took place with the low affinity binding site. This low affinity binding site was later named as the activity site (A-site).

It is now well established that the overall activity of most RNRs is regulated by binding of ATP and dATP to the A-site located in the ATP-cone (Figure 1.2, p 8) [30,

108]. The ATP cones of RNRs have the signature sequence VXKRDG. The ATP cone is present in all non-viral eukaryotic RNR1 sequences. In prokaryotic RNRs, ATP binding cones are only universally found in Class 1a and Class Ic [2]. Also when an ATP binding cone is present , dATP binding is not always inhibitory; examples include Trypanosoma brucei [109] in Class Ia and Thermoplasma acidophilum [110] in Class II. Duplication or triplication of the ATP-cone domain in P. aeruginosa and C. tracomatis have been reported and N-terminal deletion studies of such organisms have revealed that the most

N-terminal ATP cone confers the allosteric function [111]. The importance of the A-site in regulating the level of dNTP pools has been confirmed by both in vitro and in vivo experiments. For example, the D57N mutation in the ATP-binding cone in S. cerevisiae has been shown to increase the dNTP pools 30-fold over the wild type RNR during response DNA damage [112].

Initial binding studies with E. coli enzyme showed that the A-site has a relatively high affinity for dATP relative to ATP[32]. The A-site bound dATP with a dissociation constant (Kd) of 0.6 µM, and dATP was able to overcome the stimulatory effects of ATP.

Similar observations were made with mammalian RNRs. Chimploy and co-workers reported the Kd for dATP binding at the A-site to be 54.3 μM, while Kashlan and co- workers reported the Kd to be 1.1 μM with mouse RNR [113, 114]. In contrast, the Kd for

ATP binding at the A-site was shown to be 94 μM [113, 114]. Since the A-site has a higher affinity for dATP than ATP, dATP can out-compete ATP, resulting in enzyme inhibition even though ATP is still present [115, 116].

Until recently, it remained largely unknown why the A-site has higher affinity for dATP than ATP. The only structure with nucleotide bound at the A-site was the AMPNP- bound E .coli RNR [30]. This structure provided the first evidence of where the A-site was located in the N-terminal portion of RNR. However, the AMPPNP-bound structure was not useful for interpreting the functional difference between ATP and dATP. The structures of dATP and ATP bound to the A-site in human RNR1 (HuRRM1) revealed for the first time, the structural basis of the differential action of these two effectors

(Figure 1.5, p 22) [108]. Both ATP and dATP bind to the ATP binding cone formed by the first 90 residues of the protein. ATP binds less deeply inside the four-helix bundle

(FHB), covering less surface area, and adopts a 2’-endo conformation (Figure 1.5 A, p

22). In contrast, dATP binds deep inside the FHB and its ribose sugar adopts a half chair conformation (Figure 1.5 B, p 22). These structures provided for the first time, a plausible explanation of why dATP has a higher affinity than ATP [108] .

Figure 1.5 ATP and dATP binding to the ATP cone

Activity site regulation requires the binding of ATP or dATP to the four-helix cone, which is also known as the ATP cone. The A-site is formed by helices H1-H4, which form a four-helix bundle, one end of which is covered by a β-sheet cap. (A) The four- helix cone bound with ATP (PDB ID 3HNE). (B) The four–helix cone bound with dATP. Both nucleotides are shown in stick representation (PDB ID 3HNF).

It was believed for a long time that the subtle changes caused by ATP and dATP in the structure of the α2β2 heterotetramer determine the respective stimulatory and inhibitory effects of these two nucleotides. ATP or dATP binding is a pre-requisite for stimulatory or inhibitory effects on RNR. This pre-requisite is evident from the ATP binding cone mutants D57N in mouse and in budding yeast [117], and H59A and H88A mutants [118] in E. coli , which are insensitive to dATP inhibition. The changes in oligomeric states of RNRs with respect to different concentrations of ATP/dATP have been proposed as a basis for their stimulatory and inhibitory effects [114]. Nucleotide induced oilgomerization was initially observed with the RNR1/RNR2 complex of E. coli.[1, 119]. These experiments showed that dATP increased the degree of oligomerization, and that effect was reversed by ATP. This initial observation of enhanced subunit association of R1 and R2 in the presence of dATP was further confirmed by Surface Plasmon Resonance (SPR) studies, which showed 100-fold stronger interaction [120]. In addition, mutational studies of the ATP binding cone of E. coli identified hydrogen-bonded H59 and H88 as important residues which trigger negative allosteric effects upon binding dATP [118]. Studies with mouse RNR enzyme also suggested enhanced R1 and R2 subunit interactions in the presence of dATP and

ATP with subunit stoichiometry of 1:1. Disruption of the long range electron transport between tightly interacting subunits was proposed as the basis of dATP mediated inhibition of the enzyme activity [121].

Based on a coordinated set of enzymatic activity measurements, molecular mass determination and ligand binding experiments, Kashland and co-workers observed mouse

RNR forms higher order oligomers with increasing concentration of dATP and ATP [44,

114]. They postulated the existence of two different forms of tetramers designated as

R14a and R14b upon binding of ATP or dATP to the A-site. At physiologically relevant concentration, ATP drove the α subunit to form hexamers by binding to the hexameric site (h-site). However, dATP was only able to form tetramers at its physiological concentration. The dATP-induced tetramers were shown to be less active than the ATP- induced hexamers when combined with the small β subunits. Based on these results,

Kashlan et al. proposed the comprehensive model for RNR regulation, in which dATP binding at the A-site forms inactive tetramers, while ATP binding at the A-site initially forms tetramers with low activity. The comprehensive model further proposed that high concentration ATP binds to a third allosteric site known as h-site inducing the formation of active hexamers [44].

However, the recent characterization of oligomeric states of mouse RNR by

GEMMA (Gas-phase Electrophoretic Macromolecule Mobility Analysis) revealed only the existence of dATP induced hexamers at physiological concentrations of dATP. The same study also found that the ATP/dATP-induced hexamers can associate to form an

α6β2 oligomer which is either active or inactive depending on whether ATP or dATP is bound, respectively.

Activity site regulation is better understood in Class I RNRs. However, recent studies indicate significant differences in A-site regulation between prokaryotic and eukaryotic RNRs [47]. Unlike eukaryotes, prokaryotes seem to form different types of holoenzyme complexes with dATP and ATP. Based on combined GEMMA and enzyme activity assays, it has been found that E.coli Class Ia RNR forms an inhibited α4(β2)2 complex in the presence of dATP and an active α2β2 complex in the presence of ATP.

Unlike the dATP bound α4(β2)2 complex, the ATP-bound α2β2 complex showed much weaker subunit interactions leading to an equilibrium with free α and β subunits[47].

Consistent with these findings, a recent study has described a dATP-bound α4(β2)2 complex which forms a ring-like structure composed of alternating α2 and β2 subunits

[122]. The formation of a ring-like structure in the presence of dATP also suggests the disruption of the radical transfer pathway from β2 to α2. The same study also suggests the existence of an ATP/CDP or a dTTP/GDP based α2β2 complex which is active and can be rapidly converted from an α4(β2)2 complex in the absence of dATP [122].

Apart from these natural nucleotides, certain nucleotide analogs such as gemcitabine and clofarabine have been shown to induce subunit oligomerization. The phosphorylated form of gemcitabine (F2CDP) acts as a mechanism based inhibitor and , when incubated with E. Coli or human RNR, it induced the formation of tight α2β2 or

α6β6 complexes with the E coli and human enzymes, respectively [45, 123]. A recent study has shown both clofarabine-5-di- and triphosphates inhibit human RNR by hexamerization of the large subunit [124]. Section 1.5.2 gives a more detailed account of these inhibitors.

1.4.1.2 Specificity Site Regulation

A remarkable feature of all RNRs is their ability to reduce four different NDPs

/NTPs using an intricate mechanism of substrate selection. The RNRs accomplish this by an elegant allosteric mechanism requiring the coordination of two allosteric effector binding sites and the catalytic site (Figure 1.2, p 8). The basic mechanism of substrate selection was first elucidated using E. coli RNR [20, 34]. Based on rapid dialysis and gel

25 filtration binding studies, it was proposed that the large α subunit of E. coli. contains two separate allosteric effector binding sites, one site for high affinity binding of dATP, ATP, dGTP and TTP that determined the substrate specificity, and another site with low affinity for dATP and ATP that regulated the overall activity [32]. In the current literature, these sites are now named the substrate specificity site (S-site) and the activity site (A-site). The S-site binds dGTP, TTP, ATP and dATP. ATP or dATP binds at the S- site and selects either CDP or UDP to be reduced at the C-site (Figure 1.6 A, p 28). The product dUDP is subsequently dephosphorylated to dUMP and further metabolized by thymidylate synthetase to form TMP, which is then phosphorylated to TTP [3]. TTP binds at the S-site and selects for GDP to be reduced at the C-site. Nucleotide diphosphate kinase (NDK) converts the dGDP to dGTP, which in turn binds the S-site and selects for ADP as the substrate for reduction. The conversion of dADP to dATP is also accomplished by NDKs. The relatively high affinity of dATP compared to ATP

[113, 114] means it competes with ATP, displacing it from the A-site leading to inhibition of the enzyme (Figure 1.5, p 22).

The selection rules mentioned above are shared by both Class I and Class II enzymes [31, 63]. In contrast, Class III enzymes use a slightly different set of substrate selection rules. The analog of A-site is known as the pyrimidine site and can bind both

ATP and dATP. Binding of ATP to the pyrimidine site stimulates reduction of pyrimidine ribonucleotides. The purine site resembles the S-site in Class I and Class II enzymes.

Binding of dGTP or TTP bind at the purine site selects for ATP or GTP at the C-site, respectively. However, the binding of dATP to either the purine or the pyrimidine site is always inhibitory.

Several X-ray structures with allosteric effectors bound to the S-site have been reported for Class Ia [30, 31, 83, 108], Class Ib [125] Class II [63, 65] and Class III

RNRs [126]. The structure of TTP bound to the large subunit of E. coli RNR revealed for the first time that Loop 2 is an important structural motif which bridges the gap between the effector and substrate binding sites. The binding modes of three different effectors (i.e dTTP, dATP and dCTP) at the S-site were later described in the structure of the Class Ib

RNR1 from S. typhimurium [125]. Both of these structures were determined in the absence substrates and it was not clear how different effectors selected their respective substrates.

However, the molecular basis for substrate selection rules became evident with two crystallographic studies of α subunit in S. cerevsiae (Class Ia) and in T. maritima

(Class II). These structures, both containing bound substrate and allosteric effectors, demonstrated for the first time an elegant structural mechanism known as specificity cross-talk operating between the S- and C-site for substrate selection [31, 63]. Structures of effector-substrate pairs (dATP-UDP, dATP-CDP, dGTP-ADP, dTTP-GDP) of the

Class II RNR of T. maritima revealed the importance of the different conformations of

Loop 2 in substrate selection [63]. In these structures, Loop 2 adopted a β-hairpin like structure for purine-pyrimidine pairs (Figure 1.6B, p 28). However, the β-hairpin was excluded in the dGTP-ADP structure because the larger base moiety required more space.

Several key conserved residues in Class II RNRs are involved in S-site regulation. The most notable residues include A210 and R207 which are important for clamping the substrate in the active site and residue Q203 which is involved in base recognition.

Figure 1.6 Specificity site regulation in eukaryotes.

(A) Binding of ATP or dNTPs (dATP, TTP, dGTP) at the S-site selects the cognate substrates to be reduced at the C-site. Positive feedback loops are indicated by dashed lines. (B) The structural basis of S-site regulation has been studied in both Class I and

Class II enzymes. Substrate selection is based upon the different conformations of Loop 2 which moves upon binding of effectors: Substrate (UDP), Loop 2 and effector

(AMPPNP) are shown for AMPPNP-UDP. Loop 2 is shown for apo (black), AMPPNP only (gray), and AMPPNP-CDP (orange).

The conservation of key residues of Loop 2 in Class II and Class I RNR suggested a common mechanism of S-site regulation (Figure 1.6B, p 28) [63]. This became evident when the structures of Class Ia RNR of S. cerevisiae were determined with cognate effector-substrate pairs [31]. These studies emphasized the importance of effector- binding at the S-site as a prerequisite for the dimerization of the large subunit, which is essential for substrate selection. Specificity cross-talk between the S-and C-sites takes place thorough rearrangements of Loop 1 and Loop 2. The effectors (ATP/dNTPs) bound at the S-site only interact with Loop 1. In contrast, Loop 2 mediates allosteric communication between subunits by connecting the S-site on one subunit with the C-site of the adjacent subunit and while doing so it contacts the bases of the effector and the substrate. In the apo enzyme, Loop 2 occupies a position that sterically restricts substrate binding and this observation is consistent with biochemical data which shows that only

10% of the activity is retained without effector binding [42, 87]. When effectors bind at the S-site, Loop2 moves towards the S-site, which pulls it away from the C-site, creating space for substrates to bind. Once the substrate binds the C-site, Loop 2 shifts part-way back towards the C-site. In the process, Loop 2 adopts a unique conformation for each effector-substrate complex (Figure 1.6B, p 28).

The substrate selection rules first proposed by Brown and Reichard [32, 34], are maintained at the molecular level by specific interactions made between residues of Loop

2 and the substrate. In particular, Arg 293 and Gln 288 are crucial for substrate recognition. Specifically, these residues appear to be crucial for ADP selection. Arginine

293 forms a hydrogen bond and makes stacking interactions with the adenine ring, while

Gln 288 forms a hydrogen bond. Mutation of Arg 293 in the yeast enzyme was recently

29 shown to be synthetically lethal [127]. Furthermore, the sequence of Loop 2 is identical in yeast, mouse and human enzymes indicating a common mechanism of substrate selection.

The structural basis of specificity regulation in Class IIIRNR has been studied with phage T4 RNR complexed to four different effectors [128]. These structures show some notable differences from S-site regulation of Class I & II RNR. As in Class I and

II, effectors bind at the dimer interface in Class III, but the effector binding site is more than 25 Å away from the closest active site and the interaction between effector and substrate is probably mediated by helix αB in addition to Loop 2. None of these structures have any substrates bound at the C-site, so the specific details of substrate recognition are unknown, however modeling studies of Class III substrate binding based on Class II structures indicate a similar mode of substrate recognition [2] .

1.4.2 Transcriptional Regulation of Ribonucleotide Reductase

Transcriptional induction of RNR subunits occurs during the S-phase of the cell cycle and in response to DNA damage. This section gives a brief account of transcriptional regulation of RNR in yeast and mammalian cells.

In S. cerevisiae the minimally active holoenzyme consists of an α2ββ’ heterotetramer where α is ScRR1, β is ScRR2, and β’ is ScRR4. The large subunit (α) is encoded by two highly homologous genes termed RNR1 and RNR3 [129]. The small subunit of budding yeast RNR is a heterodimer composed of ScRR2 and ScRR4 subunits, which are encoded by two different genetic loci [130-132]. Unlike other RNRs, ScRR3

30 and ScRR4 are yeast specific and are homologous to large and small subunits, respectively.

ScRR1 is essential for survival, whereas ScRR3 is non-essential, but strongly induced after DNA damage. The N-terminal 780 residues are highly conserved between

ScRR1 and ScRR3 and among the large subunits of other eukaryotic RNRs. ScRR3 is expressed at extremely low levels during normal cell growth and its mutants have no observable phenotype [129]. However, over-expression of ScRR3 is able to rescue yeast cells which have a null mutation for ScRR1 [129, 133, 134]. After DNA damage, or when cells experience stress, the expression of ScRR3 is up-regulated 50-100 fold, however, its protein level reaches only one tenth of the level of ScRR1 [133] . ScRR3 shows less than 1% of the activity of ScRR1 when assayed with ScRR2●ScRR4.

However, ScRR3 shows a synergistic effect with both catalytically active and inactive

ScRR1. Interestingly, the allosteric inhibitor dATP was shown to have a stimulatory effect on ScRR3 [133].

ScRR2 is encoded by the RNR2 gene which is essential for mitotic cell viability and is expressed at higher levels after DNA damage [130, 135]. As might be expected for a eukaryotic protein, ScRR2 has a higher percentage of sequence identity with the same subunit from higher order eukaryotes than with prokaryotic and viral R2 subunits [130].

The small subunit of RNR contains 16 residues that are conserved from E. coli to mammals. Among these residues is histidine 179, which co-ordinates the iron required for generation of the free radical, and tyrosine 183 which stores the free radical. ScRR4 which is encoded by the RNR4 gene, and like its counterpart, ScRR2, is also essential for mitotic cell viability. Sequence identity between ScRR2 and ScRR4 is 56% but ScRR4

31 lacks 50 amino acid residues in its N terminus when compared to ScRR2. ScRR4 has also lacks 6 of the 16 residues conserved from E. coli to mammals in ScRR2 [132]. ScRR4 has tyrosine residue (Y131), equivalent to Y183 in ScRR2, which harbors the free radical

However, ScRR4 lacks the two conserved histidines and the glutamate residues required for iron co-ordination. Two tyrosines (Y127 and Y223) are substituted for histidines, and arginines are substituted for the glutamate residues [136]. Deletion of the RNR4 gene is lethal and RNR activity is not detectable in the RNR4-Δ1 genetic background.

Bacterially expressed ScRR4 dramatically increased RNR activity upon addition to yeast extract lacking endogenous ScRR4. However, substitution of Y131 with phenylalanine in

ScRR4 doesn’t confer cellular lethality [136]. Stubbe and co-workers has suggested, based on their studies, that ScRR4 functions as a chaperone in the assembly of the diiron cluster in ScRR2 [137]. However, no RNR activity was observed when ScRR1 was assayed with either ScRR2 or ScRR4 alone but RNR was found to be highly active upon addition of co-expressed ScRR2●ScRR4. Lack of any iron or organic free radical in

ScRR4 strongly suggests it plays a non-catalytic role in RNR activity. Consistent with these observations, CD spectroscopic studies indicate that ScRR4 is required for proper folding of ScRR2 [138]. X-ray crystal structures are available for both homodimeric

ScRR2 and ScRR4 and as well as for the heterodimer [139, 140]

RNR genes of S. cerevisiae are induced at the G1/S boundary during the cell cycle and the induction of these genes is dependent on the Mbp1/Swi6 pathway [141,

142]. This protein complex recognizes the Mlu1 cell cycle box (MCB) element in the promoter region of the RNR genes [143, 144]. Mbp1 seems to increase ScRR1 mRNA transcripts several fold [129, 145] while it has a minimum impact on ScRR2 and ScRR4

32 mRNA transcription levels [145]. In addition to the Mbp1/Swi6 pathway in S. cerevisiae, the Mec1-Rad53-Dun1 pathway also maintains an adequate supply of dNTP by regulating the activity of RNR mainly in response to DNA damage, but also during the normal cell cycle [146, 147]. Mec1 and Rad53 are required for S-phase checkpoint and transcriptional induction of RNR genes and G1 and G2 arrest in response to DNA damage

[148, 149]. Dun1 is also activated in response to DNA damage and replication blocks, and is dependent on Mec1 and Rad53 [148]. The key targets of the Mec1-Rad53-Dun1 pathway include the transcription factor Crt1, a small protein regulator named Dif1 that mediates nuclear retention of ScRR2●ScRR4 and a small protein inhibitor named Sml1

[150]. S. cerevisiae induces transcription of all RNR genes during the normal cell cycle and after DNA damage. However, transcription of RNR2, RNR3 and RNR4 genes outside the S-phase of the cell cycle is repressed by the transcription factor Crt1, an X-Box specific DNA binding protein [151]. Crt1 confers repression by recruiting the Tup1-Ssn6 corepressor complex to the promoters of the RNR genes [151]. Hyperphosphorylation of

Crt1 by Dun1 kinase relieves this repression. However, the Mec1/Rad53/Dun1 pathway does not control RNR1 gene transcription (Figure 1.7, p 37) [151]. Instead, the transcription factor Irx1 interacts with the RNR1 gene promoter to control the expression of RNR1. Deletion of Irx1 leads directly to decreased dNTP levels due to reduced RNR1 gene expression (Figure 1.7, p 37). Furthermore, elevation of ScRR1 levels in response to DNA damage does not depend on Dun1 but only on Mec1 and Rad53 [150].

Periodic expression of RNR genes has been described in other organisms as well.

Unlike in budding yeast, transcription of both large (Cd22) and small (Suc22) subunits of

33 fission yeast RNR is induced during the cell cycle by Dsc1, a transcription factor similar to Mbp1/Swi6 [152-155].

A similar periodic pattern of expression of RNR has been observed in mammalian cells. In S-phase, the mammalian RNR enzyme is mainly composed of the large α subunit and the small β subunit which can exist in either α6β2 or α2β2 oligomeric states.

This is because of S-phase specific transcription of the RNR1 and RNR2 genes [156,

157]. Levels of both RNR1 and RNR2 transcripts were shown to be very low in cells arrested in the G0/G1 phase [157]. Measurement of mRNA levels in centrifugally elutriated cells and in cells synchronized by serum starvation has shown pronounced increase of RNR transcripts as the cells progress through the S-phase. RNR1 and RNR2 transcript levels then gradually declined as the cells entered into the G2/M –phase [157].

S-phase specific transcription of the RNR1 gene is based on the TAA-less gene promoter and has been extensively studied in mouse. Four different promoter elements named β, α,

Inr and γ are involved in the transcription of the RNR1 gene and these elements are also highly conserved in the human RNR1 gene promoter [158, 159]. Many transcription factors including YY-1 are shown to bind to these promoters and the cell cycle specific expression of RNR1 is primarily controlled by the Inr and γ elements [160]. In contrast,

S-phase specific transcription of the RNR2 gene is dependent on the binding of the E2F4 transcription factor to the upstream promoter activating region and the proximal promoter repressive element [161, 162]. R1 protein has a long half-life and therefore its concentration remains constant throughout the cell cycle [163]. Since the RNR2 protein is expressed in the S-phase and has a half-life of 3 hours due to degradation in late

34 mitosis, the RNR activity in cycling cells is principally governed by the R2 protein levels[163, 164].

Transcription of the RNR1 gene is also induced in response to DNA damage in resting fibroblasts [165]. In contrast, transcription of the RNR2 gene is not influenced by

DNA damage or replication block in resting cells [166]. Instead expression of another protein known as p53R2 is induced by DNA damage [167, 168]. Studies with recombinantly expressed mouse and human p53R2 have shown that the p53R2 protein can form a tyrosyl radical similar to R2 proteins and can combine with human and mouse

R1 protein to form fully active RNR complexes [165]. Induction of p53R2 is p53 dependent. Studies with cycling cells and synchronized cultures of mammalian cells indicated that p53R2 expression is constitutive and remains low irrespective of the phase of the cell cycle [169]. Since hydroxyurea treated cells in G0/G1-phase show significant reduction in dNTP pools, it is evident that the RNR1-p53R2 complex plays a vital role in supplying dNTPs for basal DNA repair and mitochondrial DNA synthesis.

1.4.3 Regulation of RNR by Subunit Compartmentalization

Subunit compartmentalization also regulates RNR activity in S. cerevisiae

Immunofluorescence studies on subcellular localization of RNR subunits in non-dividing cells indicate that ScRR1 and ScRR3 are primarily located in the cytoplasm, whereas

ScRR2 and ScRR4 are predominantly in the nucleus [170]. Nuclear localization of the small subunit during the G1 and G2/M stages of the cell cycle depends on two different

Wtm proteins, known as Wtm1 and Wtm2 [171, 172]. Both proteins seem to physically interact with ScRR2 and ScRR4. Deletion of Wtm1 causes loss of nuclear localization of

ScRR2 and ScRR4 at all stages of the cell cycle. Recent studies have shown another protein known as Dif1 is directly involved in importing the small subunit to the nucleus from the cytoplasm independent of Wtm1 and its nuclear importin, Kap122 [173]. Upon

DNA damage or during the normal cell cycle ScRR2 and ScRR4 undergo redistribution from the nucleus to the cytoplasm as a heterodimer [172]. This redistribution of the heterodimer is Mec1/Rad53/Dun1 dependent and is mediated via phosphorylation of Dif1 and its subsequent degradation via proteolysis (Figure 1.7, p 37) [173].

Figure 1.7 Regulation of yeast RNR

Regulation of yeast RNR is dependent on Mec1/Rad53/Dun 1 Kinase during normal growth and after DNA damage. Crt1, Sml1 and Dif are direct targets of Dun1 kinase.

Crt1 acts as a repressor for the transcription of RNR2, RNR3 and RNR4 genes. Crt1 recruits general inhibitors Ssn6 and Tup1 to inhibit transcription of RNR genes. Sml1 binds to ScRR1 and inhibits RNR activity during G1 and G2/M phase. Dif1 facilitates the nuclear localization of the ScRR2•ScRR4 subunit. RNR1 gene expression is dependent on Irx1.

1.4.4 Regulation of RNR by Small Protein Inhibitors

Regulation of RNR by the small inhibitory protein known as Sml1 was first described in budding yeast. Sml1 inhibits S.cereviasie ScRR activity by binding to its large subunit. This inhibition is relieved upon degradation of Sml1 in response to DNA damage or upon cells entering the S–phase of the cell cycle. The Sml1 (suppressor of mec1 lethality) gene was first isolated during a genetic screening of a mec1-1 homozygous strain that could still survive. Deletion of MEC1 and RAD53 is normally lethal in yeast [174, 175]. This mec1-1 strain was able to survive because the mutation of the SML1 gene suppressed the lethality of the double deletion. The SML1 gene is located on chromosome XIII and encodes a small protein of 104 amino acids [176].

Several lines of evidences suggested that Sml1 is a negative regulator of S. cerevisiae RNR. One noticeable phenotype of Sml1 deletion is a reduced frequency of petite formation. In contrast, expression of an extra copy of Sml1 resulted in increased frequency of petite formation. Lack of functional, replicating mitochondria is one of the reasons for petit formation. Since mitochondria lack their own version of RNR, mitochondrial DNA replication is negatively affected by reduced dNTP pools caused by

Sml1 [176]. Deletion of the Sml1 gene (Sml1Δ) plainly showed the effect of Sml1 on dNTP pools. Cells with Sml1Δ had 2.5 fold higher cellular dNTP levels without altering transcription of RNR genes. Furthermore the Sml1Δ deletion also alleviates reduced dNTP levels in Dun1 Δ strains after DNA damage. Sml1’s negative regulatory effect on dNTP pool sizes prompted Rothstein and co-workers to probe for RNR Sml1 interactions. Both in vivo and in vitro analysis using yeast two hybrid and co immunoprecipitation assays showed that Sml1 indeed interacts with the ScRR1 subunit

38 of RNR but not with the ScRR2● ScRR4 subunit. Sml1 also interacts with ScRR3 in vivo but the biological significance of this interaction has been not been pursued [176].

Thelander and co-workers later demonstrated that Sml1 inhibits the S. cerevisae

RNR in vitro activity assay using recombinant proteins. Furthermore they observed no competition between Sml1 and ScRR2●ScRR4 for binding to ScRR1. Using a Surface

Plasmon Resonance (SPR) assay, they determined the dissociation constant (Kd) between

Sml1 and ScRR1 to be 0.25-0.4 µM [177]. Sml1 is able to interact with RNR of heterologous species. The large subunit of the human RNR enzyme has been shown to interact with Sml1 using a yeast two hybrid assay [178]. An SPR assay demonstrated interaction of Sml1 with the large subunit of mouse RNR. However, the mode of Sml1 inhibition of mouse RNR seems to be different since the small subunit of the mouse RNR competes with Sml1 for binding to its large subunit[177].

Sml1 protein levels fluctuate during the cell cycle and decrease 6-fold compared to G1 arrested cells during S phase. Additionally, they peak 9-fold higher than their S- phase levels during G2/M phase. Initial studies showed that degradation of Sml1 after

DNA damage or replication block is Mec1/Rad53 dependent and only the unbound form of Sml1 is targeted for degradation [176, 178]. The involvement of Dun1 kinase in regulating Sml1 was evident from three lines of genetic studies of Sml1 Δ and Dun1 Δ mutants. The Sml1 Δ mutant not only suppressed the DNA damage sensitivity of Dun1 Δ strains without affecting transcription but also suppressed the loss of telomere position effect in Dun1 Δ strains [179]. Furthermore over expression of Dun1 also suppressed the

Mec1 Δ and Rad53 Δ lethality [147]. Rothstein and co-workers later showed that Dun1 and Sml1 interact in vivo using a yeast two hybrid analysis. The inability of Dun Δ strains

39 to phosphorylate Sml1 and the inability of Dun1-D238A mutant to phosphorylate Sml1 in vitro further confirmed that Sml1 is a target of the Mec1/Rad53/Dun1 pathway (Figure

1.7, p 37) ( [180].

Site directed mutagenesis and mass spectrometry have revealed the phosphorylation sites on Sml1. Dun 1 kinase phosphorylates Sml1 between residues 52 and 64, in a sequence containing three serine residues (56, 58 and 60) in close proximity

[181]. Under in vivo conditions, Dun 1 kinase phosphorylated the same set of serine residues. Recent genetics studies have shown that phosphorylated Sml1 is recognized by the Rad6/Urb3/Mub1 E2/E3 ligase complex leading to its ubiquitination. Impairment of

Sml1 protein degradation in the temperature sensitive 26S proteasomal mutants, pre1-1 or pre2-2 shows that Sml1degradation is dependent on the 26S proteasome complex (Figure

1.8, p 42) [182].

Figure 1.8 Regulation of Ribonucleotide Reductase by Sml1

Sml1 is an intrinsically disordered small protein inhibitor of RNRs of budding yeast.

Sm1l interacts with ScRR1 and inhibits its activity. During G1/S transition or DNA damage, Dun1 kinase phosphorylates Sml1 and relieves its inhibition. Phosphorylated

Sml1 is ubiquitylinated by Rad6/Urb3/Mub1 E2/E3 ligase and subsequently undergoes

26S proteasomal dependent degradation.

Sml1 has also been subjected to extensive biophysical characterization.

Sedimentation equilibrium analysis has shown that Sml1 is a dimer at high concentration in solution. The presence of reducing agents during protein purification and mutation of the sole cysteine to serine (C14S) did not result in any change of the oligomeric state of the protein. Furthermore, sedimentation equilibrium analysis of Δ 8 and Δ 20 deletion mutants of Sml1 showed that the dimerization domain resides between residues 8 and 20

[183]. Structural studies with NMR spectroscopy have revealed Sml1 is a loosely folded structure with three regions that have a high degree of backbone order. The ordered regions extend from residues 4 to 14, 20 to 35 and 61 to 80 of Sml1 with the first and last regions adopting an alpha helical structure [183]. Translational diffusion NMR experiments confirmed the dimeric nature of Sml1 and determined the dimerization Kd to be 0.1 mM. Spin labeling experiments at the C14 position of Sml1 revealed that N terminus of Sml1 interacts with residues spanning 60 to 70 and 85 to 95 and folds onto the C-terminal domain. Protease digestion studies combined with mass spectrometry further indicate that the N-terminal domain protects the C terminus of Sml1 from proteolytic digestion [183].

