Interactions of Dna Polymerase Theta and Ku70/80 With

INTERACTIONS OF DNA POLYMERASE THETA AND KU70/80 WITH

OXIDATIVE DNA DAMAGE

by Daniel Laverty

A dissertation submitted to Johns Hopkins University in conformity with the requirements for the degree of Doctor of Philosophy

Baltimore, MD Submitted June 2018

Abstract

Oxidized abasic sites (L, C4-AP, and DOB) are formed by ionizing radiation, reactive oxygen species, and some chemotherapeutics. Like abasic sites (AP), these lesions are cytotoxic and mutagenic and must be repaired, primarily by base excision repair (BER).

If left unrepaired, abasic lesions stall replication and induce mutations. Repair of oxidized abasic lesions exhibits unique challenges, however. C4-AP and DOB inactivate the lyase activity of the repair enzymes DNA polymerase β and λ. Recently, several other enzymes were shown to possess lyase activity, allowing them to excise abasic lesions. Among these are DNA polymerase θ (Pol θ) and Ku70/80 (Ku). As Pol θ promotes resistance to cancer therapies which form oxidized abasic sites, the repair and replication of these lesions by

Pol θ is potentially important. Ku is a core factor for non-homologous end-joining and removes AP from double strand breaks (DSBs). The interaction of Ku with oxidized abasic sites near DSB termini is potentially important for the response to ionizing radiation, which is used as a cancer treatment.

Synthetic oligonucleotides containing abasic and oxidized abasic sites were prepared, and their repair or replication by Pol θ was analyzed. Pol θ bypasses C4-AP and

L with reduced efficiency relative to AP and has a strong propensity to induce frameshift mutations during bypass of AP, C4-AP, L, and the oxidized nucleobase, thymidine glycol.

Studies on the repair of C4-AP and DOB by Pol θ showed that Pol θ is inactivated by pC4-

AP, which covalently modifies Lys2383, but not by DOB. Site-directed mutagenesis showed that Lys2383 is essential for both polymerase and lyase activity of Pol θ. These results have unveiled the primary nucleophile (Lys2383) responsible for Pol θ lyase activity.

The repair of oxidized abasic sites by Ku was also analyzed. Ku exhibits differential repair capability on different oxidized abasic lesions. C4-AP and DOB are excised more efficiently than AP, yet Ku cannot excise L. Failure to remove this lesion can potentially inhibit repair of a DSB. Unlike Pol θ which was modified by pC4-AP, modification of Ku was not detected following repair of DOB or C4-AP.

Advisor: Prof. Marc M. Greenberg

Readers: Prof. Steven Rokita

Prof Rebecca Klausen

iii

Acknowledgements

I’m very thankful to all the people who have helped me through this journey. My parents, my brother, and my sister have always been extremely supportive, and I couldn’t have done it without them. Thanks to all my friends here in Baltimore and my friends back home. Thanks to all the members of the Greenberg lab, past and present, for their tremendous help. Special thanks to Liwei Weng and to Rakesh for their great help when I was getting started in the lab. Thanks very much to Kun for his help with cloning and protein purification and for helpful discussions. Thanks to Arnab and Josh for their friendship, support, and advice. I’m also very grateful to my friend Liwei Zheng who’s been here every step of the way. I couldn’t have asked for a better friend to have by my side on this journey. Thanks to Shelby and Marco for their support and for a lot of laughs along the way. Thanks to Sahil for support and encouragement. Thanks to all other members of the lab, past and present, for their support. I’m very thankful to Dr. Greenberg for his mentorship through this journey. I’m continually impressed by his dedication to ensuring that his students get the most out of their graduate experience. I’m especially thankful for his encouragement in pursuing my interests when they differed from our original goals. Thanks to the readers of my thesis, Dr. Rokita and Dr. Klausen. Thanks to

Dr. Townsend for encouragement and advice during my search for a post-doctoral position.

Table of Contents

Abstract ...... ii

Acknowledgements ...... iv

List of Tables ...... x

List of Figures...... xii

List of Abbreviations ...... xix

1. Introduction ...... 1

2. Background ...... 3

2.1. DNA Damage ...... 3

2.1.1 Abasic Sites ...... 4

2.1.2 Oxidative DNA Damage ...... 5

2.1.2.1 Thymidine Glycol ...... 6

2.1.2.2 Sugar Oxidation ...... 8

2.1.2.3 C4’-Oxidized Abasic Site ...... 9

2.1.2.4 2-Deoxyribonolactone ...... 11

2.2 Double Strand Breaks ...... 13

2.2.1 Double Strand Break Formation by Ionizing Radiation ...... 14

2.2.2 Double Strand Break Formation by Antitumor Antibiotics ...... 15

2.2.3 Repair of Double Strand Breaks ...... 17

2.2.3.1 Non-Homologous End Joining ...... 17

2.2.3.2 Homologous Recombination ...... 21

2.2.3.3 Alternative End Joining ...... 22

2.3.1 Mechanism of Base Excision Repair ...... 24

2.3.2 Substrate Scope of Base Excision Repair ...... 26

2.3.3 Glycosylases ...... 27

2.3.4 Ape1 ...... 29

2.3.5 Base Excision Repair Polymerases ...... 29

2.3.6 Base Excision Repair of Oxidized Abasic Sites ...... 32

2.4 DNA Damage Tolerance by Translesion Synthesis ...... 34

2.4.1 Mechanism of Translesion Synthesis ...... 35

2.4.2 Discovery of Translesion Synthesis Polymerases ...... 37

2.4.3 Translesion Synthesis Polymerases in Eukaryotes ...... 38

2.4.3.1 General Features of Translesion Synthesis Polymerases ...... 39

2.4.3.2 Rev1 ...... 41

2.4.3.3 Pol η ...... 42

2.4.3.4 Pol κ ...... 43

2.4.3.5 Pol ι ...... 45

2.4.3.6 Pol ζ ...... 46

2.4.3.7 Other Polymerases Involved in Translesion Synthesis ...... 47

2.4.4 Translesion Synthesis of Abasic Sites and Oxidized Abasic Sites ...... 49

2.4.5 Translesion Synthesis of Thymidine Glycol ...... 53

2.4.6 Regulation of Translesion Synthesis in Eukaryotes ...... 54

2.5 DNA Polymerase Theta (Pol θ) ...... 56

2.5.1 Discovery and Cloning ...... 56

2.5.2 Structure and Unique Characteristics of Pol θ ...... 58

2.5.3 Function of Pol θ ...... 59

2.5.4 Pol θ as a Cancer Target ...... 62

2.6 Ku70/80 ...... 63

2.6.1 Ku Structure ...... 64

2.6.2 Structural Role of Ku in NHEJ ...... 65

2.6.3 Ku Lyase Activity ...... 65

2.6.4 Ku Function at Telomeres ...... 67

2.6.5 Ku Removal from DNA ...... 68

3. Results and Discussion ...... 69

3.1 Bypass of Abasic and Oxidized Abasic Lesions by Pol θ...... 69

3.1.1 Expression and Purification of Pol θ Catalytic Core (residues 1792-2590)

...... 70

3.1.2 Pol θ Polymerase Active Site Titration ...... 72

3.1.3 Nucleotide Insertion Opposite Abasic Lesions by Pol θ ...... 73

3.1.4 Full-Length Bypass of Abasic and Oxidized Abasic Lesions by Pol θ ...... 76

vii

3.1.5 Formation of Deletions During Abasic Lesion Bypass ...... 79

3.1.6 Kinetic Analysis of Extension Past Abasic Lesions ...... 82

3.1.7 Formation of Deletions with dG Opposite AP and C4-AP ...... 90

3.1.8 Direct Detection of Deletions During AP bypass by DNA sequencing ...... 93

3.2 Bypass of Thymidine Glycol by Pol θ...... 104

3.2.1 Primer-template design and kinetic analysis of TLS for Tg...... 105

3.2.2 Steady-state analysis of Tg bypass: error-free extension and template

misalignment...... 109

3.2.3 Full-Length Bypass of Thymidine Glycol ...... 112

3.2.4 Direct Detection of Deletions During Thymidine Glycol Bypass...... 113

3.2.5 Implications for Translesion Synthesis by Pol θ ...... 115

3.3 Pol θ Lyase Activity on 5’-dRP, DOB, and pC4-AP ...... 118

3.3.1 Pol θ Lyase Active Site Titration ...... 119

3.3.2 Removal of 5’-dRP, DOB, and pC4-AP by Pol θ Lyase ...... 120

3.3.3 Inactivation of Pol θ Lyase by pC4-AP ...... 121

3.3.4 Attempted Identification of the Catalytic Nucleophile of Pol θ Lyase by

Reductive Trapping and LC-MS/MS ...... 125

3.3.5 Modification of Pol β Lys72 by pC4-AP ...... 130

3.3.6 Identification of Lys2383 as the Major Nucleophile for Pol θ Lyase Activity

...... 133

viii

3.4 Expanded Substrate Scope of Pol θ and Pol β Lyase...... 149

3.4.1 Lyase Activity of Pol θ on Abasic Lesions Near DSB Termini ...... 149

3.4.3 Probing the Scope of Pol θ Lyase on Various AP Substrates ...... 155

3.4.4 Lyase Activity of Pol θ and Pol β on Clustered Lesions ...... 156

3.5 Lyase Activity of Ku on Abasic and Oxidized Abasic Lesions ...... 170

3.5.1 Purification of Ku ...... 170

3.5.2 Lyase Activity of Ku on AP, C4-AP, and L in the Penultimate Position 173

3.5.3 Lyase Activity of Ku on DOB, pC4-AP, and 5’-dRP in the Terminal Position

...... 175

3.5.4 Attempted Detection of DNA-Protein Crosslinks Between Ku and Oxidized

Abasic Lesions ...... 179

3.5.5 Attempted Detection of Ku Modification by LC-MS/MS ...... 181

4. Future Experiments ...... 191

5. Experimental ...... 192

6. Appendix ...... 226

7. References ...... 290

8. Curriculum Vitae ...... 340

List of Tables

Table 1. Steady-state kinetic analysis of nucleotide insertion opposite lesions by Pol θ. 75

Table 2. Steady-state kinetic analysis of extension past abasic and oxidized abasic lesions by Pol θ...... 83

Table 3. Steady-state kinetic analysis of extension past abasic lesions flanked by 3'-TTA.

...... 83

Table 4. Steady-state analysis of extension past AP flanked by 3'-TAA...... 89

Table 5. Steady-state analysis of extension past dG•dX base pairs as a function of downstream flanking sequence...... 93

Table 6. Results of bypass sequencing for AP...... 102

Table 7. Steady-state kinetic analysis of nucleotide insertion opposite T and Tg (48 and

49)...... 107

Table 8. Steady-state kinetic analysis of insertion opposite T and Tg (50 and 51)...... 108

Table 9. Steady-state kinetic analysis of extension past T and Tg (52 and 53)...... 111

Table 10. Steady-state kinetic analysis of extension past T and Tg (54 and 55)...... 111

Table 11. Sequencing of Pol θ bypass products for T and Tg...... 114

Table 12. Removal of abasic and oxidized abasic lesions by Pol θ under single turnover conditions...... 121

Table 13. Fragmentation of modified peptide 63...... 132

Table 14. Comparison of predicted and observed fragments for peptide 65...... 138

Table 15. Comparison of steady-state rate constants for Pol θ variants on 48...... 141

Table 16. Comparison of lyase activity (on 53) and DNA binding (on 66) of Lys2383 variants...... 142

Table 17. Comparison of lyase activity on 58 of Lys2383 variants from a second, independent preparation of each variant...... 144

Table 18. Comparison of Lyase Activity (on 53) and DNA binding (on 66) of Pol θ variants...... 148

Table 19. Comparison of single turnover rate constants for Pol θ lyase on ternary complexes (58-60) and DSBs (67-69)...... 150

Table 20. Single turnover rate constants for Pol β and Pol θ lyase on abasic lesions at a

DSB terminus...... 153

Table 21. Dissociation constants for Pol θ and Pol β with 66 and 70...... 154

Table 22. Lyase activity of Pol θ on alternative substrates 71-74...... 156

Table 23. Lyase Activity of Pol θ and Pol β on clustered lesions containing AP...... 161

Table 24. Lyase activity of Pol θ and Pol β on clustered lesions containing C4-AP. .... 161

Table 25. Dissociation constants for Pol θ and Pol β on model clustered lesions...... 162

Table 26. Comparison of single turnover rate constants for wild type (WT) and K2383R

Pol θ on negative polarity clustered lesions...... 169

List of Figures.

Figure 1. Representative forms of DNA damage...... 3

Figure 2. Generation of DSBs by Ape1 action on clustered lesions...... 14

Figure 3. Mechanism of DSB repair by NHEJ...... 18

Figure 4. Repair of a DSB by homologous recombination...... 21

Figure 5. Mechanism of alternative end joining utilizing internal microhomologies or near- terminal microhomologies...... 22

Figure 6. Mechanism of SP-BER initiated by a monofunctional glycosylase...... 24

Figure 7. Mechanism of SP-BER initiated by a bifunctional glycosylase...... 25

Figure 8. Representative BER substrates...... 27

Figure 9. Domain organization of Pol β and Pol λ...... 30

Figure 10. Polymerase switching and gap-filling models of TLS...... 36

Figure 11. Domain organization of prokaryotic and eukaryotic Y-family polymerases. 39

Figure 12. Hydrogen bonding of dCTP and Arg234 in the Rev1 active site ...... 41

Figure 13. Crystal structure of Pol κ...... 44

Figure 14. Hoogsteen hydrogen bonding in the Pol iota active site...... 45

Figure 15. Bypass of AP in yeast...... 51

Figure 16. Replication past Tg by human translesion synthesis polymerases...... 53

Figure 17. Regulation of TLS by post-translational modification of PCNA...... 55

Figure 18. Domain organization of human Pol θ...... 57

Figure 19. Location of insertion loops in polymerase domain of Pol θ...... 58

Figure 20. Crystal structure of Pol θ in a ternary complex with DNA and ddATP...... 59

Figure 21. Ku crystal structure viewed down the DNA helix and in a side view...... 64

xii

Figure 22. Substrate scope of Ku lyase activity...... 66

Figure 23. Proposed mechanism for Ku removal from repaired DSBs...... 69

Figure 24. Analysis of Pol θ purity at each step of the purification by 10% SDS-PAGE.

...... 71

Figure 25. Representative active site titration of Pol θ...... 72

Figure 26. Generation of abasic and oxidized abasic sites for analysis of TLS by Pol θ. 74

Figure 27. TLS of abasic and oxidized abasic sites by Pol θ...... 77

Figure 28. Effect of dNTP ratios on deletion formation by Pol θ...... 80

Figure 29. Dependence of Watson-Crick base pairing on deletion formation by Pol θ.. 81

Figure 30. Deletion formation by Pol θ with downstream 3’-TTA sequence...... 87

Figure 31. Effect of nucleotide opposing AP on deletion formation...... 91

Figure 32. Generation of AP by photolysis from AP-ONv precursor or by UDG treatment of dU precursor...... 94

Figure 33. Bypass of AP or control (T or dU) in primer-templates for LC-MS experiments.

...... 95

Figure 34. Workflow for bypass sequencing experiments...... 97

Figure 35. Determining reaction conditions for bypass sequencing experiments...... 98

Figure 36. Analysis of PCR by 16% native PAGE with ethidium bromide staining. ... 100

Figure 37. Purification of digested pBluescript plasmid by agarose gel...... 101

Figure 38. Nucleotide insertion by Pol θ on primer-templates containing T and Tg..... 106

Figure 39. Single nucleotide insertion during extension past Tg...... 109

Figure 40. Full-length bypass of T and Tg by Pol θ...... 113

xiii

Figure 41. Generation of 5’-dRP, DOB, and pC4-AP from respective photochemical precursors...... 118

Figure 42. Active site titration of Pol θ with 5’-dRP...... 120

Figure 43. Removal of 5’-dRP, DOB, and pC4-AP by Pol θ under multiple turnover conditions...... 122

Figure 44. Stepwise inhibition of Pol θ by pC4-AP...... 122

Figure 45. Ethanolamine does not prevent inactivation of Pol θ by pC4-AP...... 123

Figure 46. Trapping of Schiff base intermediate for 5’-dRP, DOB, and pC4-AP...... 126

Figure 47. Trapping of Schiff base intermediate for DOB by NaCNBH3...... 127

Figure 48. Isolation of reduced DPC by SDS-PAGE...... 127

Figure 49. Coverage map for in-gel trypsin digestion of unmodified Pol θ (control) and

Pol θ-reduced DPC (analyte)...... 128

Figure 50. Comparison of observed and predicted spectra for modified peptide 63. .... 131

Figure 51. MS/MS of modified peptide 63...... 132

Figure 52. Coverage maps for Pol θ digestion by trypsin or Glu-C...... 134

Figure 53. Coverage map for Pol θ digestion by trypsin Lys-C...... 136

Figure 54. MS/MS of peptide 65 containing Lys2383 modification...... 138

Figure 55. SDS-PAGE analysis of Pol θ Lys2383 variants...... 139

Figure 56. Polymerase activity of Pol θ Lys2383 variants on native DNA...... 140

Figure 57. Comparison of single turnover rate constants for lyase activity of Pol θ Lys2383 variants...... 142

Figure 58. Trapping yield of Pol θ Lys2383 variants relative to wild type...... 145

xiv

Figure 59. Crystal structure of Pol θ in a ternary complex with a primer-template and incoming ddGTP...... 145

Figure 60. SDS-PAGE analysis of all Pol θ variants...... 146

Figure 61. Trapping yield of all Pol θ variants relative to wild type...... 146

Figure 62. Inhibition of Pol θ lyase by pC4-AP in a 5’ overhang...... 151

Figure 63. Stepwise inhibition of Pol θ lyase by pC4-AP in a 5’ overhang...... 152

Figure 64. Crystal structure of Pol β with gapped DNA...... 164

Figure 65. Crystal structure of lyase domain of Pol β with gapped DNA with relevant lysines indicated...... 166

Figure 66. Crystal structure of Pol β bound to gapped DNA with Lys234 indicated...... 167

Figure 67. Crystal structure of Pol θ bound to a primer template with the O-helix highlighted...... 168

Figure 68. Representation of construct used for Ku expression...... 171

Figure 69. Analysis of Ku purity by SDS-PAGE...... 172

Figure 70. Generation of abasic and oxidized abasic lesions by photolysis...... 173

Figure 71. Ku excision of C4-AP, AP, and L...... 174

Figure 72. Ku excision of pC4-AP, DOB, and 5’-dRP...... 176

Figure 73. Lyase activity of Ku as a function of concentration...... 177

Figure 74. Effect of pre-incubation on Ku lyase activity...... 177

Figure 75. Assaying for Ku crosslinking to C4-AP or L...... 180

Figure 76. Assaying for Ku crosslinking to C4-AP or DOB...... 181

Figure 77. Coverage map for Ku digestion by trypsin...... 183

Figure 78. Coverage map for digestion of Ku by trypsin/Lys-C.

...... 184

Figure 79. Coverage map for Ku incubated with C4-AP (91) or DOB (92) and digested by trypsin/Lys-C followed by chymotrypsin...... 187

xvi

List of Schemes

Scheme 1. Formation of AP by hydrolysis and ring-opened and ring-closed forms of AP.

...... 4

Scheme 2. Formation of Tg from T by HO•...... 7

Scheme 3. Mechanism of Tg formation from T by HO•...... 8

Scheme 4. Formation of strand breaks or C4-AP from a C4’-radical generated by bleomycin...... 10

Scheme 5. Activation of NCS chromophore by thiol...... 11

Scheme 6. Formation of L by NCS chromophore...... 12

Scheme 7. Products resulting from C5' sugar oxidation...... 13

Scheme 8. Proposed mechanism for formation of DOB by C5’-hydrogen abstraction. .. 13

Scheme 9. Formation of a DSB by bleomycin...... 16

Scheme 10. Mechanism of monofunctional and bifunctional glycosylases...... 28

Scheme 11. Catalytic mechanism of Ape1...... 29

Scheme 12. Proposed mechanism for excision of 5’-dRP by Pol β...... 31

Scheme 13. Generation of pC4-AP from C4-AP by Ape1...... 32

Scheme 14. Inactivation of Pol β by DOB and pC4-AP...... 34

Scheme 15. Bypass of C2-AP in E. coli by Pol II and Pol IV...... 52

Scheme 16. Bypass of L in E. coli by template misalignment...... 53

Scheme 17. Proposed mechanism for template misalignment during bypass of abasic lesions...... 80

Scheme 18. Watson-Crick base pairing between the nucleotide opposite the lesion and an adjacent template nucleotide dictates which extension product is formed...... 92

xvii

Scheme 19. Desilylation of Tg precursor...... 105

Scheme 20. Mechanism for formation of 1-nucleotide deletions and two-nucleotide deletions during Tg bypass by Pol θ...... 110

Scheme 21. Reductive trapping of Schiff base intermediate formed during lyase reaction..

...... 125

Scheme 22. Proposed mechanism for modification of Pol β by pC4-AP...... 133

Scheme 23. Proposed function of Lys2383 and Lys2575/Lys2577 in 5’-dRP excision by Pol θ.

...... 148

xviii

List of Abbreviations

DNA deoxyribonucleic acid

AP apurinic/apyrimidinic site (abasic site)

C4-AP C4’-oxidized absic site

DOB 5′-(2-phosphoryl-1,4-dioxobutane)

L 2-deoxyribonolactone

Tg thymidine glycol

DSB double strand break

SSB single strand break

BER base excision repair

SP-BER short patch-base excision repair

NCS neocarzinostatin

Alt-EJ alternative end joining

NHEJ non-homologous end joining

HR homologous recombination

Pol DNA polymerase

Ku Ku70/80

AP endonuclease 1 Ape1 dA 2’-deoxyadenosine dG 2’-deoxyguanosine dC 2’-deoxycytidine

T thymidine

xix

HO• hydroxyl radical dNTP 2’-deoxynucleotide triphosphate dATP 2’-deoxyadenosine triphosphate dGTP 2’-deoxyguanosine triphosphate dCTP 2’-deoxycytidine triphosphate

TTP thymidine triphosphate

8-oxo-dG 8-oxo-2’-deoxyguanosine

DTT dithiothreitol

EDTA ethylene diamine tetraacetic acid

Tris tris(hydroxymehyl) amino methane

BME β-mercaptoethanol

PAGE polyacrylamide gel electrophoresis

SDS sodium dodecyl sulfate

MS mass spectrometry

MALDI TOF matrix-assisted laser desorption

ionization time of flight

ESI electrospray ionization

UPLC ultra-performance liquid chromatography

ARP aldehyde reactive probe hNth1 human endonuclease III protein 1 hOGG1 human 8-oxoguanine DNA glycosylase 1 hNEIL human Nei-like DNA glycosylase

XRCC X-ray cross complementing protein

DNA-PK DNA-dependent protein kinase

DNA-PKcs DNA-dependent protein kinase catalytic

subunit

TLS translesion synthesis

PCNA proliferating cell nuclear antigen

PIP PCNA interacting peptide

UBM ubiquitin binding motif

UBZ ubiquitin binding zinc finger dG-BPDE dG-benzo[a]pyrene dihydrodiol epoxide

DPC DNA-protein crosslink yOGG1 Yeast 8-oxoguanine DNA glycosylase

UDG uracil DNA glycosylase

SMUG1 single-strand selective monofunctional

uracil DNA glycosylase

xxi

1. Introduction

Essential cellular information is stored within the biopolymer deoxyribonucleic acid (DNA). This essential information is frequently threatened, as the DNA in every cell is damaged from 10,000 to 100,000 times per day.1,2 DNA damage has myriad negative consequences, ranging from interference with replication—a potentially cytotoxic event— to potentially carcinogenic changes in sequence.3 It is therefore unsurprising that cells possess multiple DNA repair and damage tolerance pathways, each specialized to cope with a subset of DNA damage.4 The importance of these repair processes is highlighted by evidence that dysregulation or inactivation of DNA repair pathways can lead to the development of cancer.5–7 Elucidating the details of DNA damage response pathways may therefore give further insight into the development and prevention of cancer

DNA damage comes from many endogenous and exogenous sources including reactive oxygen species, ionizing radiation, industrial pollutants, and some natural products. The DNA lesions formed by these agents are numerous and include breaks in the phosphodiester backbone, damage or loss of the nucleobase, and damage to the deoxyribose sugar.8 Dedicated repair pathways exist for many DNA lesions, and in many cases, these pathways act with remarkable efficiency, processing thousands or tens of thousands of lesions in each cell every day.4 Some lesions; however, present challenges for

DNA repair systems. One such class of lesions are oxidized abasic sites, which are formed by ionizing radiation and some chemotherapeutics.9,10 This class of lesions includes the

C4’-oxidized abasic site (C4-AP), 2-deoxyribonolactone (L), and 5′-(2-phosphoryl-1,4- dioxobutane) (DOB). These lesions are repaired by the base excision repair (BER) pathway, which also processes ubiquitous apurinic/apyrimidinic (AP) sites.11,12 In contrast

1 with AP, oxidized abasic sites present unique challenges for BER, with both C4-AP and

DOB irreversibly inactivating the BER polymerases Pol β and Pol λ.13–15 Such inactivation could reduce the ability of cells to repair DNA damage, potentially contributing to the cytotoxicity of cancer treatments which produce these lesions.

The effects of oxidized abasic sites on eukaryotic repair enzymes and polymerases other than Pol β and Pol λ have not been examined in detail. Several enzymes have recently been shown to possess lyase activity, potentially contributing to the repair of abasic lesions in eukaryotic cells. Two such enzymes are DNA polymerase θ (Pol θ) and Ku70/80

(Ku).16,17 Pol θ is an interesting multifunctional enzyme which possesses lyase, translesion synthesis, and double strand break (DSB) repair activities.18 Expression of Pol θ is upregulated in some cancers and correlates with poor prognosis, making Pol θ a potential cancer drug target.19 Additionally, Pol θ promotes resistance to some DNA-damaging cancer treatments which produce oxidized abasic lesions.20 Therefore, the effect of these lesions on Pol θ lyase and translesion synthesis (TLS) activities is of interest. Similarly,

Ku70/80 is an important DSB repair enzyme which also possesses lyase activity, excising

AP near DSB termini.16 Because oxidized abasic lesions can be formed at complex DSBs generated by ionizing radiation and some anti-cancer agents, the effect of these lesions on

Ku70/80 is of interest.21

In the research presented, the ability of Pol θ lyase activity to repair AP, C4-AP, and DOB was examined. Pol θ was irreversibly inactivated by C4-AP, which covalently modified Lys2383, but was not inactivated by DOB. Further studies showed that Lys2383 is essential for efficient lyase activity, suggesting that it is the primary nucleophile responsible for lesion excision. The TLS activity of Pol θ was examined on templates containing AP, C4-AP, and L, along with the oxidized nucleobase lesion, thymidine glycol

(Tg). Pol θ was shown to form sequence-dependent frameshift mutations during TLS of these lesions. Additionally, the interactions of Ku70/80 with the oxidized abasic lesions

C4-AP, DOB, and L were studied in comparison with AP.

2. Background

2.1. DNA Damage

Figure 1. Representative forms of DNA damage.

DNA damage is a ubiquitous and complex process, with the DNA of every cell being damaged at least tens of thousands of times per day.2 The resulting DNA lesions include damage or loss of the nucleobase, breaks in the sugar-phosphate backbone, and damage to the sugar moiety (Figure 1).8 These lesions exhibit differential toxicity and mutagenicity. Some lesions are repaired efficiently, while others lead to very deleterious outcomes.4,22 Understanding the frequency of each lesion, how it is generated, and how it effects cellular processes is important to understanding basic cellular functions. It is also important for understanding cancer disposition as well as the toxicity and mutagenicity of various environmental or industrial pollutants or exposures.8

2.1.1 Abasic Sites

Scheme 1. Formation of AP by hydrolysis.

AP sites are formed by hydrolytic loss of the nucleobase moiety of DNA (Scheme

1). Purines and pyrimidines undergo hydrolysis, but this reaction is considerably faster for

23 the former (t1/2: 730 yrs) than for the latter (t1/2: 14,000 years). Although both reactions are quite slow, the large size of the human genome (3.2 gigabases) means that at least

10,000 AP lesions are formed in each cell every day by spontaneous hydrolysis.23 Damage to the nucleobase, which occurs under endogenous conditions as well as during treatment with alkylating or oxidizing agents, increases the lability of the N-glycosidic bond, leading to formation of more AP sites.24 The cyclic hemiacetal of AP is in equilibrium with the 4 ring-opened aldehyde, although this equilibrium lies heavily on the side of the hemiacetal, with approximately 1% aldehyde present (Scheme 1.25 The aldehyde form is prone to β- elimination due to the acidity of the C2’ proton. Indeed, AP sites undergo spontaneous β- elimination to form cytotoxic single-strand breaks, although this reaction is relatively slow

26 under physiological conditions (t1/2~1000 h).

AP also interferes with replication by blocking progression of replicative polymerases which are inefficient at bypassing this lesion.27,28 Additionally, the lack of a Watson-Crick base pairing face means that AP is non-instructive to polymerases during replication, making

AP a pro-mutagenic lesion.29

Steady-state levels of AP have been measured in different cell types using probes which are reactive towards aldehydes, typically alkoxyamines similar to the aldehyde reactive probe (ARP).30 These experiments reported steady-state AP levels of 10,000-

20,000 per cell in human cell culture, although a similar investigation reported that AP levels in human tissue samples ranged from 50,000 to 200,000 depending on the cell type.31–33 A recent investigation coupled ARP treatment with LC-MS detection, allowing

AP sites to be distinguished from other aldehyde-containing lesions. Using this assay, levels of approximately 5,000 intact AP sites and 9,000 β- elimination intermediates were reported.34 The high steady- state levels of AP, even in the presence of efficient repair processes suggest that AP is likely the most frequently formed lesion in human cells.

2.1.2 Oxidative DNA Damage

Intracellular oxidizing agents are commonly formed as a byproduct of cellular

•− respiration. Small amounts of superoxide (O2 ) are generated by leakage from the electron

35 •− transport chain (Equation 1). Detoxification of O2 by superoxide dismutase forms

36 •− hydrogen peroxide (H2O2) (Equation 2). Although H2O2 is less toxic to cells than O2 and can be detoxified by catalase, it generates hydroxyl radical (HO• ) in the presence of iron via the Fenton reaction (Equation 3).37

- •− Equation 1: O2 + e → O2

•− + Equation 2: 2 O2 + 2H → O2 + H2O2

2+ 3+ • − Equation 3: Fe + H2O2 → Fe + HO + HO

HO• is also formed by the direct ionization of water by γ-radiolysis.38 This process contributes to the cytotoxicity of radiation therapy, creating bursts of deleterious HO• within tumors. HO• is a highly reactive species which attacks the nucleobase moiety of

DNA and abstracts hydrogens from the deoxyribose sugar, although reactions with the nucleobase account for approximately 90% of HO• reactivity, while sugar oxidation accounts for less than 10%.39 Many oxidative DNA lesions are cytotoxic and/or mutagenic, and may contribute to the development of cancer and other diseases. Therefore, the manner in which various lesions are formed and their effect on cellular processes is of interest, from both a mechanistic and clinical perspective.40

2.1.2.1 Thymidine Glycol

Scheme 2. Formation of Tg from T by HO•.

Thymidine glycol (Tg) is a common DNA lesion under normal cellular growth conditions, with approximately 300 Tg lesions formed in each human cell every day.41,42

Additionally, Tg is a major thymidine oxidation product in cells exposed to ionizing radiation.43 Interestingly, Tg levels are considerably higher in rats than in humans. As rats have the greater metabolic rate, Tg levels correlate with metabolic rate, suggesting that Tg is a biomarker of oxidative stress.41 Tg can exist as four diastereomers, but γ-radiolysis of duplex DNA primarily forms the two cis diastereomers (Scheme 2). Epimerization at the

C6 position results in interconversion between cis- and trans-diastereomers, although the equilibrium favors the cis-diastereomers by 3−5-fold.44,45 Therefore, Tg is typically present as a mixture of four diastereomers in duplex DNA.

Tg is formed by attack of HO• upon the nucleobase of thymidine (Scheme 3).

Hydroxyl radical primarily attacks the C5 position of thymine, although C6 attack occurs as well. Thus, the 6-OH-5yl radical (1) and the 5-OH-6yl radical (2) are formed in an approximately 2:1 ratio.46,47 Under aerobic conditions, both radicals are trapped by oxygen to form the respective hydroperoxides (3 and 4).48 Reduction of each hydroperoxide forms

Tg. Under anaerobic conditions; however, only the 5-OH-6yl radical 2 gives rise to Tg.

This occurs by oxidation of 2 to give 5, which is attacked by water to from Tg. The 6-OH-

5yl radical on the other hand, is reduced to give 6-hydroxy-5-hydrothymidine (6).

Scheme 3. Mechanism of Tg formation from T by HO•.

2.1.2.2 Sugar Oxidation

The deoxyribose sugar of DNA is also oxidized by HO• and certain DNA damaging agents. Due to the high reactivity of HO•, hydrogen atom abstraction from the deoxyribose

8 sugar is largely governed by accessibility.49 As such, the 5’ position is the most frequently attacked, accounting for greater than 50% of HO• reactivity. The C4’ position accounts for more than 20% of reactivity, while the other positions are approximately equal in reactivity

(10-15%).49 Radiomimetics such as bleomycin, copper-phenanthroline, and the enediynes also oxidize the sugar moiety of DNA, often demonstrating considerably greater selectivity for the hydrogens which they abstract.

2.1.2.3 C4’-Oxidized Abasic Site

C4-AP is formed by abstraction of the C4’ hydrogen from the deoxyribose sugar moiety. This occurs in cells exposed to ionizing radiation, as well as during treatment with bleomycin. Bleomycin is a glycopeptide antibiotic produced by the bacterium

Streptomyces verticillus and used for the treatment of testicular cancer and lymphoma.50

Bleomycin acts in a metal-dependent fashion, primarily coordinating iron to form a reactive radical, likely an Fe5+-oxo species, which abstracts the C4’ hydrogen from deoxyribose

(Scheme 4).51,52 The proportion of C4-AP formed from 7 is dependent upon the availability of oxygen, as the radical can be trapped by oxygen to form 8 or oxidized to a carbocation.53

Reduction of 8 gives the C4’ hydroperoxide (9) which gives rise to 10 by a Criegee rearrangement. Subsequent eliminations forming a single nucleotide gap with 3’- phosphoglycolate and 5’-phosphate termini. Under anaerobic conditions; however, 7 is oxidized to the C4’ carbocation (11), which is trapped by water to form 12. The free nucleobase is released to form C4-AP.54

Scheme 4. Formation of strand breaks or C4-AP from a C4’-radical generated by bleomycin.

Bleomycin exhibits some sequence selectivity, abstracting the C4’ hydrogen from the pyrimidine nucleotide in 5’-GT and 5’-GC sequences.55 Formation of C4-AP has been quantified in DNA irradiated or treated with bleomycin in vitro, where 4 lesions per 106 nucleotides per Gy or 12 lesions per 106 per μM were detected.56 C4-AP formation has not been quantified in cells, either under endogenous conditions or in response to damaging agents; however, a cytosine adduct likely formed by reaction with C4-AP on an opposing strand is present at endogenous levels of 30-100 per cell, suggesting that endogenous levels of C4-AP are at least that high.57 C4-AP is also formed by enediyne natural products, although it is a minor product of these drugs, which primarily do C1’ and C5’ chemistry.

Interestingly, C4-AP formation by the enediynes requires molecular oxygen, indicating

10 that formation of this lesion proceeds through different mechanisms depending on the drug.9 In the case of enediynes, the C4’ hydroperoxyl 9 is likely reduced to give C4-AP.

2.1.2.4 2-Deoxyribonolactone

The 2-deoxyribonolactone lesion (L) is generated by oxidation of the C1’ position. The C1’ hydrogen of deoxyribose is buried in the minor groove and is relatively inaccessible to solvent, accounting for the smallest proportion of deoxyribose reactivity with HO• in vitro, at approximately 11%.49 Similarly, quantification of L in cells exposed to ionizing radiation showed that L accounts for approximately 7% of sugar oxidation products under these conditions.58 Although the C1’ position is a minor target for HO•, it is susceptible to attack by minor groove-binding radiomimetics such as neocarzinostatin (NCS), produced by the bacterium Streptomyces macromyceticus.9 NCS is a complex consisting of a chromophore

(NCS chromophore) and a 113 amino acid protein which protects the chromophore from degradation. NCS belongs to the enediyne class of antibiotics, a group of molecules which undergo cyclization reactions to generate diradical species which abstract hydrogens from

DNA (Scheme 5). NCS chromophore is activated by nucleophilic attack of a thiol, forming a reactive enyne cumulene system (13).59 Cyclization of 13 generates the 2,6 diradical 14.

Scheme 5. Activation of NCS chromophore by thiol.

NCS primarily generates single stranded breaks containing a 5’-aldehyde and a 3’- phosphate; however, on certain sequences it generates bistranded lesions.60 These bistranded lesions typically involve formation of L on one strand and are thought to be a major contributor to the cytotoxicity and mutagenicity of NCS.61 NCS abstracts the C1’- hydrogen from the deoxyribose sugar, generating C1’-radical 15 (Scheme 6). Trapping by

62 O2 yields peroxyl radical 16, which is followed by loss of superoxide to give cation 17.

Trapping by H2O followed by base-catalyzed loss of the nucleobase gives L.

Scheme 6. Formation of L by NCS chromophore.

2.1.2.5 Dioxobutane

Dioxobutane (DOB) is formed by hydrogen atom abstraction from the C5’-position of the deoxyribose sugar by HO• and radiomimetic agents such as NCS.63,64 The C5’- position of deoxyribose is the primary site for abstraction with HO•; however, DOB accounts for only 4% of the sugar oxidation products formed by γ-radiolysis.63 This is because 5’-nucleoside-5’-aldehydes are the primary product of C5’ oxidation, and indeed the primary sugar oxidation product in cells exposed to γ-radiolysis, accounting for 40% of all sugar oxidation products (Scheme 7).58

Scheme 7. Products resulting from C5' sugar oxidation.

The mechanism of DOB formation from the C5’ radical (18) is unclear, although it was proposed to result from trapping by oxygen followed by reduction of the peroxy radical to the C5’-hydroperoxide 19 (Scheme 8). Rearrangement of 19 gives 20, which is followed by decomposition to hemiacetal 21.64 Loss of the nucleobase then forms DOB.

Quantification of DOB in duplex DNA treated in vitro showed that γ-radiolysis produces

∼6 lesions per 106 nt per Gy, while calicheamicin forms 9 lesions per 106 nt per μM.63 DOB levels in cells are unknown, however.

Scheme 8. Proposed mechanism for formation of DOB by C5’-hydrogen abstraction.

2.2 Double Strand Breaks

2.2.1 Double Strand Break Formation by Ionizing Radiation

DSBs are formed infrequently, with estimates ranging from 10-50 DSBs per cell each day under normal conditions.65–67 Many of these DSBs are formed by collapse of replication forks when they encounter DNA lesions.68 Another major cause of DSB formation is through the action of HO•, which forms sugar-centered radicals that give rise to oxidized abasic sites (section 2.1.4), along with breaks in the sugar-phosphate backbone of DNA.69,70 However, formation of a DSB requires the formation of two single strand breaks in close proximity (within two helical turns). Therefore, DSBs are generated infrequently by HO• under endogenous conditions. Ionizing radiation; however, generates bursts of ∙OH along the track of radiation.69,70 These bursts of hydroxyl radical form DSBs and clustered lesions, sites containing 2 or more DNA lesions within 20 nucleotides. As the linear energy transfer of the radiation increases, the frequency and complexity of clustered lesions also increases.71,72 Some of these clustered lesions are refractory to repair and are thought to be a significant contributor to the cytotoxicity of radiation therapy.71,73

However, when the cell repairs these clustered lesions, it can form DSBs.

Figure 2. Generation of DSBs by Ape1 action on clustered lesions. X=AP.

DNA repair is generally considered to be advantageous to cell survival and function, but depending on the context, it can sometimes be deleterious to the cell. This is particularly evident in the repair of bistranded and clustered lesions, which can be converted to more deleterious lesions, such as DSBs, by DNA repair enzymes. For example, siRNA knockdown of hNTH1 or hOGG1, two base excision repair glycosylases, leads to the production of fewer DSBs, indicating that clustered lesions are aberrantly repaired to give DSBs.74 Experiments using purified enzymes and site-specifically incorporated DNA lesions showed that the spacing and positioning of lesions in a damage cluster influences the repair outcome. For example, Ape1 exhibits a considerable polarity preference for cleaving bistranded lesions (Figure 2).75,76 Bistranded lesions containing AP sites 3’ to one another on opposite strands (spaced either 1, 3, or 5, nt apart) are readily cleaved to generate DSBs (Figure 2A). However, when AP bistranded lesions are spaced apart by 1 nt in the 5’-direction, Ape1 slowly cleaves only one strand and is unable to incise the other (Figure 2B). When AP sites are 3 nt apart in the 5’-direction, neither lesion is incised (Figure 2C). However, this polarity preference is diminished as the AP sites are spaced further apart, so DSBs are generated from bistranded lesions spaced by 5 or more nucleotides in the 5’-direction (Figure 2D).

2.2.2 Double Strand Break Formation by Antitumor Antibiotics

Radiomimetic antitumor compounds produce DNA damage products similar to those generated by ionizing radiation. In contrast with the stochastic DSBs generated by bursts of hydroxyl radical, some radiomimetic compounds produce DSBs in a concerted fashion, where a single molecule produces a break on each strand of the DNA.9 For instance, bleomycin produces DSBs in a concerted fashion (Scheme 9).10,77

Scheme 9. Formation of a DSB by bleomycin (BLM). B=nucleobase.

Activated bleomycin abstracts the C4’-hydrogen from the deoxyribose sugar on one strand generating a C4’-radical on that strand along with an Fe4+ hydroxyl bleomycin species (22). Trapping of the C4’ radical by oxygen yields a C4’ peroxyl radical (23) which can reactivate bleomycin to the Fe5+-oxo form (24). This activated form of bleomycin abstracts the C4’-hydrogen from the opposite strand, giving a C4’-radical, which is trapped by oxygen and then reduced (25). Cleavage of each strand proceeds via a Criegée rearrangement (Scheme 4). DSBs are a minor product of bleomycin, accounting for less than 10% of damage products; however, DSBs are believed to be a major source of the cytotoxicity and antineoplasticity of bleomycin.9,50

Enediynes such as NCS and calicheamicin also produce DSBs in a concerted fashion. Unlike bleomycin, which damages DNA via a single radical species and must be reactivated to initiate bistranded damage, enediynes generate diradical species upon activation (Scheme 5). While NCS primarily

16 generates bistranded lesions containing a SSB and L, calicheamicin produces a high proportion of DSBs (SSB:DSB ratio=1:3), a fact which likely explains the high cytotoxicity of this compound.78 This has largely precluded the use of enediynes in cancer treatment, as their high cytotoxicity is not limited to cancer cells and plagues healthy cells as well.79 However, there is some promise in using these highly cytotoxic molecules in antibody drug conjugates to achieve cancer cell-specific cytotoxicity.80,81

2.2.3 Repair of Double Strand Breaks

Although DSBs are formed infrequently by most damaging agents, a single unrepaired DSB can cause cell death via apoptosis.82 To avoid these serious consequences, cells have three main repair pathways for DSBs: non-homologous end joining (NHEJ), homologous recombination (HR), and alternative end-joining (alt-EJ), also called microhomology mediated end joining (MMEJ). The pathway chosen for DSB repair is dependent upon many factors, including the stage of the cell cycle. The cell cycle is divided into G1, S, G2 and M phases. Cells grow during G1 phase, replicate their DNA in S phase, continue to grow in G2 phase, and divide during M phase.83 NHEJ is active at all stages of the cell cycle and functions via the direct ligation of a double strand break, while HR utilizes the complementarity of the sister chromatid and is therefore only active in late S and G2 phase.68,84,85 Alt-EJ is an error-prone ligation pathway which is most active in early

S phase.86 It is thought to repair fewer than 10% of DSBs in healthy cells, although it makes significant contributions to DSB repair in cells deficient in either NHEJ or HR.86–88

2.2.3.1 Non-Homologous End Joining

Figure 3. Mechanism of DSB repair by NHEJ.