Currently the binding region of Sml1 to the ScRR1 is known. Scanning mutagenesis studies and yeast two hybrid screens of Sml1 have identified a series of point mutations in the last 33 residues of the C-terminus of Sml1 important for its interaction with ScRR1. Moreover, these mutants showed reduced inhibition of ScRR1 when compared to wild type Sml1. However, N-terminal deletion mutants that removed residues 2 to 39 and 28 to 50 did not affect the ability of Sml1 to inhibit RNR, demonstrating that this region is not necessary for binding of Sml1 to ScRR1 [183]. A

42 recently concluded genetic study has suggested that Sml1 prevents active site regeneration by the CX2C motif of ScRR1 [88].

A similar type of small protein inhibitor known as Spd1 has been described in fission yeast[184]. Over-expression of Spd1 inhibits G1/S progression[185]. Unlike

Sml1, Spd1 has a less conserved R1B domain and interacts with both the large (Cdc22) and small (Suc22) subunits of fission yeast RNR [186]. Binding of Spd1 to Suc22 regulates its nuclear import [187] as Dif 1 does for ScRR2●ScRR4 nuclear transport.

Recent FRET studies of subunit interaction between Cdc222 and Suc22 in the presence of Spd1 suggest that this intrinsically disordered protein allows the formation of an inactive holoenzyme complex [188].

1.4.5 Regulation of RNR by Selective Protein Degradation

Selective protein degradation as a mode of RNR regulation has been studied in mammalian cells. The small subunit of mammalian RNR is subject to degradation by anaphase promoting complex-Cdh1 (APC-Cdh1) in late mitosis. Mammalian RNRs form holoenzyme complexes that differ according to the phase of the cell cycle for this reason.

Degradation of RNR2 protein in late mitosis is dependent on an N-terminally located conserved KEN box sequence and requires APC-Cdh1, one of the ubiquitin protein ligases that have central roles in cell cycle regulation [189]. Cdh1 has a KEN box identifying motif and an activating module for APC [190]. The degradation of RNR2 protein is thought to be vital in preventing unscheduled DNA replication since inhibition of APC-Cdh1 in cultured cells causes premature entry into S-phase and deregulated

RNR2 protein increases malignant transformation in Blab/3T3 cells[191] .

1.5 Inhibitors of Ribonucleotide Reductase

RNR is an attractive target for both cancer chemotherapy and antiviral therapy

[192]. During the last few decades, a considerable amount of effort has been devoted to developing specific and novel inhibitors of this enzyme [193]. In this section I will focus on inhibitors currently used in clinics for cancer chemotherapy. I will discuss these inhibitors under three broad categories; translational inhibitors, inhibitors of the large subunit of RR, and inhibitors of the small subunit of RR.

1.5.1 Translational inhibitors

Translational inhibitors of RNRs are complimentary oligonucleotides that bind mRNA of either RNR1 or RNR2. Once complexed, these oligonucleotides either block translation or degrade the mRNA by activating RNase H. GTI-2040 and GTI-2501 are two promising examples of this approach; they are 20-mer phosphorothioate oligonucleotides that have undergone clinical trials [194]. GTI-2501 targets the coding region of RNR1 and reduces both mRNA and RNR1 protein levels in a dose-dependent manner. Both in vitro and in vivo studies have shown that GTI-2501 significantly inhibits the growth of various human cancer types [195].

1.5.2 Inhibitors of the Large Subunit of RNR

RNRs of Class Ia have four drug interaction sites (Fig. 1). They are: 1) the A-site,

2) the S-site, 3) the C-site and 4) the P-site [196, 197]. The first three sites are ribonucleotide and deoxyribonucleotide binding sites, while the fourth site is the peptide

44 binding site. The A-site and S-site bind ribonucleoside/deoxyribonucleoside triphosphates while the C-site binds ribonucleoside diphosphates. Most of the nucleotide- based drugs that target RNR work by binding at either the A-site or the C-site. Most of the analogs that bind the A-site also bind the S-site. Now I will briefly describe the drugs currently in clinical practice.

1.5.2.1 A-Site Analogs

Fludarabine [198], cladribine [199, 200], and clofarabine [201] are clinically used drugs whose metabolites target the A-site of RNR1. These appear to be non-covalent inhibitors that bind the A-site. Of these three, the best characterized is clofarabine, which is used to treat childhood leukemias [201-203]. Clofarabine triphosphate is an analog of dATP. Like dATP, clofarabine can also hexamerize RNR1 [124]. Stubbe and coworkers showed that the clofarabine inhibition is reversible. They also showed that clofarabine triphosphate inhibits hRR1 by binding at the A-site with a Ki of 40 nM. After initial inactivation, however, the enzyme recovers 50% of its activity. Furthermore, clofarabine diphosphate had a slightly lower Ki of 17 nM at the C-site and also induced RNR1 hexamerization in the study. As previously mentioned in the section on allosteric regulation by ATP and dATP, hexamerization is important for both activation and inactivation of the enzyme. In particular, dATP at physiological concentrations causes

RNR1 to hexamerize. Using site directed mutagenesis, we have shown that dimers of hRR1 cannot be inhibited by dATP [108]. Hence, nucleoside analogs that retain the ability to hexamerize the RNR1 subunit in the same way as dATP are likely to be potent inhibitors of ribonucleotide reductase.

1.5.2.2 C-Site Analogs

The most well studied substrate analog that binds the C-site is gemcitabine [45, 123].

Gemcitabine is a billion dollar drug that is a major component of standard chemotherapy for various cancers such as lung and pancreatic carcinomas [204, 205]. Gemcitabine, an analoge of deoxycytidine (2’-2’-difluorodeoxycytidine, F2dC), is sequentially phosphorylated to the 5’-monophosphate (F2dCMP) by deoxycytidine kinase, and to difluorodeoxycytidine 5’-diphosphate (F2dCDP) possible by uridylate-cytidylate monophosphate kinase (UMP/CMP kinase) [206]. However, the role of UMP/CMP kinase in phosphorylation of F2dCMP is controversial , since the metabolite levels of gemcitabine remain unaffected in cell lines overexpressing or underexpressing this enzyme [207]. In the presence of reductants, F2dCDP covalently modifies RNR1. In the absence of reductants, with pre-reduced RNR1 and RNR2, loss of the tyrosyl radical in

RNR2 leads to inhibition [45]. F2dCDP inactivates human RNR by generating a tight

α6β6 complex. F2dCDP has recently been shown to inhibit p53R2 (β’) but unlike in α6β6, the α6β’6 complex appears to be much weaker than α6β6 one [123]. Inhibition of RNR by

F2dCDP leads to reduction of the pool of dNTPs available for DNA synthesis, presumably favoring incorporation of the gemcitabine triphosphate metabolite by DNA polymerase  into growing DNA strands. [208].

Radiation sensitization by gemcitabine has been shown to correlate with dATP depletion through RNR inhibition and S-phase accumulation [209]. Schewach and colleagues hypothesized that radiosensitization to F2dCDP is due to dNTP pool imbalances leading to nucleotide misincorporations which increase cell death following irradiation. The misincorporation rates become significant when mismatch-repair

46 deficient cells are irradiated and treated with F2dCDP. Disruption of allosteric regulation of RNR can lead to dNTP pool imbalances. In the case of F2dCDP, nucleotide pool imbalances probably occur through the inactivation of RNR at the catalytic site and disruption of the allosteric communication between the specificity and catalytic sites. The latter is described below.

Although, there are no structural data for the quaternary structure of RNR in complex with gemcitabine, the structure of F2dCDP bound at the C-site of ScRR1 has been solved [210].

1.5.3 Inhibitors of the Small Subunit of RNR

The inhibitors of the RNR2 subunit are either radical scavengers or iron chelators.

Hydroxyurea (HU) is the best characterized tyrosyl radical scavenger and has been used to treat various neoplastic and non-neoplastic conditions [211, 212]. Its action is reversible and is restricted to the S-phase of the cell cycle. Hydroxyurea is used to treat a wide variety of neoplasms, including primary brain cancer, renal cell carcinoma, melanoma, breast cancer, chronic myeloproliferative disorders and chronic myeloid leukemia [213, 214]. Recently its therapeutic spectrum has been expanded to include non-neoplastic diseases such as sickle cell anemia. Hydroxyurea has limited clinical effectiveness as an anticancer drug because of its relatively short half-life and its low affinity towards RNR2 in humans.

Another class of inhibitors currently under consideration is the iron chelators

[215]. Since tyrosyl radical generation in the RNR2 subunit requires iron, iron chelators are considered as one of the most potent inhibitors of RNR activity. Deferoxamine is an

47 iron chelator that has shown promise in cancer therapy [216]. It has been shown to inhibit

RNR activity by reducing the intracellular pool of iron, rather than directly attacking the tyrosyl radical of RNR2 [217, 218]. Thiosemicarbazones are another group of iron chelators that inactivate RNR2. Of these, triapine (3-aminopyridine-2-carboxaldehyde thiosemicarbazone, 3-AP) is the most promising iron chelator due to its low toxicity. 3-

AP has shown higher inhibitory potency than HU and F2dCDP in a wide variety of cancer cell lines [219]. Furthermore, cells exposed to 3-AP appear to have enhanced sensitivity to treatment with radiation and/or cisplatinum [220]. Currently, 3-AP is undergoing extensive clinical trials to treat cancers as a monotherapy or a combination therapy [221-

225].

1.6 Summary, Rationale and Aims of the Current Study

RNR plays a central role in the cellular metabolism by providing the adequate and balanced pool of dNTPs required for DNA replication and repair. Any imbalance in the cellular dNTP pool increases mutation frequency and thereby predisposes cells to genomic instability. Cells avoid such catastrophic dNTP pool imbalances by regulating the transcription of nrd genes and allosterically controlling the RNR enzyme.

The allosteric regulation of RNR has been intensively studied over the last five decades. Structural and biochemical studies of specificity site regulation in both Class I and Class II RNR have been elucidated in great details in recent years and have provided insight into how cognate effector-substrate pairs are selected based on specificity cross talk. Of these two allosteric mechanisms, activity site regulation in RNRs is less well understood. Several biochemical and mutagenesis studies performed in vitro and in vivo

48 identified that both ATP and dATP binds with different affinities to a low affinity binding site on RNR and exerts either a stimulatory or inhibitory effect depending on whether

ATP or dATP is present at the A-site, respectively.

Until recently, two questions remained unanswered. Although it was known that dATP has a higher binding affinity than ATP to the A-site in order to overcome the stimulatory effect of ATP, the structural basis of this differential affinity remained obscure. The Class Ia E.coli RNR with AMPPNP bound at the A-site was the only structure available with an effector bound at the ATP-cone. Recent crystallographic studies carried out in our lab by soaking dATP and ATP into the A-site of HuRRM1 have revealed the structural basis of the binding affinity difference between ATP and dATP.

The other unresolved issue is why ATP acts as an activator and dATP acts as an inhibitor.

It was generally accepted that binding of ATP or dATP to the A-site brought about subtle structural changes in the α2β2 heterotetramer resulting in either stimulatory or inhibitory effects. Based on these studies it has now become clear that at least among the Class Ia

RNRs, binding of dATP/ ATP and subsequent induction of conformational changes in the interacting subunits are not enough for enzymatic activity, but formation of higher order oligomeric complexes is also required. Biochemical and biophysical studies with prokaryotic and eukaryotic enzymes have revealed major differences between ATP and dATP based oligomeric complexes. In prokaryotes, dATP induces an octameric holoenzyme with a subunit composition of α4 (β2)2, while the ATP based holoenzyme has an α2β2 subunit composition. In contrast, both ATP and dATP based holoenzymes in eukaryotes form active or inactive octameric α6β2 complexes.

Lack of structural information distinguishing between active and inactive octameric α6β2 complexes in eukaryotes prompted us to determine the structural basis of formation of these complexes. We hypothesized that hexamerization is a pre-requisite for dATP based inhibition of the RNR enzyme. Chapter 3 of this thesis describes in detail the biochemical and biophysical characterization of dATP and ATP based holoenzyme complex formation by human and yeast RNR.

RNRs are regulated at many levels, of which their allosteric and transcriptional regulation mechanisms are well studied. Some of these regulatory mechanisms are restricted to certain members of a given class of RNR. For example, RNR regulation by small proteins such as Sml1, Spd1 and Dif1 in yeast has been described. This novel mode of inhibition seems to exist only in some of fungal species. Chapter 4 & 5 of thesis are devoted for studies carried out to determine the molecular mechanism by which Sml1 inhibits the RNR enzyme of S. cerevisiae.

The number of agents available to treat fungal infections has dramatically increased over the past decade, yet there are still only about 15 antifungal agents belonging to polyenes [226], echinocandins [227], triazoles [228] and flucytosine [229] currently approved for clinical use. Even though the greater number of anti-fungal agents available allows for therapeutic choices, significant differences exist in their spectrum of antifungal activity, bioavailability, drug interactions and side effects. Since the incidence of infections with invasive mycoses continues to increase with increasing immunosuppressed patient populations, there is a greater need for novel anti-fungals that offer more effective and less toxic therapies. Therefore, understanding the mechanistic

50 and structural basis of ScRR inhibition by Sml1 is important in part because it may lead to the development of novel antifungal agents.

Chapter 2: Materials and Methods

2.1 Protein Expression and Purification

2.1.1 Site Directed Mutagenesis

Expression plasmids for hRRM1 mutants (D16R, H2E and D57N) were constructed commercially by TOP Gene Technologies (Quebec, Canada). The primers used are listed in Table 2.1 (p, 54).

2.1.2 Preparation of the Peptide Affinity Column

Purification of ScRR1 and HuRRM1 was done by using peptide affinity chromatography. Two different peptide affinity columns based on ScRR4

(KEINFDDDF) and HuRRM2 (NSFTLDADF) peptides were made by coupling them to

NHS-Activated Sepharose 4-Fast Flow Resin (GE Life Sciences, Piscataway, NJ) using a modified protocol [230]. All the peptides were synthesized at the W. M. Keck Facility.

Briefly, each peptide was suspended separately in 1 ml of 50 mM Sodium bicarbonate at pH=7.0 to a concentration of 20 mM and dissolved by addition of a small volume (1/200 total volume) of 1M NH4OH. The pH of the peptide solution was adjusted to 7.0 by adding a small volume of 500 mM NaH2PO4 pH 4.5 while checking the pH using small amounts of the solution on a litmus paper.

Then, 1 ml of the NHS-Activated Sepharose 4 Fast Flow resin was added to a

Ecnocolumn (Biorad, CA) with a glass bed support. The resin was washed with 10 column volumes of ice cold 1 mM HCl followed by 5 column volumes of ice cold phosphate buffer saline solution (137 mM NaCl, 2.5 mM KCl, 4.3 mM Na2HPO4, 1.4

52 mM K H2PO4 at pH=7.4). Then one end of the column was sealed and 1 ml of the peptide solution was added to the resin. The peptide solution was then incubated overnight with the resin at 4ºC while swirling. During this step, primary amines on the peptide will react with N-hydroxy succinimide (NHS) esters and be covalently linked to the resin. After 12 hours of incubation, the resin in the column was washed with 10 column volumes of 50 mM Tris at pH=8.0. The peptide coupled resin was incubated for an additional 12 hours at 4ºC with 2 column volumes of 50 mM Tris at pH=8.0. This step was essential to inactivate the remaining NHS esters in the resin. The column was subsequently washed two times with 3 column volumes of 100 mM acetic acid at pH=3.5. After each purification, peptide columns were washed with a 6M guanidium hydrochloride solution followed by 50 column volumes of ultrapure water and stored in 20% ethanol at 40C.

2.1.3 Protein Concentration Determination

Concentrations of the protein solutions that contain ATP or other dNTPs were determined either by Bradford assay using Coomassie® Plus Protein Assay Reagent Kit (PIERCE

Rockford, IL). The same assay was also used to measure the concentrations of wild type

Sml1 and C14S Sml1 and. Protein concentrations of the large and small subunits of the

RNRs were determined by measuring their absorbance at a wavelength of 280 nm [231].

Extinction coefficients of the respective proteins were determined using the Expasy

ProtParam server (http://web.expasy.org/protparam/). When determining the concentrations of Alexa 350 labled C14S S60C Sml1 and 6X His-GFP-Sml, in addition to the above methods, the extinction coefficients of 17,000M-1cm-1 at 344 nm and 56,000

M-1cm-1 at 484 nm were used.

Table 2.1 Primers used in this Study

2.1.4 Expression & Purification of ScRR1

S. cerevisiae ScRR1 protein was expressed in E. coli BL21 (DE3) pLysS

(Invitrogen, Carlsbad, CA). Briefly, the plasmid pWJ751-3 conferring ampicillin resistance was transformed into BL21 (DE3) pLysS (Invitrogen, Carlsbad, CA) cells. A single colony was picked from the transformants and was expanded in an overnight culture of 50 mL of LB medium containing 100 µg/ml ampicillin and 34 µg/ml chloramphenicol. The Initial overnight culture was then scaled-up by diluting it 1:100 by inoculating 10 ml of it into 1 L of cultures containing Terrific Broth (TB) media, 100 mg/L ampicillin and 34 mg/L chloramphenicol. The expanded cultures were then incubated at 37˚C while shaking at 200 RPM until the optical density (OD600) reached a value of 0.6. The cultures were chilled to 15˚C on an ice-water mixture for 30 minutes and expression of the ScRR1 was induced by adding IPTG to a final concentration of 0.5 mM. The expressing cultures were incubated for 16-20 hours at 15˚C while shaking at

200 RPM. Harvest was performed by centrifugation at 5,000 x g for 20 minutes.

Following centrifugation, cells were flash-frozen in liquid nitrogen and stored at -80˚C.

Cells were thawed and re-suspended in lysis buffer (50 mM HEPES pH 7.0, 5 mM MgCl2, 5% glycerol v/v, 5 mM DTT, 1 mM PMSF, 1x COMPLETE EDTA Free

Protease Inhibitor (Roche Biochemicals, Indianapolis, IN). Cells were then lysed using a

French press at 15,000 pounds per square inch (psi). The lysate was then centrifuged at

30,000 x g for 40 minutes to remove cellular debris. The resulting supernatant after centrifugation was collected and the ribonucleoproteins were precipitated by adding streptomycin sulfate to a final concentration of 1.5% w/v and incubating the mixture at

4˚C for 30 minutes. The precipitated proteins were then removed by centrifugation at

30,000 X g for 20 minutes. Supernatant was then collected and ammonium sulfate was added to a final concentration of 40% w/v over a 15 minute period and allowed to incubate at 4˚C for 30 minutes. Precipitated proteins were collected by centrifugation at

30,000 x g for 30 minutes. Supernatant was discarded and the protein pellet containing

ScRR1 was stored at -80˚C.

The ammonium sulfate pellet of ScRR1 was re-suspended in the lysis buffer and de-salted using PD10 columns (GE Life sciences, Piscataway, NJ). The de-salted protein solution was incubated with 1 ml of ScRR4 9-mer peptide-affinity resin loaded into the

Econo-column (Biorad, CA) for one hour at 4˚C. After incubation, the column was washed with 5 column volumes of Buffer A (50 mM HEPES pH 7.0, 5 mM MgCl2, 5% glycerol, 5 mM DTT). The column was washed with Buffer A until OD280 reached 0, followed by another wash of 80% Buffer A + 20% Buffer B (Buffer A + 1 M KCl) to remove weakly and non-specifically bound proteins. ScRR1 was then eluted from the column with 15 column volumes Buffer B. Purity was assessed by 12% SDS-PAGE

(Figure 2.1, p 57). Pure fractions were pooled and concentrated using Centricon and

Microcon centrifugal concentrators (Millipore, Bedford, MA) with a molecular-weight cutoff of 50 kDa. The protein sample was diluted 1:10 with Buffer A to bring the concentration of KCl down to 0.1 M and the diluted sample was then re-concentrated to the desired concentration.

Figure 2.1 Purification of ScRR1

Wild type ScRR1 was purified using peptide affinity chromatography based on yeast

RNR4 peptide (KEINFDDDF). Bound proteins were eluted using a 1M KCl gradient

(green line). A similar procedure was used to purify the mutants (i.e K10E, K17E and

K10EK17E). Protein absorbance (blue line) was measured at 280 nm and the fractions (

) were collected and concentrated. The purity of the concentrated fractions was tested using 12% SDS-PAGE (inset) and the protein identity was confirmed by comparing it with molecular weight standards.

2.1.5 Expression & Purification of ScRR2●ScRR4

BL21CodonPlus(DE3) RIL cells were co-transformed with HisX6 ScRR2 and

ScRR4 plasmids kindly provided by JoAnne Stubbe (MIT) and plated on an LB agar plate with 50 µg/ml kanamycin and 100 µg/ml ampicillin [137, 232, 233]. A single colony was chosen and an overnight culture of 50 ml was grown to saturation. The overnight culture was scaled-up 100-fold in TB medium with 50 µg/ml kanamycin and

100 µg/ml ampicillin and grown at 37˚C to an OD600 of 0.6. Prior to induction cultures were treated with 1 mM EDTA for 15 minutes at 37˚C. The cultures were induced at

15˚C by adding 1 mM IPTG, and cells were harvested 6 hours after induction. For the purification of ScRR2●ScRR4, cells were re-suspended in lysis buffer (50 mM HEPES pH 7.4, 5% glycerol, 1mM PMSF, 1X COMPLETE EDTA-free protease inhibitor cocktail) and lysed by passing through a French Press at 15,000 PSI, followed by centrifugation at 30,000 x g for 30 minutes. The supernatant was treated with 2% (w/v) streptomycin sulfate. The supernatant after centrifugation was incubated with TALON cobalt resin (BD Bioscience) for 1 hour at 4˚C. Impurities were removed by washing the column with 50 column volumes of buffer containing 50 mM HEPES, 5% glycerol, 100 mM KCl and 10 mM imidazole at pH 7.4. The ScRR2●ScRR4 protein was eluted with buffer containing 100 mM imidazole in 50 mM HEPES, 5% Glycerol and 100 mM KCl at pH 7.4. Protein-containing fractions were identified by 10% SDS-PAGE, and imidazole was removed using a PD-10 column equilibrated with 50 mM HEPES pH 7.4,

5% glycerol and 100 mM KCl (Figure 2.2 p, 59).

Figure 2.2 Purification of ScRR2•ScRR4

ScRR2•ScRR4 was purified using Cobalt affinity chromatography. Bound proteins were eluted using 100 mM Imidazole in washing buffer. Protein absorbance (blue line) was measured at 280 nm and the fractions ( ) were collected and concentrated. The purity of the concentrated fractions was tested using 12% SDS-PAGE (inset) and the protein identity was confirmed by comparing it with known molecular weight standards.

2.1.6 Expression & Purification of Sml1

Yeast Sml1p was expressed in BL21 (DE3) pLysS cells. The overnight culture was scaled-up in large culture flasks containing liquid Terrific Broth (TB) media, 100 mg/l ampicillin, and 34 mg/l chloramphenicol. Cells were grown in liquid TB media at

37˚C until OD600 reached 0.6. Cells were then induced with 0.5 mM IPTG for 3 hours at

37˚C and harvested by centrifugation at 5,000 x g for 30 minutes. Harvested cells were re-suspended in lysis buffer (50 mM TRIS pH 7.4, 10% glycerol, 5 mM DTT, 1 mM

PMSF, 1 mM EDTA, 1x COMPLETE EDTA Free Protease Inhibitor (Roche

Biochemicals, Indianapolis, IN), and 1:10,000 dilution of Benzonase®Endonuclease

(EMD Biosciences, San Diego, CA). Re-suspended samples were flash-frozen in liquid nitrogen and stored at -80˚C.

Cells were thawed on ice and were subjected to ultra-centrifugation at 150,000 x g for 1 hour in an Ultima Max ultracentrifuge (Beckman, Fullerton, CA). The supernatant was collected, and ammonium sulfate was gradually added at 40C and stirred for 30 minutes to a final concentration of 25% w/v. Precipitated protein was separated from the supernatant by centrifugation at 30,000 x g for 30 minutes. The ammonium sulfate pellet was then re-suspended in 2.5 ml buffer (50 mM HEPES pH 7.4, 5% glycerol, 100 mM

KCl, 5 mM DTT, 1 mM PMSF, 1x COMPLETE EDTA Free Protease Inhibitor (Roche

Biochemicals, Indianapolis, IN) and loaded onto a HiLoad 16/60 Superdex 75 (GE

Lifesciences, Piscataway, NJ) size exclusion column. Protein peaks were analyzed by

15% SDS-PAGE and the fractions containing Sml1p were pooled (Figure 2.3 p, 62).

Contaminating DNA was removed by anion exchange chromatography using a DEAE

Sepharose column (GE Lifesciences, Piscataway, NJ) on the pooled Sml1p fractions

60 using 50 mM HEPES, 5% glycerol, 100 mM KCl and 5 mM DTT at pH 7.4. After anion exchange, the Sml1 solution was concentrated using an Amicon (Millipore, Bedford,

MA) ultrafiltration device with a molecular weight cut-off of 10 kDa.

GFP-Sml1 was expressed in ArcticExpress (DE3) E. coli cells (Agilent

Technologies, CA) using auto-induction. Cells were initially grown in ZYP-5052 rich medium at 37˚C until the OD600 reached 0.5, cooled to 15˚C and allowed to grow for 12 hours in the same auto-induction medium containing lactose. Expression of the GFP-

Sml1 mutant was verified by analysis on 12% SDS-PAGE gels, and cultures were then allowed to grow for another 24 hours before the cells were harvested by centrifugation.

GFP-Sml1was purified using Cobalt affinity chromatography followed by Superdex 75

10/300 GL SEC.

Figure 2.3 Purification of Sml1

Wild type Sml1 was purified using Superdex 75 size exclusion chromatography (SEC).

Protein absorbance (blue line) was measured at 280 nm and the fractions ( ) were collected and concentrated. Excess DNA was removed by DEAE anion exchange chromatography. A similar procedure was used to purify C14S S60C Sml1. The purity of the concentrated fractions was tested using 12% SDS-PAGE and the protein identity was confirmed by comparing it with known molecular weight standards

2.1.7 Expression & Purification of HuRRM1

Wild type cDNA for human RNR1, termed HuRRM1 was kindly provided by Dr.

Yen (City of Hope Hospital, Los Angeles, CA). The cDNA was subcloned into a pET-

21a (+) vector. Expression plasmids for hRRM1 mutants (D16R, H2E and D57N) were constructed commercially by TOP Gene Technologies (Quebec, Canada). The primers used are listed in Table 2.1. Wild-type and D57N HuRRM1proteins were expressed recombinantly in E. coli BL21-CodonPlus-RIL (Stratagene). Cells were grown in TB medium with 100 mg/L ampicillin, and 34 mg/L chloramphenicol at 37˚C until the culture reached an OD600 of 0.6. Cells were then chilled to 15˚C for 30 minutes and induced with 0.5 mM IPTG for 18 hours at 15˚C while shaking at 220 RPM. Cells were harvested by centrifugation at 5000 x g for 30 minutes, and cell pellets were stored at -

80˚C.

D16R mutant protein was expressed in ArcticExpress cells (Agilent Technologies,

CA) cells. D16R cells were grown in a 50ml pre-culture for 12 hours in LB medium containing 100 mg/l ampicillin and 30 mg/l kanamycin. The pre-culture was scaled up in

TB medium without the antibiotics at 30˚C until it reached an OD600 of 0.6. Cells were chilled to 11˚C, and protein expression was induced by the addition of 0.5 mM IPTG for

24 hours while shaking at 220 RPM. Purified D16R hRRM1 protein was confirmed by

Western blot analysis and mass spectrometry.

H2E mutant protein was expressed in RosettaTM 2 E. coli cells (EMD Milipore,

Germany) using auto-induction [234][234]. Cells were initially grown in ZYP-5052 rich medium at 37˚C until the OD600 reached 0.5, cooled to 15˚C and allowed to grow for 12 hours in the same auto-induction medium containing lactose. Expression of the H2E

63 mutant was verified by analysis on 12% SDS-PAGE gels, and cultures were then allowed to grow for another 24 hours before the cells were harvested by centrifugation.

Cells were thawed on ice and re-suspended in lysis buffer (50 mM Tris pH 8.0,

5% glycerol, 5 mM MgCl2, 10 mM DTT, 1 mM PMSF, 1 x COMPLETE EDTA-free protease inhibitor (Roche), 1:10,000 dilution Benzonase). Cells were lysed using a

French Press at 15000 PSI, followed by centrifugation at 30,000 x g for 30 minutes to produce cleared lysate. After centrifugation, the supernatant was collected and ammonium sulfate was added to a final concentration of 50% (w/v) and incubated at 4˚C for 30 minutes with stirring. Precipitated protein was collected by centrifugation at

30,000 x g for 30 minutes. The supernatant was discarded, and the ammonium sulfate pellet was stored at -80˚C.

Ammonium sulfate pellets were thawed on ice and re-suspended in the same lysis buffer without Benzonase. The re-suspended ammonium sulfate pellet was desalted using

PD10 disposable desalting columns (GE Life Sciences). Desalted protein solution was allowed to incubate with peptide affinity resin for 1 hour (see below for construction of the peptide affinity column). After incubation, the column was loaded onto an AKTA

Purifier (GE Healthcare). The column was washed with buffer A (50 mM Tris pH 8.0,

5% glycerol, 5 mM MgCl2, 10 mM DTT) and the protein was eluted using a linear gradient from 100% buffer A to 100% buffer B (buffer A + 2 M NaCl). Fractions were analyzed for purity using a 12% SDS-PAGE gel, and the fractions containing hRRM1 were pooled and concentrated.

2.1.8 Expression & Purification of HuRRM2

BL21 (DE3) cells were transformed with hRRM2 plasmids kindly provided by

JoAnne Stubbe (MIT) and plated on an LB agar plate with 50 µg/ml kanamycin. A single colony was chosen and an overnight culture of 50 ml was grown to saturation. The overnight culture was scaled up in 4 liters of LB medium with 50 µg/ml kanamycin and grown at 37˚C to an OD600 of 0.9. Prior to induction cultures were treated with 1 mM

EDTA for 15 minutes at 37˚C. They were then induced at 30˚C by adding 1 mM IPTG, and cells were harvested 6 hours after induction. For the purification of hRRM2, cells were re-suspended in lysis buffer (50 mM NaH2PO4 pH 7.0, 1% Triton X-100 , 0.1 mM

PMSF, 1 x COMPLETE EDTA-free protease inhibitor and 10 mM β-mercaptoethanol) and lysed by passing through a French Press at 15000 PSI, followed by centrifugation at

30,000 x g for 30 minutes. The supernatant was treated with 2% (w/v) streptomycin sulfate. The supernatant after centrifugation was incubated with PrepEase® (High Yield)

Resin (USB) for 1 hour at 4˚C. Impurities were removed by washing the column with lysis buffer containing 800 mM NaCl and 50 mM imidazole. The hRRM2 protein was eluted with buffer containing 100 mM imidazole in 50 mM NaH2PO4 pH 7.0, 1% Triton

X-100, 10 mM β-mercaptoethanol, 300 mM NaCl. Protein-containing fractions were identified by 10% SDS-PAGE, and imidazole was removed using a PD-10 column equilibrated with 50 mM HEPES pH 7.6, 5% glycerol, 100 mM KCl.

2.2 Labeling of Sml1

All fluorescent studies were done with C14S S60C labeled Sml1. C14S/S60C

Sml1 was labeled with Alexa 350 C5-maleimide (Alexa 350, Molecular Probe, Eugene,

OR). Sml1 concentration was kept at 50 µM before conjugating the fluorescence probe.

The conjugation reaction was carried out in a buffer containing 50mM HEPES, 5% (v/v) glycerol, 0.5mM TCEP and 0.1M KCl at pH 7.2. Throughout the protocol light exposure was minimized. Alexa 350 was dissolved in ultra-pure water to a final concentration of

10 mM and centrifuged for 10 minutes at 20,000 x g at room temperature. The concentration of Alexa 350 C5-maleimide was determined from the diluted sample based

- on absorbance at 444 nm and extinction coefficient of Alexa 350-amide (ε444=17,000cm

1 -1 M ). Alexa 350 C5-maleimide dissolved in water was added to the C14S/S60C Sml1 solution in 10-fold molar excess. The mixture was kept in dark and incubated at 4ºC for

12. After incubation, DTT was added to 15 molar excess of Alexa 350 C5-maleimide to inactivate unreacted fluoropore. The sample was concentrated to 100 µl using Amicon

Ultra-15 centrifugal filter units (Milipore, MA) with molecular weight cut-off 10 kDa and the sample was then injected to Superdex 75 10/300 gel filtration column (GE

Lifesciences, Piscataway, NJ pre-equilibrated with 50mM HEPES, 5% glycerol and 0.1M

KCl at pH 7.2 to further purify the sample.

Fractions containing the labeled Sml1 were further concentrated to 1-3 mg/by

Amicon Ultra -15 centrifugal units with a molecular weight cutoff of 10 KDa. The concentrated sample was divided into aliquots and frozen in liquid nitrogen. The samples were wrapped with aluminum foil and kept at –80ºC until they were used. The concentration of Sml1 was determined by Coomassie Plus Protein Assay Kit (Thermo

Scientific, IL). A molar ratio of Alexa 350 to Sml1 of >0.9 demonstrated that over 90% of Sml1 was conjugated with the fluorescence probe.

2.3 Determining the Enzymatic Activities of RNRs

2.3.1 Iron Loading to the Small Subunit of RNR

The apo form of HuRRM2 or ScRR2●ScRR4 was expressed and purified as described.

Before loading iron, the buffer solution (50mM HEPES / 50mM Tris 5% glycerol 0.1M

KCl at pH 7.6) was deoxygenated on a Schlenk line by repeated cycles (>5) of evacuation

(1 min) followed by flushing with argon (30 Sec). The DNA content of the protein solutions was quantified by measuring the absorbance ratio at 260nm/280nm since iron loading requires this ratio to be < 0.6. As the protein solutions (a 400µl aliquot at 40µM concentration in a 10ml flask) were deoxygenated, great care was taken to avoid protein precipitation by direct exposure to the vacuum. This was achieved by continuous turning

(i.e. 180% rotation from one closed position to next closed position) of the stopcocks of the flasks containing protein solutions, thus exposing the protein solutions to vacuum for less than one second at a time. In general, this continuous turning of the stopcock was done 10 times during one cycle of evacuation and then the protein solution was immediately exposed to argon for 30 seconds. The deoxygenated buffers, the protein solutions and open Ependroff tubes containing 6 mg of (NH4)2Fe(SO4)2. 6H2O, 14.8 mg of ascorbic acid and 4.9 mg of ferozine were taken inside a glove box (M. Braun, NH) at

0 4 C. Inside a glove box, 6 mg of Fe(NH4)2SO4 , 4.9 mg of ferrozine and 14.8 mg of ascorbic acid were dissolved in 1 ml of deoxygenated buffer. Fe (II) concentration in the solution was determined using a modified Ferrozine assay [235] under reducing and non- reducing (i.e. with and without ascorbic acid) conditions (Table 2.2 , p 70). The amount

67 of Fe (II) was determined by measuring the absorbance of the purple color complex

-1 -1 (ε562=27870 cm M ) at 562 nm after 2-fold dilution of the solution.

The following equations based on Beer-Lambert Law were used to calculate the concentration of the iron solution (i.e. [Fe (II)])

With ascorbic acid (Tube A; reducing condition),

[Fe (II)]= [A562 *2* (200+200+200+5+40)/5]/ 2.787* 10^4) Eqn- (2.1)

Without ascorbic acid (Tube B; non-reducing condition),

[Fe (II)] = [A562 *2* (200+200+200+5+40)/5]/ 2.787* 10^4) Eqn- (2.2)

The iron solution was only used for making the free radical, if the Fe (II) concentrations determined as above under reducing and non-reducing conditions were nearly identical

(i.e. differences between the two absorbance values should not be > 0.1 AU). Based on the Ferozine assay, a total of 5 equivalents of Fe(II) per HuRRM2 or ScRR2●ScRR4 dimer from (NH4)2Fe(SO4)2 solution was added drop wise to the protein solutions and incubated at 4ºC in the glove box [45]. Diferric tyrosyl radical generation requires O2 and the amount of O2 required was calculated using Eqn – (2.3) While the protein solutions were being incubating inside the glove box, 10ml of ice cold buffer (50mM HEPES/Tris,

0.1M KCl, 5% glycerol pH 7.6) was saturated with O2. The amount of O2 needed to convert Fe (II) to Fe (III) was 3.5 equivalents per RNR2 dimer and the required volume of O2 containing buffer was calculated using the following equation.

Required volume of O2 = ([RNR2 dimer] * volume of the protein solution* 3.5) / (1.9*

10^3) Eqn- (2.3)

The protein solutions were then removed from the glove box and the 8 volumes O2 containing buffer calculated from Eqn (2.3) was added. Subsequently O2 was blown over the surface of the protein solution for 30 seconds. Excess iron was removed by Superdex

S200 10/300 size exclusion chromatography and the proteins were concentrated to the desired concentration and aliquots were flash frozen. The efficiency of iron loading was determined by measuring the RNR2 specific activity (See below). The reconstruction of the ScRR2●ScRR4 and the subsequent determination of the ScRR1 and ScRR2●ScRR4 specific activities were carried out in similar fashion.

Table 2.2 Components of the Ferrozine assay

Components (µl) Tube A Tube B

Fe solution 5 5

Ascorbic Acid 40 0

Buffer 0 40

1M HCl 200 200

Ammonium Acetate pH 5.00 * 200 200

Ferrozine solution 200 200

* Saturated ammonium acetate solution.

2.3.2 Preparation of Boronate Columns

Anion exchange beads (AG-X8: Biorad, CA) was packed in a column and washed with 10 column volumes of 1M NaOH followed by another 10 column volumes of deionized (Mili Q : 18.2 MΩ) water. Beads were then stored in 0.6 M saturated potassium tetraborate solution. To make the column for the RNR activity assay, the bottom of a Pasteur pipette (o.6 cm diameter) was plugged with glass wool and 2.00 ml of the boronate resin was put in the pipette packed by gravity. Packed columns were then washed with 10 column volumes of Mili Q deionized water.

2.3.3 Preparation of Radioactive Stocks

Unless otherwise specified, all radioactive stocks were made in the specific activity range of 2,000-4,000 cpm/nmol. Radioactive [3H]-CDP and [14C]-ADP were purchased from Vitrax Radiochemicals (Placentia, CA) and PerkinElmer (Waltham,

MA), respectively. For the preparation of [3H]-CDP, 100 µl of the 0.05 µCi/µl stock was lyophilized to complete dryness. Likewise, 250 µl of 0.02 mCi/ml of [14C]-ADP was also lyophilized. The dried samples were then dissolved in 185 µl of 30 mM cold CDP or

ADP prepared in 20mM HEPES at pH 7.0. Radioactive stocks were diluted in 1/1000 in

20mM HEPES buffer at pH 7.0 and the concentration of the individual stocks was determined from a diluted solution taking into the consideration the extinction

-1 -1 -1 -1 coefficients (€260 ADP = 15, 400 M cm and €270 CDP = 9,000 M cm ). To determine the specific activity of each stock, the original stocks were diluted (1/100) and varying volumes (2,4, 6 and 8 µl) of the diluted solutions were mixed with aqueous scintillation cocktail (SX-8, Fisher) and the radioactive decay events (cpm) were counted. The volume

71 was plotted against the cpm, and analyzed using linear regression to determine the concentration.

2.3.4 Determining the Specific Activities of the Large & Small Subunits of RNRs

The specific activities of both the large and small subunits of RNR were determined using an in vitro activity assay using either 3H-CDP or 14C-ADP. When determining the specific activity of the small subunit (i.e. HuRRM2 or ScRR2●ScRR4), the reaction mixture contained 0.3 µM of the small subunit and 2.1 µM of the large subunit (i.e. HuRRM1 or ScRR1) in an activity assay buffer of 50 mM HEPES, 15 mM

3 MgCl2, 1 mM EDTA, 100 mM KCl, 5 mM DTT, 3 mM ATP and 1 mM [ H]-CDP

(~3000 cpm / noml) at pH 7.6. When assaying ADP reductase activity, the assay buffer contained 100 µM dGTP and 1 mM [14C]-ADP (~3000 cpm/nmol) with or without 3mM

ATP. When determining the specific activity of the large subunit (i.e.HuRRM1 and

ScRR1), the reaction mixture contained 0.3µM of the large subunit and 2.1 µM of the small subunit while the other components were the same as mentioned above.

When assaying for human RNR activity ( wild type and D16R and H2E mutants), the reaction mixture was pre-incubated for 3 min at 37˚C, and 30 µl aliquots were sampled at fixed time intervals over a period of 12 min after reaction initiation. ScRR activity was measured at 30 ˚C. Product formation was linear over the 12 min time course. Reactions were quenched by immersion in a boiling water bath, cooling, and treatment with 10.2 U/ml of alkaline phosphatase (Roche). Product [3H]-dCDP or [14C]- dADP that formed during the reaction was separated from substrate [3H]- CDP or [14C]-

ADP using boronate affinity chromatography[45]. The amount of 3H-dCDP or 14C-dADP

72 formed was quantified by liquid scintillation counting using a Beckman LS6500 liquid scintillation counter.

2.3.5 Determining the Effect of dATP On Human & Yeast Enzyme Activity

The effect of dATP on HuRR and ScRR activity was measured using the in vitro activity assays described above. The reaction mixtures contained increasing concentrations of dATP (0, 5, 10, 20 and 50 µM). Control experiments were carried out by substituting a corresponding amount of buffer (50mM HEPES 15mM MgCl2, 1mM

EDTA and 100mM KCl) for each dATP concentration.

2.3.6 Determining the Mode of Inhibition of ScRR Activity by Sml1

Enzyme kinetics experiments were performed with 0.5 µM ScRR1dimer and 5

µM ScRR2•ScRR4. When ATP/CDP was used as the effector-substrate pair, [3H]-CDP

(~ 900 cpm/nmol to ~3000 cpm/nmol) concentrations were varied from 0.05 mM to1 mM

(0.05, .10, .28, .40, 1.0 mM). Similarly when dGTP / ADP were used as the effector- substrate pair, ADP (~3000 to 9000 cpm/nmol) concentrations were varied from 0.05 mM to1 mM (0.05, 0.10, 0.28, 0.40, 1.0 mM). Initial rates of substrate reduction (i.e. amount of dCDP or dADP formed per unit time) at different concentration of [3H]-CDP and [14C]-ADP were first determined in the absence of Sml1. For the inhibition studies with [3H]-CDP, we used 1, 1.5 and 2 µM final concentrations of Sml1 monomer. For

[14C]-ADP, we used 1, 2, and 3µM final concentrations of Sml1 monomer.

Likewise, to determine the effect of Sml1 on ScRR1 specific activity, 0.3 µM

ScRR1 was assayed with 2.1 µM ScRR2•ScRR4 in the above activity assay buffer with 4

µM Sml1 monomer. To test the effect of Sml1 dATP hexamer the reaction mixture contained 20 µM dATP in addition to the components mentioned in [3H]-CDP / [14C]-

ADP assay. Specific activities of ScRR1 mutants (K10E, K17E, K10E / K17E) were determined under similar conditions.

2.3.7 Determining the IC50 Value of HuRRM1 for Sml1

To determining IC50 value for human RNR, HuRRM1 activity was measured with increasing concentration of Sml1 (0.3, 3, 10 and 30) µM. The activity assay mixture contained 50mM HEPES 15mM MgCl2, 1mM EDTA, 0.3µM HuRRM1, 2.1 µM

HuRRM2, 3mM ATP , 1mm [3H]-CDP (~3100 cpm/nmol). Products were analyzed as described above. The IC50 value obtained above was converted to the corresponding Ki value using Cheng-Prusoff equation.

2.3.8 Enzyme Kinetics and Data Analysis

Modes of enzyme inhibition were determined by visual inspection of data series intercepting at x- and y- axes using double reciprocal plot analyses. Enzyme kinetics parameters were determined through fit of the data to Eqn (4.1)-(4.8) (Section 4.2.4, p

131) using non-linear least squares available in the statistical programming language R

[236]. Goodness-of-fit penalized for the number of estimated model parameters (to prevent over-fitting) was evaluated for the models using Akaike’s Information Criterion,

AIC=2k-2LogL where k is the number of estimated parameters and LogL is the maximum value of the log-likelihood function. Among a set of candidates, the model with the minimum AIC is considered to be the best.

2.4 Size Exclusion Chromatography

Molecular weight estimation of monomeric and oligomeric proteins was carried out using size exclusion chromatography (SEC). For estimation of molecular weights less than 75 kDa, Superdex 75 16/60 HR (GE Lifesciences) or Superdex 75 10/ 300 GL (GE

Lifesciences) columns were used. The columns were calibrated using molecular weight standards consisting of ovalbumin (44 kDa), carbonic anhydrase (29 kDa) and ribonuclease A (13.7kDa). For estimation of molecular weights greater than 75 kDa,

Superdex 200 16/60 HR (GE Lifesciences) or Superdex 200 10/ 300 GL (GE

Lifesciences) columns were used. The columns were calibrated using a standard curve based on molecular weight standards consisting of ribonuclease A (13 kDa) myoglobin

(17 kDa), ovalbumin (44 kDa), conalbumin (75 kDa), γ-globulin (158 kDa), ferritin (440 kDa), and thyroglobulin (670 kDa). The exclusion limits of the columns were determined by Blue dextran. Based on the elution volumes (Ve) of these standards, Kav values were determined using the void volume (Vo) and the total bed volume (Vt) of the individual columns as indicated in the Eqn 2.4.1. For each columns, standard curves were created that plot the log10 (molecular weights) and the Kav values against each other. From these plots the equations for the relationships between the two parameters were determined using least square linear regression. These equations were used to determine the molecular weights of the unknown species using their elution volumes.

(Eqn 2.4)

2.4.1 Characterization of HuRRM1 Oligomers

Higher-order oligomers of hRRM1 were characterized by SEC using a Superdex

200 10/300 GL (S200) column (GE Lifesciences). The S200 column was pre-equilibrated with 50 mM Tris pH 7.6, 5 mM MgCl2, 100 mM KCl, and equilibrated with a fixed amount of dATP in the range of 1-20 µM prior to each run. ATP-induced oligomers of hRRM1were analyzed in the presence of 3 mM ATP. The nucleotide concentration in protein samples was adjusted to that of the S200 running buffer for each run. Protein samples (50 µl) were injected onto the column and eluted in equilibration buffer at a flow rate of 0.5 ml/min. Molecular weights of the eluting oligomeric species were calculated using a standard curve based on molecular weight standards.

2.4.2 Purification of the ScRR1 Hexamer

ScRR1 was expressed and purified using peptide-affinity chromatography as previously described [31, 196]. Purified ScRR1 (5 µM) was incubated with 25 μM dATP in 50 mM ammonium acetate pH 7.0, 5% glycerol, 5 mM MgCl2, 100 mM KCl, and 5 mM DTT. A Superdex-200 10/300 GL column (GE-Life Sciences) was equilibrated with the above buffer. The ScRR1 sample (200 µl) 3was injected, and the eluted fractions were collected. The molecular weight of the ScRR1-dATP hexamer was determined using a calibration curve based on molecular weight standards.

2.4.3 Purification of the SCRR1●dATP Holo Complex

Pure ScRR1 hexamer was prepared as described above. The ScRR2●ScRR4 (ββ’) heterodimer was purified using cobalt affinity chromatography [140]. The α6●ββ’●dATP holo complex was prepared by first incubating α6 with ββ’ in 50 mM ammonium acetate pH 7.4, 50 μM dATP, 100 μM CDP, 5% glycerol, 5 mM MgCl2 and 100 mM KCl. The molar ratio between α6 and ββ’ was 1:2. The complex was loaded onto a Superdex-200

10/300 GL column (GE-Life Sciences) that was pre-equilibrated with the above buffer.

The molecular weight of the holo complex was calculated based on the standard curve using ribonuclease A (13 kDa), conalbumin (75 kDa), γ-globulin (158 kDa) and thyroglobulin (670 kDa) as standards. The presence of the α6●ββ’ complex was confirmed by collecting fractions corresponding to the molecular weight of the holo complex and running them on 4-12% SDS-PAGE (Invitrogen). The peak fraction of the holo complex was immediately used for EM imaging as described below.

2.4.4 Purification of the ScRR1● TTP● Sml1 Complex

Protein samples (ScRR1●Sml1 complex and ScRR1) for protein footprinting were purified in buffer pre-equilibrated with 10 mM Sodium cacodylate, 5mM MgCl2 ,

0.1 M KCl 1mM TCEP and 100 µM TTP. The molar ratio between ScRR1●TTP (50µM) and Sml1 monomer (100 µM) was 1:4 and the mixture was incubated for 10 minutes on ice. TTP concentration of the sample was adjusted to 100 µM. Then 100 µl of sample was injected onto Superdex S200 10/300 column and 250 µl fractions were collected.

The peak fraction was analyzed by 4-20% SDS-PAGE and visualized by Commassie blue

77 stain for the presence of both proteins. Samples were sent on ice to BNL for exposure to the radiation at Beamline X 28-C.

2.4.5 Purification of the dATP Induced ScRR1 Hexamer●Sml1 Complex

Complex formation of ScRR1 and Sml1 was investigated using an S200 SEC column pre-equilibrated with 50 mM HEPES 5% (v/v) Glycerol, 5 mM MgCl2, 100 mM

KCl, and appropriate nucleotides prior to each run. ATP/dATP-induced oligomers of

ScRR1 were analyzed in the presence of 3 mM ATP or 50µM dATP respectively. The nucleotide concentration in the protein samples was adjusted to that of the S200 running buffer before each run. The dATP-induced hexamer-Sml1 complex was prepared by first incubating 10 µM of ScRR1 and 50 µM dATP for 10 minutes and then adding 40 µM of

S60C Sml1 to a final volume of 100µl, after which the resulting mixture was incubated for an additional 10 minutes on ice. A similar procedure was used to purify GFP-Sml1

ScRR1 dimer and hexamer complexes. Similarly for the ATP-induced hexamer C14S

S60C Sml1 complex we used 3 mM ATP instead of 50 µM dATP in the reaction mixture while keeping the order of addition of components and the incubation times are the same.

The resulting complexes in aliquots of 100 µl were injected onto the column and eluted in equilibration buffer at a flow rate of 0.5 ml/min. Eluting complexes were monitored at

280 nm and 344 nm in the case of dATP and at 290 nm and 344 nm in the case of ATP.

Molecular weights of the eluting oligomeric species were determined using a standard curve based on molecular weight standards consisting of ribonuclease A (13 kDa) myoglobin (17 kDa), ovalbumin (44 kDa), γ-globulin (158 kDa), conalbumin (75kDa), and thyroglobulin (670 kDa).

2.5 Crystallization Techniques

Crystallization of the ScRR1 Hexamer

ScRR1 hexamers were crystallized by the hanging drop vapor diffusion method at room temperature. To prepare ScRR1●dATP hexamers for crystallization, the SEC- purified hexamer fractions were concentrated to 15 mg/ml in the presence of 20 µM dATP, 50 mM ammonium acetate pH 7.0, 5% glycerol, 5 mM MgCl2, 100 mM KCl, and

5 mM DTT. The well buffer contained 5% (v/v) tacsimate pH 7.0, 0.1 M HEPES pH 7.0,

10% PEG monomethyl ether 5000. Hanging drops contained 1 µl well buffer and 1 µl protein solution.

2.6 Protein Foot Printing

2.6.1 Selection of Buffers

The ideal buffer for protein foot printing experiments was determined using rate of Alexa 488 fluorophore oxidation upon exposure to the synchrotron radiation at

Brookhaven National Laboratory (BNL) [237]. These experiments were performed in collaboration with Dr. Mark Chance and Dr. Sayan Gupta. Buffers exposed to synchrotron radiation are listed in section 5.2.1 (See p, 146). The 10 mM Na-cacodylate buffer at pH 7.0 with 5mM MgCl2 0.1 M KCl and 100 µM TTP showed the highest degree of Alexa 488 modification and was used for further radiolytic experiments [238].

The ScRR1 and ScRR1-Sml1 complex were purified in the above buffer using

Superdex S200 10/300 Size Exclusion Chromatography (SEC) as described in section

2.4. The radiolytic experiments were performed at BNL by Dr. Sayan Gupta. Aliquots of

10µM solutions of ScRR1●TTP and the ScRR1●TTP●Sml1 complex were irradiated with the white synchrotron X-Ray beam in duplicates at 0, 8, 15, and 30 ms time intervals. After the reaction, 10 mM Met-NH2 was added to the samples to quench the reaction, which were frozen immediately.

2.6.2 Proteolysis & Mass Spectrometric Analysis

Radiolyzed protein samples were alkylated and digested with sequencing grade modified trypsin at an enzyme to protein ratio of 1:20 (wt/wt) at 37 0C for 12 hours.

Reactions were terminated by adding 0.1 % acetic acid to each sample. The peptide mixtures were then desalted by C18 ZiTips and the resulting elutes were dried using a centrifugal vacuum concentrator (SpeedVac) [239] .

Mass Spectrometric analysis was performed by Dr. Serguei Ilchenko at the Center for Proteomics and Bioinformatics at Case Western Reserve University. Prior to LC-

MS/MS analysis samples were reconstituted with 0.1% TFA. Samples were then injected onto a trapping column (C18, PepMap100, 300 µm X 5mm, 5µm particle size, Dionex,

Germany) followed by reverse phase chromatography on a column (C18, 75 µm X

150mm, 3µm, 100 Å; Dionex) using the nano-LC system (Ultimate 3000, Dionex,

Sunnyvale, CA). Peptides were eluted using mobile phases A (95% water, 5% acetonitrile, 0.1% TFA) and B (5% water, 95% acetonitrile, and 0.1% TFA). Peptides eluting from the reverse phase column with were directed to the high resolution Fourier

80 transform ion cyclotron resonance mass spectrometer (LTQ-FTICRMS) operated in positive ion mode. The peptides were infused at a flow rate of 300 nL/min via the silica non-coated PicoTip emitter at a voltage of 2.4 kV. Full MS spectra were recorded in the

FTICR and the tandem mass spectra of the 8 most intense ions were recorded by the LTQ ion trap at normalized collision energy of 35 eV [240].

2.6.3 Identification of the Sites of Modification

Tandem mass spectra (MS/MS spectra) of tryptic peptides were compared to an in silico peptide library database calculated from ScRR1 and Sml1 in the Mascot Program

(Matrix Science Ltd., UK) to identify the sites and types of oxidative modifications. Each

MS/MS spectrum was also interpreted manually with the aid of the Protein Prospector program (University of California, San Francisco, CA).

2.6.4 Calculation of Peptide Modification Rates

We extracted the ion chromatogram for each modified peptide, and calculated the area under the ion signals for both oxidized and unoxidized peptides at individual time points (0, 8, 15, 30 ms). The unmodified fraction of each peptide was calculated from the ratio of the area under the ion signal for the unoxidized peptide to the sum of the area under ion signal for the unoxidized peptide and its oxidized product. The area under the peak corresponding to background modifications seen for methionine containing peptides in the unexposed samples was subtracted from the area of total modification. The fraction unmodified for each individual peptide at each time point was normalized to the fraction

81 at 0 ms. Dose–response curves were calculated for each peptide by plotting the fraction unmodified as a function of exposure time. To determine the rate of oxidation of any given peptide, data for the fraction-unmodified peptide at different exposure times was

-kt fitted to the equation Y=Y0 e using Origin 8.0 software. Y and Y0 are the fraction unmodified at a time t and at 0 (ms), respectively, and k is the first order rate constant

[241].

Solvent accessibility of the side chains of ScRR1 was calculated using the

VADAR computer program (PENCE, University of Alberta, Canada. The TTP-GDP bound ScRR1 structure was used for this purpose.

2.7 Chemical Cross-Linking

A combination of zero length cross linkers, EDC and sulfo-NHS were used to cross link [242] Sml1 and ScRR1. To initiate the cross-linking reaction, a 25µM solution of C14S Sml1 was incubated for 15 min at room temperature with 2.5mM EDC and sulfo-NHS in 50 mM MES, 5mM MgCl2 and 5% glycerol containing buffer at pH 6.5.

The DTT free ScRR1 solution of 10 µM in 50mM HEEPS 5mM MgCl2 5% glycerol,

0.1M KCl and 100 µM TTP at pH 7.0 was added directly to the activated Sml1. Excess cross-linking reagents were removed by spin desalting of the activated Sml1 before addition of ScRR1 in order to minimize EDC and sulfo-NHS reactions with the carboxylic groups of ScRR1 that would lead to cross linking of ScRR1 itself. Aliquots were taken at every 30 minute intervals and the cross-linking reactions were quenched by adding 10mM DTT solution. Unreacted activated carboxylic acid residues were

82 regenerated by adding 1M KOH to a final concentration of 50 mM. The cross-linked products were analyzed using 4-20% SDS-PAGE, and the bands corresponding to the predicted molecular weight of the cross-linked ScRR1●Sml1 complex (ScRR1●Sml1X) were subjected to in-gel digestion with TPCK modified trypsin. Control reactions were carried out with the cross-linker added only to ScRR1 (ScRR1X) and Sml1 (Sml1X) separately. Subsequent analysis was performed as described in the section on proteolysis and MS analysis. Identification of the cross-linked was done using Mass Matrix software

[243].

2.8 Fluorescence Spectroscopy

Fluorescence experiments were performed with a PTI fluorescence spectrometer

(Photon Technology International, Birmingham, NJ). Initially, the excitation and emission spectra of labeled C14S S60C-Sml1 were determined at 344 nm and 444 nm respectively. The fluorescence anisotropy (FA) and fluorescence quenching (FQ) experiments were performed at room temperature. When FA experiments were performed, 1 ml solution of 40 nM C14S S60C-Sml1 (50mM HEPES, 5% Glycerol,

5mM MgCl2, 5mM DTT, 0.1M KCl and 100 µM TTP) was placed in the fluorescence cuvette and excited at 344 nm to measure the FA over 2 minutes. Then the solution was titrated with increasing concentrations of ScRR1 (40 to 3000 nM). After each addition the solution was left to equilibratione for 5 minutes and the FA was measured. The FA was plotted against the ScRR1 concentration and a simple hyperbola was fitted to the data

83 using Graphpad Prism. The FQ experiments were used to determine the Kd for peptides.

As for the FA experiments, 1 ml solution of 40 nM C14S/S60C-Sml1 was titrated with the peptide corresponding to the binding epitope and the fluorescence quenching was monitored at 444 nm.

Chapter 3: Nucleotide Induced Oligomerization of Human and Yeast

Ribonucleotide Reductase

The work presented in Chapter 3 was published in,

Nat Struct Mol Biol. 2011 Mar; 18 (3):316-22

3.1 Introduction

An intricate allosteric mechanism regulates RNR through the S-site and A-site.

Specificity site regulation maintains a balanced pool of dNTPs through an elegant cross talk mechanism while activity site regulation determines the overall dNTP pool size. Of these allosteric mechanisms, specificity site regulation in Class I and Class II has been extensively studied [31, 63, 108]. The substrate selection rules of specificity site regulation have been independently confirmed by both biochemical and structural studies. The least understood mechanism is that of A-site regulation. Recent reports suggest that both dATP and ATP regulate RNR activity by altering its oligomeric state in an ATP/dATP concentration-dependent manner. It is widely accepted that under physiological conditions eukaryotic RNRs form an α6β2 octameric complex which is either active or inactive depending on whether ATP or dATP is bound at the ATP cone

[41]. The only structure of the holoenzyme currently known is that of the α2β2 holo complex from the bacterium Salmonella typhimurium [244]. However, this structure shows an asymmetric complex, in which the C-terminus of only one of the two RNR2 subunits makes the expected interactions with RNR1, suggesting that it might not represent a catalytically active form. 85

However, the structural basis of dATP and ATP induced oligomerization remains to be elucidated. In this chapter, I will describe our findings on the structural basis of dATP induced hexamerization of the large subunit of human RNR and how it differs from the ATP-induced hexamer. I will also describe the structural characterization of the first ever dATP –induced octameric holoenzyme complex.

3.2 Results

3.2.1 Purification of HRRM1 and HRRM2

The proteins, HuRRM1 (i.e. wild type and mutant) and HuRRM2 were expressed and purified in E. coli cells, as described in the section 2.1.7. The purity of each subunit was assessed by SDS-PAGE and in both cases the proteins was more than 95% pure

(Figure 3.1 p, 87 & Figure 3.2 p, 88). The specific activities of HuRRM1 and HuRRM2 were tested in vitro using [3H]-CDP and [14C]-ADP (Section 2.3.4 p, 72) (Table 3.1 p,

89).

Figure 3.1 Purification of HuRRM1 Subunit

(A) Wild type HuRRM1 was purified using peptide affinity chromatography based on human RNR2 peptide (NSFTLDADF). Bound proteins were eluted using a 2M KCl gradient (green line). A similar procedure was used to purify the mutants (i.e D16R, H2E and D57N) .Protein absorbance (blue line) was measured at 280nm and the fractions (

) were collected and concentrated. (B) The purity of the concentrated fractions was tested using 12% SDS-PAGE and the protein identity was confirmed by comparing it with molecular weight standards.