NHEJ begins with binding of the DSB by Ku70/80 (Ku), a heterodimeric protein composed of Ku70 and Ku80 subunits. Ku is a toroidal protein which allows it to encircle and tightly bind duplex DNA in a sequence-independent manner.89,90 Its high affinity for duplex DNA (KD = 2 nM), along with its high intracellular concentration (400,000-500,000 molecules per cell) allow one molecule of Ku to bind to each end of a DSB within seconds of their formation.67,91,92 Ku then recruits several other proteins to process and ligate the break. DSBs are produced with a variety of non-ligatable groups such as 3’-phosphate, 5’-

OH’, 3’-phosphoglycolate, or damage to flanking bases. Consequently, processing of DSB termini is sometimes necessary prior to ligation. Traditionally, NHEJ has been viewed as 18 a sequential process, where Ku binds DNA and recruits DNA-dependent protein kinase catalytic subunit (DNA-PKcs), which associates with Ku and forms a complex which tethers both ends of the DSB and serves to protect the DNA from degradation by nucleases.93,94 DNA-PKcs also plays an important role in dictating the subsequent steps of

NHEJ. Autophosphorylation in trans (each molecule of DNA-PKcs phosphorylates the other, primarily at the Ser2056 cluster) causes DNA-PKcs to dissociate from the break, allowing end ligation without end processing.95 If processing of DNA ends is necessary prior to ligation, DNA-PKcs is phosphorylated, primarily at the Thr2609 cluster, by ATM kinase. This leads to recruitment of processing factors such as polynucleotide kinase (PNK) and/or Artemis, which process the break.95 PNK removes 3’-phosphates and phosphorylates 5’-OH’s, while Artemis nuclease removes hairpins or 3’- phosphoglycolates.96,97 If necessary, Pol X family polymerases such as Pol λ or Pol μ fill add nucleotides to create complementary ends or fill in gaps, and then XLF (sometimes referred to as Cernunnos) and the XRCC4-Ligase IV complex seal the DSB.98,99 Single molecule FRET studies generally support this mechanism, showing that DNA-PKcs is required for long-range synapse formation, and the catalytic activity of DNA-PKcs is required for formation of a short-range complex.100 These studies also revealed that XLF and XRCC4-Ligase IV are essential for short range complex formation. However, live-cell imaging experiments suggest that DNA-PKcs may not be required for the ligation of all breaks.

Simple DSBs (those possessing chemically ligatable end groups) induced by low

LET radiation are repaired quickly and efficiently in a DNA-PKcs-independent fashion and without the aid of other processing enzymes.101 Meanwhile, complex DSBs induced

19 by high LET radiation are repaired more slowly in a DNA-PKcs-dependent fashion.102 This suggests that NHEJ follows a damage-dependent model, where processing enzymes are recruited only to complex DSBs that cannot be ligated without their assistance.103

NHEJ is often regarded as an error-prone repair pathway because of its propensity to induce insertions or deletions during DSB repair. This classification is somewhat narrow; however, as the NHEJ machinery is capable of accurately repairing chemically ligatable DSBs lacking base damage or mispairs.104 The apparent error-proneness of NHEJ is largely attributable to the necessity of processing non-ligatable end groups or joining non-complementary ends. Ligation-blocking end-groups are usually removed by exonucleases, resulting in a loss of sequence information.105 Additionally, activity of Pol

μ or Pol λ, which promotes joining of non-complementary ends can sometimes result in insertions.106 As radiation-induced DSBs often contain additional damage, the inability to process non-ligatable end groups would result in substantial radiosensitivity, even to background radiation.104

Unrepaired DSBs are lethal to the cell. Although NHEJ may induce mutations which could be deleterious to cell survival and function, the activity of NHEJ is largely beneficial to the cell. Indeed, Ku suppresses chromosomal translocations, the exchange of

DNA between non-homologous chromosomes, as well as chromosomal fusions, the aberrant joining of two chromosomes.107 Additionally, if NHEJ cannot join a DSB, it may be repaired by alt-EJ, which results in greater loss of sequence information and has a higher propensity to form chromosomal translocations. Therefore, it appears as though the possibility of inducing small insertions or deletions by NHEJ is often an acceptable price to pay for avoiding potentially malignant transformations or cell death.82

2.2.3.2 Homologous Recombination

The most accurate pathway for DSB repair is homologous recombination (HR). HR uses the undamaged homologous chromosome as a template for accurate repair of DSBs. The necessity for the homologous chromosome means that, in humans, HR is active only in late

S and G2 phases of the cell cycle after DNA has been replicated.108 Although HR is more accurate than NHEJ, it is also considerably slower, with HR taking more than 7 h to complete, while NHEJ can be accomplished within 30 min.109 The mechanism of HR is complex and requires a number of proteins. A brief mechanism is presented here (Figure 5).110,111

Following formation of a DSB, an exonuclease generates single stranded overhangs. The Rad51 recombinase coats the single stranded region and searches for a homologous sequence, promoting invasion of the sister chromosome. A DNA polymerase uses the homologous chromosome as a template, synthesizing new

DNA.112,113 This process is repeated for the second single stranded region of the break, generating a double Figure 4. Repair of a DSB by homologous recombination. Newly synthesized DNA is shown in 114 Holliday junction. This intermediate is red..

21 topologically linked and must be unhooked by resolvases, in some cases leading to crossing over, a process where the DNA sequence from one chromosome is transferred to the other.114

2.2.3.3 Alternative End Joining

Figure 5. Mechanism of alternative end joining utilizing internal microhomologies or near-terminal microhomologies. A) alt-EJ utilizing internal microhomologies. B) alt-EJ utilizing near-terminal microhomologies. DNA synthesized by Pol θ is shown in red. Microhomologies are represented by a blue box.87,115 The existence of a DSB repair pathway that is independent of NHEJ and HR was determined by early studies in S. cerevisiae, where inactivation of Ku caused cells to utilize an error-prone end-joining pathway.116 End-joining products from this pathway were characterized by large deletions up to hundreds of bp and contained regions of microhomology (1-4 homologous nucleotides) between the two molecules. Studies in

22 human cells deficient in XRCC4, an essential component of NHEJ, found that alt-EJ operates in human cells at high levels when NHEJ is knocked out.117 Alt-EJ also makes a prominent contribution to DSB repair in cells deficient in HR.88 More recent studies have shown that alt-EJ is active in cells proficient in both NHEJ and HR, repairing 10-20% of

DSBs.86 Additionally, the majority of chromosomal translocations and rearrangements that occur during DSB repair are catalyzed by alt-EJ.

Alt-EJ begins with DNA end resection by exonucleases (MRN complex and CtIP) to generate 3’ single stranded overhangs (Figure 4).86 This end resection step is common between HR and alt-EJ, although HR requires resection of hundreds of nucleotides to generate long stretches of single stranded DNA which can be annealed to the homologous sister chromatid. End resection in alt-EJ is typically shorter, requiring the digestion of shorter sequences (25 nt) to expose regions of microhomology. These microhomologies are short complementary sequences, which can be as short as a single nucleotide in mammals.118 Pol θ promotes annealing of these microhomologies, which can be present at the DSB terminus or at an internal position.119 Annealing of internal microhomologies forms single strand flaps which are trimmed by an exonuclease, creating deletions (Figure

4A). Pol θ fills in the remaining single stranded regions using the other strand as a template and ligase I or ligase III seals the breaks.115 When microhomologies are present near the

DSB termini, the mechanism and outcome differ somewhat. Pol θ promotes annealing of microhomologies and extends the primer to stabilize the annealed structure (Figure

4B).20,88 In this case, flap trimming is unnecessary, and Pol θ generates templated insertions. Finally, the break is sealed on each strand by either ligase I or ligase III to complete repair.120

2.3 Base Excision Repair: Repair of Base Damage and Abasic Lesions

The base excision repair (BER) pathway repairs nucleotides damaged by alkylation, oxidation, or depurination.121 BER also repairs AP and oxidized abasic sites

(C4-AP, DOB, and L).11,13,122 BER operates at all domains of life, and there are numerous differences between prokaryotic and eukaryotic BER. Additionally, BER can be divided into short-patch and long-patch BER pathways.123 For brevity and relevance to this research, only eukaryotic short-patch base excision repair (SP-BER) will be discussed.

2.3.1 Mechanism of Base Excision Repair

Figure 6. Mechanism of SP-BER initiated by a monofunctional glycosylase. The general mechanism of SP-BER involves four steps: 1) removal of the damaged base to generate AP 2) removal of AP to generate a single-nucleotide gap 3) filling in of the gap by a polymerase leaving a single strand break 4) ligation of the break to restore

24 intact DNA (Figure 6). Damaged bases are removed from DNA by damage-specific glycosylase enzymes, which typically flip the lesion out of the duplex and into the enzyme active site where the N-glycosidic bond is hydrolyzed.124,125 This generates AP as an intermediate, which is recognized by the major AP endonuclease, Ape1, in humans.126

Ape1 cleaves the phosphodiester bond 5’ to AP, generating a strand break and 5’- deoxyribose phosphate (5’-dRP). This is removed from DNA by a lyase activity present in a BER polymerase, primarily Pol β in humans.127 The lyase reaction proceeds by attack of a nucleophilic amine, usually a lysine residue, on the C1’ aldehyde of 5’-dRP. The resulting iminium ion intermediate increases the acidity of the C2’ proton, promoting strand cleavage by β-elimination, forming a single-nucleotide gap. This gap is filled in by the polymerase activity, utilizing the opposite strand as a template.128. Ligation of the two

DNA ends then rejoins the intact DNA duplex.129

Figure 7. Mechanism of SP-BER initiated by a bifunctional glycosylase.

SP-BER can also proceed via a different mechanism depending upon the identity of the glycosylase which removes the damaged nucleobase (Figure 7). Glycosylases such as hNTH1 and the hNEIL family, which are specialized for oxidative DNA damage, are bifunctional.130–132 These enzymes possess an intrinsic lyase activity in addition to their glycosylase activity. In the case of hNEIL, this reaction proceeds via a β-elimination and then a subsequent δ-elimination, removing the sugar and leaving a 3’ phosphate terminus.131 This obviates the requirement of Ape1 or the lyase activity of the BER polymerase. The 3’ phosphate is removed by the phosphatase activity of PNKP, and the gap filling and ligation steps proceed in the same fashion as described above.133 In the case of hNTH1; however, the lyase activity catalyzes a β-elimination, leaving an unsaturated sugar (26) at the 3’ terminus of the DNA (Figure 7).130 This is removed by Ape1, generating a single nucleotide gap. The gap is filled in by a BER polymerase followed by ligation to intact DNA.

Inactivation or imbalance of base excision repair proteins has serious consequences for an organism. For example, inactivation of alkyl adenine glycosylase in mice leads to increased cancer risk, a single nucleotide polymorphism in the gene encoding hOGG1 increases lung cancer risk, and loss of Pol β, the major BER polymerase, is embryonically lethal.5,6,134 Therefore, BER is an important mechanism for protecting against the deleterious consequences of DNA damage.

2.3.2 Substrate Scope of Base Excision Repair

Figure 8. Representative BER substrates. Base excision repair primarily recognizes and removes non-bulky DNA lesions

(Figure 8). These include 3-methyl dA, a cytotoxic, replication-blocking lesion formed by methylating agents.24 Oxidative forms of DNA damage such as 8-oxo-7,8-dihydro-2'- deoxyguanosine (8-oxodG) and Tg are also removed by BER.135,136 Uracil, formed by deamination of cytosine as well as via misincorporation during replication, is removed by the base excision repair pathway.124 Additionally, some mismatches are removed by BER, such as dA when it is misincorporated opposite 8-oxodG.137 Each of these lesions or incorrect nucleotides is removed by specialized glycosylase enzymes which have evolved to recognize particular substrates. AP sites and oxidized abasic sites are also substrates for

BER, although as they lack a damaged nucleobase, glycosylase activity is not necessary for repair.

2.3.3 Glycosylases

Mammalian cells have at least 11 glycosylase enzymes which remove damaged or incorrect nucleobases to create an AP site.138 Each glycosylase acts upon a subset of lesions, recognizing unique structural features of the damaged nucleobase that distinguish it from undamaged bases.139,140 Most glycosylases insert a hydrophobic residue into DNA to kink the duplex and then flip the damaged base out of the duplex and into a substrate

27 recognition pocket of the glycosylase active site.141 142,143Aromatic residues stabilize the nucleobase in an extrahelical position by π-π stacking, while a hydrophobic residue inserted into the duplex kinks the DNA to allow for flipping out of the damaged base.140 AlkD glycosylase, present only in prokaryotes and some single-celled eukaryotes, is a notable— and perhaps singular—exception to the base-flipping mechanism, utilizing CH-π interactions to stabilize the transition state during hydrolysis of intrahelical nucleobase.

Scheme 10. Mechanism of monofunctional and bifunctional glycosylases. A) Monofunctional glycosylases. B) Bifunctional glycosylases.

Monofunctional glycosylases, which catalyze only a single reaction, the removal of the nucleobase from the DNA to generate AP, typically utilize an activated water molecule to cleave the N-glycosidic bond (Scheme 10A).144 Bifunctional glycosylases typically utilize the ε-amino group of a lysine residue to attack the C1’ position of the sugar. In addition to catalyzing the cleavage of the N-glycosidic bond, this also forms a Schiff base to promote cleavage of the phosphodiester backbone 3’ to the lesion (Scheme 10B).145

2.3.4 Ape1

Ape1 is homologous to the bacterial AP endonuclease, exonuclease III (Xth).

Knockout of Ape1 is embryonically lethal in mice, and Ape1-deficient embryos displayed increased sensitivity to ionizing radiation, indicating a defect in BER.146,147 Ape1 activity is magnesium-dependent, with the magnesium playing an important role in coordinating the oxygen of the scissile P-O bond (Scheme 11).148 Ape1 is highly proficient at incising the phosphodiester backbone 5’ to AP, with a single-turnover rate of at least 850 s-1 for this

149 process. The steady-state rate constant for this reaction is much slower than the single- turnover rate constant at 2 s-1.149 This considerable reduction in steady-state rate relative to single turnover rate is common for BER enzymes and indicates that the rate-limiting step occurs after the chemical reaction and is likely product release. Ape1 also recognizes oxidized abasic sites, although the catalytic efficiency is reduced by approximately 5-fold and 10-fold (relative to AP) for incision of L and C4-AP, respectively.13,150

Scheme 11. Catalytic mechanism of Ape1.

2.3.5 Base Excision Repair Polymerases

Specialized polymerases are essential for BER. For instance, knockout of Pol β, the primary BER polymerase, is embryonically lethal in mice.134 BER polymerases play two roles in this pathway: 1) removing 5’-dRP which is generated by the action of a monofunctional glycosylase and Ape1 and 2) subsequent gap-filling DNA synthesis, a necessary step for BER by monofunctional or bifunctional glycosylases (Figure 6 and 7).127

The major BER polymerases in humans, Pol β and Pol λ, belong to the X family.151 These enzymes are characterized by a lack of proofreading exonuclease activity and a relatively high error rate (10-4 compared to 10-6-10-8 for replicative polymerases) on undamaged

DNA.151 Additionally, Pol β and Pol λ both possess a distinct 8 kDa lyase domain that is separate from the polymerase domain (Figure 9). Pol λ is larger than Pol β and possesses an additional N-terminal BRCT domain, which is important for protein-protein inteactions in NHEJ, that is absent in Pol β.151 Despite these differences, the two proteins share considerable sequence identity (32%) and the polymerase domain bears more than 50% sequence homology.152,153

Figure 9. Domain organization of Pol β and Pol λ.

Among BER polymerases, the mechanistic details for 5’-dRP excision are best characterized for Pol β, although some details are still unclear. Excision of 5’-dRP proceeds

30 through a Schiff base intermediate, with the nucleophilic lysine, Lys72 in Pol β, attacking the C1’-position of 5’-dRP (Scheme 12).127,154,155 5’-dRP exists in equilibrium between the ring open aldehyde and the ring closed hemiacetal, with the hemiacetal form being heavily favored (99:1).25 The aldehyde form is generally assumed to be the electrophilic species which is attacked by the ε-amino group of the catalytic lysine, although this has not been demonstrated conclusively.156 Pol β may promote ring opening to the aldehyde form by donating a proton from the catalytic lysine.156,157 Attack on the aldehyde by the catalytic lysine forms an iminium which lowers the pKa of the C2’-proton. This promotes β- elimination via deprotonation of the C2’-position, which is likely catalyzed by a nearby

156 carboxylate such as Glu26. Other BER polymerases follow a similar mechanism, although the specific residues which are important for each of these enzymes are unknown, with the exception of Pol λ, for which Lys310 has been identified as the primary nucleophile.158

Scheme 12. Proposed mechanism for excision of 5’-dRP by Pol β.156,157 P=phosphate

Other nuclear polymerases such as Pol θ, Pol ι, and Rev1, as well as the mitochondrial polymerase Pol γ, also possess lyase activity and may participate in

BER.17,159–163 Reductive trapping experiments indicate that 5’-dRP excision proceeds through a Schiff base intermediate for each of these enzymes, similar to Pol β and Pol λ.

However, the major catalytic nucleophile has not been identified for any of these other

BER polymerases. None of these enzymes are known to possess a distinct lyase domain like Pol β and Pol λ, and limited proteolysis has mapped the catalytic lysine residue to the polymerase domain for Pol θ, Pol ι, and Rev1.17,161,162 The physiological function of these enzymes as 5’-dRP lyases is unclear; however. Pol θ was proposed to contribute to repair

159 of oxidative DNA damage, as Pol θ knockout sensitizes chicken DT40 cells to H2O2.

Additionally, cell extracts deficient in Pol θ showed compromised repair of 8-oxodG.159

Similar to Pol θ, Pol ι has been implicated in the repair of oxidative DNA damage by BER based upon several observations: Pol ι knockdown sensitizes cells to H2O2 treatment and reduces BER of a uracil-containing oligonucleotide duplex.164 Furthermore, immunoprecipitation confirmed that Pol ι interacts with XRCC1, a scaffolding factor in

BER.164 The physiological function of Rev1’s lyase activity is even less clear than that of

Pol θ and Pol ι. It has been proposed to play a role in somatic hypermutation, although this has not been demonstrated.162

2.3.6 Base Excision Repair of Oxidized Abasic Sites

Scheme 13. Generation of pC4-AP from C4-AP by Ape1.

Similar to AP, oxidized abasic sites are substrates for BER. Ape1 recognizes and incises the phosphodiester backbone 5’ to C4-AP, generating pC4-AP, a substrate for BER polymerases (Scheme 13).13 DOB is generated simultaneously with a single strand break and does not require phosphodiester backbone incision by Ape1 to generate a substrate for a BER polymerase (Scheme 7). Pol β and Pol λ act upon both of these oxidized abasic

-1 lesions. Under single turnover conditions, Pol β activity on pC4-AP (kobs: 3.6 s ) is reduced

-1 13 slightly relative to 5’-dRP (kobs: 6.2 s ). This reduction is more pronounced,

-1 122 approximately 100-fold, for DOB (kobs: 0.04 s ). Pol λ exhibits a 7-fold reduction in

-1 -1 activity for pC4-AP ((kobs: 0.01 s ) relative to 5’-dRP (kobs: 0.07 s ) and an approximately

-1 15-fold reduction for DOB (kobs: 0.006 s ). In addition, DOB and pC4-AP inactivate Pol β and Pol λ.14,15 Studies using monoaldehyde derivatives of DOB show that the 1,4- dicarbonyl moiety is necessary for inactivation of Pol β lyase.122 This inactivation stems from covalent modification of the protein, both in the form of a DNA-protein crosslink

(DPC), as well as formation of a lactam adduct (Scheme 14A).14 Inactivation of Pol β by pC4-AP is slightly less efficient than by DOB, requiring 7-8 turnovers before complete enzyme inactivation (as opposed to 4 turnovers for DOB).13 Additionally, inactivation of

Pol β by pC4-AP proceeds solely through lactam formation and not via formation of a DPC

(Scheme 14B). Pol λ is inactivated primarily by lactam formation during excision of both

DOB and pC4-AP.15

Scheme 14. Inactivation of Pol β by DOB (A) and pC4-AP (B).

2.4 DNA Damage Tolerance by Translesion Synthesis

Despite the existence of efficient repair pathways to cope with many DNA lesions, some lesions remain unrepaired such that they are still present during replication.

Depending on the nature of the lesion, this can pose a considerable burden upon the cell, as the replication machinery may be unable to bypass the lesion. This is especially true for bulky or helically-distorting lesions which cannot be accommodated within the tight active sites of the polymerases which replicate the majority of the genome.165 Prolonged stalling of the replication fork can lead to replication fork collapse, dissociation of the replication machinery from DNA, leaving an exposed DSB.166 Although replication can be restarted by repair processes, aberrant repair of the DSB can cause chromosomal fusions or translocations, the aberrant exchange of genetic information between non-homologous chromosomes.167 Therefore, it is advantageous for cells to avoid replication fork collapse through damage tolerance mechanisms such as translesion synthesis (TLS) or template

34 switching.168 Template switching utilizes the core homologous recombination machinery to switch replication from the damaged template to the undamaged sister chromatid.169 This pathway results in the error-free replication past the DNA lesion. On the other hand, TLS involves the recruitment of specialized, error-prone polymerases which replicate past certain DNA lesions.165 This pathway is often mutagenic, a consequence of both the miscoding nature of some DNA lesions, as well as the low fidelity of most TLS polymerases. It is important to note that these damage tolerance mechanisms are not a form of DNA repair; they allow the replication machinery to continue replicating past the lesion but do not remove or repair the lesion.

2.4.1 Mechanism of Translesion Synthesis

It is no small feat for cells to replicate myriad DNA lesions which exhibit considerable structural and chemical diversity and are often intrinsically difficult to replicate. To do this, cells have evolved polymerases which are specialized for certain lesions. In humans, at least ten polymerases possess TLS activity.165 Many DNA lesions impose an impediment to both nucleotide insertion opposite the lesion as well as extension past it.170–172 Therefore, TLS can be divided into two distinct steps: 1) nucleotide insertion and 2) extension. In many cases, each step is carried out by a different polymerase, with lesion bypass requiring the sequential action of both polymerases.173–175 Notable exceptions to the two-polymerase paradigm exist; however, with some polymerases, such as Pol θ, possessing a unique ability to conduct both insertion and extension for several lesions (section 2.4.3.7).176–179

Figure 10. (A) Polymerase switching (B) and gap-filling models of TLS. Template DNA is indicated by a black solid line, while newly synthesized DNA is indicated by a blue arrow.