Figure 3.2 Purification of HuRRM2 Subunit

(A) HuRRM2 was purified using Cobalt affinity chromatography. Bound proteins were eluted using 100 mM Imidazole in washing buffer. Protein absorbance (blue line) was measured at 280nm and the fractions ( ) were collected and concentrated. (B) The purity of the concentrated fractions was tested using 12% SDS-PAGE and the protein identity was confirmed by comparing it with known molecular weight standards.

Table 3.1 Specific Activities of Wild-type and Mutant HuRR

Specific Activity (nmol-1 min-1 mg-1) Subunit [dATP] µM CDP Reductase ADP Reductase HuRRM2 1,101.70 648.6 HuRRM1 0 440.1 244.1 20 104.5 40.3 D16R 0 242.1 163.5 20 253.9 167.4 H2E 0 246.4 136.4 20 240.4 125.4

3.2.2 Size Exclusion Chromatography of ScRR1 and HuRRM1

To study the oligomeric states of hRRM1 and ScRR1 in the presence of varying dATP /ATP concentrations, Superdex S200 size exclusion chromatography was used

(Section 2.4.1 p, 76). The void volume of the column was determined to be7.9 ml and was determined using Blue dextran. Biorad molecular weight standards ranging from 10 kDa to 670 kDa were used to calibrate the column. The elution profile of the molecular weight standards is shown in (Figure 3.3 p, 91). According to the profile, thyroglobulin with a molecular weight of 670 kDa was eluted at 8.9 ml while vitamin B12 with a molecular weight of 1.3 kDa was eluted at 20.4 ml. Other molecular weight standards eluted according to their expected molecular weights and their elution volumes were distributed between thyroglobulin and vitamin B12.

Figure 3.3 Standard Curve for the Determination of Molar Masses (Mr) of RNR

The black dots indicate the partition coefficients (Kav) obtained for the molecular weight markers as determined by a Superdex S200 10/300 GL size exclusion column. Kav was determined by the equation Kav = (Ve-Vo) / (Vt-Vo), where Ve = elution volume, Vo = void volume and Vt = total volume. HuRRM1 forms monomers (α) without dATP (blue dot). HuRRM1 forms dimers (α2) and hexamers (α6) with increasing concentrations of dATP.

3.2.3 dATP-Induced Oligomerization

Size exclusion chromatography (SEC) was used to investigate the dATP concentration-dependent oligomerization of wild type HuRRM1 and wild type ScRR1

(Figure 3.4 p, 93). Molecular weights of the eluting species were determined using calibration standards. In the absence of any effector molecules, HuRRM1 has an elution volume that corresponds to the molecular weight of a monomer (Figure 3.4A, blue line).

As dATP concentration increased to 5 µM, a reduction in monomer population with concomitant emergence of dimers and hexamers was observed. (Figure 3.4A, red line).

When dATP concentration increased to 20 µM, the concentration during S-phase, the elution profile was dominated by hexamers with fewer dimers and very few monomers present (Figure 3.4A, green line). Similarly at 20µM dATP ScRR1 also predominantly formed hexamers with very little dimers (Figure 3.4.B p, 93). These results have been independently corroborated with multi-angle light scattering (MALS) experiments

(Appendix 1A p, 194). At 20 µM dATP, MALS shows peaks at 521 kDa and 185 kDa corresponding to hRRM1 hexamers and dimers, respectively.

When the first 74 residues of HuRRM1 that belong to the ATP-binding cone are deleted, dATP no longer induces the formation of hexamers. Instead, in the presence of

HuRRM2, dATP induces the formation of an α2β2 holo RNR complex. This result further illustrates the importance of the A-site for oligomerization (Appendix 2A p, 196).

Figure 3.4 SEC Analysis of HuRRM1 and ScRR1 Oligomers

(A). HuRRM1 forms monomers without dATP (blue trace) and a mixed population of monomers (α), dimers (α2) and hexamers (α6) at 5 µM dATP (red trace). At 20 µM dATP, hexamers are the dominant species with a small amount of dimer (green trace).

(B) ScRR1 forms hexamers in the presence of 20 µM dATP. 12% SDS-PAGE was used to determine the identity of the peak fractions (inset).

3.2.4 The Effect of Subunit Oligomerization on Enzyme Activity

The activity of nucleotide-induced oligomers was assessed by in vitro assays measuring the product formation of either [14C]-dADP or [3H]-dCDP in the presence of increasing concentrations of dATP (Section 2.3.5 p, 73). The specific activities of both

HuRRM1 and ScRR1 in the presence of dATP are listed in the Table 3.1. The specific activity of both HuRRM1 and ScRR1 decreased gradually with increasing concentrations of dATP. At 20 µM dATP, when HuRRM1 exists mainly as hexamers, its specific activity was only ~20% of the specific activity in the absence of dATP (Figure 3.5 A p,

96). A similar trend was observed with the specific activity of ScRR1 in the presence of

20 µM dATP (Figure 3.5 B p, 96).

Figure 3.5 Specific Activity Measurement of HuRRM1 and ScRR1 in the Presence of dATP

(A) The specific activity of HuRRM1 was measured at increasing concentrations of dATP using [3H]-CDP (blue) and [14C]-ADP (red). Note that the specific activity declined gradually with increasing concentration of dATP. (B) The specific activity of

ScRR1 in the presence of 20 µM dATP showed a lesser degree inhibition with dATP.

3.2.5 The Structural Basis of RNROligomerization

To obtain insight into the structural basis of RNR1 oligomerization, ScRR1 hexamer was purified using Superdex S200 10/300 GL SEC and concentrated to

10mg/ml (Section 2.4.2 p, 76 & Section 2.5 p, 79). ScRR1 crystalized in a hexagonal space group (P63), and diffracted to 6.6 Å resolution (Figure 3.6A p, 98). A low resolution crystal structure of the ScRR1●dATP hexamer was determined by Dr. James

Fairman using the HuRRM1●dATP●TTP structure as the search model for molecular replacement. Although the structure of the ScRR1 hexamer at 6.6 Å doesn’t provide the details at atomic resolution, it is useful for understanding the packing arrangement of the

ScRR1 hexamer in the crystals.

The packing of ScRR1 hexamer in the crystal accommodated two packing models, referred to here as A and B. In both models, the ScRR1 hexamer packs as a trimer of dimers. The three dimers in each hexamer are related to each other by a three- fold axis (Figure 3.7 p, 99). However, there are important differences between these models. The packing arrangement of subunits in model B accommodates a relative large central pore, while the pore size is smaller in model A. In addition, models A and B also differ in the formation of the hexamer interface. In model A, only three dATP-bound

ATP-binding cones form the hexamer interface. In model B, all six dATP-bound ATP- binding cones form the hexamer interface (Figure 3.7 p, 99). Within the RNR1 dimer itself, the two dATP-bound ATP—binding cones contact each other in an antiparallel fashion and are related to each other by 2-fold symmetry.

Figure 3.6 A Photograph of ScRR1●dATP Hexamer Crystals

The hexagonal symmetry of the crystals was evident when the crystals were viewed under the microscope. Crystals were typically 200 µm in diameter in their longest dimension

Figure 3.7 Hexameric RNR1 based on the Low Resolution X-Ray Crystal Structure of the ScRR1 Hexamer

The RNR hexamer can take two possible packing arrangements namely model A (Right) and B (Left). ScRR1 monomers are dark green and light green or blue and cyan or dark brown and light brown. The ATP binding cones are red. This structure was solved by Dr.

James Fairman.

3.2.6 Validation of The dATP-Induced Hexamer by Site-Directed Mutagenesis

The hexamer interfaces for both models were inspected for interactions that could be targeted by mutagenesis to disrupt hexamerization (Table 2.1 p, 54). D16R and H2E mutations of HuRRM1 were designed based on the interface as seen in model B (Figure

3.8, p 101). Packing interactions in model A are poor compared to model B, and only a single site-directed mutation, D182R, was chosen to test this model. Initial attempts to express D16R and H2E in BL21 (DE3) RIL cells resulted in inclusion body formation.

Hence, the D16R mutant was expressed in Arctic BL21 (DE3) cells with IPTG induction and the H2E mutant was expressed in Rosetta BL21 (DE3) cells with auto-induction

(Section 2.1.7 p, 63) (Figure 3.9A p, 102). Purified proteins were enzymatically active.

The D16R and H2E mutant proteins retained 55% and 56% of the wild type activity for

CDP reduction and 67% and 56% for ADP reduction, respectively (Figure 3.9B p, 102)

(Table 3.1 p, 89).

100

FIGURE 3.8 Ribbon Diagram of the Hexamer Interface of HuRRM1 in Model B

His 2 D16

Phe 15

Arg 12 Val 13 Met 14 Met 14 Val 13 Phe 15 Arg12

D16 His 2

Two ATP cones that constitute the hexamer interface are colored in magenta and green.

Potential side chains that may be present in the interface are labeled. To test the model B, we selected the H2E and D16R mutants.

101

FIGURE 3.9 Purification of the HuRR1 Mutants andTheir Specific Activities

(A) Expression of HuRRM1 mutant H2E was not possible with 1mM IPTG. Its expression requires auto induction ( ) at 150C using minimal media prepared with lactose. Similarly, expression of D16R was achieved with Arctic BL21 (DE3) cells at 90C

(B) The specific activities of the H2E and D16R mutants were reduced ~40-55% when compared to wild type HuRRM1. The activity assay was performed using [3H]-CDP and

[14C]-ADP as substrates.

102

Using SEC, each mutant was tested for its ability to form hexamers 20 µM dATP, the concentration present during the S-phase of the cell cycle (Section 2.4.1 p, 76)

(Figure 3.10A p, 104). As previously observed, most of the wild type protein formed hexamers at this dATP concentration (Figure 3.10A, blue trace). The H2E mutation resulted in a shift of the equilibrium from mainly hexamer towards a mix with more dimer and less hexamer, while the D16R mutation completely disrupted hexamer formation (Figure 3.10A, red and green traces). The molecular weights of the oligomers formed by the D16R mutant protein were independently derived from MALS to be 190 kDa and 88 kDa, corresponding to a dimer and monomer, respectively

(Appendix 3A p, 197). To further confirm that Asp 16 was involved in hexamerization, the D16R mutation was also examined in ScRR1. As expected, ScRR1 D16R mutant protein also failed to form dATP-induced hexamers, showing that the same mechanism must underlie dATP-induced hexamerization of ScRR1 and hRRM1. Interestingly, the aforementioned D57N mutant that is not inhibited by dATP also forms dimers but not hexamers at physiological concentrations (Figure 3.10B p, 104). The D182R mutation at the hexamer interface of model A in ScRR1 did not disrupt dATP-induced hexamerization, and the mutant protein had 61% of wild type activity and was inhibited by physiological concentrations of dATP.

103

Figure 3.10 SEC Analysis of HuRRM1 Mutants

(A) Wild type HuRRM1 and H2E (orange line) and D16R (purple line) mutants were tested for their ability to form hexamers at 20 µM dATP using SEC. D16R only forms dimers while H2E showed a shift in equilibrium towards dimer formation. (B) The D57N mutant is insensitive to dATP inhibition. The mutant only formed dimmers in the presence of 20 µM dATP

104

Since the D16R mutation abolished the ability of hRRM1 to form dATP-induced hexamers, we hypothesized that like D57N, D16R would also prevent allosteric inhibition of HuRRM1 by dATP at physiological concentrations. To test this hypothesis, purified wild type, D16R, and H2E proteins were subjected to in vitro activity assays in the presence and absence of 20 µM dATP. Interestingly, the addition of 20 µM dATP to the reaction did not further reduce the activity of the D16R mutant, and the protein retained a statistically similar specific activity in the absence of dATP (Figure 3.11 A).

To ensure the D16R HuRRM1 mutant was folded, the protein was examined by circular dichroism, which showed a strong α-helical signal. The D16R and D57N mutants are not inhibited or hexamerized (3.10 B p, 104) at physiologically relevant concentrations of dATP. Both SEC and in vitro activity assays provide experimental evidence that hexameric RNR1 in solution packs as seen in model B (Figure 3.7 p, 99). Site-directed mutations designed to disrupt interactions seen in model A, did not interfere with either.

105

Figure 3.11 Activity of D16R in the Presence of dATP

D16R does not form hexamers in the presence of 20 µM dATP, nor was any significant difference of specific activity observed in the presence of 20 µM dATP.

106

3.2.7 ATP Hexamers Differ from dATP Hexamers

Since ATP and dATP bind the same site in the ATP binding cone, we investigated whether the ATP hexamer shares the same hexamer interface as the dATP hexamer. We investigated ATP-induced hexamer formation of HuRR1 in the presence of physiological concentrations of ATP. Wild type HuRRM1 formed a hexamer with a molecular weight of 580 kDa, slightly higher than the molecular weight of the corresponding dATP based hexamer. We used the same set of mutants (D16R, H2E and D57N), which disrupt the hexamer formation in the presence of dATP, to test whether ATP would induce them to form hexamers. Surprisingly, all of these mutants were able to form hexamers in the presence of ATP. This confirms that ATP –induced hexamers form a different interface than dATP-induced hexamers (Figure 3.12 p, 108).

107

Figure 3.12 The ATP hexamer Interface is Different from that of the dATP hexamer

SEC was used to assess the ability of wild type HuRRM1 (red trace) and D16R mutant

(blue trace) to form hexamers in the presence of 3mM ATP. Eluting species were monitored at 290nm. Both wild type and D16R mutants were able to form hexamers in the presence of ATP which is suggestive of a different interface for ATP-induced hexamer.

108

3.2.8 Purification and EM Analysis Of ScRR1●dATP Holo Complex

Computational modeling studies of the dATP hexamer and the small β subunit indicated that model A can accommodate 3 molecules of β2 while model B can only accommodate a single β2 subunit. To further examine how hexamer models A and B are or are not consistent with the α6β2 holo complex, the ScRR●dATP holo complex was isolated using SEC (Section 2.4.3 p, 77). The molecular weight (702kDa) derived from

SEC was consistent with an α6ββ’ complex (Figure 3.13 p, 110). The purified

ScRR●dATP holo complex, treated with a negative stain, was examined with Electron

Microscopy (EM) by Dr. Thomas Walz’s Group at Harvard University (Figure 3.14 A-C p, 111). EM analysis revealed the holo complex is consistent with an α6ββ’ arrangement and the ScRR2●ScRR4 hetero-dimer is clearly located inside the ring formed by the six

ScRR1●dATP subunits, which supports model B for the hexameric arrangement of

ScRR1.

109

Figure 3.13 Purification of the ScRR1●dATP Holo Complex

(A) The retention volume of the holo complex corresponds to a molecular weight of 703 kDa, which was consistent with an α6●ββ’ complex. Note the small amount of soluble aggregate in the void volume. (B) Analysis of the holo complex peak by 4-12% SDS-

PAGE. Lane 1: Precision Plus Protein Standards from Bio-Rad, lane 2-4: elution fractions of the holo complex revealing the presence of the α6●ββ’ complex.

110

Figure 3.14 Electron Micrograph of ScRR Holo Complex

(A) Raw image of the holocomplex in negative stain. Scale bar, 50 nm. Right panels, representative class averages. Side length of individual panels is 35 nm. (B) Different views of the 28-Å density map calculated using the random conical tilt approach with

50°/0° image pairs of cryonegatively stained holo complex. Scale bar, 5 nm. (C) Model of α6–ββ′–dATP holocomplex. Gray contour, crystal structure of RR1–dATP hexamer resolution-filtered to 28 Å. Golden contour, difference density obtained by subtracting density of RR1–dATP hexamer from EM density map of holocomplex. Green ribbon diagram, RR1–dATP hexamer; red ribbon diagram, yeast ββ′ heterodimer (PDB 1JKO)

42 docked into the difference peak. EM data collection and analysis was performed by

Dr. Zongli Li under the direction of Dr. Thomas Walz.

111

3.3 Discussion

Due to the important role RNR plays in balancing nucleotide pools, it is regulated by an intricate allosteric mechanism [32, 112]. In this study, we elucidated the structural basis of dATP mediated inhibition of RNR using a combination of structural, biochemical and biophysical techniques. Recent reports suggest that both dATP and ATP regulate RNR by altering its oligomeric state in an ATP/dATP concentration-dependent manner [41, 114]. SEC studies with hRRM1 and ScRR1 showed that the two proteins undergo oligomerization in response to increasing dATP concentration. At 20 µM dATP, both proteins predominantly form hexamers with very little dimer formation. The gradual decline of the specific activities of both hRRM1 and ScRR1 in the presence of increasing dATP concentrations further indicates that hexamer formation is a pre-requisite for enzyme inhibition (Figure 3.4 p, 93 & Figure 3.5 p, 96). Our SEC and enzyme activity data are consistent with the GEMMA findings of Rofugaran and co-workers who did not observe dATP induced tetramers but instead saw the dATP induced hexamers with low activity [41, 44].

These findings deviate from a previously reported study on mouse RNR. In the study reported by Kashlan and co-workers [44,114], the dATP-induced mRRM1 tetramer was the least active, and the hexamer was only observed at non-physiologically high dATP concentrations. This discrepancy may not be due to the dynamic light scattering results but rather to their interpretation. Since batch (i.e. un-fractionated) dynamic light scattering experiments (DLS) lack the capability to resolve radius differences that are less than 4-5 times of magnitude the reported result may well have been an average mass of all species in solution. In practice the radius difference observed between different

112 oligomeric states of a protein typically falls well below the resolution limit of batch DLS experiments and therefore, it is possible that the measured value of ~380 kDa in the presence of 10 – 100 µM dATP reported by Kashlan and co-workers [44, 114] represents a mixture of dimers and hexamers, rather than just tetramers as they suggested.

To better understand the molecular basis of dATP mediated inhibition of RNR,

ScRR1 dATP hexamer was crystallized and the structure was solved to 6.6 Å resolution by molecular replacement using HuRRM1●TTP●dATP structure (Figure 3.6 p, 98). The structure of HuRRM1●TTP●dATP was used as the search model because the ScRR1 hexamer was expected to be similar to the HuRRM1 hexamer as both proteins form inactive hexamers at 20 μM dATP (Figure 3.4 p, 93 & Figure 3.5 p, 96), and the structures of the dimers are similar. Our low resolution hexamer structure showed the very first time the arrangement of dimers within the hexamers.

The ScRR1 dATP hexamer structure shows two possible models, A and B, for the packing arrangement of ScRR1 hexamers in the crystal (Figure 3.7 p, 99). Since multiple crystal packing arrangements, as seen for example with EGFR [245], are reasonably common, both models were tested for their ability to form hexamers in the presence of dATP using site directed mutagenesis and SEC. The mutants D16R and D57N only formed dimers at physiological concentrations of dATP (Figure 3.10 A p, 104).

Furthermore, the D16R mutant was not inhibited at 20 µM dATP (Figure 3.11 p, 106).

These results confirm that the hexamerization is a prerequisite for HuRR inhibition.

As both ATP and dATP can induce hexamers, how can dATP be an allosteric inhibitor while ATP is an allosteric activator? Our SEC data on the mutants suggest that dATP and ATP form different types of HuRRM1 hexamers (Figure 3.12 p, 108). If the

113

ATP and dATP hexamers have different packing arrangements, as our data suggest, this may offer clues to why ATP and dATP exert opposite allosteric effects on RNR. Several groups have reported data on the quaternary structure of mammalian RNRs, including an

α6β2complex for ATP and dATP, as well as ATP-induced α6β6 holocomplexes. The structure of the yeast RNR–dATP holocomplex shows for the first time that only a dimer of the small subunit can be accommodated inside the hexamer pore, forming an

α6β2 complex, thus confirming the findings of GEMMA study [41]. The conformational changes accompanying dATP hexamerization may disrupt free-radical transfer from the small subunit to the active site. Indeed, although a higher-resolution structure will be needed to be certain, our low-resolution model of the yeast RNR holocomplex indicates that the small subunit (ScRR2●ScRR4) may bind farther away from ScRR1 compared with the S. typhymurium RNR holo complex ( Figure 13.4 p, 111) which is considered to be an intermediate of the active form. The packing of the RR–ATP holo complex, on the other hand, may only lead to conformational changes that promote the activation of RR.

On the basis of structure-function data, we present a model that accounts for the down regulation of RNR activity by dATP-induced oligomerization. Our SEC and MALS data show that there is a dynamic equilibrium between the α, α2 and α6 forms of

HuRRM1 and that the hexamer population increases with increasing dATP concentrations (Figure 3.4 p, 93). Without nucleotide effectors, RNR1 exists as an inactive monomer. Binding of the effectors ATP / dNTPs to the S site causes RNR1 to form dimers, which can then form α2β2 heterotetramers. Incomplete inhibition of the human enzyme at 50 µM dATP may be attributed to the α2β2 heterotetramers (Figure 3.5

A p, 96). Partial occupation of the A site by dATP may cause two α2 subunits to associate

114 into a transient tetramer intermediate that is not observable in our SEC and MALS experiments. This is presumably because once the S site is occupied by dNTP effectors and the α2 subunit is formed, higher-order oligomerization will occur by the association of pairs of RNR1 dimers rather than monomers binding dimers. Hence, transient tetramers may be formed by the association of two RNR1 dimers. The tetramers may be very unstable, and either fall apart again or immediately pick up an additional dimer to form a hexamer, making them difficult to observe. However, we cannot rule out the possibility that three dimers may associate to form a hexamer without the requirement of a tetramer as an intermediate step. RNR activity will also continue to diminish with the increase in dATP-induced hexamers in response to increased dATP levels (Figure 3.5 p,

96). Consistent with previous data and our EM structure, α6 can associate with β to form an α6β2 holocomplex in the presence of dATP (Figure 3.14 p, 111). Finally, when dATP levels become depleted during DNA replication or repair, dATP will dissociate from the

A site and active ATP-bound RR oligomers will be formed to replenish the dNTP supply.

115

Chapter 4: Kinetic Mechanism of ScRR Inhibition by Sml1

This work has been submitted for publication in Journal of Biological Chemistry

4.1 Introduction

As discussed in chapter 1, in S. cereviase, dNTP synthesis rates are regulated by a complex interplay between RR gene transcription, Sml1 inhibition, and dATP feedback inhibition. This 104-amino acid protein that contains a conserved Sml domain and a Rnr-

1 binding (R1B) domain. Sml domain is found in Dif1 and Ar122c proteins. R1B domain is also found in Aer122C, but it is less conserved in Spd1 [178, 186]. Even though interaction between Sml1 and ScRR1 has been studied in great detail, especially at the genetic level, its mechanism of inhibition of ScRR1 remains unclear. In this chapter, I describe the interaction between Sml1 and ScRR1 hexamer and also the possible kinetic mechanism of ScRR1inhibition by Sml1.

116

4.2 Results

4.2.1 Characterization of C14SS60C Sml1

We used C14S S60C labeled Sml1 to monitor interaction between ScRR1 and Sml1.

Labeling at S60 position was selected since it is neither located within the N-terminal dimerization domain nor the C-terminal RR binding domain (R1B) of Sml1 (Section 2.2 p, 65). The mutant C14S S60C Sml1 was labeled with Alexa 350 C5- malemide and purified using superdex S75 10/300 GL SEC (Figure 4.1 A p, 118). More than 90% of the C14S S60C Sml1 was conjugated with the fluorescent probe. Labeled protein has an absorption maximum at 340 nm and emission maximum at 444 nm as measured by fluorescence spectroscopy. ScRR1 has a specific activity of 340 nmol.min-1.mg-1 for CDP reduction, in the absence of C14SS60C Sml1. The specific activity of ScRR1 was reduced to 21 nmol.min-1.mg-1 when assayed with C14S S60C Sml1 (Figure 4.1 B p,

118).

117

Figure 4.1 Purification of C14S S60C Sml1

(A) Purification of C14S S60C Sml1 labeled with Alexa 350 C5 malemide using

Superdex G75 10 /300 SEC. 15% SDS-PAGE was used to determine the identity of the peak fractions (inset). (B) The C14SS60C Sml1 protein inhibited the ScRR activity (21 nmol. min-1.mg-1) like its wild type counterpart. The uninhibited ScRR1 enzyme has the specific activity of 340 nmol.min-1.mg-1.

118

4.2.2 Sml1 Binds to the ScRR1 Hexamer

The impact of Sml1 binding to ATP-induced ScRR1 hexamers and dATP-induced

ScRR1 hexamers was explored using size exclusion chromatography (SEC) (Section

2.4.6 p, 77). Both dATP and ATP induced the formation of hexamers (Figure 4.2 A & B p, 120). Molecular weights of the complexes were determined using known molecular weight standards (Figure 4.2 C p, 120). The size of the dATP hexamer is 630 kDa and is independent of protein concentration. The size of ATP hexamer is 594 kDa at a concentration of 2 µM ScRR1. Increasing the protein concentration (~ 5 fold) in the presence of 3 mM ATP results in the peak shifting to a higher molecular weight of 750 kDa (Figure 4.2 D p, 120). This is a concentration based aggregation phenomenon, which has been observed with the mouse RR1 ATP complex (Andreas Hofer, personal communication). Binding of Sml1 to pre-formed ATP/dATP based hexamers was investigated using Alexa350 labeled C14S/S60C Sml1 (Figure 4.1A p, 118) in the presence of 3mM ATP or 50 µM dATP. Binding of labeled Sml1 to ScRR1 was detected at 344 nm. Representative data provided in Figure 4.3 A & B showed the Sml1 was able to bind both ATP and dATP hexamers. ScRR1 ATP hexamer bound to Sml1 appears to elute at a higher molecular weight of 802 kDa while the Sml1 bound dATP hexamer elutes at a molecular weight of 660 kDa (Figure 4.3A &B p, 121).

To test whether Sml1 affects hexamer formation, a preformed dGTP-

ScRR1dimer-Sml1 complex was injected into a Superdex S200 column pre-equilibrated with 50 µM dATP. Only the formation of hexamer was observed, and significant dimer populations were not detected. These results further indicate that Sml1 does not interfere with hexamer formation (Figure 4.4 p, 122).

119

Figure 4.2 SEC Analysis of ATP and dATP-induced ScRR1 Hexamer

(A) ScRR1 ( i.e 2 µM) forms hexamers in the presence of 20 µM dATP (top) (B) ScRR1

( i.e 2 µM) forms hexamers in the presence of 3mMATP (bottom). (C) Standard curve used to measure the molecular weights of ATP-induced and dATP-induced ScRR1 hexamers. (D) Molecular weight of ATP-loaded ScRR1 hexamer is protein concentration dependent. Note the shift in molecular weight towards the void volume when the ScRR1 concentration is 10 µM. Red dotted line indicates the position of ATP-loaded hexamer at

2 µM ScRR1

120

Figure 4.3 SEC Analysis of ScRR1 Hexamer•Sml1 Interactions Using C14S S60C

Sml1

(A) Alexa 350 labeled Sml1 was used to monitor interactions with ScRR1 hexamer. Sml1 bound to the dATP ScRR1 hexamer was monitored at 344 nm (red trace). Complex formation was also monitored at 280 nm (blue trace). (B) Co-elution of Alexa 350 labeled Sml1 with ATP-induced hexamer. Complex formation was also monitored at 290 nm (blue trace), in order to reduce background by 3mM ATP.

121

Figure 4.4 Effect of Sml1 on Hexamer Formation

Sml1 has no effect on hexamer formation. Injection of preformed dGTP ScRR1 dimer–

Sml1 complex (blue line) to the gel filtration column pre-equilibrated with 50 µM dATP does not interfere with hexamer formation 122

4.2.3 Enzymatic Activity of ScRR1 with Sml1

The inhibitory effect of Sml1 on ScRR activity was measured in the presence of

3 mM ATP and [3H]-CDP (Section 2.3.6 p, 73). The specific activity of ScRR2 was 524 nmol/min/mg. The specific activity (SA) of [3H]-CDP reduction of ScRR1 enzyme in the presence of 3 mM ATP was ~331 nmol/min/mg (Table 4.1 p, 124). Addition of Sml1 resulted in almost complete inhibition of the enzyme (Figure 4.5 A p, 125). The inhibitory effect of Sml1 was also tested in the absence or presence of 20 µM dATP using [3H]-CDP. In the presence of 3 mM ATP and 20 µM dATP, the SA of ScRR1 was reduced by ~50% (Figure 4.5 B p, 125). This is because dATP is an allosteric inhibitor of ScRR1 [108]. The addition of Sml1 to the ATP/[3H]-CDP or ATP/dATP/[3H]-CDP containing reaction mixtures completely inhibited the ScRR activity (Figure 4.5 B p,

125). A similar trend was observed with dGTP/[14C]-ADP assay (Table 4.1 p, 124).

123

Table 4.1 Specific Activity of ScRR1 in the Presence of dATP and Sml1.

Specific Activity ( nmol min-1 mg-1) dATP [µM] Sml1 [ µM] CDP Reductase a ADP Reductase b S. cerevisiae ScRR2 0 0 524.32 ScRR1 0 0 331.6 222.2 20 0 169.83 91.2 20 4 13.44 3.8 0 4 26.8 5.4 a 3 mM ATP b 3 mM ATP 100 µM dGTP

124

Figure 4.5 Sml1 Inhibition of ATP and dATP –induced Holo Enzymes

(A) Percentage of specific activity of ScRR1 in the presence and absence of Sml1 with

ATP-[3H] CDP. (B) ScRR1 showed ~50% inhibition in the presence of 20 µM dATP.

Complete inhibition of ScRR in the presence of 20 µM dATP requires the presence of

Sml1.

125

4.2.4 Mode of Inhibition of ScRR1 by Sml1 is non-Linear

To better understand the mechanism of ScRR1 inhibition by Sml1, we investigated the mode of inhibition by measuring rates of product formation over a range of concentrations of NDP (i.e. [14C]-ADP or [3H]-CDP) substrates taken at four different concentrations of Sml1 (Section 2.3.6 p, 73). Double-reciprocal inhibition plots using

CDP as the substrate showed a series of parallel lines with no apparent point of intersection on either the y- or x-axes (Figure 4.6 A p, 127), suggestive of uncompetitive inhibition by Sml1. The slope (S) replot was linear while the intercept (I) replot was parabolic (Figure 4.6 A inset). The non-linear nature of the latter was also evident from the Dixon plot obtained by varying Sml1 concentrations (Figure 4.6 B p, 127). The double-reciprocal plot obtained by varying ADP at several fixed concentrations of Sml1 was indicative of a non-competitive mode of inhibition (Figure 4.7A p, 129). ADP reduction showed a similar trend to that of CDP reduction in the slope and the intercept replots (Figure 4.7A inset) and also in the Dixon plots (Figure 4.7 B p, 129).

126

Figure 4.6 Mode of Inhibition of ScRR1 by Sml1 in the Presence of dGTP and [3H]-

CDP

127

(A) Double reciprocal plot for uncompetitive mode of inhibition of SCRR by Sml1 in the presence of dGTP [14C]-ADP. Sml1 monomer concentrations used in the experiment are

1, 2 and 3 µM. Slope and intercept replots are also shown to the right. Slope replot is linear while intercept replot is parabolic. (B) Dixon plot for [14C]-ADP reduction. Note the parabolic nature of the plot.

128

Figure 4.7 Mode of Inhibition of ScRR1 by Sml1 in the Presence of dGTP and [14C]-

ADP

129

(A) Double reciprocal plot for uncompetitive mode of inhibition of SCRR by Sml1 in the presence of dGTP [14C]-ADP. Sml1 monomer concentrations used in the experiment are

130

Different modes of inhibition by Sml1 in the presence of differing cognate effector-substrate pairs suggest that Sml1 can inhibit RNR via a mechanism that depends on the oligomeric state of the enzyme. The different modes of inhibition were represented by different models and further examined by fits of the data using non-linear least squares regression available via the function nls() in the statistical programming environment

R [236] (Section 2.3.8. p, 74) The models fitted included representations of competitive inhibition (Eqn 4.1), non-competitive inhibition (Eqn 4.2), uncompetitive inhibition (Eqn

4.3), mixed inhibition (Eqn 4.4), and S-linear I-parabolic non-competitive inhibition (Eqn

4.5) [246]. Estimates of the kinetic parameters of the fitted models are summarized in

Table 2. Goodness of fit penalized for the number of model parameters estimated was evaluated using the Akaike Information Criterion (AIC) [247]. The S-linear I-parabolic noncompetitive inhibition model had the lowest AIC for both substrates examined and was thus considered to be the best model of the data observed (Table 4.2 p, 133).

131

132

Table 4.2: Estimation of Kinetic Parameters for Different Models of ScRR1

Inhibition by Sml1. * Note Subunit Stoichiometry was not included in the Initial

Model in Eqn (5)

133

The S-linear I-parabolic model assumes that ATP hexamers are composed of three identical dimers and that the binding of Sml1 to each dimer occurs independently of the other dimers in the hexamer (Figure 4.8 A p, 136). However, interactions within dimers are permitted so that the second molecule of Sml1 can bind to the ScRR1 dimer with a different affinity (Kii) from the first (Ki). This model assumes the same binding constant

(Ki) for Sml1 binding to ScRR1 dimers with or without substrate and that the NDP substrate binding constant Km is identical across enzyme states (Figure 4.8 A&B p,

136). A previous study using sucrose gradient centrifugation showed that the binding stoichiometry between ScRR1 and Sml1 is 1:1 [248]. This is consistent with up to two molecules of Sml1 binding to the ScRR1 dimers as in model 5.

We next quantified the increase in per-site binding affinity by considering the equilibrium between 0 to 1 Sml1 molecules bound to an ScRR1 dimer versus 1 to 2 Sml1 molecules per bound ScRR1 dimer (Figure 4C p, 136). We rewrote the S-linear I- parabolic model in terms of per-site binding constants, defined in terms of per-site koff and kon rate constants as Kd = koff/kon. This introduces stoichiometric factors of 2 into the model. For example, an unoccupied dimer (i.e. no Sml1 bound) has two empty sites, which can bind a Sml1 molecule, but after the dimer binds one molecule of Sml1, it has one site from which it can lose a Sml1 as in Figure 4.8C & Eqn – (4.6). Finally, a saturated dimer has two Sml1 molecules that can dissociate. However, once one molecule of Sml1 dissociates, the ScRR1 dimer has only one site at which it can rebind a Sml1 molecule as indicated in Figure 4C and Eqn – (4.8).

134

This yields the following form of Eq. 4.5

135

Figure 4.8 Models Describing [3H] CDP and [14C] ADP Reduction

(A) Binding of Sml1 to the ScRR1 hexamer. Sml1 and substrate are indicated by (red dotted circle) and (purple dot), respectively. Ki and Kii indicate the dissociation constants of the first and second Sml1 molecule binding to the ScRR1 dimer. Note the second molecule of Sml1 binds (i.e.0.7 +0.35 µM) with higher affinity than the first one. (B)

Binding of Sml1 to the ScRR1 dimer. Sml1 and substrate are indicated in (red dotted circle) and (light blue dot). (C) Derivation of stoichiometric factor of 2. Equilibrium between unoccupied ScRR1 dimer (top) versus Sml1 and occupied ScRR1 dimer

(bottom) versus Sml1 is shown.

136

The advantage of this form over the form in Eqn. (4.5) is that under a null hypothesis of equal binding affinities for the first and second Sml1 bound to ScRR1 dimers, the value of Ki equals Kii. Fitting Eqn (4.8) to our data, the first molecule of Sml1 binds to its site in a hexamer (using ATP/CDP conditions) with a Ki of 8.7 + 2.1 µM whereas the second molecule of Sm1l then binds with a 12-fold higher affinity of Kii =

0.7 + 0.35 µM (Figure 4.8 A p, 136). A similar mechanism of positive cooperativity for

Sml1 binding to its site in dimers (under dGTP/ADP conditions) yields Ki = 1.6 + 0.5 µM and Kii = 0.2 + 0.1 µM, i.e. an 8-fold higher affinity for the second site (Figure 4.8 B p,

136). These results indicate that Sml1 binds to ScRR1 dimers in hexamers with less affinity compared to independent dimers, i.e., those not present as hexamers.

4.2.5 Identification of Sml1 like Proteins and Implication of Sml1 in Therapeutics

In order to identify Sml1 orthologs in other fungal species, we carried out a systematic search using the Fungal BLAST search engine. We identified Sml1 orthologs belonging to several Saccharomyces species as well as to distant relatives of S. cerevisiae like Candida glabrata, Kluyveromyces lactis, Ashbya gosypii, Lachancea thermotolerans and Zygosaccharomyces rouxii. The highest conservation of sequence identities was located at the Sml and Rnr-1-Binding (R1B) domains of these proteins. Most of the sequence similarities and identities are well conserved, especially among the

Saccharomyces species (Figure 4.9 A p, 139). Structural predictions using the Jpred3 server (http://www.compbio.dundee.ac.uk/www-jpred/) revealed that these proteins are intrinsically disordered and predominately consist of loosely folded helices. Our analysis

137 also revealed that Sml1 orthologs do not exist in many other fungal species and in eukaryotes including mammals.

However, ScRR1 has 67% sequence identity and 83% sequence similarity to the large subunit of human and mouse RNR. Therefore we tested the extent to which human

RNR can be inhibited by Sml1.We determined the IC50 value for Sml1 inhibition of human RNR (Section 2.3.7 p, 74). Sml1 inhibited human RNR with an IC50 value of 20

+ 2µM (Figure 4.9 B p, 139). It is thus reasonable to suspect that it may also inhibit RNR in other species that similarly do not contain an Sml1 ortholog.

138

Figure 4.9 Sml1 orthologs in fungi and Sml1 and human RNR interactions

139

(A) Multiple sequence alignment of Sml1 orthologs. Sml and R1B domains are indicated in blue and red boxes with dotted lines. Identical residues in the Sml and R1B domains are colored red (B) IC50 value for human RNR inhibition by Sml1 was measured at different concentrations of Sml1 monomer (0.3, 3, 10 and 15) µM.

140

4.3 Discussion

The genomic stability of organisms depends on high fidelity DNA replication and repair. The balanced dNTP pool maintained by RNR is crucial for both of these processes

[3]. Eukaryotes use multiple mechanisms to regulate RNR activity and some of these mechanisms are organism specific. In budding yeast, for example, one mechanism for regulating ScRR is inhibition by Sml1 [176]. While the inhibitory effects of Sml1 on

RNR are well documented, both the molecular interactions involved in the mode of binding and the enzymatic mechanism of inhibition remain unclear.

Previous reports by Chabes et al. showed that Sml1 binds to the ScRR1 dimer with 1:1 stoichiometry [248]. Recent studies suggest that the active and inactive forms of eukaryotic RNR1 exist as hexamers under physiological conditions [41, 108]. Despite this, binding of Sml1 to the ATP and dATP-induced hexamers has not been characterized to date. Using SEC, we demonstrated that Sml1 binds to both dATP and ATP-induced hexamers (Figure 4.3 A&B p, 121). Although Sml1 binds to each hexamer, Sml1 does not disrupt hexamer formation induced by dATP and ATP (Figure 4.4 p, 122). Thus, our data suggests that Sml1 regulates RNR activity not by altering the oligomeric state of the enzyme inside the cell, i.e., hexamer to dimer, but through a different mechanism.

We recently determined the structure of ScRR• dATP holo complex, and found it has the subunit composition α6β2 [108]. As revealed by GEMMA studies, mouse RNR

ATP holo enzyme also has a similar subunit composition [41]. Our enzyme inhibition studies showed that Sml1 is needed for the complete inhibition of both ATP and dATP holoenzymes (Figure 4.5 p, 125). Furthermore, the inability of the allosteric inhibitor, dATP, to abolish RNR activity is consistent with the relaxed feedback inhibition

141 proposed by Chabes et al. This further emphasizes the importance of Sml1 in regulating

ScRR activity during the cell cycle [112]. Sml1 levels fluctuate during the cell cycle and in particular are low during S-phase but increase during G1 and G2/M phase of the cell cycle [178]. Since the level of ATP is presumably relatively constant throughout the cell cycle, we can assume that during G1 and the beginning of S-phase, the ScRR ATP holocomplex is likely to exist as an α6(β2)n oligomer. [n=2, 4, 6]. Our enzyme inhibition data revealed that ScRR1 is fully inhibited in the presence of 3mM ATP and 4 µM Sml1.

Therefore, binding of Sml1 to the ATP bound holoenzyme prevents dNTP synthesis prior to the S-phase of the cell cycle. Sml1 is degraded in a Mec1 and Rad53 dependent manner as cells enter S-phase, to allow dNTP synthesis to proceed as needed for DNA replication [178]. At the beginning of S-phase, dATP accumulates to a setpoint to down regulate RNR and thus controls dNTP flux supply to equal its demand due to DNA replication. We speculate that the enzyme is converted into a dATP based holoenzyme complex. In a previous study we showed that ScRR1 hexamerizes at 20 µM dATP and exists as an α6–ββ′–dATP holo complex [108]. ScRR1 is 50% inhibited at this dATP concentration due to the aforementioned relaxed dATP feedback inhibition. Thus Sml1 levels increase at this point, as cells enter G2, to fully inhibit the α6–ββ′–dATP holocomplex (Figure 4.5 B p, 125), in effect functionally substituting for the KEN box found in human R2 but not in ScRR2• ScRR4.

Enzyme kinetic experiments were carried out to better understand the mechanism behind the inhibition by Sml1. Our results show that Sml1 exhibits an S-linear I– parabolic non-competitive mode of inhibition with both the ADP and CDP used as substrates (Figure 4.6 p, 127 & Figure 4.7 p, 129) (Table 4.2 , p 133). The non-

142 competitive nature of this non-linear inhibition mode indicates that Sml1 does not compete with the binding of CDP or ADP at the catalytic site. Instead, this mode of inhibition supports a model in which Sml1 interferes with the C-terminal tail of ScRR1 for active site redox state regeneration as suggested by the genetic studies carried out by

Zhang et.al. [88]. The parabolic nature of the inhibition arises due to the cooperative binding of two molecules of Sml1 to each of the three dimers of the ScRR1 hexamer

(Figure 4.6 p, 127 & Figure 4.7 p, 129). Enzyme systems with an S-linear I-parabolic non-competitive mode of inhibition have been previously described [249, 250]. For example, the hydrolysis of (S)-hydroxynitrile to benzaldehyde and HCN follows an S- linear I-parabolic type of inhibition[250]. The reaction has been shown to proceed via an ordered uni bi type of reaction scheme with HCN forming a dead-end complex with the enzyme [250]. We propose that ScRR1 follows a similar ordered uni bi reaction mechanism in which H2O is released upon CDP or ADP reduction. According to this mechanism, Sml1 inhibition of ScRR1 results in a dead-end complex formation. Binding of one Sml1 molecule to an RNR dimer positioned within the RNR hexamer effects positive cooperativity for a second Sml1 binding event (Figure 4.8 p, 136). Furthermore, we found that Sml1 binds to the ScRR1 dimer (when ATP is absent) with higher affinity than it does to the hexamer (when ATP is present). We propose that the difference in affinity of Sml1 to the ScRR1 dimer and hexamer is due to conformational differences.

Binding of Sml1 to ATP hexamers with less affinity may be physiologically important since Sml1 needs to be degraded upon cells entering into S-phase. Therefore, less tight binding of Sml1 to ATP hexamer may facilitate dissociation of Sml1 from ScRR1, when unbound Sml1 is ubiquitinated and degraded by proteasome.

143

It is interesting to note that several fungal species do not possess any Sml1 like proteins (Figure 4.9, p 139). Most notable among this list are Candida albicans,

Histoplasma capsulatum and Blastomyces dermatitidis, which are known to be opportunistic pathogens that can cause life-threatening infections in immunocompromised humans. It remains to be determined whether RNRs in these organisms can be inhibited by Sml1. If so it may be possible to develop Sml1-based compound(s) to combat infections caused by these organisms. Similarly, the ability of

Sml1 to inhibit human RNR (Figure 4.9B p, 139) may also aid the development of novel anticancer agents that may be used against cancers that were once responsive but have become resistant to gemcitabine, e.g. via mutations in deoxycytidine kinase.

144

Chapter 5: Structural and Biochemical Characterization of Sml1 ScRR1

Interactions

5.1 Introduction

In Chapter 4, I described the kinetic mechanism of ScRR1 inhibition by Sml1. In this chapter, I will be describing the experiments we carried out to map the binding site of

Sml1 on ScRR1. We used protein foot printing and chemical cross linking as two main structural methods to identify the binding site.

5.1 Results

5.1.1 Identification of Buffers for Foot Printing

Ideal buffer conditions to perform protein foot printing experiments were assessed by exposing buffers containing Alexa 488 fluoropore to synchrotron radiation (Section

2.6.1 p, 79). To characterize the ScRR1•TTP•Sml1 interaction, five different buffer conditions were chosen. (Figure 5.1 p, 146) Na cacodylate containing buffer solutions increased Alexa 488 oxidation, while the presence of 1% glycerol and higher concentrations of TTP (i.e 1mM) reduced it. The presence of TCEP in the buffer systems led to a slightly faster rate of fluoropore oxidation. Based on these results, the buffer F

(10 mM Na cacodylate 5mM MgCl2, 0.1 M KCl 1mM TCEP and 100 µM TTP) was used in subsequent experiments.

145

Figure 5.1 Identification of Buffers for Protein Foot Printing

-1 Buffer A (10mM HEPES 5 mM MgCl2 1mM TCEP 1mM TTP): k = 38.25 sec

Buffer B (10mM HEPES 1% glycerol (v/v) 5 mM MgCl2 1 mM TCEP 1 mM TTP): k =

34.99 sec-1

Buffer C (10mM Na cacodylate 1% glycerol (v/v) 5 mM MgCl2 1mM TCEP 1mM TTP): k = 51.83 sec-1

Buffer D (10mM Na cacodylate 1% glycerol (v/v) 5 mM MgCl2 1mM TTP): k = 50.42 sec-1

Buffer E (10mM HEPES 1% glycerol (v/v) 5 mM MgCl2 1mM TTP): k = 20.57 sec-1

-1 Buffer F (10mM Na cacodylate 5 mM MgCl2 1mM TCEP 100 µM TTP): k = 59.74 sec

146

Testing of buffers that do not contain glycerol was not attempted since the purification of these proteins resulted in aggregation. Please refer to the section 5.2.2.1.

5.2.2 Protein Foot Printing

5.2.2.1 Purification of ScRR1•TTP•Sml1 Complex for Protein Foot Printing

The ScRR1•Sml1 complexes were purified by size exclusion chromatography (SEC)

(Section 2.4.4 p, 77) (Figure 5.2 p, 148). In the presence of 100 µM TTP, ScRR1 formed a 210 kDa complex which corresponds to the dimer. Similarly, ScRR1-dimer was incubated with 4 fold molar excess of Sml1 and the resulting complex eluted at a molecular weight of 263kDa. The presence of both proteins in the complex was verified by 4-12% SDS-PAGE. The peak fraction of the protein complex was collected for X-ray exposure. Isolation of either ATP ScRR1 hexamer Sml1 complex or dATP SCRR1 hexamer Sml1 complex in the absence of glycerol by SEC was not possible due to protein aggregation.

147

Figure 5.2 Purification of ScRR1•TTP•Sml1 Complex

(A) Right, Purification of Sml1 bound to ScRR1 dimer. Elution buffer contained no glycerol. Samples for foot-printing were collected from the peak fraction and kept at 40C till X-ray exposure. Left, 4-20% SDS-PAGE gel of the SEC purified proteins in the peak fraction. Protein concentration of the peak fraction is 5 µM. Note the presence of both

ScRR1 and Sml1 in the complex.

148

5.2.2.2 Proteolytic Digestion of the X-ray Exposed Samples

Samples that had been exposed to synchrotron radiation were proteolyzed using

TPCK treated trypsin and analyzed by liquid chromatography coupled tandem mass spectrometry (LC-MS/MS) (Section 2.6.2 p, 80). Based on our LC-MS/MS analysis, we determined the sequence coverage of the proteins in the complex and in the ScRR1 dimer samples. As expected, highest sequence coverage was observed with samples that were not exposed (i.e. 0 ms) to X-ray (Figure 5.3 A p, 150). However, the sequence coverage of the proteins gradually decreased with increasing exposure times to around 52%

(Figure 5.3 B p, 150).

149

Figure 5.3 Sequence Coverage of X-ray Exposed ScRR1 upon Proteolysis

Samples exposed to synchrotron radiation were digested with trypsin and resulting peptides were desalted using C18 ZipTips. Peptide sequences identified by mass spectrometry are colored red. (A) Sequence coverage for ScRR1 at 0 ms was 72%. (B)

Sequence coverage is dramatically reduced to 52% upon exposure to X-rays for 30 ms.

Most of the sequence is missing in the β/α- barrel domain.

150

5.2.2.3 Protein Foot Printing Data Analysis

Protein foot printing experiments were carried out to identify the Sml1 binding surface of ScRR1. In foot printing experiments, hydroxyl radicals are allowed to react with the side chains of the solvent-exposed protein regions, whereas areas of protein not exposed to the solvent are protected against these modifications (Section 2.6.3 p, 81)

[241] . This difference allows the protein binding interfaces to be mapped with high resolution. Based on our LC-MS/MS analysis, we found 10 peptides of ScRR1 that contain oxidative modifications. The sites of oxidations within those peptides were confirmed by tandem mass spectrometry. Solvent accessible surface areas (ASA) were calculated for each amino acid side chain to identify which primary target residues were likely to be oxidized in the ScRR1 protein (Table 5.1 p, 152). Several regions are not visible in the crystal structure of the ScRR1 due to poor electron density; we could not calculate the ASA for these regions. The extent of side chain modification of the affected peptides was analyzed by LC-MS and dose-response curves were obtained by plotting the fraction remaining unmodified for each peptide as a function of exposure time. As such, first-order rate constants could be measured from this analysis for ScRR1 protein and

ScRR1-Sml1 complex, and these values are summarized in Table 5.1.

151

Table 5.1 Modification Rates of the ScRR1 Peptides in the Presence and Absence of

Sml1

Modification Rate (s-1) Degree ASA Of Region Modification site (Å2) R RS Protection 11 to 21 11EPVQFDKITA 194 7.0+0.7 8.3+0.9 0.8 81 to 96 81IAISNLHKQTTKQFSK 31.8 2.4+0.2 2.3+0.1 1.04 159 to 173 159INGQVAERPQHLIMR 0.6,0 3.4+ 0.4 5.5+0.3 0.62 469 to 475 469KLHEIAK 13.3 0.7+ 0.5 0.6+ 0.4 1.17 502 to 520 502HRPIALGVQGLADTFMLLR 9.6 5.1+0.2 7.5+0.7 0.68 694 to 702 694TIINMAADR 3.9 5.3+0.3 4.9+0.5 1.08 738 to 744 738TGMYYLR 0 3.1+0.4 2.4+0.6 1.29 814 to 827 814EASPAPTGSHGSLTK * 7.3+0.4 10.3+0.9 0.71 828 to 849 828GMAELNVQESKVEVPEVPAPTK * 34.6+2.1 31.1+3.2 1.11 854 to 871 854AAPIVDDEETEFDIYNSK * 14.9+2.1 22.2+1.6 0.67

Modifications within the peptide are shown in red. The degree of protection was defined as the ratio of a residue's modification rate in ScRR1 to its modification rate in the

ScRR1-Sml1 complex. ASA: Accessible surface Area. (*) denotes the residues that are not visible in the structure, and whose ASAs therefore cannot be calculated

152

A parameter known as degree of protection was defined as the ratio of modification rate of ScRR1 to ScRR1-Sml1 complex (Section 2.6.4 p, 81). Based on the degree of protection of any given peptide, two groups of regions were identified; peptides with values less than 1, and peptides with values greater than 1. However, we did not identify any regions of ScRR1 with significant protection (>4 fold) by Sml1. The regions that included residues 11-21, 81-96, 694-702 and 828-849 showed no changes in oxidation rate, within experimental error, within the experimental error, Sml1 binding

(Figure 5.4 A&B p, 154). Several other regions including peptides 469-475 and 738-744 of the ScRR1 sequence exhibited a moderate protection (between 1.17 and 1.29 folds).

These peptides are not clustered together to provide a true binding surface. Thus a modest protection may result from partial burying of these regions due to conformational changes that result from binding of Sml1. The four peptides comprised of sequences 159-173,

502-520, 814-827 and 854-871, which are distributed throughout the ScRR1 structure, exhibited a modest increase in the oxidation rate upon complex formation, which ranged from 0.62 to 0.71 fold. Two of these peptides 814-827 and 854-871 are confined to the C- terminal insert (CI) of ScRR1. This implies that the CI of the ScRR1 does not interact with Sml1.

153

Figure 5.4 Protein Foot-printing of ScRR1 and the Sml1•ScRR1 Complex

154

(A) The MS/MS spectra of the peptide, EPVQFDKITA of ScRR1. Both y and b type ions are shown. (B) Structure of ScRR1 (PDB ID 1ZYZ) showing the residues with oxidative modifications. Regions that exhibit mild to moderate protection and negative protections are colored in magenta and blue respectively. A modified residue (F15) in the ATP binding cone is shown red.

155

5.2.3 Chemical Cross-Linking and Mass Spectrometry.

Chemical cross-linking in combination with mass spectrometry was used as an independent method to map the interaction surface between ScRR1 and Sml1 (Section

2.7. p, 82). The zero-length cross-linker EDC in combination with sulfo-NHS was used as the cross-linker of choice (Figure 5.5 A p, 157) [251]. EDC reacts with carboxylic acid groups of the protein to form a short-lived active O-acylisourea intermediate and the

Sulfo-NHS was used to extend the half-life of the activated carboxylate [251]. EDC when used in combination with sulfo–NHS leads to the formation of an amide bond between carboxylic groups (i.e. aspartic acid / glutamic acid / C-terminus) and primary amine groups (lysine / the N-terminus). As illustrated in Figure 5.5 B, a cross-linked complex corresponding to a molecular weight of 112 kDa appeared as early as 30 minutes after exposure to these agents. The cross linking seems to be selective as we did not observe any higher molecular weight gel bands that might reflect protein aggregation caused by excessive cross-linking (Figure 5.5 B p, 157).

156

Figure 5.5 Chemical Cross-linking of ScRR1 and Sml1

157

(A) Activation of the carboxylic acid group of an amino acid by EDC results in an unstable amine reactive intermediate, which reacts with the primary amino group of a second amino acids to yield an amide cross-link. (B) Separation of cross-linked ScRR1-

Sml1 complex on the 4-20% SDS-PADE gel. Left, time dependent cross-linking of

ScRR1and Sml1, sampled over the course of 2 hours. Right, the gel shift of the cross- linked ScRR1•Sml1 complex (red arrow) is shown.

158

The bands corresponding to cross-linked ScRR1-Sml1 were excised and subjected to in gel digestion followed by liquid chromatography coupled tandem mass spectrometry

(LC-MS/MS). The identification of the cross-linked peptides was carried out using Mass

Matrix software. The peptides corresponding to both ScRR1 and Sml1 were first observed at the 30 minutes time point with MS experiments. However, the first cross- linked peptides of ScRR1 and Sml1 appeared at the 2 hour time point. The sequence coverage of ScRR1 and Sml1 were found to be 73% and 21 %, respectively. Two unique cross-linked fragments were observed between ScRR1 and Sml1 (Table 5.2 p, 160). In both cross-linked peptides, the C-terminal tail of Sml1 was cross-linked to a region in the

N-terminal ATP cone of the ScRR1. One inter-molecular product has a m/z value of

948.47 Da which corresponds to a peptide containing D79 of Sml1 cross-linked to K17 of the ScRR1. The other peptide has both intermolecular and intramolecular cross linking between glutamic acid and lysine residues. In this case, E89 of Sml1 is cross-linked with

K10 of ScRR1 while an intermolecular crosslink is observed between E101 and K98 of

Sml1. The identity of both peptides was confirmed by MS/MS experiments (Figure 5.6 p, 161). However, chemical cross linking data does not corroborate our findings from the protein foot printing data.

159

Table 5.2 Cross-linked ScRR1•Sml1 Peptides Identified by Mass Spectrometry

ScRR1 peptides are marked in red and Sml1 peptides are marked in blue. Cross linking sites are indicated with ($). ($1): intermolecular cross links, ($2): intramolecular cross links. The peptides with m/z values of 817.13 Da and 948.47 Da showed unambiguous and unique cross-linking patterns.

160

Figure 5.6 Mass Spectrometric Analysis of Cross-link Fragments

161

(A) Total ion chromatogram of the cross-linked peptide with m/z of 814.13. The m/z display of the (M+4H)4+ charge state of the cross-linked peptide and its corresponding isotopic distribution (Inset). (B) MS/MS fragmentation of the cross-linked peptide. The fragmentation pattern revealed two unique peptides, VQF and NQGK, which belonged to

ScRR1 and Sml1 respectively. The cross-linked sites are marked in red. R denotes

ScRR1 and S denotes Sml1.

162

5.2.4 Characterization of N-terminal Mutants

Based on our cross-linking data, we targeted the ScRR1 and Sml1 interface using site directed mutagenesis to disrupt complex formation. We generated two deletion mutations in the ATP cone of ScRR1 designated Δ22 and Δ90 in which the first 22 and

90 amino acids respectively were deleted, from the N-terminus. Of these two mutants, the purification of Δ90 ScRR1 yielded ~50 µg of protein for 8 liters of culture medium.

The ability of these mutants to form complexes with Sml1 was investigated using size exclusion chromatography (SEC). The mutant Δ22 did not form hexamers in the presence of either ATP or dATP. Both mutants formed dimers in the presence either TTP or dGTP.

Sml1 did not co-elute with either mutant when the corresponding fractions were analyzed by the 4-20% SDS-PAGE (Figure 5.7 p, 164).

163

Figure 5.7 Characterization of the N-terminal Mutant Δ22 ScRR1

(A) SEC analysis of Δ22 ScRR1 mutant binding to Sml1. (B) Fractions containing the putative complex were collected and analyzed by 4-20% SDS-PAGE. Note the absence of Sml1in lane 1 of the eluting fraction (C) C14S S60C Sml1 (red trace) did not co-elute with Δ20 ScRR1 dimer.

164

To quantify the interactions of these mutants with Sml1, we developed a novel fluorescence based assay to measure their binding affinities using fluorescence anisotropy

(FA). For this purpose, we generated a C14S S60C Sml1 double mutant and labeled the protein with Alexa 350 fluorophore as described in the method section 2.2. The initial anisotropy value of S60C-Sml1 increased upon addition of ScRR1 in the presence of 100

µM TTP until the reaction reached complete saturation. The data was best fitted to a one- site binding model (hyperbola) with a Kd of 0.081 + 0.02 µM. (Figure 5.8 A p, 166).

However, S60C-Sml1 does not bind with either the Δ22 or Δ90 ScRR1 mutants (Figure

5.8 A p, 166). These negative results support the notion that the Sml1 binding site is at the N terminus of ScRR1. To further validate our results, we synthesized a peptide called the 10-21 N-terminal epitope (KEPVQFDKITAR) that was shown to bind Sml1 by our cross-linking experiments. We tested its affinity to S60C-Sml1 using fluorescence quenching spectroscopy. This peptide binds C14S S60CSml1 with a Kd of 0.11 + 0.014

µM (Figure 5.8 B&C p, 166). A random peptide (KEINFDDF) was used as a negative control which did not bind to C14S S60C-Sml1.

165

Figure 5.8 Binding of ScRR1 or its Peptides to C14S S60C Sml1 using Fluorescence

Spectroscopy

(A) Binding affinities of wild-type and Δ22 mutant ScRR1 to C14S S60C Sml1 were measured using fluorescence anisotropy (FA). Kd for the wild type and Δ22 mutant

ScRR1 was calculated for a one site binding model using Graphpad Prism. (B)

Fluorescence quenching pattern of ScRR1 peptide (KEPVQFDKITAR) bound to C14S

S60C Sml1. Different concentrations of the peptides used are indicated. (C) The Kd of the

10-21 ScRR1 peptide was calculated for a one site binding model using Graphpad Prism.

(D) A random peptide does not show binding to ScRR1.

166

5.2.5 Peptide Inhibition Assay and Mutant Activity

Both deletion mutants (i.e Δ22 and Δ90) were remained inactive. Hence, we were not able to test the inhibitory effect of Sml1 on these deletion mutants. Therefore, we tested whether the N-terminal peptides corresponding to the putative Sml1 binding region of ScRR1 could overcome Sml1inhibition of ScRR1 using a standard activity assay.

However, even at 1,000-fold molar excess, the N-terminal peptide (KEPVQFDKITAR) did not significantly overcome the ScRR1 inhibition by Sml1 (Figure 5.9A p, 168). We also tested whether mutation of the lysines K10 and K17 in the N-terminal cone, implicated in binding by the cross-linking experiments, reduced the ability of Sml1 to inhibit the mutant proteins (Section 2.3.6, p 73). The degree of inhibition of K10E and

K17E mutants by Sml1 was 74% and 78% respectively, when [14C]-ADP was used as the substrate. When [3H]-CDP was used as a substrate, the degree of Sml1 inhibition was

69% and 81%, respectively. Since Sml1 was less able to inhibit ScRR1 that had had either lysine mutated, but still showed some degree of inhibition we tested whether the double mutant, i.e. K10E / K17E in region would be even less subject to Sml1 inhibition.