There are two prevailing mechanistic models for TLS in cells, the polymerase switching model and the gap filling model (Figure 10).180 There is evidence for each model, and it is likely that both contribute to TLS in human cells.181 In the polymerase switching model, a DNA lesion blocks progression of the replication fork. Stalling of the fork leads to the recruitment of specialized TLS polymerases which switch with the replicative polymerase to bypass the lesion (Figure 10a). The TLS polymerase then dissociates from

DNA or is displaced in some fashion to allow the replicative polymerase to associate with

36 the replisome and continue replication.182 A notable feature of the polymerase switching model is that TLS occurs during S phase, as DNA is being replicated. In the gap-filling model of TLS, DNA lesions present in the template are skipped during replication, leaving a single stranded gap (Figure 10b). The gap is then filled in by TLS polymerases later in the cell cycle, either in late S phase or in G2/M phase. Evidence for the gap-filling model stems from the observation that the important TLS polymerase, Rev1, is upregulated in

G2/M phase of the cell cycle.183 Additionally, quantitative TLS assays have established that TLS occurs both in S phase and in G2/M phase, indicating that the polymerase- switching and gap-filling models of TLS both contribute to cellular lesion bypass.181 There is also evidence that the choice between polymerase-switching and gap-filling is dictated by the type of DNA lesion.184

2.4.2 Discovery of Translesion Synthesis Polymerases

Early genetic experiments in E. coli identified genes involved in the SOS response, a response to DNA damage that involves the upregulation of proteins involved in DNA repair and mutagenesis.185 The dinB and umuC genes, which are now known to encode

TLS polymerases, were identified as essential for a mutator phenotype, where the mutational frequency increased in response to DNA damage.186 At the time; however, the lack of sequence similarity of these genes with any known polymerases precluded their identification as such. Instead, the dinB and umuC genes were believed to encode accessory factors which interacted with replicative polymerases, relaxing their fidelity and allowing them to bypass DNA damage.187 This paradigm was challenged; however, when the yeast

Rev1 protein, which promotes mutagenic bypass and is homologous to bacterial umuC, was identified as a dC transferase which inserts dC opposite abasic sites.188 This raised the

37 possibility that mutator genes identified in both prokaryotes and eukaryotes could encode non-canonical polymerases. Shortly after the discovery that Rev1 is a polymerase, a complex of the umuC and umuD genes in bacteria was shown to be an error-prone polymerase, Pol V, which bypasses abasic sites in vitro.189,190 Mutation of a conserved aspartate residue abolished polymerase activity, confirming that Pol V is a bona fide polymerase and excluding the possibility that the observed activity was due to contaminating polymerases.190 Pol IV, the product of the dinB gene, was purified shortly after this and was also shown to encode an exonuclease-deficient polymerase with the ability to extend from mismatched primer termini.191 These two polymerases belong to the

Y-family and are characterized by their lack of exonuclease activity, low processivity, and ability to replicate past certain DNA lesions. Within a few short years, additional Y-family polymerases were discovered in eukaryotes based on sequence homology to the bacterial umuC and dinB genes.192,193 These polymerases have been extensively characterized and shown to play important roles in DNA damage response and mutagenesis.

2.4.3 Translesion Synthesis Polymerases in Eukaryotes

There are many different polymerases in nature. E. coli contain at least five polymerases, S. cerevisiae at least ten, and humans at least 17.165,194,195 All of these catalyze the same nucleotidyl transfer reaction, yet polymerases differ considerably in their structure and sequence and can be classified into at least six different families on this basis:

A, B, C, D, X, and Y.196,197 Family C and D polymerases are absent in eukaryotes, where family B polymerases conduct the majority of replication.196 Family A and X polymerases are majorly involved in DNA repair but also function in TLS.165 The major TLS polymerases belong to the Y-family, with lower eukaryotes such as S. cerevisiae containing

38 two Y-family polymerases, Rev1 and Pol η. Mammals, on the other hand, contain Rev1 and Pol η along with Pol κ and Pol ι.180 Another polymerase, Pol ζ of the B-family, is also important for TLS in eukaryotes. Additional TLS polymerases have been identified in recent years, although these polymerases are generally not as well characterized as the Y- family polymerases and Pol ζ. The importance of TLS in humans is underscored by the defects observed in cells deficient in this process. For instance, loss of Pol η, which bypasses UV-induced cyclopyrimidine dimers, leads to the disorder xeroderma pigmentosum, which is marked by increased sensitivity to UV-irradiation and an increased risk of developing skin cancer.198 Knockout of Rev3 (the catalytic subunit of Pol ζ) is embryonically lethal in mice, and Rev3-/- embryonic fibroblasts are sensitized to a number of DNA damaging agents.199 Interestingly, TLS polymerases share many features in common but also possess unique characteristics which allow them to fulfil disparate roles in cells.

2.4.3.1 General Features of Translesion Synthesis Polymerases

Figure 11. Domain organization of prokaryotic and eukaryotic Y-family polymerases. Reproduced from Yang, W. Biochemistry 2014, 53, 2793–2803. Not subject to copyright.

Translesion synthesis polymerases are unified by their low fidelity on undamaged

DNA, a lack of exonuclease activity, and a relatively open active site which enables them to accommodate DNA lesions.200 Most TLS polymerases also exhibit low processivity, a feature which prevents them from synthesizing long stretches of DNA in cells and increasing the mutational frequency.165 Of the TLS polymerases, those of the Y-family are the best characterized at the structural, biochemical, and genetic level.201 The Y-family encompasses Pol η, Pol κ, Pol ι, and Rev1 in human cells. Each of these polymerases is composed of a polymerase domain and a regulatory domain (Figure 11).200 These regulatory domains have varied functions ranging from mediation of protein-protein interactions to modulation of activity based on the status of post-translational modifications.202,203 All Y-family polymerases also contain a ubiquitin-binding region. Pol

κ and Pol η bind ubiquitin by a ubiquitin binding zinc finger (UBZ) while Rev1 and Pol ι utilize a ubiquitin binding module (UBM).200,204,205 These interactions are important for the regulation of TLS (section 2.4.5). Additionally, all Y-family polymerases possess proliferating cell nuclear antigen (PCNA) interacting peptide (PIP) domains.165,200 PCNA is a clamp protein which acts as a necessary processivity factor for replication, interacting with the replicative polymerase Pol δ.206,207 Interaction of Y-family polymerases with

PCNA stimulates their activity on both damaged and undamaged DNA.208,209 TLS polymerases of the A-family (Pol θ and Pol ν) and of the X-family (Pol β and Pol λ) share important characteristics with Y-family polymerases such as a lack of exonuclease activity and low fidelity and processivity.165 However, they lack the domain organization that is characteristic of Y-family polymerases. Additionally, it is unclear how they are recruited for TLS, whether they interact with PCNA, and in many cases it is unclear which lesions

40 these polymerases bypass in cells. The current understanding of Y-family polymerases, along with the well-characterized B-family TLS polymerase Pol ζ is summarized here to contextualize the relative dearth of information about the function of Pol θ in TLS.

2.4.3.2 Rev1

Rev1 primarily inserts dC opposite template dG or some template lesions such as

AP.188 The basis for this nucleotide selectivity is the presence of an arginine Figure 12. Hydrogen bonding of dCTP and residue (Arg234) in the polymerase active Arg234 in the Rev1 active site site which hydrogen bonds with the incoming dCTP (Figure 12).210 Rev1 can therefore more accurately be classified as a dC transferase rather than a true polymerase.

Considerable evidence suggests that Rev1 rarely employs its dC transferase activity; however, often functioning in a regulatory role instead of inserting nucleotides.211–213 For example, genetic experiments identified a role for yeast Rev1 in promoting TLS through UV- induced 6-4 T-T photoproducts. Biochemical experiments; however, showed that Rev1 cannot insert a nucleotide opposite this lesion nor can it conduct extension.214 Furthermore, a Rev1 mutant lacking dC transferase activity still promotes TLS of 6-4 photoproducts.215 These observations suggest that Rev1 primarily plays a non-catalytic role in TLS. The full details of this role remain to be discovered; however, considerable evidence suggests that Rev1 mediates important protein-protein interactions. Rev1 interacts with all of the Y-family TLS polymerasess: Pol κ, Pol ζ, Pol η, and Pol ι.202 These interactions are mediated by a C-terminal portion of Rev1 that consists

41 of approximately 100 residues (RIR docking domain, Figure 11).200 Rev1 also interacts with Rad5, an important enzyme in the template switching pathway, via its C-terminal domain, playing an essential noncatalytic role in TLS of UV-induced DNA damage.211

Rev1 also possesses a BRCA1 C-terminal (BRCT) domain which is important for cellular response to UV irradiation.213 Although the exact function of this domain in TLS is unclear, it has also been proposed to mediate protein-protein interactions.214,216 These data suggest that the primary role for Rev1 in TLS is to act as a scaffold protein which interacts with other TLS polymerases as well as additional protein factors.

2.4.3.3 Pol η

Pol η, encoded by the Rad30 gene in S. cerevisiae and the XPV gene in humans, is a highly error prone polymerase (10-2 to 10-3) when replicating undamaged

DNA.201 Biochemical experiments showed that Pol η efficiently bypasses UV-induced cyclobutane pyrimidine dimers (CPDs), exhibiting the same fidelity on CPDs as on undamaged DNA.217 Cellular experiments have shown that the fidelity of Pol η during CPD bypass is even greater than this; however, exceeding the limit of detection (10-3) in a previous report.64 These results indicate that although TLS polymerases often have low accuracy on undamaged DNA, they contribute to genomic stability and, in some cases, to fidelity, by accurately replicating lesions which other polymerases cannot tolerate. Indeed, knockout or depletion of Pol η increases the mutational frequency in cells treated with UV light, and mutations in the XPV gene are linked with the cancer predisposition syndrome xeroderma pigmantosum.198,219

Although TLS polymerases exhibit low fidelity on undamaged DNA, they are more than just error-prone; they are specialized to replicate certain DNA lesions—in many cases in a largely error-free manner. However, this ability often comes at the cost of fidelity on undamaged DNA. The crystal structure of Pol η in complex with CPD-containing DNA reveals this specialization, showing that Pol η is able to accommodate both nucleotides of the CPD in its active site and act as a molecular splint to stabilize the lesion in a position where base stacking is maintained.172 Pol η is also capable of conducting TLS opposite lesions such as AP, Tg, and 8-oxodG in vitro.220 However, cellular experiments have shown that Pol η is not required for bypass of these lesions, indicating that Pol η is majorly involved in bypass of UV-induced lesions.177,221,222 Aside from TLS, Pol η is also important for somatic hypermutation, where its low fidelity on undamaged

DNA is used to generate targeted mutations in B cell receptor

223 genes, an essential process for adaptive immunity.

2.4.3.4 Pol κ

Pol κ, identified by homology of the human POLK gene to bacterial DinB, is the most accurate of the Y-family polymerases on undamaged DNA, misincorporating nucleotides with a frequency ranging from 10-3 to 10-

4.193,224 Pol κ is uniquely proficient at extending mispaired primer termini and can also extend past O6- methylguanine, 8-oxodG, and a CPD if another polymerase inserts a nucleotide opposite the lesion.225,226 The crystal structure shows that this ability stems from an α-helical region, termed the “N-clasp,” present at the N-terminus

43 of Pol κ (Figure 13A).227 This N-clasp is unique among Y-family polymerases and enhances the binding affinity of Pol κ to primer-template complexes and may serve to lock the polymerase onto the DNA (Figure 13B). This could prevent the polymerase from dissociating before the primer has time to align into the correct position for nucleotide insertion.227 Consistent with a role for Pol κ as an extender polymerase, it also extends past an acrolein-derived dG adduct, N2-(3-oxopropyl)-dG, along with 3-methyl dA, with Pol ι acting as the inserter polymerase in both cases.178,228

Figure 13. Crystal structure of Pol κ. A) Crystal structure of Pol κ bound to DNA. B) Comparison of Pol κ crystal structure to Pol ι and Pol η. Reprinted from Molecular Cell, Volume 25/Issue 4, Lone, S.; Townson, S. A.; Uljon, S. N.; Johnson, R. E.; Brahma, A.; Nair, D. T.; Prakash, S.; Prakash, L.; Aggarwal, A. K, Human DNA Polymerase κ Encircles DNA: Implications for Mismatch Extension and Lesion Bypass, 601-614. Copyright (2018), with permission from Elsevier 227

Interestingly, cellular studies have shown that, despite Pol κ’s proficiency in conducting the extension reaction, it often functions in the insertion step in cellular TLS while another polymerase acts as an extender. Pol κ functions in a largely error-free manner during the insertion opposite Tg and a dG-benzo[a]pyrene dihydrodiol epoxide (dG-BDPE) adduct, with Pol ζ acting as the extender polymerase.173,174 Pol κ also conducts the insertion step in the mutagenic TLS of cisplatin intrastrand crosslinks.173 Growing evidence suggests

44 that Pol κ has additional cellular roles outside of TLS such as in interstrand crosslink repair by a replication-independent pathway and may function in promoting microsatellite stability.229,230

2.4.3.5 Pol ι

Pol ι was discovered based on sequence homology of the human Rad30B gene to the Saccharomyces cerevisiae Rad30 gene encoding Pol η.231 Pol ι displays surprising DNA synthesis properties, replicating template dA relatively accurately (error rate of 3×10-4), while replicating template dT with a remarkably high error rate, inserting dG several times more efficiently than correct dA.231

The crystal structure of Pol ι in a ternary complex with a primer-template containing dA and an incoming TTP reveals that the unique fidelity of Pol ι stems from Hoogsteen base pairing in the enzyme active site, where the template base is flipped into the syn conformation (Figure 14).232 Template dA is replicated with reasonable fidelity because it can form two hydrogens bonds with the incoming T, even when dA is in the syn conformation. Although dG can also form two hydrogen bonds, this base pair is larger and is excluded from the active site, making misincorporation of dG opposite dA inefficient.232

Meanwhile, the inaccuracy of T replication by Pol ι stems from the lack of hydrogen bonding groups on the

Hoogsteen face of T, which does not allow Pol ι to correctly discriminate Figure 14. Hoogsteen hydrogen bonding in the Pol between dA and dG insertion. iota active site.

Biochemical experiments utilizing deaza analogues of dA and dG, which cannot Hoogsteen base pair effectively, inhibit synthesis by Pol ι, further confirming the role of Hoogsteen base pairing during DNA synthesis by this enzyme.233

The Hoogsteen base pairing mechanism utilized by Pol ι makes this polymerase especially well-suited to insert nucleotides opposite minor groove adducts of purines, such as N2 adducts of dG or N3 adducts of dA. By positioning the adducted template base in the syn conformation, the adducted N2 or N3 position is then positioned in the minor groove where there is less steric interference.232 Indeed, Pol ι conducts error-free insertion opposite both N2-(3-oxopropyl)-dG and 3-methyl dA in human cells.178,228 Pol ι also functions as a mutagenic, backup polymerase in the bypass of CPD lesions in Pol η- deficient cells, conducting the insertion step, while Pol ζ conducts extension.234,235

2.4.3.6 Pol ζ

Pol ζ is the only B-family polymerase involved in TLS, a unique classification, as other B-family polymerases include the highly accurate Pol δ, Pol ε, and Pol α.236 The structure of Pol ζ differs considerably from the Y-family polymerases in that it is a heterotetramer, originally believed to compose only the Rev3 catalytic subunit and the

Rev7 regulatory subunit but now known to be made up of the additional subunits POLD2 and POLD3.237–239 Most biochemical evidence about Pol ζ function has been gleaned through studies utilizing yeast Pol ζ due to the difficulty in purifying human Pol ζ.240

Considerable evidence suggests that Pol ζ is uniquely specialized to extend beyond a lesion after another polymerase has inserted opposite that lesion.28 For example, biochemical experiments show that yeast Pol ζ is extremely inefficient at insertion opposite an AP site, with the catalytic efficiency being decreased by several thousand-fold relative to

46 undamaged DNA.28 However, Pol ζ is highly efficient at extending past the lesion, exhibiting only a five-fold reduction in efficiency relative to undamaged DNA. Pol ζ also functions in the extension past Tg, 3-methyl dA, dG-BPDE, a (6-4) photoproduct, and a cisplatin intrastrand crosslink.173–175,178 The role for Pol ζ in extension past cisplatin intrastrand crosslinks explains how the upregulation of Pol ζ expression mediates acquired cisplatin resistance in tumors.241

Interestingly, knockout of the Rev3 subunit of Pol ζ is embryonically lethal, with embryos failing to survive past 12 days.242 Rev3-/- mouse embryonic fibroblasts can be generated if p53 is simultaneously disrupted, although p53 disruption alone is insufficient to rescue viability of embryos.243,244 These cells display increased sensitivity to DNA damage in the form of UV irradiation or cisplatin treatment. The molecular mechanism for the lethality of Pol ζ knockout is not entirely clear. It is possible that loss of Pol ζ severely compromises TLS, consistent with the requirement for Pol ζ for extension past many lesions. Compromised TLS could result in replication fork collapse, leading to lethal double strand breaks which cause p53-induced apoptosis.243 However, Pol ζ has also been implicated in homologous recombination and mismatch repair in addition to TLS.112,245

Therefore, embryonic lethality of Pol ζ knockout could result from a convergence of failures in several DNA repair and damage tolerance pathways.

2.4.3.7 Other Polymerases Involved in Translesion Synthesis

In addition to the four Y-family polymerases and Pol ζ, several other polymerases such as Pol β and Pol λ of the X-family and Pol ν and Pol θ of the A-family have been implicated in human TLS. The error-prone nature of these polymerases along with their lack of exonuclease activity suggested a possible role in TLS; however, the contributions

47 of these polymerases to cellular lesion bypass are not as well understood as those made by

Y-family polymerases. Pol β and Pol λ conduct in vitro TLS on templates containing non- bulky lesions such as 8-oxodG and AP sites.246,247 The physiological relevance of TLS by

Pol β has not been demonstrated; however, Pol λ plays an important role in the bypass of

8-oxodG in human cells.248

Similar to the X-family polymerases, Pol ν TLS activity has been demonstrated in vitro, but little is known of its TLS capacity in cells. Pol ν is unable to conduct TLS on templates containing AP, CPD, or a 6-4 TT photoproduct.249 Pol ν bypasses thymidine glycol quite efficiently in vitro, but does not contribute to its bypass in cells.177,249 Most interestingly, Pol ν bypasses a large DNA-peptide crosslink linked through the N6 position of dA.250 Pol ν also bypasses an interstrand crosslink between N6 positions on dA nucleotides on opposing strands. This ability is absent; however, on N2-dG crosslinks, suggesting a preference for bypassing major groove bulky adducts and crosslinks over those present in the minor groove.250

Slightly more is known about TLS by Pol θ, for which there is growing evidence of a significant role in cellular TLS. Pol θ bypasses AP in vitro, although it is unknown if it contributes to AP bypass in human cells.176 Recent genetic and biochemical studies show that Pol θ conducts both the insertion and extension steps during TLS of 3-methyl dA, Tg, and N2-3-oxopropyl dG.177–179 This ability to conduct both the insertion and extension steps is quite rare for a TLS polymerase, especially as it has been demonstrated for several lesions.

Unlike Y-family polymerases, TLS by Pol θ does not depend on interaction with Rev1.179 If the recent surge in information about TLS by Pol θ is any indication, it is likely that future studies will unveil additional roles for Pol θ in TLS.

2.4.4 Translesion Synthesis of Abasic Sites and Oxidized Abasic Sites

Abasic and oxidized abasic sites are the prototypical pro-mutagenic lesions, lacking a Watson-Crick base pairing face to direct nucleotide insertion. Abasic site bypass has been investigated in both prokaryotes and eukaryotes, using E. coli and S. cerevisiae as model organisms, respectively. AP bypass in E. coli is largely dependent on the SOS response, where DNA damage induces the expression of Pol II, Pol IV, and Pol V.185,251 Pol V appears to be the major TLS polymerase in AP bypass in E. coli; however, Pol IV and Pol

II also contribute to lesion bypass.252 Pol IV and Pol V are capable of both insertion and extension past AP, while Pol II has primarily been implicated in the extension step.253–255

Nucleotide insertion follows the A-rule, where dA is preferentially inserted opposite the lesion, perhaps due to more favorable π-stacking compared to other nucleotides.252

Abasic site bypass in S. cerevisiae differs considerably from that in E. coli. Studies using gapped duplex plasmids showed that AP bypass follows a “C-rule” where dC is preferentially inserted opposite AP.256 In this process, Rev1 utilizes its dC transferase activity to insert dC opposite the lesion while Pol ζ extends past the lesion.257 Pol η may make a minor contribution, likely acting as an extender, although in vitro studies have shown Pol η to be highly inefficient at both the insertion and extension steps for AP.258

Studies utilizing heteroduplex plasmids found quite different results, where AP bypass followed the A-rule, with dA being preferentially inserted opposite the lesion.259 This preferential insertion of dA opposite AP is consistent with observations in Ape1-deficient yeast treated with alkylating agents.28 During AP bypass on heteroduplex plasmids, Rev1 is required for TLS, but its catalytic activity is dispensable, consistent with a structural role for this polymerase.259 Pol ζ, the extender polymerase, is required for efficient AP bypass, but Pol η is also dispensable. These data suggest that, in yeast, a replicative polymerase inserts dA opposite the lesion and Pol ζ extends (Figure 15A). These observations are consistent with the ability for Polδ/Polζ to bypass abasic sites in vitro.28 A minor pathway involves nucleotide insertion by TLS polymerases such as Pol η, with Pol ζ conducting extension (Figure 15B). The basis for the differential dependence on Rev1 catalytic activity that was observed using different plasmid-based approaches is unclear. It could reflect a different mechanism of TLS in each case with TLS of the gapped plasmid occurring via the gap-filling model of TLS, where a lesion present in a heteroduplex plasmid may be bypassed the polymerase-switching model (Figure 9). Notably Pol θ is absent in yeast, so studies about AP bypass in yeast could not shed light on the role for Pol θ in bypassing this lesion.

Figure 15. Bypass of AP in yeast. A) Pol δ and Pol ζ (major pathway). B) Bypass of AP by Pol η (or other TLS polymerases) and Pol ζ (minor pathway).