To our surprise we did not observe any additive effect of the double mutant (Figure 5.9

B p, 168).

167

Figure 5.9 Specific Activity Measurements with Mutants

168

(A) Specific activity was measured using a dGTP / [14C]-ADP reduction assay. The N- terminal peptide 10-21 was tested to determine the degree to which it could protect

ScRR1 from inhibition by Sml1. Peptide at 1,000-fold molar excess over Sml1 was not unable to relieve inhibition. The peptide itself reduced the specific activity of ScRR1 by

20%. (B) Percentage of specific activities of the single (K10E, K17E) and double (K10E

K17E) mutants in the presence of CDP (grey) and ADP (black).

169

5.2.6 Determining the cryo-EM Structure of ScRR1 Hexamer• Sml1 Complex

Chemical cross-linking and protein foot printing experiments described in the sections 5.2.2 and 5.2.3 were carried out under conditions that promote Sml1 binding to

ScRR1 dimer. Inability to purify Sml1 bound dATP and ATP-induced hexamers in the absence of glycerol led us to explore the possibility of using cryo-EM to map the binding site. Hence we isolated the ScRR1 hexamer Sml1 complex using SEC (Section 2.4.5 p,

78) (Figure 5.10A p, 172). The complex eluted at a molecular weight of 660kDa and the presences of both proteins was confirmed using 12% SDS-PAGE. The peak fraction corresponding to ScRR1 hexamer Sml1 complex was processed for cryo-EM analysis.

Initial cryo-EM analysis performed by Dr. Thomas Walz group at Harvard University has revealed that Sml1 may bind around the α/β barrel domain of the ScRR1 (Appendix 4A p, 198). Since the resolution of this map is around 13 Å, no further attempts were made to identify the interactions between the main chain atoms of ScRR1 and Sml1 at this resolution.

In order to precisely identify the binding site of Sml1 in the dATP-induced hexamer, an N-terminally 6X His tagged green fluorescent protein (GFP) was added to

Sml1to make a fusion protein with a molecular weight of 39 kDa. The expression of the

GFP-Sml1in BL21 (DE3) plysS cells resulted in inactive protein. Therefore GFP-Sml1 protein was expressed in Arctic BL21 (DE3) cells with auto induction. The GFP-Sml1 fusion protein was purified using cobalt affinity chromatography and was able to inhibit the activity of ScRR1 dimer, but not hexamer, like wild type Sml. GFP-Sml1 was further purified with SEC (Figure 5.10B-D p, 172). The purified GFP-Sml1 was then tested for its ability to interact with ScRR1 dimer and hexamer using SEC, and monitoring the

170 eluate at 280 nm and 488 nm. GFP-Sml1 was only observed to co-elute with the ScRR1 dimer (Figure 5.11A-C p, 173). Analysis of GFP-Sml1-ScRR1 dimer complex by cryo-

EM has revealed that only a fraction of ScRR1 dimers carrying the fusion protein

(Appendix 5A p, 199).

171

Figure 5.10 Purification of dATP-induced ScRR1 Hexamer•Sml1 Complex and GFP

Sml1

(A) dATP-induced ScRR1 hexamer-Sml1 complex was purified using SEC. The complex eluted at a molecular weight of 672 kDa. Presence of both ScRR1 and Sml1 was verified using 4-12% SDS-PAGE (inset).The peak fraction was processed for cryo-EM without freezing. (B) Purification of 6X-His tagged GFP-Sml1 expressed in ArcticExpress BL21

(DE3) cells with auto-induction. Initial purification was done with cobalt affinity chromatography and the fractions were concentrated and re-purified using Superdex 75

10/300 GL SEC. (C) Fractions (1) and (3) were analyzed by 4-12% SDS-PAGE (D)

Specific activities of fractions (1) and (3) were tested using a [14C]-ADP reduction assay in the presence of dGTP. Only fraction (3) inhibited ScRR1.

172

Figure 5.11 Characterization of GFP-Sml1 and ScRR1-dimer Interactions

(A) GFP-Sml1•ScRR1 dimer was purified using S200 SEC. Eluting species were simultaneously monitored at 280 nm and 488 nm. Note the co elution of the GFP-Sml1 and ScRR1 dimer. (B) 4-12% SDS-PAGE analysis of protein fractions eluted in SEC.

Note the presence of both GFP-Sml1 and ScRR1 in fraction-4. (C) Cryo-EM analysis revealed the presence of GFP-Sml1 bound ScRR1. Red arrow indicates the density for

GFP-Sml1 bound ScRR1. Cryo-EM analysis was performed by Dr. Zongli Li at Harvard

University.

173

5.3 Discussion

The genomic stability of organisms depends on high fidelity DNA replication and repair. Both of these processes require maintenance of a balanced dNTP pool, which is mainly regulated by RNR. Eukaryotes have evolved multiple mechanisms to regulate

RNR activity and some of these mechanisms are organism specific. In budding yeast, one mechanism for regulating the dNTP pool is through the inhibition of ScRR by Sml1 [112,

176]. While the inhibitory effects of Sml1 on RNR are well documented, the molecular interactions involved in the mode of binding remain unclear. In this report, we tried to characterize the binding interface between Sml1 and ScRR1 using a wide variety of biochemical and biophysical methods.

Our attempts to co-crystalize ScRR1-Sml 1 complex was unsuccessful. Therefore, we explored the use of mass spectrometry to map out the binding surface of Sml1 on

ScRR1. Although a common method used in such studies is chemical cross linking with tandem MS, it is prone to result in non-specific cross linking [251]. A more reliable method is the protein foot printing combined with tandem mass spectrometry [241].

Exposure of protein samples to white synchrotron X-rays for up to 30 ms time scale predominantly produces oxidative modifications in the side chains of proteins compared to cross linking events [240]. In this study we used a protein foot printing method to determine the location of Sml1’s interaction with ScRR1, which we tried to independently corroborate by chemical cross-linking and site-directed mutagenesis.

However, we were not able to identify a common region of Sml1 binding on ScRR1 using these two techniques.

174

Our protein foot printing data was not able to identify any peptides with significant protection (i.e. > 4-fold) by Sml1 to define a true binding surface (Table 5.1 p, 152). The principal reason behind this is the lack of sequence coverage of ScRR1 in our foot printing experiments. We were not able to detect the peptides corresponding to functionally important regions of ScRR1 especially in the α/β barrel domain. The lack of sequence coverage dramatically increased with increased exposure times to the X –ray

(Figure 5.3 p, 150). However, based on our limited sequence coverage we were able reach some important conclusions about the interaction of ScRR1 and Sml1.

It has been reported with yeast that residues W688 and E689 located near the three helix insert (THI) motif, influenced Sml1 binding to ScRR1 [252]. Mutations

(W688A/G and E689D/A) at these two positions in the THI of ScRR1 has shown to enhance the lethality of Sml1 inhibition [88]. However, our foot printing data did not indicate that this region is directly or indirectly involved in Sml1 binding. Neither techniques employed in this study to map the interaction surface identified these residues as being of interest (Table 5.1 p, 152 and Table 5.2 p, 160). These two mutations are flanked by two lysine residues (K685 and K693), which can be cross-linked with activated glutamic or aspartic acid residues of Sml1 during our cross-linking experiments.

Furthermore, the W689 residue of ScRR1 can be modified by the hydroxyl free radical.

The ASA calculation shows that the W689 has a higher degree of solvent accessibility than the M697, which is the nearest residue to W688 and E689, and which showed oxidative modification in both ScRR1 and ScRR1●Sml1 complex. However, the degree of protection of M697 in ScRR1●Sml1 complex was not significant. Therefore, the non- existence of any cross-links or protection against oxidative modifications in this study

175 suggests that Sml1 doesn’t directly interact with ScRR1 near the THI region. However, we can’t rule out that this region may be involved in a long range allosteric effect which contributes to Sml1 inhibition of ScRR1. In contrast, the claim that W688 E689 region of

ScRR1 may be involved in Sml1 interaction can’t be completely excluded because in vivo nature of the yeast genetic experiments. It is possible that Sml1 may be interacting with WE motif in a different oligomeric state of ScRR1 (i.e. α6).

Another functionally important region of ScRR1 is the C-terminal region, which is important for the regeneration of the active site cysteines of ScRR1 at the end of each catalytic cycle. Our foot printing data identified two ScRR1 peptides (814-827 and 854-

871) in the C-terminal region that experienced increased rates of oxidation. This implies that the C-terminus of ScRR1 does not interact with Sml1. A recently concluded yeast genetic study in yeast also agreed with our observation that Sml1 is still able to interact with ScRR1 when the CI region (788-888) of ScRR1 is deleted.

Our foot printing data also identified two regions of ScRR1 with a moderate degree of protection by Sml1. One of these peptides (738-744) includes the two tyrosine residues that receive the free radical from the small subunit. This modest degree of protection (< 4 fold) may contribute to ScRR1 inhibition resulting from partial burying of this region due to a conformational change upon binding of Sml1. Consistent with these results, enzyme kinetics studies by Chabes et al. have shown that Sml1 does not compete with ScRR2●ScRR4 subunit for binding to the ScRR1 subunit.

Chemical cross linking identified two unambiguous peptides (Table 5.2 p, 160) in the N-terminal region extending from amino acid residues 10 to 21 of ScRR1 that are

176 involved with Sml1 binding. But our foot printing data ruled out the possibility that this

N-terminal region was not protected by Sml1 binding since the oxidation rates were similar in the complex and in the ScRR1 alone. In order to independently confirm whether the residues 10-21 form the region involved in binding Sml1 we conducted site directed mutagenesis. We initially made two deletion mutants Δ22 and Δ90 in which the first 22 residues and the entire ATP cone were deleted. Furthermore, we constructed two single mutants, K10E and K17E and a double mutant (K10E/K17E) from the same region. We used a combination of fluorescence anisotropy / quenching, SEC and enzyme activity assays to test these mutants.

We determined the dissociation constant of Sml1 binding to ScRR1 dimer to be

0.083 + 0.02 M using fluorescence anisotropy. Previous studies using BIACORE analysis have shown that Sml1 specifically binds to ScRR1 with a dissociation constant of 0.4 + 0.1 μM [248]. The discrepancy in the Kd determination may result from the absence of nucleotides in the biosensor samples and the inherent differences between two techniques. For instance, the biosensor samples are restrained by solid support at one end while the fluorescence method has no such restraints. In our experiments using fluorescence anisotrophy, the two deletion mutants Δ22 and Δ90 were shown not to bind to ScRR1 (Figure 5.7 p, 164). To further corroborate these results, we tested the ability of Sml1 to bind a peptide epitope (KEPVQFDKITAR) using fluorescence quenching studies. Our data show Sml1 binds to the 10-21 peptide with a Kd of 0.12 +0.02 µM. The random peptide used as negative control did not bind to the Sml1.

We independently verified that the N-terminal deletion of Δ22 ScRR1 abolished

Sml1 binding. These mutants did not co-elute with Sml1 in a SEC experiment further

177 confirming our findings that the N10-22 epitope is involved in Sml1 binding. The two deletions (Δ22 and Δ90) in the ATP binding cone completely abolished the ScRR activity. In addition, the two point mutants (K10E & K17E) showed considerable degree of inhibition by Sml1 when compared to wild type ScRR1 (Figure 5.9 p, 168). It is surprising that the double mutant (K10E/K17E) in the ATP binding cone showed a comparable degree of inhibition to that of wild type ScRR1. Even though the 10-21

ScRR1 peptide binds to C14S S60C Sml1 with nanomolar affinity, it was unable, even at

1,000-fold molar excess, overcome the Sml1 inhibition of ScRR1 (Figure 5.7 p, 166).

Taken together, these results did not provide conclusive evidence to support the notion that Sml1 binds to the 10-21 ScRR1 peptide.

The different results obtained for the Sml1 binding site on ScRR1 using the above methods prompted us to further probe these discrepancies. The measurement of the alpha helical content in the Δ22 mutant by CD spectroscopy revealed that the mutant protein lost ~ 20% of its alpha helical content (Appendix 6A p, 200). A loss of alpha helical content of this magnitude would have a dramatic effect on the structure of ScRR1. As revealed by our fluorescence and SEC experiments, Sml1 may not interact with ScRR1 with an altered structure. The observed cross links between the N-terminus of ScRR1 and

Sml1 reported in our experiments may be due to non-specific chemical cross linking.

We also used a cryo-EM based approach to identify the binding site of Sml1 on

ScRR1. Initial cryo-EM analysis reveals it may bind to the catalytic core of ScRR1

(Appendix 4A p, 198). Due to the low resolution of the cryo-EM structure it was not possible to identify the interacting residues. We tried to enhance the visual resolution by mapping the binding site using GFP-Sml1, but complex formation between GFP-Sml1

178 and dATP-induced ScRR1 hexamer was not successful. GFP-Sml1 also failed to inhibit

CDP reduction by the hexamer; we postulate that steric clashes between the GFP moieties within the hexamer ring may be responsible for the results. Complex formation between GFP-Sm11 and ScRR1 dimer was detected and GFP-Sml1 inhibited the ADP reduction by the dimer. However, cryo-EM image analysis revealed very few ScRR1 dimers bound to GFP-Sml1 ((Figure 5.11 p, 173) (Appendix 5A p, 199).

Even though cryo-EM analysis does not directly confirm the binding site, protein foot printing, chemical cross linking and mutagenesis data indirectly suggests that Sml1 binding does require neither the N-terminus nor the entire C-terminal insert of ScRR1.

Further cryo-EM studies are currently under way along with complementary biochemical and biophysical approaches to determine the Sml1 binding site. It seems that the hybrid structural method hold the most promise for identifying the Sml1 binding site on ScRR1.

179

Chapter 6: Summary and Future Directions

6.1 Inhibition of RNR by dATP

RNR plays a central role in dNTP metabolism and has been considered an attractive target of anticancer and antimicrobial therapy. All known RNRs are subject to multiple modes of regulation and among these regulatory mechanisms are allosteric regulation and transcriptional regulation of ndr genes. There were two overall goals of this dissertation research. The first goal (dealt with in Chapter 3) is to understand the structural basis of activity site regulation by dATP and / ATP. The second goal (dealt with in chapter 4 & 5) of this dissertation is to identify the binding site of Sml1 on yeast

RNR and its kinetic mechanism of inhibition.

We observed oligomerization of the large subunit of HuRRM1 and ScRR1 and decreasing activity of both enzymes with increasing concentration of dATP. At low dATP concentration (i.e. 2 µM) HuRRM1 predominantly formed dimers but at physiological concentrations of dATP HuRRM1 mainly formed hexamers as determined by SEC and MALS. Therefore, we hypothesized that hexamerization is a prerequisite for enzyme inhibition. Previous studies have reported that these oligomeric complexes have subunit compositions of α6β2 but this the first study that describes the subunit arrangement in the hexamer. We were able to successfully crystalize the dATP hexamer in the P63 hexagonal space group. These crystals diffracted to 6.6 Å and at this resolution we were only able to understand the packing arrangement of individual ScRR1 molecules. Our observation suggested two different packing models of the dATP hexamer, model A and B. Upon examination of these models, we selected for targeted

180 mutagenesis amino acid residues D16 and H2 from the hexamer interface seen in model

B and D182 from the hexamer interface seen in model A. We mutated these residues in

HuRRM1 to give rise to the D16R, H2E and D182R mutants, and tested their ability to form hexamers. The two mutants, D16R and H2E that disrupted the hexamer interface seen in model B did not form hexamers suggesting that model B is correct. Furthermore, these two mutants provided an opportunity to test our hypothesis which states that hexamerization is a prerequisite for enzyme inhibition. As expected, our activity assay results showed that dATP does not inhibit the D16R and H2E mutants. This observation also confirms that the dATP does not inhibit the dimeric form of the enzyme.

Another unresolved issue in activity site regulation of RNR is how ATP and dATP which differ by a single atom, are able to bind to the same activity site and bring about hexamer formation, yet have totally opposite effects on RNR activity. We further hypothesized that though ATP and dATP bring about hexamer formation, changes in the hexamer interface cause inhibitory and stimulatory effects. We tested the ability of the mutants D16R, H2E and D57N, to form hexamers in the presence of ATP. ATP induced hexamer formation in all of these mutants suggesting that ATP-induced and dATP- induced hexamers utilize different interfaces for hexamer formation.

We also attempted to crystalize the dATP holoenzyme complex but the crystals were too small for further crystallographic studies. Purified dATP holo enzyme complex has the subunit composition of α6β2 as determined by SEC. Structure determination by cryo-EM microscopy revealed the similar subunit composition of the complex. The small

ββ’ subunit was located inside the hexamer ring and was positioned in a way that would

181 not result in the formation of the proper free radical pathway from the small subunit to the large subunit.

Based on our experiments, we proposed a model for inhibition of RNR by dATP.

In the absence of any nucleotide triphosphates, RNRs mainly exists as monomers. Upon binding of ATP, dATP, dGTP or TTP to the S-site of the large subunit, the enzyme forms a dimer. These dimers can acquire a small subunit of RNR to form an active tetrameric

(α2β2) complex. These dimers can undergo further oligomerization upon binding of dATP to the A-site resulting in hexamer formation. Binding of one small subunit to the dATP hexamer then results in the formation of inactive α6β2 complex. It is to be noted that even at physiological concentrations of dATP, the enzyme is not fully inhibited. This is solely due to the existence of the α2β2 complex.

In this work we described the low resolution structure of the dATP-induced hexamer and the dATP ScRR holenzyme complex. We have shown that HuRRM1 also forms hexamers at physiological concentrations of ATP. Interestingly the ATP-induced hexamer adopts a different hexamer interface. Therefore it is important to determine the structural details of the ATP-induced hexamer interface. Since the ATP-induced hexamers is the most active form of RNR, determining its structure may provide a novel site for drug development.

Structure determination of the ATP-induced hexamer and ATP-bound holoenzyme require both crystallographic and cryo-EM approaches. Unlike the dATP- induced hexamer, the ATP-induced hexamer is less stable and seems to comprise a heterogenous mixture. This phenomenon is also seen in the case of the E. coli ATP- induced oligomer [122]. The stability of the ATP-induced hexamer may be improved by

182 changing purification parameters like ATP concentration, buffer composition and inclusion of additives like substrates UDP or CDP. It is important to note that direct addition of ATP to the protein results in sample precipitation. This may be a result of impurities present in the ATP itself (i.e. ADP). Therefore, it may be necessary to purify

ATP using ion exchange chromatography (i.e. Mono Q). Recent studies by the Stubbe lab have shown clofarbine diphosphate can stabilize hexamers in the presence of ATP

[124]. Therefore crystallization of the ATP-induced hexamer should be attempted in the presence of ATP, dGTP and clofarabine diphosphate. In parallel to crystallization, the

ATP clofarabine hexamer complex must be subjected to cryo-EM analysis. Our preliminary studies with cryoEM showed that ATP hexamers forms rod like structures with varying lengths. These may well represents breakdown of the putative closed ring structure of the ATP-induced hexamer.

It is very unlikely that the ATP-induced hexamer adopts structure completely different from that of the dATP-induced hexamer. Provided that the ATP-induced hexamer adopts slightly different hexamer interface, it is possible to improve the stability of the interface using disulfide cross-linking. Based on the structural information about the dATP-induced hexamer interface available from model B, a series of cysteine mutants can be designed in the hexamer interface. These cysteine mutants must be first tested for their ability to form ATP /dATP hexamers. Purified mutant ATP-hexamers must be further dialyzed in the absence of reducing agents and in the presence of dATP/ATP and oxidizing agents like Cu(II)-1-10-phenanthroline. One potential drawback of this method is the dimer formation between free C-terminal tails of ScRR. In order to prevent such tandem dimers, it is advisable to use truncated ScRR1 lacking the

183

CXXC motif and the N-terminal cysteines for disulfide cross-linking. After dialysis these mutants must be tested for their ability to form hexamers in the absence of ATP. If these mutants are able to form hexamers in the absence of ATP, such mutants can be subjected to cryo-EM analysis

These mutants may also be used to get useful structural information. Briefly, a given mutant can be labeled independently with two different FRET pairs (Alexa Fluor

350 as donor and Alexa Fluor 488 as acceptor) [253]. Then these individually but differently labeled mutants can be mixed in equimolar concentrations in the presence of

ATP. Incorporation of two different tags (i.e. a 6XHis tag and a GST tag) to the C- terminus of each protein may allow purification of ScRR1 dimers with two different fluoropores attached to their ATP cones. Then one can do a detailed FRET pair analysis to obtain the distances between the fluoropores. The FRET phenomenon will be only observed between ATP cones with two different fluoropores attached within 10-100 Å of each other. Distances between the two FRET pairs can be calculated under different conditions (i.e. ATP-induced hexamers vs. dATP-induced hexamers) and these distances can be used as constraints to model the hexamer interface.

As an alternative method to Cys-mutants, one can incorporate unnatural amino acids with different side chains to pre-determined positions using Shultz method [254].

The method allows cotranslational incorporation of unnatural amino acids into proteins with high fidelity and efficiency by means of a unique codon and corresponding orthogonal tRNA-amainoacyl-tRNA synthetase pair. Incorporation of these unnatural amino acids has allowed labeling of proteins in a position specific manner to probe conformational changes using techniques like FRET [255]. Since this method has been

184 used successfully to incorporate these unnatural amino acids into RNR, it is quite possible to perform these studies.

Current information regarding RNR oilgomerization has been acquired almost exclusively from in vitro studies. It is imperative that the ATP or dATP-induced hexamers be characterized in vivo. The simplest method would be to incorporate an amino acid like L-Photo-Methionine into RNR and allow them to cross-link under UV light [256]. The amino acid L-photo-Methionine contains a diazirine ring, which generates reactive carbene intermediates that irreversibly cross-link proteins. These amino acids can be readily incorporated into proteins in cell culture media lacking methionine. Cells fed these amino acids can be exposed to UV light at different stages of the cell cycle in hopes of purifying cross-linked ATP and dATP-induced hexamers from them. Such cross-linked complexes could be subsequently analyzed by SEC, sucrose density gradient centrifugation and mass spectrometry to determine their identity. The next frontier would be to investigate RNR complexes in real time during the normal cell cycle and after DNA damage. In recent years advances have been made in high resolution optical microscopy, giving rise to a technique known as stochastic optical reconstruction microscopy (STORM) that has overcome the diffraction limit. Using photo switchable fluoropores (i.e Cy5 and Cy3) to label the large and small subunits of RNR it may be possible to study the dynamics of these subunits in cell culture systems in nanometer resolution [257].

185

6.2 Inhbition of Yeast RNR by Sml1

Inhibition of RNR is not restricted to dATP but in some organisms, especially among fungi, small inhibitory proteins like Sml1 and Spd1 regulate RNRs. In this thesis, we explored the inhibitory mechanism of Sml1 and its binding site on ScRR1 using an approach based on enzyme kinetics and structural biology. Previous studies have shown

Sml1 is a potent inhibitor of ScRR. Binding of Sml1 to ScRR1 dimer has been characterized. Here we investigated Sml1 binding to ATP and dATP- induced hexamers.

For this purpose, we labeled a mutant Sml1 protein (C14S/S60C Sml1) with Alexa 350 fluoropore to monitor its interaction with ScRR1 hexamers. Using this technique, we showed that Sml1 can bind to both dATP and ATP-induced hexamers. Our enzyme activity assays showed that Sml1 almost completely inhibits the activities of both ATP and dATP-induced hexamers. It is well known that dATP-induced hexamers are not fully inhibited at physiological concentrations of dATP. We also observed that ScRR1 retained

50% of specific activity in the presence of 20 µM dATP and full inhibition of the enzyme requires the presence of Sml1.

We investigated the kinetic mechanism of ScRR1 inhibition by Sml1 using the initial rates method. We tested the inhibitory effect of Sml1 under with hexameric

(ATP/CDP) and dimeric (dGTP/ADP) ScRR1. It was revealed that ScRR inhibition by

Sml1 follows an S-linear I-parabolic mode of inhibition irrespective of the oligomeric state of the enzyme. Binding of Sml1 to ScRR1 shows positive cooperativity. According to our model, in the case of the ATP-induced hexamer, the first molecule of Sml1 is bound with 12-fold less affinity then the second molecule of Sml1. Likewise the first molecule of Sml1 is bound to the dGTP based dimer with 7–fold less affinity than the

186 second molecule of Sml1. We also found that binding affinity of Sml1 is greater towards the ScRR1 hexamers than it is toward the dimers. This may be physiologically significant since Sml1 can easily dissociate from the ATP-induced hexamer and undergo degradation at the beginning of the S-phase, allowing cells to synthesize dNTPs.

Database sequence search reveals Sml1 orthologs in many fungal species, some of which are close relatives of S. cerevisiae. Most notably Sml1 like proteins were absent from many fungi which cause life threatening infections, especially among immunocompromised patients. The lack of Sml1 like orthologs, may pave the way to develop a novel antifungal agent based on Sml1 peptides to treat those fungal infections.

Our discovery that Sml1 also inhibits human RNR with an IC50 value of 20 + 2 µM, may aid the development of novel anticancer agents.

In this study we used a multidisciplinary approach to identify the binding site of

Sml1 on ScRR1. Protein foot printing and chemical cross linking studies did not reveal any conclusive evidence of Sml1 binding on ScRR1. Lack of sequence coverage was the major problem using the foot printing method to identify the binding surface precisely.

Even though the chemical cross linking data provided unambiguous chemical cross-links between N-terminus of ScRR1 and C-terminus of Sml1, it was later revealed, based on our mutageneis, fluorescence spectroscopy and circular dichroism studies, that these are non-specific cross links. However, taken together these data provide indirect evidence that the Sml1 binding site does not involve either the N-terminal ATP-binding cone or the

C-terminal insert of ScRR1. We did not find any supporting evidence that the THI of

ScRR1 is involved in Sml1 interaction as suggested by genetic studies. Both of our cross linking and foot printing studies failed to identify any peptides in the THI that are either

187 cross linked to Sml1 or protected by Sml1. These observations left only the α/β barrel domain by default as the key region of ScRR1 that may be involved in interaction with

Sml1. Our cryo-EM data on ScRR1 hexamer●Sml1 complex seems to support this notion. However, at 13 Å resolution of the cryo-EM data, we were not able to confirm the interacting residues. Attempts to improve the resolution of the complex using GFP-Sml1 were not successful probably due to steric clashes between the GFP moieties within the central hole of the dATP-induced hexamer.

The interaction of Sml1 and ScRR1 has been studied in detail in vitro and in vivo.

However it still remains unclear where Sml1 interacts on ScRR1, Determining the exact region of binding is important in developing novel Sml1 based therapeutics against fungal diseases and cancers. Since attempts to crystallize ScRR1●TTP ●Sml1 have not been successful in the past, it is important to undertake alternative and hybrid structural approaches to identify the binding region at high resolution.

Currently, a cryo-EM based approach is being pursued in order to identify the region. However, the current resolution of the cryo-EM map is lingering around 13 Å, and at this resolution, it may be quite difficult to precisely map out the binding region.

Lack of a high resolution structure of Sml1 further complicates the interpretation of the cryo-EM results. It has been observed that GFP-Sml1 is bound to only a small percentage of ScRR1 dimers. Since reliable high-resolution structure determination of the ScRR1

Sml1 complex by cryo-EM requires a structurally and compositionally homogeneous sample, the proportion of ScRR1 dimer particles bound to GFP-Sml1 must be improved using a sample preparation method like GraFix [258]. This protocol calls for

188 sedimentation of ScRR1 dimers in a sedimentation gradient that includes a weak chemical fixative agent.

In parallel to this, attempts must be made to crystallize the dATP-induced ScRR1 hexamer bound to Sml1. Since Sml1 is very flexible, and ScRR1 has many flexible parts this might seem unlikely to succeed. However binding to the ScRR1 hexamer may stabilize Sml1 enough to make determining a crystal structure possible. It is unlikely, however, that a high resolution structure of the complex will be attainable. Therefore, one must consider using re-engineering ScRR1 to be less flexible. Previous genetic studies have identified a double mutant (W688A/G and E689D/A) in the three helix insert (THI) of ScRR1, which has enhanced lethality for yeast when Sml1 is present suggesting it binds Sml1 more tightly. Genetic studies have also identified the flexible C-terminal tail is not important for Sml1 inhibition of ScRR1. Deletion of residues up to 90 amino acids from the C-terminal tail does not perturb the structure and ScRR1 activity (personnel communication with Dr. Jim Fairman). Therefore, when crystallization of ScRR1 hexamer●Sml1 complex is attempted, it is recommended that this ScRR1 double mutant with C terminal tail deletion be used in place of wild type ScRR1. Even a low resolution crystal structure of the complex (i.e. ~6 Å) may improve the chance of identification of the potential binding region Sml1 on ScRR1 by cryo-EM.

Even if the combination of cryo-EM and X-ray structure yields a relatively low resolution structure, it is possible to improve the structure to the atomic resolution level by analyzing the Sml1 and ScRR1 interaction using alternative techniques. One such technique is protein foot printing (PF). The work described in this thesis was not able to identify the ScRR1dimer-Sml1 interaction by PF. One of the main problems associated

189 with this method is the lack of sequence coverage of ScRR1. Provided that a higher degree of sequence coverage is obtained by the PF method, it would be possible to map out the binding region at atomic resolution, since tandem mass spectrometry is able to precisely indicate the site of oxidative modifications, which enables identification of the residues within the region on ScRR1 protected from solvent by Sml1 binding. If PF is pursued as an alternative strategy, a great deal of attention must be devoted to improving the sequence coverage of ScRR1 especially in the α/β-barrel domain. In this study, we used only trypsin as the sole proteolytic enzyme and therefore testing other proteolytic enzymes such as endoproteinase Glu-C and endoproteinase Asp-C could potentially improve the sequence coverage. The other parameter that needs to be optimized is the X- ray exposure time of the samples since higher exposure times may lead to sample aggregation and thereby affect sequence coverage.

Another approach that one may explore for mapping the interaction surface between the two proteins is heteronuclear NMR spectroscopy [259]. Since ScRR1 has a molecular weight of 100 kDa, the ScRR1 dimer is small enough to be used in NMR experiments. For this purpose ScRR1 protein has to be labeled with either 15N or 13C isotopes. This can be easily achieved by expressing the protein in minimal media

15 2 containing NH4Cl. At the same time, Sml1 must be labeled with H isotope to make it invisible to NMR spectroscopy. An inverse NMR experiment known as HSQC

(Heteronuclear Single Quantum Correlation) can be performed on labeled ScRR1 in the presence and absence of deuterated Sml1. Since each signal in HSQC spectrum represents a proton that is directly bound to a nitrogen atom (HN) and since there is only one backbone HN atom per amino acid each HSQC signal represents a single amino acid.

190

Therefore, it is possible to identify amino acids that experience chemical shift (ppm) in the presence and absence of deuterated Sml1 and these amino acids represent the residues directly involved in the binding surface or those that experience conformational changes.

Structural information gained by these methods (NMR, mass spectrometry, EM) can be used as constraints to perform computational modeling studies to further improve the structural details of the ScRR1●Sml1 interaction surface. Once the interface between

Sml1 and ScRR1 is established with great certainty, the validity of the results must be further explored by site-directed mutagenesis. Our foot printing data has been able to determine that the ATP binding cone, the C- terminal insert of ScRR1 and the WE motif in the THI are not involved in interactions with Sml1. Therefore it seems most likely that

Sml1 binds to the α/β barrel domain. ScRR1 mutants in the binding interface (single, double or triple) should be tested for their activity and their ability to be inhibited by

Sml1. If these interface mutants are not active enough for Sml1 inhibition to be tested, peptides corresponding to these regions of ScRR1 should be tested in a competitive inhibition assay. As an alternative solution, these ScRR1 mutants can be tested for their ability to interact with Sml1 using a binding assay. The work described in this thesis developed a fluorescence anisotropy based binding assay using labeled C14S S60C Sml1.

The same assay can be utilized to measure the binding affinity of the mutants towards

Sml1. Finally, these mutants should be introduced into yeast cells and their phenotypes investigated. For example, one investigates the effects of these mutations on dNTP pools

[47]. In addition, direct interaction of these mutant ScRR1 and Sml1 can be tested using yeast two-hybrid assays [88].

191

Our enzyme kinetic experiments identified that Sml1 binds ScRR1 hexamer with less affinity than the dimer. We could not independently verify this observation because the fluorescence anisotropy (FA) based assay yielded a binding isotherm, which widely deviated from standard curves of data fitting for the ScRR1 hexamers. This unexpected finding may be the result of protein aggregation. Since FA based assays are inherently sensitive to protein aggregation, this technique may not be suitable for assaying Sml1 binding to ScRR1 hexamers. As an alternative to a FA based assay, binding interactions can also be assayed using SPR. The protein Sml1 can be immobilized on the sensor chip and pre-formed hexamers or dimers can be flowed over the chip to determine the Kd.

We also found that Sml1 like analogs are not present in a number of pathogenic fungal species and in mammals. However, these fungal species carry a copy of the RNR gene and Sml1 has been shown to inhibit RNR of other species. Therefore, Sml1 based peptides or peptidomimetics could be used as antifungal agents. As a preliminary step,

RNRs of these fungal pathogens (i.e. Candida albicans, Histoplasma capsulatum,

Blastomyces dermatitis) which have a high sequence homology to yeast RNR can be cloned and expressed in E .coli. Then, the varying length of the C-terminal peptides (i.e. in the last 30 residues) of Sml1 can be tested for their inhibitory effect using the standard activity assay to determine their Ki value. In addition, the binding affinity of these peptides towards the selected fungal RNR can be estimated using either isothermal titration calorimetry (ITC) or R1- peptide binding assay as described by Pender et al.

Initial attempts can be made to develop Sml1 peptide (i.e. R1B domain) based therapeutics against fungi that causes skin infections like Candida albicans, which lacks

Sml1 like analogs. Similarly, if these Sml1 based peptides are successful in inhbiting the

192

RNR of fungal pathogens like Histoplasma capsulatum, those peptides may also be combined with antifungal agents like amphotericin B not only to reduce its toxicity, but also to enhance the synergistic effect of both drugs.

193

Appendix

1A MALS Analysis of dATP-induced HuRRM1 Hexamer

MALS analysis of wild type HuRRM1 at 20 μM dATP shows that the HuRRM1 forms predominantly hexamers (~520 kDa) and very little dimers (~185 kDa). Note: LS = light scattering detector and dRI = refractive index detector.

As mentioned in Chapter 3, accurate determination of molecular weight of HuRR1 oligomers was carried out using MALS. This technique can be used to directly determine the molecular weights of molecules independent of molecular mass reference standards, column calibration and assumptions of molecular shape. SEC-MALS separates mixtures of oligomers and measures the absolute molecular weight of an oligomer in an elution

194 fraction, which provided important information for the present study. A Wyatt TREOS multi-angle light scattering (MALS) detector set at a wavelength of 658 nm (calibrated with toluene) and paired with a Wyatt Optilab REX differential refractive index (dRI) detector (also operating at 658 nm and calibrated with NaCl solutions) was used for the online molecular weight determination. A Wyatt SEC500, 7.8 x 300 mm SEC column was connected upstream of the MALS-RI detectors and used to fractionate the injected sample. The SEC-MALS-RI system as a whole was validated using BSA (Sigma-

Aldrich). For the wild-type hRRM1 sample, 100 μl of 0.5 mg/ml wild-type hRRM1 in 50 mM Tris pH 7.6, 5 mM MgCl2, 100 mM KCl, 20 μM dATP were injected. For the D16R hRRM1 sample, 100 μl of 0.5 mg/ml D16R mutant in 50 mM Tris pH 7.6, 5 mM MgCl2,

100 mM KCl, with 20 μM dATP were injected. The running buffer was 50 mM Tris pH

7.6, 5 mM MgCl2, 100 mM KCl, 20 μM dATP. The chromatograms and resultant molecular weight data were analyzed using the Astra 5.3 software from the Wyatt

Corporation.

195

2A SEC analysis of RNR Holo Complex with t-HuRRM1 and HuRRM2

The oligomeric states of t-hRRM1 and hRRM2 (a truncated construct starting from residue 74, with an additional N-terminal 6xHis tag) were evaluated using SEC. The t- hRRM1 in the presence of 20 μM dATP formed a dimer that eluted at a molecular mass of 186 kDa. When the two species were mixed, they eluted at a molecular mass of 278 kDa, corresponding to an α2β2 holocomplex. A Superdex 200 10/300GL column was equilibrated with 20 μM dATP in the running buffer (50 mM sodium citrate pH 5.5, 150 mM NaCl, 10% glycerol, 2 mM TCEP). Protein samples were prepared in the same running buffer with 20 μM dATP. The t-hRRM1 was first run alone then followed by the t-hRRM1 (19.5 μM) ● hRRM2 (18.5 μM) complex. This experiment was performed at

Karolinska Institutet by Dr. Martin Welin under the supervision of Dr. Pär Nordlund.

196

3A MALS Analysis of D16R HuRRM1 Mutant

MALS analysis of D16R HuRRM1 at 20 μM dATP shows that the mutant forms only dimers (~190 kDa) and monomers (88 kDa).

197

4A Cryo-EM Analysis of dATP-induced ScRR1 Hexamer Sml1 Complex

(A) Structure analysis of the dATP-induced ScRR1 hexamer Sml1complex by cryo

EM shows a weak density in the α/β barrel domain. The resolution of the structure is 13

Å. A computational model of Sml1 was used to model the putative Sml1 density. (B) To see if Sml1 could possibly bind at the same position in the ScRR1 dimer as in the hexamer, the X-ray structure of dGTP/ADP-ScRR1 dimer (PDB ID 2EUD) was docked onto the dATP-induced ScRR1•Sml1 hexamer model using Pymol. At a resolution of 13

Å the density of the Sml1 in the ScRR1 hexamer overlapped with the catalytic pockets of the dimer of ScRR1. No attempts were made to identify the interacting residues between

ScRR1 and Sml1. Cryo-EM data collection and analysis was done by Dr. Zongli Li at

Harvard University under the guidance of Dr. Thomas Walz.

198

5A Class Averages of GFP-Sml1 bound ScRR1 dimer and hexamer

Class averages for 10,000 particles each for the (A) ScRR1 dimer with Sml1-GFP and

(B) the ScRR1 hexamer with Sml1-GFP. The red boxes indicate averages with additional density that could represent Sml1-GFP. For the hexamer, a white box marks averages, without the extra density.

199

6A Circular Dichroism (CD) Absorption Spectrum of wild-type and Δ 22 ScRR1

The Δ22 mutant ScRR1 had 20% less alpha helical content when compared to the wild type ScRR1. Note the relatively weak alpha helical signal at 222nm for Δ 22 ScRR1.

200

References

1. Brown, N.C. and P. Reichard, Ribonucleoside diphosphate reductase. Formation of active and inactive complexes of proteins B1 and B2. J Mol Biol, 1969. 46(1): p. 25-38.

2. Nordlund, P. and P. Reichard, Ribonucleotide reductases. Annu Rev Biochem, 2006. 75: p. 681-706.

3. Mathews, C.K., DNA precursor metabolism and genomic stability. FASEB J, 2006. 20(9): p. 1300-14.

4. Weinberg, G., B. Ullman, and D.W. Martin, Jr., Mutator phenotypes in mammalian cell mutants with distinct biochemical defects and abnormal deoxyribonucleoside triphosphate pools. Proc Natl Acad Sci U S A, 1981. 78(4): p. 2447-51.

5. Kunz, B.A., et al., International Commission for Protection Against Environmental Mutagens and Carcinogens. Deoxyribonucleoside triphosphate levels: a critical factor in the maintenance of genetic stability. Mutat Res, 1994. 318(1): p. 1-64.

6. Manegold, C., Gemcitabine (Gemzar) in non-small cell lung cancer. Expert Rev Anticancer Ther, 2004. 4(3): p. 345-60.

7. Ingram, G.M. and J.H. Kinnaird, Ribonucleotide reductase: A new target for antiparasite therapies. Parasitol Today, 1999. 15(8): p. 338-42.

8. Sandrini, M.P., et al., Deoxyribonucleoside kinases activate nucleoside antibiotics in severely pathogenic bacteria. Antimicrob Agents Chemother, 2007. 51(8): p. 2726-32.

9. Sandrini, M.P., et al., Nucleoside analogues are activated by bacterial deoxyribonucleoside kinases in a species-specific manner. J Antimicrob Chemother, 2007. 60(3): p. 510-20.

10. Johns, D.G. and W.Y. Gao, Selective depletion of DNA precursors: an evolving strategy for potentiation of dideoxynucleoside activity against human immunodeficiency virus. Biochem Pharmacol, 1998. 55(10): p. 1551-6.

11. Romanelli, F., C. Pomeroy, and K.M. Smith, Hydroxyurea to inhibit human immunodeficiency virus-1 replication. Pharmacotherapy, 1999. 19(2): p. 196-204.

12. Bertani, L.E., A. Haggmark, and P. Reichard, Synthesis of pyrimidine deoxyribonucleoside diphosphates with enzymes from Escherichia coli. J Biol Chem, 1961. 236: p. PC67-PC68.

201

13. Moore, E.C. and R.B. Hurlbert, Reduction of 5'-cytidylic acid to deoxycytidylic acid by mammalian enzymes. Biochim Biophys Acta, 1960. 40: p. 371-2.

14. Reichard, P., A. Baldesten, and L. Rutberg, Formation of deoxycytidine phosphates from cytidine phosphates in extracts from Escherichia coli. J Biol Chem, 1961. 236: p. 1150-7.

15. Reichard, P., Z.N. Canellakis, and E.S. Canellakis, Regulatory mechanisms in the synthesis of deoxyribonucleic acid in vitro. Biochim Biophys Acta, 1960. 41: p. 558-9.

16. Reichard, P., Formation of deoxyguanosine 5-phosphate from guanosine 5- phosphate with enzymes from chick embryos. Biochim Biophys Acta, 1960. 41: p. 368-9.

17. Reichard, P. and L. Rutberg, Formation of deoxycytidine 5'-phosphate from cytidine 5'-phosphate with enzymes from Escherichia coli. Biochim Biophys Acta, 1960. 37: p. 554-5.

18. Moore, E.C., P. Reichard, and L. Thelander, Enzymatic Synthesis of Deoxyribonucleotides.V. Purification and Properties of Thioredoxin Reductase from Escherichia Coli B. J Biol Chem, 1964. 239: p. 3445-52.

19. Laurent, T.C., E.C. Moore, and P. Reichard, Enzymatic Synthesis of Deoxyribonucleotides. Iv. Isolation and Characterization of Thioredoxin, the Hydrogen Donor from Escherichia Coli B. J Biol Chem, 1964. 239: p. 3436-44.

20. Larsson, A. and P. Reichard, Enzymatic Synthesis of Deoxyribonucleotides. Journal of Biological Chemistry, 1966. 241(11): p. 2540-2549.

21. Holmgren, A., P. Reichard, and L. Thelander, Enzymatic synthesis of deoxyribonucleotides, 8. The effects of ATP and dATP in the CDP reductase system from E. coli. Proc Natl Acad Sci U S A, 1965. 54(3): p. 830-6.

22. Brown, N.C., et al., Nonheme iron as a cofactor in ribonucleotide reductase from E. coli. Biochem Biophys Res Commun, 1968. 30(5): p. 522-7.

23. Brown, N.C., et al., Spectrum and iron content of protein B2 from ribonucleoside diphosphate reductase. Eur J Biochem, 1969. 9(4): p. 512-8.

24. Ehrenberg, A. and P. Reichard, Electron spin resonance of the iron-containing protein B2 from ribonucleotide reductase. J Biol Chem, 1972. 247(11): p. 3485-8.

25. Larsson, A. and B.M. Sjoberg, Identification of the stable free radical tyrosine residue in ribonucleotide reductase. EMBO J, 1986. 5(8): p. 2037-40.

202

26. Reichard, P., From RNA to DNA, why so many ribonucleotide reductases? Science, 1993. 260(5115): p. 1773-7.

27. Stubbe, J., et al., Radical initiation in the class I ribonucleotide reductase: long- range proton-coupled electron transfer? Chem Rev, 2003. 103(6): p. 2167-201.

28. Cotruvo, J.A. and J. Stubbe, Class I ribonucleotide reductases: metallocofactor assembly and repair in vitro and in vivo. Annu Rev Biochem, 2011. 80: p. 733- 67.

29. Jordan, A. and P. Reichard, Ribonucleotide reductases. Annu Rev Biochem, 1998. 67: p. 71-98.

30. Uhlin, U. and H. Eklund, Structure of ribonucleotide reductase protein R1. Nature, 1994. 370(6490): p. 533-9.

31. Xu, H., et al., Structures of eukaryotic ribonucleotide reductase I provide insights into dNTP regulation. Proc Natl Acad Sci U S A, 2006. 103(11): p. 4022-7.

32. Brown, N.C. and P. Reichard, Role of effector binding in allosteric control of ribonucleoside diphosphate reductase. J Mol Biol, 1969. 46(1): p. 39-55.

33. Thelander, L. and P. Reichard, Reduction of ribonucleotides. Annu Rev Biochem, 1979. 48: p. 133-58.

34. Larsson, A. and P. Reichard, Allosteric effects and substrate specificity of the ribonucleoside diphosphate reductase system from Escherichia coli B. Biochim Biophys Acta, 1966. 113(2): p. 407-8.

35. Jordan, A., et al., A second class I ribonucleotide reductase in Enterobacteriaceae: characterization of the Salmonella typhimurium enzyme. Proc Natl Acad Sci U S A, 1994. 91(26): p. 12892-6.

36. Roshick, C., E.R. Iliffe-Lee, and G. McClarty, Cloning and characterization of ribonucleotide reductase from Chlamydia trachomatis. J Biol Chem, 2000. 275(48): p. 38111-9.

37. Fontecave, M., et al., The redox centers of ribonucleotide reductase of Escherichia coli. Adv Enzymol Relat Areas Mol Biol, 1992. 65: p. 147-83.

38. Tauer, A. and S.A. Benner, The B12-dependent ribonucleotide reductase from the archaebacterium Thermoplasma acidophila: an evolutionary solution to the ribonucleotide reductase conundrum. Proc Natl Acad Sci U S A, 1997. 94(1): p. 53-8.

203

39. Nordlund, P. and H. Eklund, Structure and function of the Escherichia coli ribonucleotide reductase protein R2. J Mol Biol, 1993. 232(1): p. 123-64.

40. Holmgren, A. and M. Bjornstedt, Thioredoxin and thioredoxin reductase. Methods Enzymol, 1995. 252: p. 199-208.

41. Rofougaran, R., M. Vodnala, and A. Hofer, Enzymatically active mammalian ribonucleotide reductase exists primarily as an alpha6beta2 octamer. J Biol Chem, 2006. 281(38): p. 27705-11.

42. Kashlan, O.B., et al., A comprehensive model for the allosteric regulation of mammalian ribonucleotide reductase. Functional consequences of ATP- and dATP-induced oligomerization of the large subunit. Biochemistry, 2002. 41(2): p. 462-74.

43. Cooperman, B.S. and O.B. Kashlan, A comprehensive model for the allosteric regulation of Class Ia ribonucleotide reductases. Adv Enzyme Regul, 2003. 43: p. 167-82.

44. Kashlan, O.B. and B.S. Cooperman, Comprehensive model for allosteric regulation of mammalian ribonucleotide reductase: refinements and consequences. Biochemistry, 2003. 42(6): p. 1696-706.

45. Wang, J., G.J. Lohman, and J. Stubbe, Enhanced subunit interactions with gemcitabine-5'-diphosphate inhibit ribonucleotide reductases. Proc Natl Acad Sci U S A, 2007. 104(36): p. 14324-9.

46. Fairman, J.W., et al., Structural basis for allosteric regulation of human ribonucleotide reductase by nucleotide-induced oligomerization. Nat Struct Mol Biol. 18(3): p. 316-22.

47. Rofougaran, R., et al., Oligomerization status directs overall activity regulation of the Escherichia coli class Ia ribonucleotide reductase. J Biol Chem, 2008. 283(51): p. 35310-8.

48. Yang, F., et al., Characterization of two genes encoding the Mycobacterium tuberculosis ribonucleotide reductase small subunit. J Bacteriol, 1997. 179(20): p. 6408-15.

49. Masalha, M., et al., Analysis of transcription of the Staphylococcus aureus aerobic class Ib and anaerobic class III ribonucleotide reductase genes in response to oxygen. J Bacteriol, 2001. 183(24): p. 7260-72.

50. Jordan, A., et al., Promoter identification and expression analysis of Salmonella typhimurium and Escherichia coli nrdEF operons encoding one of two class I

204

ribonucleotide reductases present in both bacteria. Mol Microbiol, 1996. 19(4): p. 777-90.

51. Jordan, A., et al., Characterization of Escherichia coli NrdH. A glutaredoxin-like protein with a thioredoxin-like activity profile. J Biol Chem, 1997. 272(29): p. 18044-50.

52. Cotruvo, J.A. and J. Stubbe, Class I ribonucleotide reductases: metallocofactor assembly and repair in vitro and in vivo. Annu Rev Biochem. 80: p. 733-67.

53. Eliasson, R., et al., Allosteric regulation of the third ribonucleotide reductase (NrdEF enzyme) from enterobacteriaceae. J Biol Chem, 1996. 271(43): p. 26582- 7.

54. Hogbom, M., et al., The radical site in chlamydial ribonucleotide reductase defines a new R2 subclass. Science, 2004. 305(5681): p. 245-8.

55. Jiang, W., et al., A manganese(IV)/iron(III) cofactor in Chlamydia trachomatis ribonucleotide reductase. Science, 2007. 316(5828): p. 1188-91.

56. Jiang, W., et al., Formation and function of the Manganese(IV)/Iron(III) cofactor in Chlamydia trachomatis ribonucleotide reductase. Biochemistry, 2008. 47(52): p. 13736-44.

57. Logan, D.T., et al., A glycyl radical site in the crystal structure of a class III ribonucleotide reductase. Science, 1999. 283(5407): p. 1499-504.

58. Fontecave, M., Ribonucleotide reductases and radical reactions. Cell Mol Life Sci, 1998. 54(7): p. 684-95.

59. Kolberg, M., et al., Structure, function, and mechanism of ribonucleotide reductases. Biochim Biophys Acta, 2004. 1699(1-2): p. 1-34.

60. Licht, S., G.J. Gerfen, and J. Stubbe, Thiyl radicals in ribonucleotide reductases. Science, 1996. 271(5248): p. 477-81.

61. Larsson, K.M., D.T. Logan, and P. Nordlund, Structural basis for adenosylcobalamin activation in AdoCbl-dependent ribonucleotide reductases. ACS Chem Biol. 5(10): p. 933-42.

62. Booker, S., et al., Coenzyme B12-dependent ribonucleotide reductase: evidence for the participation of five cysteine residues in ribonucleotide reduction. Biochemistry, 1994. 33(42): p. 12676-85.

205

63. Larsson, K.M., et al., Structural mechanism of allosteric substrate specificity regulation in a ribonucleotide reductase. Nat Struct Mol Biol, 2004. 11(11): p. 1142-9.

64. Torrents, E., et al., Euglena gracilis ribonucleotide reductase: the eukaryote class II enzyme and the possible antiquity of eukaryote B12 dependence. J Biol Chem, 2006. 281(9): p. 5604-11.

65. Sintchak, M.D., et al., The crystal structure of class II ribonucleotide reductase reveals how an allosterically regulated monomer mimics a dimer. Nat Struct Biol, 2002. 9(4): p. 293-300.

66. Torrents, E., et al., The anaerobic ribonucleotide reductase from Lactococcus lactis. Interactions between the two proteins NrdD and NrdG. J Biol Chem, 2001. 276(36): p. 33488-94.

67. Andersson, J., et al., Cysteines involved in radical generation and catalysis of class III anaerobic ribonucleotide reductase. A protein engineering study of bacteriophage T4 NrdD. J Biol Chem, 2000. 275(26): p. 19449-55.

68. Mulliez, E., et al., Formate is the hydrogen donor for the anaerobic ribonucleotide reductase from Escherichia coli. Proc Natl Acad Sci U S A, 1995. 92(19): p. 8759-62.

69. Eliasson, R., et al., The mechanism of the anaerobic Escherichia coli ribonucleotide reductase investigated with nuclear magnetic resonance spectroscopy. Biochem Biophys Res Commun, 1995. 214(1): p. 28-35.

70. Ollagnier, S., et al., Activation of the anaerobic ribonucleotide reductase from Escherichia coli. The essential role of the iron-sulfur center for S- adenosylmethionine reduction. J Biol Chem, 1997. 272(39): p. 24216-23.

71. Padovani, D., et al., Activation of class III ribonucleotide reductase from E. coli. The electron transfer from the iron-sulfur center to S-adenosylmethionine. Biochemistry, 2001. 40(23): p. 6713-9.

72. Padovani, D., E. Mulliez, and M. Fontecave, Activation of class III ribonucleotide reductase by thioredoxin. J Biol Chem, 2001. 276(13): p. 9587-9.

73. Eklund, H., et al., Ribonucleotide reductase--structural studies of a radical enzyme. Biol Chem, 1997. 378(8): p. 821-5.

74. Stubbe, J. and W.A. van der Donk, Ribonucleotide reductases: radical enzymes with suicidal tendencies. Chem Biol, 1995. 2(12): p. 793-801.

206

75. Persson, A.L., et al., A new mechanism-based radical intermediate in a mutant R1 protein affecting the catalytically essential Glu441 in Escherichia coli ribonucleotide reductase. J Biol Chem, 1997. 272(50): p. 31533-41.

76. Mao, S.S., et al., Characterization of C439SR1, a mutant of Escherichia coli ribonucleotide diphosphate reductase: evidence that C439 is a residue essential for nucleotide reduction and C439SR1 is a protein possessing novel thioredoxin- like activity. Biochemistry, 1992. 31(40): p. 9752-9.

77. Nordlund, P., B.M. Sjoberg, and H. Eklund, Three-dimensional structure of the free radical protein of ribonucleotide reductase. Nature, 1990. 345(6276): p. 593- 8.

78. Yee, C.S., et al., 2,3-difluorotyrosine at position 356 of ribonucleotide reductase R2: a probe of long-range proton-coupled electron transfer. J Am Chem Soc, 2003. 125(35): p. 10506-7.

79. Yee, C.S., et al., Generation of the R2 subunit of ribonucleotide reductase by intein chemistry: insertion of 3-nitrotyrosine at residue 356 as a probe of the radical initiation process. Biochemistry, 2003. 42(49): p. 14541-52.

80. Ekberg, M., et al., Two conserved tyrosine residues in protein R1 participate in an intermolecular electron transfer in ribonucleotide reductase. J Biol Chem, 1996. 271(34): p. 20655-9.

81. Stubbe, J. and D. Ackles, On the mechanism of ribonucleoside diphosphate reductase from Escherichia coli. Evidence for 3'-C--H bond cleavage. J Biol Chem, 1980. 255(17): p. 8027-30.

82. Ator, M.A. and J. Stubbe, Mechanism of inactivation of Escherichia coli ribonucleotide reductase by 2'-chloro-2'-deoxyuridine 5'-diphosphate: evidence for generation of a 2'-deoxy-3'-ketonucleotide via a net 1,2 hydrogen shift. Biochemistry, 1985. 24(25): p. 7214-21.

83. Eriksson, M., et al., Binding of allosteric effectors to ribonucleotide reductase protein R1: reduction of active-site cysteines promotes substrate binding. Structure, 1997. 5(8): p. 1077-92.

84. Lin, A.N., G.W. Ashley, and J. Stubbe, Location of the redox-active thiols of ribonucleotide reductase: sequence similarity between the Escherichia coli and Lactobacillus leichmannii enzymes. Biochemistry, 1987. 26(22): p. 6905-9.

85. Aberg, A., et al., Evidence for two different classes of redox-active cysteines in ribonucleotide reductase of Escherichia coli. J Biol Chem, 1989. 264(21): p. 12249-52.

207

86. Zipse, H., et al., Structure of the nucleotide radical formed during reaction of CDP/TTP with the E441Q-alpha2beta2 of E. coli ribonucleotide reductase. J Am Chem Soc, 2009. 131(1): p. 200-11.

87. Mao, S.S., et al., A model for the role of multiple cysteine residues involved in ribonucleotide reduction: amazing and still confusing. Biochemistry, 1992. 31(40): p. 9733-43.

88. Zhang, Z., et al., Role of the C terminus of the ribonucleotide reductase large subunit in enzyme regeneration and its inhibition by Sml1. Proc Natl Acad Sci U S A, 2007. 104(7): p. 2217-22.

89. Gon, S., M.J. Faulkner, and J. Beckwith, In vivo requirement for glutaredoxins and thioredoxins in the reduction of the ribonucleotide reductases of Escherichia coli. Antioxid Redox Signal, 2006. 8(5-6): p. 735-42.

90. Holmgren, A. and R. Sengupta, The use of thiols by ribonucleotide reductase. Free Radic Biol Med. 49(11): p. 1617-28.

91. Zahedi Avval, F. and A. Holmgren, Molecular mechanisms of thioredoxin and glutaredoxin as hydrogen donors for Mammalian s phase ribonucleotide reductase. J Biol Chem, 2009. 284(13): p. 8233-40.

92. Mulliez, E., et al., Activation of class III ribonucleotide reductase by flavodoxin: a protein radical-driven electron transfer to the iron-sulfur center. Biochemistry, 2001. 40(12): p. 3730-6.

93. Holmgren, A., Hydrogen donor system for Escherichia coli ribonucleoside- diphosphate reductase dependent upon glutathione. Proc Natl Acad Sci U S A, 1976. 73(7): p. 2275-9.

94. Lillig, C.H., C. Berndt, and A. Holmgren, Glutaredoxin systems. Biochim Biophys Acta, 2008. 1780(11): p. 1304-17.

95. Lillig, C.H. and A. Holmgren, Thioredoxin and related molecules--from biology to health and disease. Antioxid Redox Signal, 2007. 9(1): p. 25-47.

96. Gallardo-Madueno, R., et al., In vivo transcription of nrdAB operon and of grxA and fpg genes is triggered in Escherichia coli lacking both thioredoxin and glutaredoxin 1 or thioredoxin and glutathione, respectively. J Biol Chem, 1998. 273(29): p. 18382-8.

97. Holmgren, A. and R. Sengupta, The use of thiols by ribonucleotide reductase. Free Radic Biol Med, 2010. 49(11): p. 1617-28.

98. Holmgren, A., Thioredoxin. Annu Rev Biochem, 1985. 54: p. 237-71.

208

99. Eriksson, S., L. Thelander, and M. Akerman, Allosteric regulation of calf thymus ribonucleoside diphosphate reductase. Biochemistry, 1979. 18(14): p. 2948-52.

100. Thelander, L., S. Eriksson, and M. Akerman, Ribonucleotide reductase from calf thymus. Separation of the enzyme into two nonidentical subunits, proteins M1 and M2. J Biol Chem, 1980. 255(15): p. 7426-32.

101. Meuth, M. and H. Green, Induction of a deoxycytidineless state in cultured mammalian cells by bromodeoxyuridine. Cell, 1974. 2(2): p. 109-12.

102. Ullman, B., et al., Deoxyadenosine metabolism and cytotoxicity in cultured mouse T lymphoma cells: a model for immunodeficiency disease. Cell, 1978. 14(2): p. 365-75.

103. Gudas, L.J., et al., Deoxyguanosine toxicity in a mouse T lymphoma: relationship to purine nucleoside phosphorylase-associated immune dysfunction. Cell, 1978. 14(3): p. 531-8.

104. Meuth, M., E. Aufreiter, and P. Reichard, Deoxyribonucleotide pools in mouse- fibroblast cell lines with altered ribonucleotide reductase. Eur J Biochem, 1976. 71(1): p. 39-43.

105. Skoog, L. and B. Nordenskjold, Effects of hydroxyurea and 1-beta-D- arabinofuranosyl-cytosine on deoxyribonucleotide pools in mouse embryo cells. Eur J Biochem, 1971. 19(1): p. 81-9.

106. Bjursell, G. and P. Reichard, Effects of thymidine on deoxyribonucleoside triphosphate pools and deoxyribonucleic acid synthesis in Chinese hamster ovary cells. J Biol Chem, 1973. 248(11): p. 3904-9.

107. Eriksson, S., et al., Evidence for genetically independent allosteric regulatory domains of the protein M1 subunit of mouse ribonucleotide reductase. J Biol Chem, 1981. 256(19): p. 10193-7.

108. Fairman, J.W., et al., Structural basis for allosteric regulation of human ribonucleotide reductase by nucleotide-induced oligomerization. Nat Struct Mol Biol, 2011. 18(3): p. 316-22.

109. Hofer, A., et al., Cloning and characterization of the R1 and R2 subunits of ribonucleotide reductase from Trypanosoma brucei. Proc Natl Acad Sci U S A, 1997. 94(13): p. 6959-64.

110. Eliasson, R., et al., Allosteric control of three B12-dependent (class II) ribonucleotide reductases. Implications for the evolution of ribonucleotide reduction. J Biol Chem, 1999. 274(11): p. 7182-9.

209

111. Bateman, A., et al., The Pfam protein families database. Nucleic Acids Res, 2004. 32(Database issue): p. D138-41.

112. Chabes, A., et al., Survival of DNA Damage in Yeast Directly Depends on Increased dNTP Levels Allowed by Relaxed Feedback Inhibition of Ribonucleotide Reductase. Cell, 2003. 112: p. 391-401.