Compared to S. cerevisiae, less is known about AP bypass in humans. Pol ζ has been well-established as an important polymerase in this process, presumably acting as an extender polymerase.173,252 Rev1 has also been shown to participate in AP bypass in human cells, likely playing a structural role as dC insertion opposite AP was not observed.252 The identity of the inserter polymerase for AP bypass in human cells is unknown, although it is evident that Pol η does not play a prominent role in this process.221 Replicative polymerases have been implicated in human AP bypass, presumably in the insertion step, on the basis that the process is sensitive to aphidicolon which inhibits replicative

51 polymerases.221 However, Pol ζ is also a member of the B-family and may be inhibited by aphidicolin. Yeast Pol ζ is insensitive to aphidicolin, although human Pol ζ differs substantially from the yeast homologue, and information on the sensitivity of human Pol ζ to aphidicolin is lacking.237 The function of other polymerases AP bypass is unclear; however, it is likely that polymerases other than Pol ζ function in bypass, as residual bypass was observed when Pol ζ was knocked down.252

Despite the importance of these lesions as products of oxidative stress and chemotherapy, little is known about TLS of oxidized abasic sites in eukaryotic cells, although replication of C2-AP, C4-AP, and L has been reported in E. coli. Interestingly, the three lesions are treated distinctly, a notable occurrence due to the lack of a Watson-Crick face. C2-AP is bypassed by a template- misalignment mechanism, which gives one-nucleotide frameshift mutations (Scheme

15).260 These frameshift mutations are dependent upon the concerted action of Pol IV and

Pol II, while Pol V suppresses frameshift mutations and promotes formation of full-length extension.

Scheme 15. Bypass of C2-AP in E. coli by Pol II and Pol IV.

Bypass of C4-AP gives rise to large amounts of three-nucleotide deletions, which— similar to bypass of C2-AP—are dependent upon Pol IV and Pol II.261 Replication of L results in considerable amounts of dG insertion, with dG insertion being preferred over dA

52 on most sequences.262 Extension beyond the lesion gives rise to template misalignment, forming frameshift mutations on appropriate sequences (Scheme 16).

Scheme 16. Bypass of L in E. coli by template misalignment.

This mechanism is distinct from that observed for C2-AP, where template misalignment occurs during nucleotide insertion opposite the lesion (Scheme 16). Information on the eukaryotic bypass of oxidized abasic sites is even more limited, comprising a single report about L bypass in S. cerevisiae. Both dA and dC are inserted opposite L, with approximately equal frequency.263 Pol η is dispensable for TLS with L present in the template, but Rev1 and Pol ζ are essential.

2.4.5 Translesion Synthesis of

Thymidine Glycol

Unlike oxidized abasic lesions,

Tg bypass in human cells is well understood. Tg is cytotoxic, but it is not highly mutagenic, indicating that polymerases replicate this lesion Figure 16. Replication past Tg by human translesion synthesis polymerases. relatively accurately, although quite

53 inefficiently in the case of replicative polymerases.174,264 Unrepaired Tg lesions which are present during replication strongly block progression of the replication fork.171 This requires the action of specialized polymerases to bypass the lesion and re-start progression of the replisome (section 2.4).174 This bypass occurs by two distinct pathways in human cells (Figure 16).177 One pathway occurs with high fidelity and involves insertion of dA opposite the lesion by Pol κ and subsequent extension by Pol ζ. The second pathway requires only Pol θ, which conducts translesion synthesis and subsequent extension past the lesion and is weakly mutagenic.

2.4.6 Regulation of Translesion Synthesis in Eukaryotes

Because TLS polymerases exhibit low fidelity on undamaged DNA, it is necessary that their access to DNA is tightly regulated to avoid excessive mutagenesis. To accomplish this, expression levels of TLS polymerases are often kept relatively low. For example, p53 promotes the depletion of Rev1 and Rev3 transcripts.265

Other TLS polymerases are generally expressed at low levels, but their expression is induced in response to DNA damage. Transcription of the Rad30 gene in yeast, which encodes Pol η, is induced by approximately four-fold in response to UV-irradiation.192

Similarly, Pol κ expression is induced by treatment of mouse cells with 3- methylcholanthrene, a polycylic aromatic hydrocarbon which forms bulky N2-dG adducts

(which are bypassed by Pol κ) after bioactivation.266 In addition to regulating expression levels of TLS polymerases, eukaryotic cells also regulate their access to the replication fork. This process is heavily regulated by post-translational modifications, both of PCNA as well as of individual polymerases (Figure 17).

Figure 17. Regulation of TLS by post-translational modification of PCNA. For clarity, the lagging strand is omitted after the first step.

Genetic experiments in the budding yeast Saccharomyces cerevisiae showed that replication fork stalling leads to monoubiquitination of K164 of PCNA by the ubiquitin ligase enzymes Rad6 and Rad18.267 Cells expressing the K164R mutant form of PCNA, which cannot be monoubiquitinated at this position, are hypersensitive to UV-induced

DNA damage and experience severe TLS defects.268 PCNA monoubiquitination is also important for TLS past CPD and a cisplatin intrastrand crosslink in human cells, although it is not essential, indicating that other TLS polymerases which do not require PCNA ubiquitination are able to compensate to some degree.269 Interestingly, monoubiquitination

55 of Pol η instead of PCNA negatively regulates TLS.203 Pol η is constitutively ubiquitylated on a PCNA-interacting region, preventing interaction of the polymerase with PCNA. UV- irradiation of cells causes deubiquitylation of Pol η, promoting its interaction with PCNA and allowing it to participate in TLS. This suggests that human cells have redundant mechanisms, both to prevent aberrant access of TLS polymerases to DNA and to specifically promote interaction of these polymerases with DNA in response to DNA damage.

An additional question regarding the regulation of TLS concerns how specific polymerases are selected to bypass a lesion. Biochemical evidence has indicated that each

TLS polymerase has a defined substrate scope, conducting TLS (either insertion, extension, or both) for a certain subset of lesions but not on other lesions.176,249,270 Therefore, it is conceivable that TLS for a specific lesion would be governed only by the ability of certain polymerases to bypass the lesion. TLS polymerases could therefore be recruited in a sequential or random manner to the lesion site until the lesion was bypassed. However, genetic evidence contradicts this model. For instance, Tg is bypassed by multiple polymerases in vitro: Pol η, Pol ν, and Pol θ can conduct both the insertion and extension steps, while Pol κ can insert opposite the lesion while Pol ζ extends past the lesion.177,220,249

Intriguingly, Pol η and Pol ν do not contribute to cellular bypass of Tg, with only Pol θ and

Pol κ/Pol ζ conducting cellular bypass.177 This suggests that TLS is tightly regulated, with specific polymerases being recruited under specific circumstances, although further insight is necessary to understand such a mechanism.

2.5 DNA Polymerase Theta (Pol θ)

2.5.1 Discovery and Cloning

Figure 18. Domain organization of human Pol θ.271

Pol θ was first identified in Drosophila when the mus308 gene was found to encode a protein that confers resistance to interstrand crosslink-inducing agents.272 Molecular cloning of mus308 predicted a 229 kDa protein, now termed Pol θ, with 55% sequence homology to E. coli Pol I.273 Mammalian Pol θ was predicted to exist based on homology of the human POLQ gene to the mus308 gene.273 The polymerase activity of Pol θ was not directly demonstrated for several years until the human POLQ gene was cloned and its gene product, a 2590 residue, 290 kDa polymerase, was purified 274 Pol θ, an A-family polymerase, bears considerably homology to E. coli Pol I, although Pol θ lacks exonuclease activity.274 Pol θ has an interesting domain organization, possessing an N-terminal helicase domain which is linked to the C-terminal polymerase domain by a large central domain which is predicted to be unstructured (Figure 18). For some time the helicase domain was predicted to be inactive; however, recent experiments revealed that Pol θ does possess helicase activity.274,275 The function of the central domain on the other hand is still poorly understood, although there is evidence that this domain interacts with Rad51 and suppresses homologous recombination.88 Interestingly, the central domain comprises approximately 1600 residues in vertebrates and only 800-1000 residues in plants and metazoans.18

2.5.2 Structure and Unique Characteristics of Pol θ

Pol θ possesses several unique characteristics that distinguish it from other members of the A-family, and indeed any other mammalian polymerases. Unlike well- characterized A-family polymerases such as Pol I, Pol θ is relatively error-prone (10-3) and

274 lacks exonuclease activity. This results from mutation of Glu357, Asp355, Asp424 which are conserved in A-family polymerases that possess exonuclease activity. Interestingly, exonuclease activity was detected in Pol θ isolated from human cells, suggesting that the polymerase may associate with an exonuclease in vivo, increasing its fidelity. Pol θ is also the only polymerase known to possess its own helicase domain. The catalytic activity of this domain was demonstrated quite recently, providing interesting implications for the function of Pol θ in alt-EJ. (section 2.5.3)

The polymerase domain of

Pol θ shares considerable homology with Pol I, and the overall structure of the polymerase domain is similar to that of Taq polymerase and Pol I. Pol

θ possesses five sequence inserts, Figure 19. Location of insertion loops in polymerase three of which are unique to Pol θ.271 domain of Pol θ.271

These sequence inserts form distinct loop structures. Two of these loops are present in the vestigial exonuclease domain, while the other three are distributed between the palm and thumb subdomains (Figure 19).276 Insertion loop 1 in the thumb subdomain plays a minor role in processivity and fidelity of the polymerase, while loops 2 and 3 in the palm subdomain are essential for TLS. These insertion loops, along with other unique structural

58 elements are responsible for the unique ability of Pol θ to conduct both insertion and extension steps of TLS during bypass of AP, 3-methyl dA, Tg, and an N2-acroline derived

177–179 dG adduct. Insertion loop 2 contains Arg2254, a positively charged residue which is highly conserved in Pol θ and mus308-like polymerases.276 This residue, absent in other

A-family polymerases, contacts the 3’-terminal phosphate of the primer strand (Figure 20).

Pol θ possesses two additional positively charged residues (Lys2181 and Arg2202) absent in other A-family polymerases, which contact the phosphodiester backbone of the primer strand (Figure 20). These three positively charged residues allow Pol θ “to “grip” the primer strand, stabilizing it in a position which allows for extension past a lesion.276

Mutation of Arg2254 eliminates the ability of Pol θ to conduct TLS on templates containing

AP, confirming the importance of this unique contact for TLS.

Figure 20. Crystal structure of Pol θ in a ternary complex with DNA and ddATP. Reprinted by permission from Springer New York, LLC: Nature. Nature Structural and Molecular Biology.276 Human DNA polymerase θ grasps the primer terminus to mediate DNA repair. Karl E Zahn, April M Averill, Pierre Aller, Richard D Wood, Sylvie Doublié, Copyright 2018.

2.5.3Reproduced Function from of Zahn. Pol276 θ

In addition to its role in TLS, vertebrate Pol θ has also been implicated in BER, as it possesses lyase activity and participates in BER of 8-oxodG in chicken cell extracts.17,159

Limited proteolysis mapped the lyase activity of Pol θ to the polymerase domain of the enzyme, although the specific residues involved in this activity are unknown.17 This lyase activity is relatively weak, with Pol θ excising 5’-dRP with a single-turnover rate constant of 0.15 s-1, approximately 1000-fold slower than Pol β.13,17 Experiments in Drosophila validated an important role for Pol θ in ICL repair, possibly in the bypass of unhooked

ICL’s, although Pol θ does not appear to contribute to ICL repair in mammalian cells.20,277,278 The best characterized role of Pol θ is in the repair of DSBs by alternative end-joining (alt-EJ).

Cellular and biochemical studies have revealed many details about the role for Pol

θ in alt-EJ. Pol θ is uniquely suited to participate in this pathway, as in vitro end joining cannot be reconstituted with other polymerases and Pol θ knockout eliminates alt-EJ in human cells.119,279 Pol θ promotes annealing of microhomologies, bringing both ends of the break into close proximity.119 Interestingly, this occurs even in the absence of sequence homology, suggesting that annealing occurs in two steps, with Pol θ promoting initial synapse formation followed by thermodynamically driven base pairing between microhomologies.119 Base-pairing between microhomologies is insufficient for joining of the DSB; however, as the polymerase activity of Pol θ is essential for alt-EJ.20,279

Presumably the minimally paired microhomologies which initiate alt-EJ cannot be ligated efficiently and must be extended by the polymerase activity of Pol θ to generate longer complementary regions before ligation. The polymerase activity of Pol θ is well suited for this role in extending from microhomologies, as Pol θ is capable of extending mismatched

60 primer termini, allowing Pol θ to promote end joining in the absence of significant homology.280

Some aspects of alt-EJ remain poorly understood, including its regulation as well as the function of Pol θ helicase domain in this pathway. The importance of the helicase domain was obscured by some studies which showed that it is dispensable for alt-EJ in human cells, where only the polymerase domain is required.20,279 In vitro studies seemed to corroborate these experiments by showing that the polymerase domain of Pol θ promotes synapse formation between the two ends of the break and is sufficient to conduct end- joining without the helicase domain.119 Furthermore, the helicase domain was believed to be inactive for some time, as original attempts to demonstrate helicase activity for Pol θ were unsuccessful.274,275 However, recent studies have revealed a function for the Pol θ helicase domain.

Most studies addressing the role of Pol θ in cellular DSB repair utilized overexpression of truncated Pol θ (for example: overexpression of only the polymerase domain) in cells where endogenous Pol θ was knocked out.20,279 Pol θ levels in cells are typically low, so overexpression of the polymerase domain may have masked the function of the helicase domain in cellular repair.281 Recently, CRISPR-Cas9 targeting was used to inactivate Pol θ helicase and ATPase activity by mutation of a single residue of the endogenous POLQ gene.281 This allowed for expression of a helicase- and ATPase-inactive form of Pol θ under the control of the endogenous promoter. This mutation reduced repair of genomic DSBs by alt-EJ, indicating that the helicase domain of Pol θ is important for alt-EJ.281 Biochemical experiments showed that the ATPase activity of Pol θ is essential for displacement of replication protein A (RPA), which inhibits alt-EJ.281 This study did

61 not directly demonstrate that helicase activity was necessary for DSB repair, as the helicase domain possesses both ATPase and helicase activity, and only the helicase activity was demonstrated in in vitro experiments. Even more recently, helicase activity was detected for Pol θ helicase domain, and this activity was shown to facilitate strand displacement synthesis by Pol θ in vitro.275 Although the importance of helicase activity has not been directly demonstrated in vivo, this may have resulted from the belief that Pol θ does not possess helicase activity. Therefore, future studies may elucidate a role for this activity in alt-EJ.

2.5.4 Pol θ as a Cancer Target

Pol θ is broadly expressed across human tissues, although expression levels are the highest in the testes and in hematopoietic cells.274 Pol θ knockout mice are viable and grow normally, indicating that Pol θ is non-essential for healthy cells.277 However, higher levels of micronuclei, indicative of DNA damage, are present when the POLQ gene is deleted.277,282 Pol θ has emerged as an attractive cancer target for several reasons.283

Knockout of Pol θ sensitives bone marrow cells to radiation, indicating that Pol θ inhibitors could be useful radiosensitizers.282 Furthermore, Pol θ expression is upregulated in breast, ovarian, and lung cancers, and this upregulation correlates with poor patient prognosis.19,284,285 Pol θ is also the only DNA polymerase known to be upregulated in either breast or non-small cell lung carcinomas.19,284 In addition to the upregulation of Pol θ expression in many cancers, Pol θ may be a good cancer target for synthetic lethality targeting approaches. Synthetic lethality occurs when the simultaneous disruption of two genes is lethal to cells, while the disruption of each of the genes individually is not.286

Depletion of Pol θ is synthetically lethal in ovarian cancer cells which have defects in

62 homologous recombination.88 These cells are hyper-dependent upon alternative end- joining to repair endogenous DSBs due to loss of homologous recombination. Therefore, inhibition of Pol θ in cancers with homologous recombination defects may be viable therapeutic strategy. As loss of Pol θ does not affect embryonic viability, development, or organismal growth in normal cells, inhibition of Pol θ may have relatively few side effects.

2.6 Ku70/80

Ku is a highly abundant (400,000-500,000 molecules per cell) heterodimeric protein composed of the Ku70 and Ku80 subunits, encoded by the XRCC6 and XRCC5 genes respectively.89,287 Ku is present in eukaryotes from S. cerevisiae and C. elegans to insects and humans.90 Ku orthologues are also present in many prokaryotes, although prokaryotic Ku is composed of only a single subunit.287,288 Ku was originally identified in the serum of humans with autoimmune diseases.289 Biochemical experiments utilizing Ku immunopurified from patients showed that Ku binds double stranded DNA.290 Cellular studies confirmed that cells deficient in Ku70 or Ku80 are hypersensitive to ionizing radiation due to their inability to repair DSBs by NHEJ.291,292 This pathway is important for the repair of DSBs, especially those formed by ionizing radiation, and is also used in B cells and T cells in VDJ recombination, a programmed form of DSB repair, that is used to diversify the genes encoding T cell receptors and immunoglobulins.293 Although Ku is essential for NHEJ, homozygous knockout of Ku is not embryonically lethal in mice, indicating that mammalian cells can survive without NHEJ.291 There is some evidence, however, that loss of Ku is embryonically lethal in humans. For example, defects in NHEJ often lead to a severely compromised immunodeficiency (SCID) phenotype due to a loss of VDJ recombination.287 Mutations in Artemis, XLF, Ligase IV, and DNA-PKcs have

63 been identified in SCID patients, but corresponding mutations in Ku have not, suggesting that these mutations impact viability.287 Additionally, human cells deficient in Ku80 undergo apoptosis after repeated cell divisions, likely due to Ku’s role in telomere maintenance (section 2.6.4).294,295

2.6.1 Ku Structure

Figure 21. Ku crystal structure viewed down the DNA helix (A) and in a side view (B).89 PDB ID: 1JEY. Ku is heterodimeric protein composed of 70 kDa and 86 kDa subunits. These two subunits associate tightly with one another, forming a stable heterodimer. Knockout of either subunit causes a considerable depletion in the soluble fraction of the other subunit, which can be rescued by ectopic expression, suggesting that heterodimerization is necessary for Ku stability.296 Comparison of the crystal structure of free Ku with that of

DNA-bound Ku shows that Ku has a central ring which is pre-formed for binding DNA

(Figure 21).89 Photo-crosslinking studies and crystallographic evidence show that when Ku loads onto dsDNA, the Ku80 end is positioned internally, with Ku70 located proximal to the DSB end.89,297 The DNA is threaded through a channel which has considerable positive

64 charge. Importantly, the protein makes no contacts with nucleobases, allowing it to bind

DNA in a sequence-independent fashion, an important characteristic for a protein which promotes NHEJ, a pathway which repairs DSBs with no sequence preference.89,298 Each subunit of Ku also possesses a C-terminal domain that is important for protein-protein interactions. For example, the C-terminus of Ku80 is essential for interaction with DNA-

PKcs, an important interaction for the repair of many DSBs.299

2.6.2 Structural Role of Ku in NHEJ

The primary function of Ku in NHEJ is to bind DSBs and act as a scaffolding factor for other repair proteins to process and join the DSB. High resolution imaging has shown that two molecules of Ku bind to a DSB, presumably one molecule loaded onto each end of the break.92 Ku then plays an important role in repair by interacting with DNA-PKcs to form the DNA-PK complex. Although lower eukaryotes such as yeast lack DNA-PKcs,

DNA-PKcs is essential in vertebrates, where it plays an important role in the choice between end processing and direct ligation without processing.95,287 Binding of Ku to a

DSB also plays an important regulatory role in DSB repair pathway choice, as Ku physically blocks the end resection necessary to promote HR.108 Interestingly, this function of Ku extends to HR-mediated restart of collapsed replication forks, where Ku regulates the length of end resection.300

2.6.3 Ku Lyase Activity

In addition to its essential role as an NHEJ scaffolding factor, Ku plays an additional role in the repair of some complex DSBs. Some DSBs, particularly those generated by ionizing radiation, contain nearby DNA lesions which can interfere with end-

65 joining.301 AP sites in particular present a strong block to end-joining, either by destabilizing the synaptic complex or interfering with ligation, and must be removed for end-joining to proceed.16 Ku removes AP from DSB termini by acting as a lyase, utilizing one of several nucleophilic lysines to attack AP and promote β-elimination.16 Biochemical analysis showed that Ku lyase activity is restricted to AP or 5’-dRP near DSB termini.302

Ku removes AP in 5’-overhangs or near the DSB terminus (Figure 22). Ku does not act on a nicked duplex containing 5’-dRP, the BER intermediate generated by Ape1, nor does Ku act on AP sites within 3’-overhangs. Comparison of Ku lyase activity with Pol β and Pol λ showed that Ku is most proficient at removing 5’-dRP at a DSB terminus, while Pol β and

Pol λ have greatly reduced activity on this substrate.16 So although removal of AP and 5’-

-1 -1 dRP by Ku is relatively sluggish (kobs: 0.1 min and 0.26 min , respectively) compared to

BER (6 s-1 for Pol β), Ku is the most efficient enzyme at removing abasic lesions from

DSB termini.13,16

Figure 22. Substrate scope of Ku lyase activity.302 Systematic mutagenesis of lysine residues predicted to reside close to AP and 5’- dRP when the protein binds damaged DNA revealed that the lyase activity possesses

303 considerable redundancy. Mutation of a single residue, Lys31, reduces activity to

66 approximately 30% of the wild type, demonstrating that Lys31 is the primary catalytic residue for Ku lyase activity. However, mutation of two additional lysines (Lys160 and

Lys164) in Ku70 and six lysines (Lys543-545, Lys565, Lys566, and Lys568) in Ku80 was necessary to completely abolish lyase activity.303 Interestingly, Ku from Xenopus laevis has negligible lyase activity.303 It is possible that other enzymes substitute for Ku lyase activity in Xenopus laevis and other organisms where Ku lacks lyase activity. DNA-PKcs is a possible candidate for this activity, as human DNA-PKcs compensates to some degree for AP removal when mutant lyase-dead Ku is utilized for in vitro experiments.303

2.6.4 Ku Function at Telomeres

Eukaryotic chromosomes are linear and therefore end in a structure called the telomere. Telomeres are long stretches (several kilobases in humans) of repetitive DNA sequences which are bound by telomere-associated proteins which protect the telomere.304

This protection of telomeres is important for cell viability, as unprotected telomeres can be degraded by nucleases or incorrectly fused to DSBs or other telomeres by end-joining processes.304 Paradoxically, Ku—an essential component of NHEJ—associates with telomeres and prevents telomere fusions.305 Disruption of the LIG4 gene in yeast has no effect on telomere shortening, indicating that Ku’s role in maintaining telomere length is independent of its role in NHEJ.306 Although Ku is not essential for viability in other vertebrates, it is essential in human cells, where homozygous knockout leads to apoptosis after repeated cell divisions.295 This results from critical telomere shortening in the absence of Ku.307 Ku’s importance for telomere maintenance appears to stem from its interaction with telomerase, the telomere-specific reverse transcriptase which synthesizes telomeric

DNA.308 Ku binds to the RNA component of telomerase in a similar manner to which Ku

67 binds to DSBs.309,310 Ku also interacts with protein subunits of telomerase such as Est1 and

Est2 to recruit telomerase to telomeres.311 Interestingly, although Ku is best characterized as an essential component of NHEJ, human cells can survive without this pathway but cannot survive without Ku-promoted telomere lengthening.