113. Chimploy, K. and C.K. Mathews, Mouse ribonucleotide reductase control: influence of substrate binding upon interactions with allosteric effectors. J Biol Chem, 2001. 276(10): p. 7093-100.

114. Kashlan, O.B., et al., A Comprehensive Model for the Allosteric Regulation of Mammalian Ribonucleotide Reductase. Functional Consequences of ATP- and dATP-Induced Oligomerization of the Large Subunit. Biochemistry, 2002. 41: p. 462-474.

115. Ormo, M. and B.M. Sjoberg, An ultrafiltration assay for nucleotide binding to ribonucleotide reductase. Anal Biochem, 1990. 189(1): p. 138-41.

116. Soderman, K. and P. Reichard, A nitrocellulose filter binding assay for ribonucleotide reductase. Anal Biochem, 1986. 152(1): p. 89-93.

117. Reichard, P., et al., Cross-talk between the allosteric effector-binding sites in mouse ribonucleotide reductase. J Biol Chem, 2000. 275(42): p. 33021-6.

118. Birgander, P.L., A. Kasrayan, and B.M. Sjoberg, Mutant R1 proteins from Escherichia coli class Ia ribonucleotide reductase with altered responses to dATP inhibition. J Biol Chem, 2004. 279(15): p. 14496-501.

119. Thelander, L., Physicochemical characterization of ribonucleoside diphosphate reductase from Escherichia coli. J Biol Chem, 1973. 248(13): p. 4591-601.

120. Kasrayan, A., et al., Enhancement by effectors and substrate nucleotides of R1-R2 interactions in Escherichia coli class Ia ribonucleotide reductase. J Biol Chem, 2004. 279(30): p. 31050-7.

121. Ingemarson, R. and L. Thelander, A kinetic study on the influence of nucleoside triphosphate effectors on subunit interaction in mouse ribonucleotide reductase. Biochemistry, 1996. 35(26): p. 8603-9.

122. Ando, N., et al., Structural interconversions modulate activity of Escherichia coli ribonucleotide reductase. Proc Natl Acad Sci U S A, 2011. 108(52): p. 21046-51.

123. Wang, J., G.J. Lohman, and J. Stubbe, Mechanism of inactivation of human ribonucleotide reductase with p53R2 by gemcitabine 5'-diphosphate. Biochemistry, 2009. 48(49): p. 11612-21.

210

124. Aye, Y. and J. Stubbe, Clofarabine 5'-di and -triphosphates inhibit human ribonucleotide reductase by altering the quaternary structure of its large subunit. Proc Natl Acad Sci U S A, 2011. 108(24): p. 9815-20.

125. Uppsten, M., et al., Structure of the large subunit of class Ib ribonucleotide reductase from Salmonella typhimurium and its complexes with allosteric effectors. J Mol Biol, 2003. 330(1): p. 87-97.

126. Logan, D.T., et al., A Glycyl Radical Site in the Crystal Structure of a Class III Ribonucleotide Reductase. Science, 1999. 283: p. 1499-1504.

127. Kumar, D., et al., Highly mutagenic and severely imbalanced dNTP pools can escape detection by the S-phase checkpoint. Nucleic Acids Res. 38(12): p. 3975- 83.

128. Larsson, K.M., et al., Structural basis for allosteric substrate specificity regulation in anaerobic ribonucleotide reductases. Structure (Camb), 2001. 9(8): p. 739-50.

129. Elledge, S.J. and R.W. Davis, Two genes differentially regulated in the cell cycle and by DNA-damaging agents encode alternative regulatory subunits of ribonucleotide reductase. Genes Dev, 1990. 4(5): p. 740-51.

130. Elledge, S.J. and R.W. Davis, Identification and isolation of the gene encoding the small subunit of ribonucleotide reductase from Saccharomyces cerevisiae: DNA damage-inducible gene required for mitotic viability. Mol Cell Biol, 1987. 7(8): p. 2783-93.

131. Hurd, H.K., C.W. Roberts, and J.W. Roberts, Identification of the gene for the yeast ribonucleotide reductase small subunit and its inducibility by methyl methanesulfonate. Mol Cell Biol, 1987. 7(10): p. 3673-7.

132. Huang, M. and S.J. Elledge, Identification of RNR4, encoding a second essential small subunit of ribonucleotide reductase in Saccharomyces cerevisiae. Mol Cell Biol, 1997. 17(10): p. 6105-13.

133. Domkin, V., L. Thelander, and A. Chabes, Yeast DNA Damage-inducible Rnr3 Has a Very Low Catalytic Activity Strongly Stimulated after the Formation of a Cross-talking Rnr1/Rnr3 Complex. J Biol Chem, 2002. 277(21): p. 18574-8.

134. de la Torre Ruiz, M.A. and N.F. Lowndes, DUN1 defines one branch downstream of RAD53 for transcription and DNA damage repair in Saccharomyces cerevisiae. FEBS Lett, 2000. 485(2-3): p. 205-6.

211

135. Elledge, S.J. and R.W. Davis, Identification of the DNA damage-responsive element of RNR2 and evidence that four distinct cellular factors bind it. Mol Cell Biol, 1989. 9(12): p. 5373-86.

136. Wang, P.J., et al., Rnr4p, a novel ribonucleotide reductase small-subunit protein. Mol Cell Biol, 1997. 17(10): p. 6114-21.

137. Nguyen, H.H., et al., Purification of ribonucleotide reductase subunits Y1, Y2, Y3, and Y4 from yeast: Y4 plays a key role in diiron cluster assembly. Proc Natl Acad Sci U S A, 1999. 96(22): p. 12339-44.

138. Chabes, A., et al., Yeast ribonucleotide reductase has a heterodimeric iron- radical-containing subunit. Proc Natl Acad Sci U S A, 2000. 97(6): p. 2474-9.

139. Sommerhalter, M., et al., Structures of the yeast ribonucleotide reductase Rnr2 and Rnr4 homodimers. Biochemistry, 2004. 43(24): p. 7736-42.

140. Voegtli, W.C., et al., Structure of the yeast ribonucleotide reductase Y2Y4 heterodimer. Proc Natl Acad Sci U S A, 2001. 98(18): p. 10073-8.

141. Koch, C., et al., A role for the transcription factors Mbp1 and Swi4 in progression from G1 to S phase. Science, 1993. 261(5128): p. 1551-7.

142. Koc, A., et al., Replication-independent MCB gene induction and deoxyribonucleotide accumulation at G1/S in Saccharomyces cerevisiae. J Biol Chem, 2003. 278(11): p. 9345-52.

143. McIntosh, E.M., MCB elements and the regulation of DNA replication genes in yeast. Curr Genet, 1993. 24(3): p. 185-92.

144. McIntosh, E.M., Expression of DNA replication genes in the yeast cell cycle. Mutat Res, 1993. 289(1): p. 61-72.

145. Spellman, P.T., et al., Comprehensive identification of cell cycle-regulated genes of the yeast Saccharomyces cerevisiae by microarray hybridization. Mol Biol Cell, 1998. 9(12): p. 3273-97.

146. Zhou, Z. and S.J. Elledge, DUN1 encodes a protein kinase that controls the DNA damage response in yeast. Cell, 1993. 75(6): p. 1119-27.

147. Elledge, S.J., Cell cycle checkpoints: preventing an identity crisis. Science, 1996. 274(5293): p. 1664-72.

148. Allen, J.B., et al., The SAD1/RAD53 protein kinase controls multiple checkpoints and DNA damage-induced transcription in yeast. Genes Dev, 1994. 8(20): p. 2401-15.

212

149. Weinert, T.A., G.L. Kiser, and L.H. Hartwell, Mitotic checkpoint genes in budding yeast and the dependence of mitosis on DNA replication and repair. Genes Dev, 1994. 8(6): p. 652-65.

150. Tsaponina, O., et al., Ixr1 is required for the expression of the ribonucleotide reductase Rnr1 and maintenance of dNTP pools. PLoS Genet, 2011. 7(5): p. e1002061.

151. Huang, M., Z. Zhou, and S.J. Elledge, The DNA replication and damage checkpoint pathways induce transcription by inhibition of the Crt1 repressor. Cell, 1998. 94(5): p. 595-605.

152. Fernandez Sarabia, M.J., et al., The cell cycle genes cdc22+ and suc22+ of the fission yeast Schizosaccharomyces pombe encode the large and small subunits of ribonucleotide reductase. Mol Gen Genet, 1993. 238(1-2): p. 241-51.

153. Harris, P., et al., Cell cycle, DNA damage and heat shock regulate suc22+ expression in fission yeast. Mol Gen Genet, 1996. 252(3): p. 284-91.

154. McInerny, C.J., Cell cycle-regulated transcription in fission yeast. Biochem Soc Trans, 2004. 32(Pt 6): p. 967-72.

155. Cunliffe, L., S. White, and C.J. McInerny, DSC1-MCB regulation of meiotic transcription in Schizosaccharomyces pombe. Mol Genet Genomics, 2004. 271(1): p. 60-71.

156. Engstrom, Y., Eriksson, S., Jildevik, I., Skog, S., Thelander, L., and Tribukait, B., Cell cycle-dependent expression of mammalian ribonucleotide reductase. Differential regulation of the two subunits. J. Biol. Chem, 1985. 260(16): p. 9114- 9116.

157. Bjorklund, S., Skog, S., Tribukait, B, and Thelander, L, S-phase-specific expression of mammalian ribonucleotide reductase R1 and R2 subunit mRNAs. Biochemistry, 1990. 29(23): p. 5452-5458.

158. Bjorklund, S., et al., Structure and promoter characterization of the gene encoding the large subunit (R1 protein) of mouse ribonucleotide reductase. Proc Natl Acad Sci U S A, 1993. 90(23): p. 11322-6.

159. Johansson, E., E. Skogman, and L. Thelander, The TATA-less promoter of mouse ribonucleotide reductase R1 gene contains a TFII-I binding initiator element essential for cell cycle-regulated transcription. J Biol Chem, 1995. 270(50): p. 30162-7.

213

160. Johansson, E., K. Hjortsberg, and L. Thelander, Two YY-1-binding proximal elements regulate the promoter strength of the TATA-less mouse ribonucleotide reductase R1 gene. J Biol Chem, 1998. 273(45): p. 29816-21.

161. Chabes, A.L., S. Bjorklund, and L. Thelander, S Phase-specific transcription of the mouse ribonucleotide reductase R2 gene requires both a proximal repressive E2F-binding site and an upstream promoter activating region. J Biol Chem, 2004. 279(11): p. 10796-807.

162. Filatov, D. and L. Thelander, Role of a proximal NF-Y binding promoter element in S phase-specific expression of mouse ribonucleotide reductase R2 gene. J Biol Chem, 1995. 270(42): p. 25239-43.

163. Engstrom, Y., et al., Cell cycle-dependent expression of mammalian ribonucleotide reductase. Differential regulation of the two subunits. J Biol Chem, 1985. 260(16): p. 9114-6.

164. Mann, G.J., et al., Ribonucleotide reductase M1 subunit in cellular proliferation, quiescence, and differentiation. Cancer Res, 1988. 48(18): p. 5151-6.

165. Guittet, O., et al., Mammalian p53R2 protein forms an active ribonucleotide reductase in vitro with the R1 protein, which is expressed both in resting cells in response to DNA damage and in proliferating cells. J Biol Chem, 2001. 276(44): p. 40647-51.

166. Chabes, A. and L. Thelander, Controlled protein degradation regulates ribonucleotide reductase activity in proliferating mammalian cells during the normal cell cycle and in response to DNA damage and replication blocks. J Biol Chem, 2000. 275(23): p. 17747-53.

167. Tanaka, H., Arakawa,H.,Yamaguchi,T.,Shiraishi,K.,Fukuda,S.,Matsui,K.,Takei,Y., and Y. Nakamura, A ribonucleotide reductase gene involved in a p53-dependent cell- cycle checkpoint for DNA damage. Nature (London), 2000. 404: p. 42-49.

168. Nakano, K., et al., A ribonucleotide reductase gene is a transcriptional target of p53 and p73. Oncogene, 2000. 19(37): p. 4283-9.

169. Hakansson, P., A. Hofer, and L. Thelander, Regulation of mammalian ribonucleotide reduction and dNTP pools after DNA damage and in resting cells. J Biol Chem, 2006. 281(12): p. 7834-41.

170. Yao, R., et al., Subcellular localization of yeast ribonucleotide reductase regulated by the DNA replication and damage checkpoint pathways. Proc Natl Acad Sci U S A, 2003. 100(11): p. 6628-33.

214

171. Lee, Y.D. and S.J. Elledge, Control of ribonucleotide reductase localization through an anchoring mechanism involving Wtm1. Genes Dev, 2006. 20(3): p. 334-44.

172. Zhang, Z., et al., Nuclear localization of the Saccharomyces cerevisiae ribonucleotide reductase small subunit requires a karyopherin and a WD40 repeat protein. Proc Natl Acad Sci U S A, 2006. 103(5): p. 1422-7.

173. Wu, X. and M. Huang, Dif1 controls subcellular localization of ribonucleotide reductase by mediating nuclear import of the R2 subunit. Mol Cell Biol, 2008. 28(23): p. 7156-67.

174. Zheng, P., et al., SPK1 is an essential S-phase-specific gene of Saccharomyces cerevisiae that encodes a nuclear serine/threonine/tyrosine kinase. Mol Cell Biol, 1993. 13(9): p. 5829-42.

175. Kato, R. and H. Ogawa, An essential gene, ESR1, is required for mitotic cell growth, DNA repair and meiotic recombination in Saccharomyces cerevisiae. Nucleic Acids Res, 1994. 22(15): p. 3104-12.

176. Zhao, X., E.G. Muller, and R. Rothstein, A suppressor of two essential checkpoint genes identifies a novel protein that negatively affects dNTP pools. Mol Cell, 1998. 2(3): p. 329-40.

177. Chabes, A., V. Domkin, and L. Thelander, Yeast Sml1, a protein inhibitor of ribonucleotide reductase. J Biol Chem, 1999. 274(51): p. 36679-83.

178. Zhao, X., et al., Mutational and structural analyses of the ribonucleotide reductase inhibitor Sml1 define its Rnr1 interaction domain whose inactivation allows suppression of mec1 and rad53 lethality. Mol Cell Biol, 2000. 20(23): p. 9076-83.

179. Craven, R.J. and T.D. Petes, Involvement of the checkpoint protein Mec1p in silencing of gene expression at telomeres in Saccharomyces cerevisiae. Mol Cell Biol, 2000. 20(7): p. 2378-84.

180. Zhao, X. and R. Rothstein, The Dun1 checkpoint kinase phosphorylates and regulates the ribonucleotide reductase inhibitor Sml1. Proc Natl Acad Sci U S A, 2002. 99(6): p. 3746-51.

181. Uchiki, T., et al., Identification of phosphorylation sites on the yeast ribonucleotide reductase inhibitor Sml1. J Biol Chem, 2004. 279(12): p. 11293- 303.

215

182. Andreson, B.L., et al., The ribonucleotide reductase inhibitor, Sml1, is sequentially phosphorylated, ubiquitylated and degraded in response to DNA damage. Nucleic Acids Res, 2010. 38(19): p. 6490-501.

183. Gupta, V., et al., Sml1p is a dimer in solution: characterization of denaturation and renaturation of recombinant Sml1p. Biochemistry, 2004. 43(26): p. 8568-78. 184. Woollard, A., G. Basi, and P. Nurse, A novel S phase inhibitor in fission yeast. EMBO J, 1996. 15(17): p. 4603-12.

185. Borgne, A. and P. Nurse, The Spd1p S phase inhibitor can activate the DNA replication checkpoint pathway in fission yeast. J Cell Sci, 2000. 113 Pt 23: p. 4341-50.

186. Lee, Y.D., et al., Dif1 is a DNA-damage-regulated facilitator of nuclear import for ribonucleotide reductase. Mol Cell, 2008. 32(1): p. 70-80.

187. Liu, C., et al., Cop9/signalosome subunits and Pcu4 regulate ribonucleotide reductase by both checkpoint-dependent and -independent mechanisms. Genes Dev, 2003. 17(9): p. 1130-40.

188. Nestoras, K., et al., Regulation of ribonucleotide reductase by Spd1 involves multiple mechanisms. Genes Dev, 2010. 24(11): p. 1145-59.

189. Chabes, A.L., et al., Mouse ribonucleotide reductase R2 protein: a new target for anaphase-promoting complex-Cdh1-mediated proteolysis. Proc Natl Acad Sci U S A, 2003. 100(7): p. 3925-9.

190. Pfleger, C.M., et al., Inhibition of Cdh1-APC by the MAD2-related protein MAD2L2: a novel mechanism for regulating Cdh1. Genes Dev, 2001. 15(14): p. 1759-64.

191. Fan, H., C. Villegas, and J.A. Wright, Ribonucleotide reductase R2 component is a novel malignancy determinant that cooperates with activated oncogenes to determine transformation and malignant potential. Proc Natl Acad Sci U S A, 1996. 93(24): p. 14036-40.

192. Wnuk, S.F. and M.J. Robins, Ribonucleotide reductase inhibitors as anti-herpes agents. Antiviral Res, 2006. 71(2-3): p. 122-6.

193. Cerqueira, N.M., P.A. Fernandes, and M.J. Ramos, Ribonucleotide reductase: a critical enzyme for cancer chemotherapy and antiviral agents. Recent Pat Anticancer Drug Discov, 2007. 2(1): p. 11-29.

194. Juhasz, A., et al., Analysis of ribonucleotide reductase M2 mRNA levels in patient samples after GTI-2040 antisense drug treatment. Oncol Rep, 2006. 15(5): p. 1299-304.

216

195. Lee, Y., et al., GTI-2501, an antisense agent targeting R1, the large subunit of human ribonucleotide reductase, shows potent anti-tumor activity against a variety of tumors. Int J Oncol, 2006. 28(2): p. 469-78.

196. Xu, H., et al., The structural basis for peptidomimetic inhibition of eukaryotic ribonucleotide reductase: a conformationally flexible pharmacophore. J Med Chem, 2008. 51(15): p. 4653-9.

197. Sun, D., et al., Structure-Based Design, Synthesis, and Evaluation of 2'-(2- Hydroxyethyl)-2'-deoxyadenosine and the 5'-Diphosphate Derivative as Ribonucleotide Reductase Inhibitors. ChemMedChem, 2009. 4(10): p. 1649-56.

198. Huang, P., S. Chubb, and W. Plunkett, Termination of DNA synthesis by 9-beta- D-arabinofuranosyl-2-fluoroadenine. A mechanism for cytotoxicity. J Biol Chem, 1990. 265(27): p. 16617-25.

199. Avery, T.L., et al., Biochemical pharmacology of 2-chlorodeoxyadenosine in malignant human hematopoietic cell lines and therapeutic effects of 2- bromodeoxyadenosine in drug combinations in mice. Cancer Res, 1989. 49(18): p. 4972-8.

200. Griffig, J., R. Koob, and R.L. Blakley, Mechanisms of inhibition of DNA synthesis by 2-chlorodeoxyadenosine in human lymphoblastic cells. Cancer Res, 1989. 49(24 Pt 1): p. 6923-8.

201. Parker, W.B., et al., Effects of 2-chloro-9-(2-deoxy-2-fluoro-beta-D- arabinofuranosyl)adenine on K562 cellular metabolism and the inhibition of human ribonucleotide reductase and DNA polymerases by its 5'-triphosphate. Cancer Res, 1991. 51(9): p. 2386-94.

202. Faderl, S., et al., The role of clofarabine in hematologic and solid malignancies-- development of a next-generation nucleoside analog. Cancer, 2005. 103(10): p. 1985-95.

203. Gandhi, V., Clofarabine : a viewpoint by varsha gandhi. Paediatr Drugs, 2005. 7(4): p. 265-6.

204. Langer, C.J., The role of new agents in advanced non-small-cell lung carcinoma. Curr Oncol Rep, 2000. 2(1): p. 76-89.

205. Squadroni, M. and N. Fazio, Chemotherapy in pancreatic adenocarcinoma. Eur Rev Med Pharmacol Sci, 2010. 14(4): p. 386-94.

206. Liou, J.Y., et al., Characterization of human UMP/CMP kinase and its phosphorylation of D- and L-form deoxycytidine analogue monophosphates. Cancer Res, 2002. 62(6): p. 1624-31.

217

207. Hu, R., et al., UMP/CMPK is not the critical enzyme in the metabolism of pyrimidine ribonucleotide and activation of deoxycytidine analogs in human RKO cells. PLoS One, 2011. 6(5): p. e19490.

208. Mini, E., et al., Cellular pharmacology of gemcitabine. Ann Oncol, 2006. 17 Suppl 5: p. v7-12.

209. Flanagan, S.A., et al., Mismatched nucleotides as the lesions responsible for radiosensitization with gemcitabine: a new paradigm for antimetabolite radiosensitizers. Mol Cancer Ther, 2007. 6(6): p. 1858-68.

210. Xu, H., et al., Structures of eukaryotic ribonucleotide reductase I define gemcitabine diphosphate binding and subunit assembly. Proc Natl Acad Sci U S A, 2006. 103(11): p. 4028-33.

211. Stearns, B., K.A. Losee, and J. Bernstein, Hydroxyurea. A New Type of Potential Antitumor Agent. J Med Chem, 1963. 6: p. 201.

212. Donehower, R.C., An overview of the clinical experience with hydroxyurea. Semin Oncol, 1992. 19(3 Suppl 9): p. 11-9.

213. Stevens, M.R., Hydroxyurea: an overview. J Biol Regul Homeost Agents, 1999. 13(3): p. 172-5.

214. Spivak, J.L. and H. Hasselbalch, Hydroxycarbamide: a user's guide for chronic myeloproliferative disorders. Expert Rev Anticancer Ther, 2011. 11(3): p. 403-14.

215. Finch, R.A., et al., Triapine (3-aminopyridine-2-carboxaldehyde- thiosemicarbazone): A potent inhibitor of ribonucleotide reductase activity with broad spectrum antitumor activity. Biochem Pharmacol, 2000. 59(8): p. 983-91.

216. Yamasaki, T., S. Terai, and I. Sakaida, Deferoxamine for advanced hepatocellular carcinoma. N Engl J Med, 2011. 365(6): p. 576-8.

217. Nyholm, S., et al., Role of ribonucleotide reductase in inhibition of mammalian cell growth by potent iron chelators. J Biol Chem, 1993. 268(35): p. 26200-5.

218. Cooper, C.E., et al., The relationship of intracellular iron chelation to the inhibition and regeneration of human ribonucleotide reductase. J Biol Chem, 1996. 271(34): p. 20291-9.

219. Zhu, L., et al., Inhibitory mechanisms of heterocyclic carboxaldehyde thiosemicabazones for two forms of human ribonucleotide reductase. Biochem Pharmacol, 2009. 78(9): p. 1178-85.

218

220. Kunos, C.A., et al., Ribonucleotide reductase inhibition enhances chemoradiosensitivity of human cervical cancers. Radiat Res, 2011. 174(5): p. 574-81.

221. Kunos, C.A., et al., Radiosensitization of Human Cervical Cancer Cells by Inhibiting Ribonucleotide Reductase: Enhanced Radiation Response at Low-Dose Rates. Int J Radiat Oncol Biol Phys, 2011.

222. Kolesar, J.M., et al., Cytotoxic Evaluation of 3-Aminopyridine-2-Carboxaldehyde Thiosemicarbazone, 3-AP, in Peripheral Blood Lymphocytes of Patients with Refractory Solid Tumors using Electron Paramagnetic Resonance. Exp Ther Med, 2011. 2(1): p. 119-123.

223. Kunos, C.A., et al., Phase I trial of pelvic radiation, weekly cisplatin, and 3- aminopyridine-2-carboxaldehyde thiosemicarbazone (3-AP, NSC #663249) for locally advanced cervical cancer. Clin Cancer Res, 2011. 16(4): p. 1298-306.

224. Choi, B.S., et al., The maximum tolerated dose and biologic effects of 3- aminopyridine-2-carboxaldehyde thiosemicarbazone (3-AP) in combination with irinotecan for patients with refractory solid tumors. Cancer Chemother Pharmacol, 2010. 66(5): p. 973-80.

225. Kolesar, J., et al., Population pharmacokinetics of 3-aminopyridine-2- carboxaldehyde thiosemicarbazone (Triapine(R)) in cancer patients. Cancer Chemother Pharmacol, 2011. 67(2): p. 393-400.

226. Kauffman, C.A., Histoplasmosis: a clinical and laboratory update. Clin Microbiol Rev, 2007. 20(1): p. 115-32.

227. Cappelletty, D. and K. Eiselstein-McKitrick, The echinocandins. Pharmacotherapy, 2007. 27(3): p. 369-88.

228. Kauffman, C.A., et al., Posaconazole. Nat Rev Drug Discov, 2007. 6(3): p. 183-4.

229. Polak, A. and H.J. Scholer, Mode of action of 5-fluorocytosine and mechanisms of resistance. Chemotherapy, 1975. 21(3-4): p. 113-30.

230. Yang, F.D., et al., The carboxyl terminus heptapeptide of the R2 subunit of mammalian ribonucleotide reductase inhibits enzyme activity and can be used to purify the R1 subunit. FEBS Lett, 1990. 272(1-2): p. 61-4.

231. Pace, C.N., et al., How to measure and predict the molar absorption coefficient of a protein. Protein Sci, 1995. 4(11): p. 2411-23.

232. Parkin, S.E., et al., Electron injection through a specific pathway determines the outcome of oxygen activation at the diiron cluster in the F208Y mutant of

219

Escherichia coli ribonucleotide reductase protein R2. Biochemistry, 1998. 37(4): p. 1124-30.

233. Ge, J., et al., Why multiple small subunits (Y2 and Y4) for yeast ribonucleotide reductase? Toward understanding the role of Y4. Proc Natl Acad Sci U S A, 2001. 98(18): p. 10067-72.

234. Studier, F.W., Protein production by auto-induction in high density shaking cultures. Protein Expr Purif, 2005. 41(1): p. 207-34.

235. Riemer, J., et al., Colorimetric ferrozine-based assay for the quantitation of iron in cultured cells. Anal Biochem, 2004. 331(2): p. 370-5.

236. Ihaka, R. and R. Gentleman, R:a language for data analysis and graphics. Journal of Computational and graphical statistics, 1996. 5: p. 299-314.

237. Xu, G. and M.R. Chance, Hydroxyl radical-mediated modification of proteins as probes for structural proteomics. Chem Rev, 2007. 107(8): p. 3514-43.

238. Gupta, S., et al., The Beamline X28C of the Center for Synchrotron Biosciences: a national resource for biomolecular structure and dynamics experiments using synchrotron footprinting. J Synchrotron Radiat, 2007. 14(Pt 3): p. 233-43.

239. Kiselar, J.G., et al., Visualizing Arp2/3 complex activation mediated by binding of ATP and WASp using structural mass spectrometry. Proc Natl Acad Sci U S A, 2007. 104(5): p. 1552-7.

240. Kamal, J.K., et al., Three-dimensional structure of cofilin bound to monomeric actin derived by structural mass spectrometry data. Proc Natl Acad Sci U S A, 2007. 104(19): p. 7910-5.

241. Takamoto, K. and M.R. Chance, Radiolytic protein footprinting with mass spectrometry to probe the structure of macromolecular complexes. Annu Rev Biophys Biomol Struct, 2006. 35: p. 251-76.

242. Sinz, A., Chemical cross-linking and mass spectrometry to map three- dimensional protein structures and protein-protein interactions. Mass Spectrom Rev, 2006. 25(4): p. 663-82.

243. Xu, H. and M.A. Freitas, MassMatrix: a database search program for rapid characterization of proteins and peptides from tandem mass spectrometry data. Proteomics, 2009. 9(6): p. 1548-55.

244. Uppsten, M., et al., The first holocomplex structure of ribonucleotide reductase gives new insight into its mechanism of action. J Mol Biol, 2006. 359(2): p. 365- 77.

220

245. Zhang, X., et al., An allosteric mechanism for activation of the kinase domain of epidermal growth factor receptor. Cell, 2006. 125(6): p. 1137-49.

246. Fromm, H.J., Reversible enzyme inhibitors as mechanistic probes. Methods Enzymol, 1995. 249: p. 123-43.

247. Akaike, H., A new look at the statistical model identification. IEEE Transactions on Automatic Control, 1974. 19: p. 716-723.

248. Chabes, A., V. Domkin, and L. Thelander, Yeast Sml1, a protein inhibitor of ribonucleotide reductase. J Biol Chem, 1999. 274(51): p. 36679-83.

249. Wang, J.H., J.I. Tu, and F.M. Lo, Effect of glucose 6-phosphate on the nucleotide site of glycogen phosphorylase b. A general approach for negative heterotropic interactions. J Biol Chem, 1970. 245(12): p. 3115-21.

250. Bauer, M., H. Griengl, and W. Steiner, Kinetic studies on the enzyme (S)- hydroxynitrile lyase from hevea brasiliensis using initial rate methods and progress curve analysis. Biotechnol Bioeng, 1999. 62(1): p. 20-9.

251. Sinz, A., Chemical cross-linking and mass spectrometry for mapping three- dimensional structures of proteins and protein complexes. J Mass Spectrom, 2003. 38(12): p. 1225-37.

252. Georgieva, B., X. Zhao, and R. Rothstein, Damage response and dNTP regulation: the interaction between ribonucleotide reductase and its inhibitor, Sml1. Cold Spring Harb Symp Quant Biol, 2000. 65: p. 343-6.

253. Yin, J., et al., Single-cell FRET imaging of transferrin receptor trafficking dynamics by Sfp-catalyzed, site-specific protein labeling. Chem Biol, 2005. 12(9): p. 999-1006.

254. Liu, W., et al., Genetic incorporation of unnatural amino acids into proteins in mammalian cells. Nat Methods, 2007. 4(3): p. 239-44.

255. Kajihara, D., et al., FRET analysis of protein conformational change through position-specific incorporation of fluorescent amino acids. Nat Methods, 2006. 3(11): p. 923-9.

256. Suchanek, M., A. Radzikowska, and C. Thiele, Photo-leucine and photo- methionine allow identification of protein-protein interactions in living cells. Nat Methods, 2005. 2(4): p. 261-7.

257. Rust, M.J., M. Bates, and X. Zhuang, Sub-diffraction-limit imaging by stochastic optical reconstruction microscopy (STORM). Nat Methods, 2006. 3(10): p. 793-5.

221

258. Kastner, B., et al., GraFix: sample preparation for single-particle electron cryomicroscopy. Nat Methods, 2008. 5(1): p. 53-5.

259. Zuiderweg, E.R., Mapping protein-protein interactions in solution by NMR spectroscopy. Biochemistry, 2002. 41(1): p. 1-7.

222