2.6.5 Ku Removal from DNA

The crystal structure of Ku bound to DNA shows that Ku encircles the DNA duplex and likely becomes topologically trapped following DSB ligation.89 Ring-shaped replication factors such as PCNA also encircle duplex DNA and must eventually be removed.312 These processivity clamps are opened by breaking the noncovalent interaction between two of the subunits, providing a mechanism for loading and unloading of these ring-shaped proteins from intact DNA.313,314 Such a mechanism is possible for Ku removal; however, the Ku70 and Ku80 subunits both encircle DNA and interact with each other extensively and tightly.89 Therefore a large conformational change is likely necessary for the Ku70 and Ku80 subunits to dissociate.89 Alternatively, Ku could be proteolytically degraded following DSB repair. Experiments in Xenopus laevis cell extracts show that Ku is polyubiquitylated by K48 branched ubiquitin and then degraded by the proteasome.315

Proteolytic activity is not required for removal of Ku from DNA; however, suggesting that polyubiquitylation triggers the unloading of Ku from DNA, which is followed by Ku degradation (Figure 23).315 It is possible that Ku is degraded by the DNA-dependent protease SPRTN, which was discovered after the original report of Ku removal by a proteasome-independent pathway, although there is presently no evidence for this.316 Ku may also be actively unloaded by VCP, an enzyme which unfolds ubiquitylated proteins.317

Proteomics studies have shown that VCP interacts with Ku in a DNA damage-, ubiquitin-

, and neddylation-dependent manner. Inhibition of Ku neddylation (modification with a ubiquitin-like protein, NEDD8) reduces Ku ubiquitylation and inhibits its removal from repaired DSBs.318 Therefore, it is likely that that Ku removal from DSBs is regulated by several post-translational modifications. The available evidence suggests that Ku is actively unfolded by VCP following repair of DSBs and is then degraded, although this has not been conclusively shown yet.

Figure 23. Proposed mechanism for Ku removal from repaired DSBs.317

3. Results and Discussion

3.1 Bypass of Abasic and Oxidized Abasic Lesions by Pol θ

3.1.1 Expression and Purification of Pol θ Catalytic Core (residues 1792-2590)

The catalytic core of Pol θ (residues 1792-2590) was used for all experiments presented here. This was previously identified as the minimal fragment of Pol θ which retained the biochemical properties of the full-length enzyme.271 This truncated protein was utilized because it can be purified in large amounts (>5 mg/L culture) from E. coli. The full-length protein cannot be expressed in E. coli. This is likely due to its large size and unique domain organization, containing an N-terminal helicase-like domain separated from the C-terminal polymerase domain by a large (1200 residues) central domain which is predicted to be largely unstructured.319 Unstructured proteins often prove challenging for bacterial expression due to a lack of chaperone proteins in bacteria. The full-length protein can be expressed in insect cells using a baculovirus expression system, but the yield (100

μg/L) is prohibitively low for conducting extensive biochemical experiments.319 After troubleshooting expression and purification conditions, expression of the catalytic core of

Pol θ gave consistently high yields, allowing for the biochemical experiments presented here.

Pol θ catalytic core (hereafter referred to as Pol θ) was encoded by a pSUMO vector containing N-terminal hexaHis and SUMO tags. The protein was expressed in E. coli

(Rosetta 2 DE3 pLysS) using autoinduction for 60-64 hr at 20 °C. For autoinduction, it is important that the culture has saturated before the cells are harvested. The expression medium contains glucose and lactose, and the glucose must be depleted to allow expression of the target protein which is under control of the lac operon. Initial attempts to purify Pol

θ were low yielding (< 500 μg/ L of culture), largely because the cultures were not seeded with enough cells and did not reach saturation. In these attempts, 5 colonies were

resuspended in 20 mL LB media and 1

mL of this suspension was added to 1 L

of expression medium, as reported

previously.271 When cells were harvested

at 60 hr, saturation had not been reached

and the protein yields were poor. If

cultures were grown until saturation (84

hr under these conditions), the protein

activity was poor. However, by seeding Figure 24. Analysis of Pol θ purity at each step of the purification by 10% SDS-PAGE. The molecular the expression culture with a greater weight (MW) of the protein ladder is shown on the left. Reprinted with permission from Laverty, D. J.; number of colonies (5 colonies per L of Greenberg, M. M. Biochemistry 2017, 56, 6726– 6733.Copyright 2018. American Chemical Society. culture), the cultures saturated by 60-64 hr, and protein yields approached 5 mg/L of culture. Pol θ was purified by Ni2+ affinity chromatography using a linear gradient to 500 mM imidazole. SUMO-tagged Pol θ eluted in a broad peak starting at approximately 150 mM imidazole. The SUMO-tagged protein migrates at approximately 100 kDa on SDS-PAGE (Figure 24). A major impurity of approximately 70 kDa was present after this step, along with lesser amounts of smaller contaminating proteins. Peak fractions (3, mL each, ~15-25 mL) were pooled and applied to Heparin column, and the protein was eluted by applying a linear gradient from 300 mM

NaCl to 1 M NaCl. Pol θ eluted in a well-defined peak at approximately 50% elution buffer.

The purity was considerably improved at this step, although minor impurities were still present. Peak fractions (2 mL each, ~6-8 mL) were pooled and incubated at 4 °C with

SUMO protease 2 to cleave the SUMO protein (which contains the hexaHis tag). The

SUMO cleavage reaction was applied to a Ni2+ column to bind the protease as well as the cleaved tag and any uncleaved Pol θ. Untagged Pol θ bound very weakly to the column and was rapidly eluted. The protein could be concentrated using an Amicon centrifugal filter

(30 K cutoff), although significant amounts of Pol θ were retained on the membrane of the centrifugal filter at this step leading to sample loss. Yields were typically in the range of 5-

10 mg of Pol θ/L of culture medium but were reduced by up to 50% when concentrated.

3.1.2 Pol θ Polymerase Active Site Titration

For most polymerases inserting the correct dNTP, the rate-limiting step occurs after the chemical step and may be either product release or a conformational change following the chemistry.320,321 Therefore, when a primer-template complex is in excess over polymerase, product formation (resulting from insertion of a single dNTP) obeys a pre-steady state

“burst” model, where the first equivalent of product is formed in a rapid burst that is faster than the steady-state rate.321 Because the amplitude of the burst phase corresponds to exactly one equivalent of catalytically active polymerase, the observed amplitude can be divided by Figure 25. Representative active site titration of the expected amplitude (based on the Pol θ. Pol θ (25 nM) was incubated with 27 (50 nM) and dATP (500 μM) for the indicated time.

72 protein concentration) to give the active fraction. For the active site titration of Pol θ, a two-fold excess of primer-template 27 was employed such that the maximum burst amplitude was 50%. A Kintek RQF-3 Rapid Quench instrument was used to vary the reaction time from 3 ms to 5 s and the polymerase active fraction was determined to be

47.5 ± 6.6% for the batch of Pol θ received from Dr. Sylvie Doublié (Figure 25).322 A later preparation of Pol θ which I isolated was 40% active in this assay.323

3.1.3 Nucleotide Insertion Opposite Abasic Lesions by Pol θ

Pol  conducts TLS on templates containing F, a stable analogue of AP, inserting primarily dA opposite the lesion and efficiently extending past the lesion.176 Although AP bypass by Pol θ has not been demonstrated in vivo, such a phenomenon may contribute to cellular resistance to DNA damaging agents.20 The ability of Pol θ to conduct TLS on templates containing oxidized abasic sites was previously unknown, even though Pol θ promotes resistance to ionizing radiation and bleomycin which produce oxidized abasic sites.20 In E. coli, oxidized abasic sites are replicated differently from AP, giving rise to deleterious frameshift mutations.260–262 Therefore, the ability of Pol θ to conduct TLS on templates containing oxidized abasic sites is of interest (Chart 1). Templates were synthesized containing dU or F (Figure 26). dU is converted to AP by treatment with uracil

DNA glycosylase (UDG), while F is used as a stable analogue of AP. Templates containing previously reported photochemical precursors to C4-AP and L were also synthesized, such that C4-AP and L were generated by photolysis at 350 nm (10 min for C4-AP and 20 min for L).324,325 Curiously, 5’-radiolabeling of the perfectly complementary primer was extremely inefficient. When the primer was lengthened by incorporation of two thymidines at the 5’-end, labeling was considerably more efficient.

Figure 26. Generation of abasic and oxidized abasic sites for analysis of TLS by Pol θ.

Chart 1. Primer-templates used for analysis of nucleotide insertion opposite each lesion.

Pol θ was previously reported to insert nucleotides opposite F following the preference of dA > dG > dT (with insertion of dC being essentially negligible).176 Steady- state kinetic analysis on primer-templates 28-30 indicated that Pol θ follows this same trend on AP and L but follows the trend dA ≈ dG > dT for C4-AP (Table 1). Incorporation of dC was not observed opposite any lesion with concentrations up to 1 mM in a 5 min reaction, indicating that insertion of dC is essentially negligible opposite each lesion (not shown).

Steady-state kinetic analysis revealed subtle differences in nucleotide insertion preference for each lesion. For AP (28), dA is preferred 7-fold over dG and 50-fold over T. The same trend is observed opposite L (29), where dA is preferred 6-fold over dG which is preferred

74 over T. Insertion of T is too inefficient to measure, suggesting that Pol θ shows the strongest preference for purines opposite L. The nucleotide preference is the most different opposite

C4-AP (30) where dA and dG are inserted with equal efficiency. Insertion of dA opposite

C4-AP is preferred over T by only 14-fold, a 3- to 4-fold reduction in preference relative to AP. The lack of preference for dA over dG and the smaller preference for purines over thymidine, indicate that C4-AP gives the widest array of nucleotide insertion events of the three lesions. Pol θ also exhibits a considerable reduction in nucleotide insertion efficiency for both oxidized abasic sites, with dA insertion opposite C4-AP only 12% as efficient as that opposite AP and insertion of dA opposite L only 8% as efficient as AP. These data indicate that Pol θ conducts TLS opposite AP, C4-AP, and L, although each lesion has distinct properties during TLS by Pol θ.

Table 1. Steady-state kinetic analysis of nucleotide insertion opposite lesions by Pol θ.

-1 kcat/Km a X dNTP kcat (min ) Km (μM) Fins (min-1•μM-1) AP (28) A 18.8 ± 0.4 12.4 ± 0.5 1.5 1 AP (28) G 9.6 ± 0.4 49.0 ± 8.8 2.0 × 10-1 0.13 AP (28) T 14.2 ± 1.3 529.0 ± 14.4 2.7 × 10-2 1.7 × 10-2 L (29) A 25.7 ± 4.0 201.5 ± 20.8 1.3 × 10-1 1 L (29) G 11.4 ± 0.9 436.5 ± 56.9 2.6 × 10-2 0.17 C4-AP (30) A 10.3 ± 0.1 84.0 ± 3.0 1.2 × 10-1 1 C4-AP (30) G 15.8 ± 0.1 130.5 ± 30.6 1.2 × 10-1 1 C4-AP (30) T 4.7 ± 0.4 537.3 ± 28.7 8.7 × 10-3 0.08 F (31) A 16.6 ± 0.5 13.0 ± 0.4 1.3 1 F (31) G 9.7 ± 0.3 55.3 ± 5.0 1.8 × 10-1 0.14 dU (32) A 70.7 ± 0.3 1.5 ± 0.3 4.7 × 101 31.3 Data are the average ± std. dev. of at least two experiments, each consisting of 3 replicates. a Fins = (kcat /Km)dNTP / (kcat /Km)dATP.

Kinetic parameters were also measured for insertion opposite F (31) to assess the effect of differing local sequences between this investigation and the previous study by

Seki et. al.176 The preference for dA insertion over that of dG was approximately seven- fold in this study compared to nine-fold in the study conducted by Seki et. al., indicating that the difference in local sequence has a minor effect on nucleotide insertion

176 preference. Km values in this study were also similar to those reported by Seki et. al. (13

± 0.4 μM vs 9.5 ± 1.8 μM for dA and 55.3 ± 5.0 μM vs 52 ± 3.0 μM for dG), consistent with the local sequence change exhibiting only a minor effect on nucleotide preference.176

Steady-state kinetic analysis on 32, containing dU (as a surrogate for T) indicated that Pol θ conducts nucleotide insertion opposite undamaged DNA considerably more efficiently than for abasic or oxidized abasic sites. In the case of AP, TLS is 30-fold less efficient than insertion opposite dU, while it is approximately 200- to 250-fold less efficient for C4-AP and L. This is in conflict with the previous report indicating that insertion of dA opposite AP is only 4-fold less efficient than that opposite template T.176 This difference could be explained by an increased efficiency for nucleotide insertion opposite dU relative to template A, G, C, or T, although this was not explored by direct comparison of nucleotide insertion on the same local sequence.

3.1.4 Full-Length Bypass of Abasic and Oxidized Abasic Lesions by Pol θ

Lesion bypass can be divided into two steps: i) insertion opposite the lesion and ii) further extension past the lesion. TLS polymerases are typically proficient at only one of these steps, yet Pol θ is uniquely capable of conducting both steps, both in vitro and in vivo.177–179 Steady-state kinetic analysis showed that Pol θ is equally proficient at nucleotide insertion and extension when AP is present in the template, making it unique

Figure 27. TLS of abasic and oxidized abasic sites by Pol θ. Pol θ (23.8 nM) was incubated with primer templates 28-32 (100 nM) and all four dNTPs (500 μM) for the indicated time. The image containing TLS of F is from a different experiment and is shown for comparison. Reprinted with permission from Laverty, D. J.; Averill, A. M.; Doublié, S.; Greenberg, M. M. ACS Chem. Biol. 2017, 12, 1584–1592. Copyright 2018. American Chemical Society. among mammalian polymerases.176 The ability of Pol θ (23.8 nM) to conduct nucleotide insertion and subsequent extension was analyzed on primer-template complexes 28-32

(100 nM) in the presence of high concentrations (500 μM) of all four dNTPs (Figure 27).

The dU-containing primer-template complex 32 was rapidly extended with one notable pause site apparent after one-minute of reaction. The location of this pause site was not determined; however, Pol θ was reported to stall at template C and may have stalled at one of several template C positions downstream of the standing-start primer.176 It is notable that the extension product for 32 migrates as a 34mer, corresponding to blunt-end addition of one nucleotide to the full-length extension product. A small amount of 35mer, corresponding to addition of a second non-templated nucleotide was observed at 30 min.

These observations are consistent with blunt-end addition activity of Pol θ.176 Insertion opposite the abasic lesions was slower than that opposite dU but was still rapid under these conditions, with the majority of the template being extended by one nucleotide within one minute. Insertion opposite C4-AP was visibly slower than opposite AP and L (Figure 27).

Although the catalytic efficiency of TLS for C4-AP and L is comparable, kcat for insertion of dA opposite L is the highest of all lesions and is more than twice that of C4-AP (Table

1). Because the reactions shown in Figure 27 were conducted with 500 μM dNTPs, all reactions proceeded at maximal or near-maximal velocity, so insertion was faster opposite

L than C4-AP.

Extension of the primer after insertion of a single nucleotide was significantly less efficient than TLS and was incomplete even at 30 min. Extension past AP, F, and L proceeded to a comparable extent (~80%); however, extension past C4-AP did not proceed past 50 % in the same span of time. While extension past dU gave a 34mer (N+1) resulting from blunt-end addition to the 33mer product, extension of lesion-containing primer- templates gave primarily the 33mer (N) extension product. Differential blunt-end addition for undamaged and damaged primer-template complexes is perplexing, as a lesion in the middle of a template would not be expected to affect blunt-end addition 15 nucleotides away at the terminus. It is therefore unclear why Pol θ added an extra nucleotide when the template contained dU but not when the template contains AP. The most striking occurrence in full-length bypass experiments was the formation of products two nucleotides shorter than the full-length product during bypass of abasic lesions. Primer- templates with AP and C4-AP had the greatest amount of this product while those containing L and F had only minor amounts (Figure 27). It is unlikely that DNA synthesis

78 by Pol θ would terminate two nucleotides before the end of the template, especially since this was not observed for dU. We hypothesized that deletions (frameshift mutations) were formed by template slippage (template misalignment) during lesion bypass. Template slippage was predicted to occur during the extension reaction, as previous studies had shown that template slippage does not occur during nucleotide insertion opposite the lesion.176

3.1.5 Formation of Deletions During Abasic Lesion Bypass

Chart 2. Primer-template complexes used for analyzing extension past abasic lesions.

Primer-templates 33-37 were prepared to examine extension past abasic lesions by

Pol θ and to probe the propensity for deletion formation (Chart 2). As dA is most frequently inserted by Pol θ opposite AP, dU, or F, Pol θ was incubated with primer-templates containing dA opposite either dU (33a), AP (34a), or F (37a) in the presence of different mixtures of dNTPs. This qualitative experiment indicated that an excess of the correct dNTP, dCTP, suppressed the formation of deletion products, giving rise to N+1 products, presumably resulting from blunt-end addition of one nucleotide to the end of the full-length product (Figure 28). When the incorrect dNTP, TTP, was present at high concentrations; however, considerable amounts of two-nucleotide deletion products were formed.

The dependence upon the presence of TTP for formation of two nucleotide deletions suggested that template-misalignment (slippage) occurred by insertion of T during lesion bypass. Watson-Crick base pairing between the nucleotide opposite the lesion and a complementary nucleotide in the 5’-direction on the template, would position template dA as the next nucleotide, directing insertion of T (Scheme 17). This mechanism is dependent upon the formation of a Watson-Crick base pair, so primer-template 38 was prepared to test for this requirement.

Figure 28. Effect of dNTP ratios on deletion formation by Pol θ. Pol θ (50 nM) was incubated with AP (34a), dU (33a) or F (37a) (50 nM) and a dNTP mix containing either dTTP or dCTP at 500 μM with the other three dNTPs at 25 μM. Reactions were analyzed by denaturing PAGE after 30 min. The relevant portion of the primer-template is shown for reference. X=AP, dU, or F.

Scheme 17. Proposed mechanism for template misalignment during bypass of abasic lesions (X).

The nucleotide opposite the lesion, dA, can base pair with template T two nucleotides away on 34a, but this Watson-Crick base pair cannot form on 38 because the

T has been replaced with dA. Consistent with the proposed mechanism for deletion formation (Scheme 17), two-nucleotide deletions are only observed during extension of

34a but not 38 (Figure 29). Additionally, Pol θ inserted the correct nucleotide (dC) on both primer-templates but only inserted T on 34a, consistent with T insertion being dependent upon the Watson-Crick base pairing shown in Scheme 17. These results strongly support

Figure 29. Dependence of Watson-Crick base pairing on deletion formation by Pol θ. Pol θ (11.9 nM) was incubated with primer-template 34a or 38 (25 nM) and 500 μM of all four dNTPs (left) or the indicated dNTP (right). Reprinted with permission from Laverty, D. J.; Averill, A. M.; Doublié, S.; Greenberg, M. M. ACS Chem. Biol. 2017, 12, 1584–1592. Copyright 2018. American Chemical Society.

81 the formation of frameshift mutations (deletions) by template misalignment during bypass of abasic lesions.

Interestingly, different amounts of deletion are apparently formed during bypass of each lesion (Figure 27). For AP (28), full-length product is formed in a small excess (60:40) over two-nucleotide deletion. This preference for full-length product is greater opposite L

(29) and F (31) (approximately 90:10 full-length:deletion) but is reversed when C4-AP is in the template (30), where full-length product is the minor product (30:70). A quantitative analysis was undertaken to compare the propensity for full-length product vs two- nucleotide deletions on 33-37.

Qualitative experiments indicated that extension past each lesion was considerably less efficient than nucleotide insertion opposite that lesion (Figure 27). Predictably, extension of 34-37 was negligible under the reaction conditions used for steady-state kinetic analysis of nucleotide insertion (1 nM Pol θ, 50 nM primer-template, not shown).

Therefore, the ratio of Pol θ to primer-template was increased (11.9 nM Pol θ, 25 nM primer-template) for extension experiments. As dC is not appreciably inserted opposite any abasic lesion, primer-templates containing dC opposite the lesion were not prepared.

Qualitative experiments on 34-36 showed that extension by Pol θ was negligible on primer- templates containing dG or T opposite either AP, C4-AP, or L and only occurred to an appreciable extent with dA opposite the lesion (not shown). Therefore, steady-state kinetic analysis was conducted only on primer-templates with dA opposite each lesion.

3.1.6 Kinetic Analysis of Extension Past Abasic Lesions

Table 2. Steady-state kinetic analysis of extension past abasic and oxidized abasic lesions by Pol θ.

5'-d-TTCCA CAG GTG CAC ACA A 3'-d- GGT GTC CAG GTG TGT XGT ATG CTC GGC CAG C

-1 kcat/Km a X dNTP kcat (min ) Km (μM) Fins (min-1•μM-1) dU (33a) C 31.0 ± 1.0 1.5 ± 0.3 20.7 -- AP (34a) C 0.19 ± 0.02 219.3 ± 32.4 8.6 ×10-4 1 AP (34a) T 0.16 ± 0.02 323.1 ± 19.4 5.0 ×10-4 0.6 C4-AP (35a) C 0.03 ± 0.001 183.6 ± 19.1 1.6 ×10-4 1 C4-AP (35a) T 0.11 ± 0.004 298.7 ± 15.1 4.0 ×10-4 2.3 L (36a) C 0.56 ± 0.04 213.6 ± 3.6 2.7 ×10-4 1 L (36a) T 0.08 ± 0.02 98.8 ± 14.7 8.1 ×10-4 0.31 F (37a) C 0.15 ± 0.001 290.8 ± 18.5 5.1 ×10-4 1 F (37a) T 0.10 ± 0.001 408.6 ± 28.4 2.4 ×10-4 0.5 Data are the average ± std. dev. of at least two experiments, each consisting of 3 replicates. a Fins = (kcat /Km)dNTP / (kcat /Km)dCTP.

Table 3. Steady-state kinetic analysis of extension past abasic lesions flanked by 3'-TTA.

5'-d-TTCCA CAG GTG CAC ACA A 3'-d- GGT GTC CAG GTG TGT XTT ATG CTC GGC CAG C

-1 kcat/Km a X dNTP kcat (min ) Km (μM) Fins (min-1•μM-1) dU (39a) A 56.9 ± 2.5 4.8 ± 1.6 11.9 -- AP (39b) A 0.22 ± 0.04 60.3 ± 12.4 3.6 ×10-3 1 AP (39b) T 0.53 ± 0.07 61.6 ± 1.7 8.6 ×10-3 2.2 C4-AP (39c) A 0.45 ± 0.01 88.3 ± 8.8 5.1 ×10-3 1 C4-AP (39c) T 0.59 ± 0.03 179.3 ± 5.6 3.3 ×10-3 0.67 Data are the average ± std. dev. of at least two experiments, each consisting of 3 replicates. a Fins = (kcat /Km)dNTP / (kcat /Km)dATP.

Primer-template complexes 33-37 were subjected to steady-state kinetic analysis

(Table 2). Insertion of dC, the correct nucleotide, was compared with insertion of T, which presumably resulted from the misalignment mechanism shown in Scheme 17. The bypass efficiency and the frequency of deletion were subtly different for AP, C4-AP, and L; however, F behaved quite similarly to AP (Table 3). The ratio of catalytic efficiencies for dC insertion varied by about ten-fold and followed the order of L > AP ~ F > C4-AP. The differences in dC insertion between lesions were mostly attributable to differences in kcat.

While the Km varied by less than two-fold between the different lesions, the kcat varied by nearly twenty-fold. The range in kcat values was considerably smaller for T insertion and varied only about two-fold, whereas the Km varied by approximately four-fold. Lesions which underwent less efficient dC insertion exhibited a relative increase in T insertion. C4-

AP, which had the lowest catalytic efficiency for dC insertion, was the only lesion that exhibited a preference for T insertion on this sequence. AP and F were intermediate in terms of efficiency of dC insertion and exhibited a two-fold preference for dC insertion over T. L, which had the highest catalytic efficiency for dC insertion, had a three-fold preference for dC insertion over T insertion.

Steady-state kinetic data indicate that extension past AP, C4-AP, L, and F is markedly less efficient than nucleotide insertion opposite that same lesion. These data are in stark contrast with a previous report showing highly efficient extension past F by Pol

θ.176 The reason for this disparity is unclear. The previous report utilized the full-length protein purified from insect cells, while this investigation utilized the previously reported catalytic core of Pol θ, purified from E. coli.271,319 It is possible that the full-length protein exhibits different catalytic parameters than the catalytic core of the polymerase. However,

84 kinetic parameters for nucleotide insertion opposite F were similar between this investigation and the previous report, suggesting that the catalytic core of Pol θ exhibits similar kinetics to the full-length polymerase. Furthermore, another report showed that expression of a C-terminal fragment of Pol θ (residues 1708-2590) in human Pol θ- knockout cells is sufficient to rescue TLS of Tg.177 This suggests that the helicase domain and the majority of the central domain of Pol θ are dispensable for TLS. However, we cannot conclusively rule out differences between the full-length enzyme and the catalytic core contributing to the strongly reduced activity of Pol θ for extension past the lesion.

Aside from a possible difference between the catalytic core and the full-length polymerase, it is unclear why extension kinetics differ so considerably from the previous investigation.

The relative preference for misalignment versus error-free bypass determined by steady-state kinetics is in general agreement with the result of full-length bypass experiments. As full-length bypass experiments were conducted with high dNTP concentrations (500 μM), the velocity for the extension reaction is at or near maximal for all lesions. Therefore, the high kcat for dC insertion past L largely accounts for the preference for full-length extension over two-nucleotide deletion, as kcat is relatively small for the latter process. The difference in kcat for dC versus T insertion during extension past

AP is more modest and accounts for the smaller preference for full-length extension observed in these experiments. The high Km for T insertion past C4-AP suggests that T insertion would proceed at less than maximal velocity, while the velocity for dC insertion would be maximal. This could therefore be expected to offset the greater than three-fold higher kcat for T incorporation, leading to the more modest preference for two-nucleotide deletion observed for C4-AP in Figure 27. The observation that extension past F gives

85 approximately 9:1 full-length: deletion when all four dNTPs are present at 500 μM is not as easily explained based on the kinetic data. The Km for dC insertion is 290.8 μM, so at a dNTP concentration of 500 μM, correct insertion would proceed at near maximal velocity.

Meanwhile, the Km for T insertion past F is 408.6 μM, so this reaction would not proceed at maximal velocity with 500 μM dTTP present but would still proceed at greater than half- maximal velocity. This means that, under the experimental conditions shown in Figure 27, the ratio of full-length to two-nucleotide deletion should be close to 3:1 instead of 9:1. The basis for the difference between kinetic parameters and full-length extension products for

F is unclear.

Chart 3. Primer-template complexes containing 3’-TTA sequence for exploring the scope of template misalignment for extension past lesions.

Having established that Pol θ forms frameshift mutations on primer-templates 33-

36 which contain the sequence 3’-GTA adjacent to the lesion, further experiments were conducted to determine the scope of deletion formation. Primer templates 39a-c were prepared where the nucleotide adjacent to the lesion in the 3’-direction was changed from

G to T (Chart 3). Qualitative full-length extension experiments on 39a and 39b showed that two-nucleotide deletion was indeed formed during AP bypass (Figure 30).

Interestingly, two-nucleotide deletion was the major product on this sequence when the

86 reaction contained all four dNTPs at either 25 μM or 500 μM (Lane 3 and 4 respectively). Two-nucleotide deletion was favored by approximately 9:1 with

25 μM dNTPs and by 4:1 with 500 μM dNTPs, differing from results on 34a where full-length product was preferred over deletion. This indicates that the propensity to form deletions differs both for different lesions as well as for the Figure 30. Deletion formation by Pol θ with same lesion in a different sequence. downstream 3’-TTA sequence. Pol θ (50 nM) was incubated with either 39a or 39b (100 nM) under Consistent with the proposed different conditions 1: dATP (500 uM), other dNTPs: (25 uM). 2: dTTP (500 uM), other dNTPs (25 uM). 3: all four dNTPs (25 uM). mechanism for deletion formation, when the correct nucleotide triphosphate (dATP) was present at 500 μM with other dNTPs present at only 25 μM, full-length product was formed exclusively (Figure 30, Lane 1). A high concentration (500 μM) of TTP, the triphosphate expected to be inserted by template- misalignment, favors the formation of two-nucleotide deletions over full-length product

(Figure 30, Lane 2), although some full-length product is still formed under these conditions. Having established that template misalignment occurs with a flanking sequence of 3’-TTA (39b, c) as well as 3’-GTA (33a-37a), steady-state kinetic analysis was conducted on 39a-c.

Kinetic parameters for dU-containing primer-template 39a were quite similar to those for 33, with the catalytic efficiency varying by less than two-fold between the two

87 sequences (Table 3). The efficiency of lesion-containing primer-templates differed considerably more between the two sequences, however. The catalytic efficiency for correct insertion of dA opposite AP on 39a is five-fold more efficient than insertion of the correct nucleotide, dC, on 34a (Table 3). This is primarily a result of a decreased Km on

39b (60.3 ± 12.4 μM) relative to 34a (219.3 ± 32.4 μM). Insertion of T is also more efficient on this sequence and is at least two-fold more efficient than insertion of dA. This preference is entirely attributable to a higher kcat for T insertion, as the respective Km values for dA and

T are nearly identical. This leads to a preference for T insertion over correct insertion of dA of more than two-fold. This preference is generally consistent with the qualitative experiment shown in Figure 30; however, the magnitudes differ somewhat as the preference for two-nucleotide deletion is at least four-fold during the full-length extension experiment. The 3’-TTA flanking sequence resulted in an even more pronounced increase in bypass efficiency for C4-AP (39c), with the catalytic efficiency of dA insertion increasing thirty-fold relative to dC insertion past C4-AP on 35a. The efficiency of T insertion increased more modestly, resulting in a Fins for T of 0.67 relative to dA. This indicates that the flanking sequence can influence both the efficiency of bypass as well as the preference for full-length extension compared to template misalignment. The flanking sequence 3’-TTA could conceivably lead to insertion of dA by template misalignment, giving rise to one nucleotide deletions. Although full-length extension experiments suggest that one-nucleotide deletions are not formed to an appreciable extent on this sequence, we cannot completely rule out some dA insertion resulting from template misalignment instead of correct insertion. Primer-template 40 was prepared to analyze the ability of Pol θ to form one-nucleotide deletions by template misalignment.

Table 4. Steady-state analysis of extension past AP flanked by 3'-TAA.

5'-d-TTCCA CAG GTG CAC ACA A 3'-d- GGT GTC CAG GTG TGT XTA ATG CTC GGC CAG C

-1 kcat/Km a X dNTP kcat (min ) Km (μM) Fins (min-1•μM-1) dU (40a) A 35.8 ± 3.9 6.6 ± 0.2 5.4 -- AP (40b) A 0.026 ± 0.003 206.5 ± 4.8 1.3 ×10-4 1 AP (40b) T 0.52 ± 0.03 300.3 ± 19.5 1.7 ×10-3 13.8 Data are the average ± std. dev. of at least two experiments, each consisting of 3 replicates. a Fins = (kcat /Km)dNTP / (kcat /Km)dATP.

Steady-state kinetic analysis of 40a showed that the catalytic efficiency for extension past dA⸱dU was similar to other primer-templates (33a and 39a) containing dU

(Table 4). The catalytic efficiency for dA insertion on 40a was slightly more than two-fold lower than 39a and about four-fold less than 33a. Correct extension past AP on this sequence was 28-fold less efficient than correct extension on 39b, even though the flanking sequences differ by only a single nucleotide (3’-TTA compared to 3’-TAA). Interestingly, extension on this sequence had a strong (nearly 14-fold) kinetic preference for misinsertion of T, presumably by template misalignment. This was in spite of a 1.5-fold higher Km for

T insertion relative to dA insertion, as the relatively large kcat value for T insertion offset the higher Km. This result suggests that Pol θ forms both one- and two-nucleotide deletions during bypass of abasic and oxidized abasic lesions. The generality of this observation was expanded to primer-templates containing dG opposite a lesion (Chart 4).

3.1.7 Formation of Deletions with dG Opposite AP and C4-AP

Chart 4. Primer-template complexes used for studies on template misalignment with dG opposite AP or C4-AP. 42 contains dA opposite AP and is used for comparison to 41b.

Although dA is preferentially inserted opposite AP, this preference is less than ten-fold, indicating that a small amount of dG insertion can reasonably be expected to occur (Table 1). Furthermore, Pol θ is equally likely to insert dA or dG opposite C4-AP, so analysis of C4-AP bypass with G opposite the lesion is as important as when dA is present opposite the lesion (Table 1). With dG opposite AP (34b), C4-AP (35b), and L

(36b) on primer-templates with a flanking template sequence of 3’-GTA, extension was negligible. Consistent with this observation, Pol θ did not conduct correct extension (dA insertion) during bypass of 41b, and full-length product was not formed (Figure 31). Pol

θ did; however, extend the primer in the presence of all four dNTPs, generating one- nucleotide deletions as the exclusive product. Apparently the very low efficiency of

90 correct extension when dG is present opposite an abasic lesion promotes formation of exclusive template misalignment products when template dC is adjacent to the lesion.

Conversely, when dA is present opposite AP in the same sequence context (42), a mixture of full-length product and two-nucleotide deletion was formed. Additionally, Pol

θ inserted both dG and T on template 42, with the former resulting from correct bypass and the latter presumably resulting from template misalignment which gives rise to the observed two-nucleotide deletions. These observations are consistent with a dependency upon formation of a Watson-Crick base pair to promote template misalignment (Scheme

18). Therefore, the nucleotide which is inserted opposite the lesion can dictate which extension product is formed based on the sequence of the adjacent template nucleotides.

A B

Figure 31. Effect of nucleotide opposing AP on deletion formation. A) Pol θ (11.9 nM) was incubated with either 42 or 41b (25 nM) with all four dNTPs (500 uM). Aliquots were removed at the indicated time and analyzed by denaturing PAGE. B) Pol θ (11.9 nM) was incubated with 42 or 41b (25 nM) and the indicated dNTP. Reactions were analyzed by denaturing PAGE after 15 min. Reprinted with permission from Laverty, D. J.; Averill, A. M.; Doublié, S.; Greenberg, M. M. ACS Chem. Biol. 2017, 12, 1584–1592. Copyright 2018. American Chemical Society.

Scheme 18. Watson-Crick base pairing between the nucleotide opposite the lesion and an adjacent template nucleotide dictates which extension product is formed. X=abasic lesion.

Similar to previous primer-templates, steady-state kinetic experiments showed that the catalytic efficiency on native DNA was quite similar between 41a and 43a (Table 5).

However, on AP-containing primer-templates, Pol θ misinserted the nucleotide directed by template-misalignment about five-fold less efficiently for one-nucleotide deletions than for two-nucleotide deletions (Table 5). This was similar to previous primer-templates, where the flanking sequence had a greater effect on bypass of abasic lesions than on synthesis of native DNA. Interestingly, the strong reduction in efficiency observed for C4-AP bypass

(relative to AP) on the 3’-GTA sequence was essentially absent on both the 3’-CTA (41) and 3’-TCA (43) sequences, where the catalytic efficiency of C4-AP bypass was quite similar to that of AP bypass. The catalytic efficiency for misalignment-mediated extension past dG⸱AP and dG⸱C4-AP was within the range of catalytic efficiencies measured for dA⸱AP and dA⸱C4-AP on other sequences. Furthermore, the inability of Pol θ to mediate

92 correct extension when dG is opposite a lesion means that extension by Pol θ is highly mutagenic when the downstream sequence permits misalignment-insertion.

Table 5. Steady-state analysis of extension past dG•dX base pairs as a function of downstream flanking sequence.

5'-d-TTCCA CAG GTG CAC ACA G 3'-d- GGT GTC CAG GTG TGT XNN NTG CTC GGC CAG C

-1 X 3'-NNN dNTP kcat (min ) Km (μM) kcat/Km (min-1•μM-1) C (41a) CTA G 40.9 ± 5.7 7.5 ± 0.4 5.5 AP (41b) CTA A 0.20 ± 0.002 123.7 ± 4.5 1.6 ×10-3 C4-AP (41c) CTA A 0.31 ± 0.02 136.6 ± 21.1 2.2 ×10-3 C (43a) TCA A 52.4 ± 4.2 8.8 ± 0.01 6.0 AP (43b) TCA T 0.75 ± 0.06 87.6 ± 8.8 8.6 ×10-3 C4-AP (43c) TCA T 0.66 ± 0.02 93.6 ± 18.0 7.1 ×10-3

Data are the average ± std. dev. of 2 experiments, each consisting of 3 replicates.

3.1.8 Direct Detection of Deletions During AP bypass by DNA sequencing

Chart 5. Primer-template complexes prepared for attempted LC-MS/MS analysis of lesion bypass.

Figure 32. Generation of AP by photolysis from AP-ONv precursor or by UDG treatment of dU precursor. Kinetic data coupled with the results of full-length bypass experiments support the formation of one- and two-nucleotide deletions during abasic lesion bypass by Pol θ. To directly confirm the misalignment-misinsertion mechanism shown in Scheme 17, primer- templates were prepared so that UPLC-MS/MS could be used to sequence Pol θ lesion bypass products (Chart 5). This technique is frequently used to study bypass of a number of DNA lesions and detailed protocols for its implementation have been reported.326

Although this method was abandoned in favor of Sanger sequencing of lesion bypass products, attempts to implement this method are presented here.

Primer-templates 44a-c were prepared. In contrast with kinetic studies which used

UDG treatment of dU-containing primer-templates to generate AP, the lesion was photochemically generated on 44c from the o-nitroveratryl-protected precursor (44b)

(Figure 32). Photochemical generation of AP was necessary because the primer strand contains dU, positioned such that UDG treatment followed by piperidine treatment cleaves the extended primer into a short oligonucleotide (7mer). Complex oligonucleotide mixtures generated during lesion bypass by Pol θ can therefore be separated by UPLC and analyzed by MS/MS. Curiously, translesion synthesis and extension were inefficient on 44c, while the control primer-template 44a containing T instead of AP was extended quite efficiently

(Figure 33). Photolysis of 44b followed by treatment with NaOH showed that photolysis

proceeded to at least 95%, so the cause of this problem is unclear (not shown). Primer-

template complexes 45a and 45b were synthesized, where UDG is used to generate AP.

Although these primer-templates would not have been useful for UPLC-MS/MS

experiments, extension of 45b was compared to that of 44b to determine whether

photochemical generation of AP was the reason for inefficient bypass. Nucleotide insertion

opposite AP was qualitatively more efficient on 45b than on 44b; however, extension was

still quite inefficient on this primer-template. This result is inconsistent with experiments

conducted on longer primer-templates used for kinetic experiments. Control reactions

containing an undamaged nucleotide (44a, 45a) proceeded rapidly to completion, so Pol θ

is clearly capable of conducting DNA synthesis on primer-templates of this length. Perhaps

the misaligned primer-template complex cannot be extended efficiently on primer-

templates of this length, although there is no evidence for this phenomenon.

Figure 33. Bypass of AP or control (T or dU) in primer-templates for LC-MS experiments. Pol θ (50 nM) was incubated with either 44a, 44c, 45b or 45a (100 nM) with all dNTPs (500 uM). Aliquots were removed at the indicated times and analyzed by 20% denaturing PAGE.

Chart 6. Primer-template complexes used for Sanger sequencing analysis of lesion bypass. (Bio= Biotin)

An alternative method, originally reported by Sabouri, was used to determine the sequence of Pol θ lesion bypass products (Figure 34).27 This method entails preparation of a primer-template complex containing a 5’-biotinylated primer (Chart 6). The template contains dU which is converted to AP by UDG treatment for one reaction and left as dU in another reaction to act as an undamaged control. Lesion bypass was conducted utilizing

Pol θ and all four dNTPs, and the biotinylated primer was isolated using streptavidin beads.

The isolated primer was PCR amplified and the PCR product was purified and cloned into a plasmid vector which was then transformed into E. coli. Individual colonies were sequenced by Genewiz.

Figure 34. Workflow for bypass sequencing experiments. A) Primer-template complexes 46a or 46b (400 nM) were incubated with Pol θ (95 nM) and 100 μM dNTPs for 30 min at 25 °C. B) The extension product is shown. C) Streptavidin-coated magnetic beads bind the biotinylated primer strand. D) The extended primer strand was isolated from the template and E) amplified by PCR. F) The PCR product and the pBluescript SK- plasmid were digested with Acc65I and EcoRI restriction enzymes. G) The digested PCR product was ligated into the pBluescript SK- plasmid and transformed into E. coli. H) Colony sequencing by Genewiz determined the sequence of Pol θ bypass products. Method adapted from Sabouri.27

The primer on 46a, b terminates directly before the lesion so that nucleotide insertion and subsequent extension past the lesion are analyzed in this experiment. Lesion bypass experiments were conducted on 5’-32P-47, which contains the same downstream 3’-

TTA sequence that is present in 46a, b (Figure 35). Two-nucleotide deletion was formed in approximately seven-fold excess over full-length product under these conditions. An apparent one-nucleotide deletion product was also visible and was present in approximately equal amounts with the full-length product.

Importantly, extension proceeded nearly to completion, confirming that these reaction conditions were adequate for the bypass sequencing experiment. Pol θ (95 nM) was incubated with 46a or 46b (400 nM) and all four dNTPs (100 μM) for 30 min to ensure that lesion bypass went to completion. The extended 5’- biotinylated primer was isolated with magnetic streptavidin beads. Control experiments using a

3’-radiolabeled, 5’-biotinylated oligonucleotide indicated that >80% of the oligonucleotide was Figure 35. Determining reaction conditions for bypass sequencing isolated from the reaction and remained bound experiments. Pol θ (50 nM) was incubated with 47 (100 nM) and dNTPs (25 μM). throughout washing steps. For the purposes of Aliquots were removed at 1 and 30 min and analyzed by 20% denaturing PAGE. this experiment; however, high recovery was not 98 necessary because PCR amplification followed product isolation. The isolated oligonucleotide, still bound to the magnetic beads, was amplified by PCR.

In these experiments, the template strand could conceivably be amplified by PCR, leading to erroneous identification of these products as Pol θ bypass events. Therefore, several methods were used to rule out this possibility. During isolation of the biotinylated primer with streptavidin beads, NaOH (0.1 M) was added to denature the duplex and to cleave the AP-containing template strand. Furthermore, a T:G mismatch was positioned near the 5’-end of the primer, allowing for discrimination of the extended primer strand from the template strand. Sequencing of bypass products revealed that the extended primer strand was the exclusive amplification product.

PCR products are typically purified from the reaction mixture by gel electrophoresis or by commercially available kits. Gel electrophoresis is more time- consuming than using a purification kit; however, it provides considerable separation between the PCR product and the primers and allows for direct visualization of whether

PCR is successful. In molecular cloning, PCR products are typically hundreds or thousands of base pairs (bp) long, much longer than primers which are typically 20-30 bp and are not usually more than 50 bp. As such, agarose gel electrophoresis is a simple and effective way of purifying PCR products and simultaneously checking for successful PCR. However, because the PCR product in this experiment was approximately 60 bp, agarose gel electrophoresis was not sufficient for separation of the product from the primers, so non- denaturing polyacrylamide gel electrophoresis was used for this purpose (Figure 36). This method gave efficient separation and showed that PCR was successful. A simpler purification method involved use of a Qiagen Nucleotide Removal Kit which separated the

Figure 36. Analysis of PCR by 16% native PAGE with ethidium bromide staining. Lane 1 and 2 are PCR reactions from AP bypass (46b) while lanes 3 and 4 are PCR reactions from dU bypass (46a). The ~60 bp PCR product migrates as expected.

PCR product from dNTPs and polymerases. Although this method likely did not remove primers from the reaction mixture, this did not appear to affect later steps. A centrifugal filter unit likely could have been used to remove primers and dNTPs (but not polymerases) from the PCR mixture, although this was not attempted.

Following gel purification, the PCR product was excised from the gel and eluted into solution. Ethidium bromide, used as a nucleic acid stain to visualize the PCR product on the gel, was removed from the DNA by Qiagen Nucleotide Removal Kit. Ethanol precipitation was originally used to remove ethidium bromide, as this compound is readily soluble in ethanol. However, the double-stranded PCR product likely denatured during ethanol precipitation, because further steps were unsuccessful when ethanol precipitation was used. Qiagen Nucleotide Removal Kit, which uses a short wash step with ethanol, was used instead of ethanol precipitation. This did not interfere with later steps, so the duplex structure of the PCR product likely remained intact. The purified PCR product was cloned

100

Figure 37. Purification of digested pBluescript plasmid by agarose gel. Lane 1: undigested plasmid. Lane 2: Acc65I digest. Lane 3: EcoRI digest. Lane 4: Acc65I and EcoRI digest. Lane 1 contains undigested plasmid which migrates faster than digested plasmid because it is supercoiled. Single digests (Lane 2 and 3) confirmed successful digestion under the reaction conditions. Double digested plasmid was purified from Lane 4. into the pBlueScript SK- plasmid. This entailed separate digestion of the PCR product and the pBlueScript SK- plasmid vector (3000 bp) with the restriction enzymes EcoRI and

Acc651. Digestion reactions were purified to remove restriction enzymes which would interfere with the ligation step. The Qiagen Nucleotide Removal Kit was used to purify the digested PCR product, while agarose gel electrophoresis was used to purify the digested plasmid (Figure 37, Lane 4).

Gel purification of the digested plasmid was necessary because undigested plasmid is expected to be mostly in the supercoiled state, while the ligated plasmid would be linear.

Supercoiled plasmids have a much higher transformation efficiency than linear plasmids, so even a small amount of undigested plasmid could lead to a considerable number of colonies containing plasmid without insert. Agarose gel electrophoresis gives good separation between linear and supercoiled plasmids, with the linearized plasmid migrating

101 at approximately 3000 bp and the supercoiled plasmid migrating faster at approximately

2000 bp (Figure 37, Lane 1). However, the Acc65I and EcoRI restriction sites are quite close to one another on the pBlueScript SK- plasmid meaning that agarose gel electrophoresis cannot separate plasmid that has been digested by one restriction enzyme but not the other. To ensure that both restriction enzymes digested the plasmid with high efficiency, control digestions with individual restriction enzymes (Acc65I and EcoRI) were carried out under identical conditions to the double digestion reaction (Figure 37, Lanes 2 and 3). These control digests indicated that both restriction enzymes had high activity and the double digestion proceeded close to completion at both restriction sites. Ligation was performed with a 3:1 molar ratio of insert to plasmid using the Quick Ligase Kit from New

England Biolabs. An aliquot (3 μL) of the ligation reaction was transformed into DH5 α cells which were grown on LB plates supplemented with ampicillin (100 μg/mL). Plates were sent directly to Genewiz for colony sequencing, which eliminated the need for plasmid amplification and isolation from individual colonies. Colony sequencing by

Genewiz entails cell lysis, random hexamer-primed amplification, and Sanger sequencing all performed in the same reaction tube.327

Table 6. Results of bypass sequencing for AP.

One- Two- Colonies dA Full-length X nucleotide nucleotide Sequenced Opposite X extension deletion Deletion dU (46a) 11 11 (100%) 11 (100%) 0 0 AP (46b) 39 37 (100%) 1 (2.5%) 1 (2.5%) 37 (95%)

102

Sequencing of bacterial colonies corroborated the template misalignment mechanism shown in Scheme 17 (Table 6). Additionally, the insertion of dA opposite AP for every colony is consistent with both a preference for dA insertion over dG and/or for highly inefficient extension if dG is inserted opposite AP. Interestingly, two-nucleotide deletion was 37 times more likely than one-nucleotide deletion and full-length product.

This is more pronounced than the 7:1:1 ratio observed when Pol θ (50 nM) was incubated with 47 (100 nM) and all four dNTPs (25 μM) (Figure 35). It is possible that the preference for two-nucleotide deletions observed in bypass sequencing experiments was skewed by a low number of colonies sequenced, although it is unclear whether this would skew the results so considerably. It is also possible that a difference in reaction conditions could lead to the differences observed between the experiment presented in Figure 35 and the bypass sequencing experiment. Although the ratio of Pol θ to primer-template (2:1) was identical in each experiment, the concentrations were 4-fold higher in the bypass sequencing experiment. Additionally, the dNTP concentration (100 μM), was four-fold greater in the bypass sequencing experiments. It is not clear how this difference in reaction conditions would lead to a difference in the ratio of full-length products to deletions, however.

Furthermore, the local sequences in 46b and 47 were identical, but the template lengths differed, with 46b containing 30 nucleotides after the termination of the primer and 47 containing 15 nucleotides. It is possible that differences in template length affect the propensity for Pol θ to induce frameshift mutations by template slippage. Again, the mechanistic basis for this possibility is unclear.

Taken together these results show that Pol θ forms sequence-dependent frameshift mutations (deletions) during the bypass of abasic and oxidized abasic sites. Although in

103 vitro experiments have shown that Pol θ is proficient at TLS of AP, it is unknown whether this polymerase contributes to cellular bypass of AP or of oxidized abasic lesions.176

Therefore, the biological significance of these observations is unclear. However, Pol θ clearly functions in the response to replication stress, even in homologous recombination- proficient cells, and functions in TLS of several other lesions.328 The formation of frameshift mutations during lesion bypass by Pol θ could provide a mechanism for increasing genetic diversity in cancer cells overexpressing this polymerase, perhaps leading to a survival advantage or treatment resistance. To further understand the biological significance of Pol θ-mediated formation of frameshift mutations, TLS by Pol θ was analyzed on templates containing Tg, a lesion known to be bypassed by Pol θ in human cells.177

3.2 Bypass of Thymidine Glycol by Pol θ

Tg is a major oxidative DNA lesion which is cytotoxic due to its strong replication- blocking properties. Replicative polymerases make little contribution to Tg bypass in human cells. Tg is instead bypassed by one of two pathways involving TLS polymerases, with one pathway being conducted by Pol κ and Pol ζ and the other by Pol θ.177 Bypass of

Tg by TLS polymerases is predominantly error-free, with only a small number of mutations resulting from insertion of an incorrect nucleotide opposite the lesion. The observation that

Pol θ forms deletions during abasic lesion bypass provided the impetus to examine whether frameshift mutations are formed during Tg bypass.

104

Chart 7. Primer-templates used for analysis of nucleotide insertion opposite Tg.

3.2.1 Primer-template design and kinetic analysis of TLS for Tg.

Scheme 19. Desilylation of Tg precursor. A) Neat triethylamine trihydrofluoride. B) Triethylamine trihydrofluoride, triethylamine, N-methyl pyrrolidinone (4:3:6 ratio by volume).

Primer-template complexes containing 5R, 6S-Tg, the major diastereomer formed in duplex DNA, were prepared using the commercially available TBDMS-protected Tg precursor reported by Iwai. 329 Curiously, the manufacturer protocol for desilylation of the precursor (neat triethylamine trihydrofluoride, 40 °C, overnight) gave a poor yield with incomplete desilylation (Scheme 19). Therefore, oligonucleotides containing Tg precursor were desilylated using a mixture of N-methyl pyrrolidinone, triethylamine, and triethylamine trihydrofluoride (4:3:6 ratio by volume, 65 °C, 1.5 h), as reported

105 previously.330,331 These conditions gave complete desilylation and improved the yield.

Primer-template complexes 48-51 were prepared for use in standing-start polymerase experiments to determine the propensity for template-misalignment during Pol θ bypass

(Chart 7). The template sequences were designed such that the template strand contained a

T either one or two nucleotides removed from Tg in the 5’-direction. Because dA is primarily inserted opposite Tg, template misalignment could be mediated by Watson-Crick base pairing between the dA inserted opposite the lesion and the template T present 1 or 2 nucleotides away, resulting in 1-nucleotide (primer-template 49) or 2-nucleotide (primer- template 51) deletions. Nucleotide insertion opposite Tg was analyzed to confirm that Pol

θ preferentially inserts dA opposite this lesion. As expected, qualitative experiments on 48-

51 showed a strong preference for dA insertion opposite Tg and T, even when high concentrations (500 μM) of each dNTP were employed (Figure 38). The magnitude of this preference was examined by steady-state kinetic analysis.

Figure 38. Nucleotide insertion by Pol θ on primer-templates containing T and Tg. Pol θ (432 pM) was incubated with 48-51 (50 nM) and the indicated dNTP (500 μM) for 5 min at 25 °C. Reprinted with permission from Laverty, D. J.; Greenberg, M. M. Biochemistry 2017, 56, 6726–6733. Copyright 2018. American Chemical Society.

Steady-state kinetic experiments showed that the efficiency of nucleotide insertion opposite Tg (49) was decreased 18-fold relative to that opposite T (48) (Table 7). This

106 difference is attributable to an increased Km, whereas kcat values are very similar for both

Tg and T. Additionally, Pol θ exhibited reduced fidelity opposite Tg. Insertion of dA opposite T was preferred over insertion of G 750-fold and insertion of T 7000-fold, while the respective preferences were reduced to 350-fold and 625-fold during insertion opposite

Tg. These results are inconsistent with the kinetic data reported by Prakash, where dA was inserted with equal efficiency opposite Tg or T, and Pol θ exhibited increased fidelity

177 opposite Tg. Additionally, kcat and Km values for insertion opposite T and Tg vary between our investigation and theirs; however, the catalytic efficiencies (kcat/Km) are comparable in every case except for insertion of dA opposite Tg (49), which is 18-fold less efficient than for T (48).

Table 7. Steady-state kinetic analysis of nucleotide insertion opposite T and Tg (48 and 49).

-1 a X dNTP kcat (min ) Km (µM) kcat/Km Fins (min-1•μM-1) T (48) A 65.1 ± 11.7 4.8 ± 0.6 13.6 1 T (48) T 2.4 ± 0.1 1216 ± 54 1.9 × 10-3 1.4 × 10-4 T (48) G 5.9 ± 0.6 336 ± 3 1.7 × 10-2 1.3 × 10-3 Tg (49) A 68.4 ± 0.8 93 ± 13 0.74 1 Tg (49) T 0.88 ± 0.06 751 ± 21 1.2 × 10-3 1.6 × 10-3 Tg (49) G 0.38 ± 0.06 432 ± 27 2.1 × 10-3 2.8 × 10-3 Data are the average ± std. dev. of at least two experiments, each consisting of 3 replicates. a Fins = (kcat /Km)dNTP / (kcat /Km)dATP.

Steady-state kinetic analysis on a different local sequence showed that dA insertion opposite Tg (51) was 22-fold less efficient than dA insertion opposite T (50). This reduction in efficiency is quite similar to the 18-fold reduction in efficiency for Tg relative to T on 48 and 49 and suggests that the relative efficiency of insertion opposite Tg is not

107 heavily influenced by changes in local sequence. In contrast with 48-49, the fidelity of nucleotide insertion opposite Tg was reduced only slightly relative to that opposite undamaged T. The preference for dA insertion over dG insertion was essentially the same for T (50) and Tg (51); whereas the preference for insertion of dA over T was reduced by only 4-fold opposite Tg relative to T. These results, as well as the results from 48 and 49, suggest that nucleotide insertion by Pol θ opposite Tg would be only weakly mutagenic, with fewer than 1% errors expected. This is inconsistent with cellular bypass of Tg by Pol

θ which showed approximately 5% mutation frequency during nucleotide insertion opposite the lesion.177 This difference between steady-state kinetic experiments and in vivo bypass results is surprising. A previous report suggested that cellular bypass is more accurate than would be expected from steady-state kinetic experiments. Whereas steady- state kinetics suggest an error rate of approximately 1×10-2 during Pol η bypass of CPDs, sequencing of cellular bypass products showed that the mutational frequency is less than

1×10-3.218

Table 8. Steady-state kinetic analysis of insertion opposite T and Tg (50 and 51).

-1 kcat/Km a X dNTP kcat (min ) Km (µM) Fins (min-1•μM-1) T (50) A 75.8 ± 4.2 8.2 ± 0.04 9.21 1 T (50) T 4.0 ± 0.6 744 ± 6 5.4 × 10-3 5.9 × 10-4 T (50) G 7.01 309 ± 53 2.3 × 10-2 2.5 × 10-3 Tg (51) A 51.0 ± 3.5 121 ± 6 0.42 1 Tg (51) T 0.45 ± 0.03 649 ± 2 6.9 × 10-4 1.7 × 10-3 Tg (51) G 0.33 ± 0.02 331 ± 21 1.0 × 10-3 2.4 × 10-3 a Data are the average ± std. dev. of at least two experiments, each consisting of 3 replicates. Fins = (kcat /Km)dNTP / (kcat /Km)dATP.

108

3.2.2 Steady-state analysis of Tg bypass: error-free extension and template

misalignment.

Chart 8. Primer-templates used for extension past Tg.

Figure 39. Single nucleotide insertion during extension past Tg. Pol θ (4.32 nM) was incubated with 53 or 55 (50 nM) and each individual dNTP (1 mM) for 5 min at 25 °C. The relevant sequence of each primer-template is shown. Reprinted with permission from Laverty, D. J.; Greenberg, M. M. Biochemistry 2017, 56, 6726–6733.Copyright 2018. American Chemical Society.

Having confirmed a preference for dA insertion opposite Tg, qualitative analysis of extension past Tg by Pol θ was conducted on 52-55 (Chart 8) where dA is present opposite

Tg or T. Even with each individual dNTP present at 1 mM, only two nucleotides were

109

Scheme 20. Mechanism for formation of 1-nucleotide deletions and two-nucleotide deletions during Tg bypass by Pol θ.

inserted for each primer-template: the correct nucleotide and the nucleotide that would result from template-misalignment followed by nucleotide misinsertion (Figure 39 and

Scheme 20). For both 53 and 55, template misalignment was qualitatively more efficient than correct extension. Steady-state kinetic experiments were conducted to assess the efficiency of error-free extension and misalignment-mediated extension on 53 and to compare this with error-free extension on the respective native template containing thymidine (52). During extension of 53, insertion of T, presumably mediated by template- misalignment, was 9-fold more efficient than insertion of the correct nucleotide, dA (Table

9). However, compared to dA insertion when undamaged T is present in the template strand

(52), the respective incorporation efficiencies were reduced by 200-fold and 1800-fold.

This indicates that extension past Tg by Pol θ is markedly less efficient than both

110 replication of undamaged DNA as well as insertion opposite Tg. Similar results were obtained on primer-template 53, where template misalignment was again preferred over incorporation of the correct nucleotide, in this case by 5-fold (Table 10). The efficiency of extension past Tg was also substantially reduced (15000-fold for correct extension and

2600-fold for misalignment) relative to 54, where Tg was replaced with T. Taken together, these data suggest that when Pol θ extends a primer past template Tg, template misalignment followed by nucleotide misinsertion is more efficient than error-free extension. However, extension is very inefficient compared to a template containing thymidine.

Table 9. Steady-state kinetic analysis of extension past T and Tg (52 and 53).

-1 a X dNTP kcat (min ) Km (µM) kcat/Km Fins (min-1•μM-1) T (52) A 33.6 ± 2.0 3.7 ± 1.6 9.1 ---- Tg (53) A 1.1 ± 0.1 215 ± 9 4.9 × 10-3 1 Tg (53) T 22.1 ± 0.6 503 ± 22 4.4 × 10-2 9.0 Data are the average ± std. dev. of two independent experiments performed in triplicate. a Fins = (kcat/Km)dNTP/(kcat/Km)dATP

Table 10. Steady-state kinetic analysis of extension past T and Tg (54 and 55).

-1 a X dNTP kcat (min ) Km (µM) kcat/Km Fins (min-1•μM-1) T (54) T 21.1 ± 4.5 1.0 ± 0.4 21.1 ---- Tg (55) T 0.68 ± 0.04 488 ± 24 1.4 × 10-3 1 Tg (55) G 1.43 ± 0.02 177 ± 12 8.1 × 10-3 5.8 Data are the average ± std. dev. of two independent experiments performed in triplicate. a Fins = (kcat/Km)dNTP/(kcat/Km)dATP

111

3.2.3 Full-Length Bypass of Thymidine Glycol

Sanger sequencing and denaturing PAGE analysis showed that template slippage during bypass of abasic lesions by Pol θ primarily gives rise to deletions. A similar mechanism was envisioned for Pol θ bypass of Tg based on the propensity for template slippage during bypass of this lesion (Scheme 20). To determine whether template misalignment does indeed give rise to frameshift mutations, full-length extension experiments were conducted on 48-51 such that Pol θ conducted both nucleotide insertion and subsequent extension (Figure 40). Extension of 49 primarily gave rise to products 1 nt shorter than in control 48, whereas extension of 51 primarily gave products 2 nt shorter than control 50. These results support the formation of deletions by misalignment-mediated lesion bypass (Scheme 20). However, interpretation of the gel in Figure 40 is complicated by the blunt-end addition of one to two nucleotides to the extension product, consistent with previous reports.176,332 To gain direct evidence for deletion formation, bypass products were sequenced using the same method employed for AP.27 To this end, a biotinylated primer was annealed to a template containing T or Tg (Chart 9).

112

Figure 40. Full-length bypass of T and Tg by Pol θ. Pol θ (20 nM) was incubated with 48-51 (50 nM) and dNTPs (500 μM). Aliquots were removed at the indicated times and analyzed by denaturing PAGE. Reprinted with permission from Laverty, D. J.; Greenberg, M. M. Biochemistry 2017, 56, 6726–6733.Copyright 2018. American Chemical Society.

3.2.4 Direct Detection of Deletions During Thymidine Glycol Bypass

Chart 9. Primer-templates used for sequencing of Pol θ Tg bypass products.

113

Primer-template complexes containing either T (56) or Tg (57) (100 nM) were incubated with Pol θ (20 nM) in the presence of all four dNTPs (100 μM). The biotinylated strand was isolated using streptavidin beads, amplified by PCR, and the PCR product was subcloned into a plasmid and transformed into E. coli for sequencing of individual colonies.

As expected, control 56 bearing T gave only full-length products where dA was inserted opposite T (Table 11). Nucleotide insertion opposite Tg (57) was predominantly error-free, with a single mutational event corresponding to T insertion opposite Tg (3% mutational frequency). This frequency is greater than expected from steady-state kinetic experiments, which showed that misinsertion opposite Tg is expected to occur in less than 1% of bypass events based on steady-state kinetics (Table 7 and 8). Instead, the misinsertion frequency opposite Tg is more similar to a previous report which showed 5% mutational frequency for Pol θ mediated TLS of Tg in human cells.177

Table 11. Sequencing of Pol θ bypass products for T and Tg.

One- Two- Template # Colonies dA opposite Full-length nucleotide nucleotide nucleotide sequenced Xa extension deletion deletion T (56) 15 15 (100%) 15 (100%) 0 0 Tg (57) 36 35 (97%) 8 (22%) 2 (6%) 26 (72%) a A single bypass event involved T insertion opposite Tg followed by full-length extension.

Extension beyond Tg by Pol θ followed the preference 2-nt deletion (72%) > full- length (22%) > 1-nt deletion (6%). This differed from bypass of AP on the same sequence, where two-nucleotide deletions were nearly the exclusive product (95%). This suggests Pol

θ bypasses Tg more accurately than it bypasses AP, although analysis of bypass on additional template sequences is necessary to substantiate this hypothesis. These results suggest that on sequences containing a downstream T within one or two nucleotides, Tg

114 bypass by Pol θ is primarily mediated by a misaligned primer-template complex which gives rise to deletions. Interestingly, the single event involving T insertion opposite Tg gave rise to full-length product without template misalignment. This is consistent with the inability of this primer-template to form a Watson-Crick base pair with a flanking template nucleotide within 2 base pairs when T is inserted opposite Tg.

3.2.5 Implications for Translesion Synthesis by Pol θ

The data presented here raise interesting questions about Tg bypass by Pol θ and about TLS in general. Nucleotide insertion opposite Tg is relatively efficient, yet extension past Tg is highly inefficient, with the efficiency for correct extension of 52 reduced by

1800-fold relative to an otherwise identical complex containing thymidine (53). Correct extension past Tg is even less efficient on 55, where it is reduced by 15000-fold relative to the native primer-template 54. In spite of this strong reduction in efficiency for in vitro extension, Pol θ conducts both insertion and extension in vivo, accounting for approximately 50% of Tg bypass events in human cells.177 Interestingly, Pol θ also shows a 1000-fold reduction in efficiency during nucleotide insertion opposite 3-methyl dA, yet

Pol θ bypasses this lesion in vivo.

These data suggest that inefficient in vitro bypass of a lesion does not necessarily argue against a role for a polymerase in cellular TLS of that lesion. The factors which allow

Pol θ to bypass Tg despite its low efficiency at doing so, are unclear. For instance, it is possible that accessory factors modulate the efficiency of Pol θ, allowing it to conduct extension more efficiently in vivo than it does in vitro. Presently there is no evidence that this occurs, although the TLS activity of Pol θ has only come into focus relatively recently, so such factors would likely not have been uncovered yet. It is also possible that bypass

115 efficiency is not modulated in vivo, and Pol θ simply conducts inefficient bypass.

Interestingly, two other polymerases, Pol η and Pol ν can also bypass Tg in vitro, with Pol

η being particularly efficient.220,249 It is therefore unclear why Pol θ is utilized instead of these polymerases in human cells. In addition to the inefficiency of extension past Tg by

Pol θ, in vitro bypass is also highly mutagenic on sequences containing T within 1 or 2 nucleotides. A recent investigation into the bypass of Tg in human cells utilized a sequence which could have given rise to one-nucleotide deletions during Tg bypass by Pol θ, yet deletions were not reported.177 This raises two important questions: 1) are frameshift mutations (deletions) formed by Pol θ during bypass of Tg? And 2) Why do cells utilize

Pol θ to bypass Tg if it is inefficient at extension and prone to forming mutagenic frameshift mutations?

The aforementioned report utilized a heteroduplex plasmid containing a site- specific Tg lesion within the reading frame of the lacZ gene to investigate TLS in human cells.177 The strand complementary to the lesion-containing strand contained an additional one nucleotide bulge opposite Tg. If the Tg lesion were repaired by BER or NER, the complementary strand would be used as a template during the DNA synthesis step, leading to insertion of an extra nucleotide and shifting the reading frame of the lacZ gene. Plasmids which were correctly replicated possessed an in-frame lacZ gene, giving blue colonies after transformation into bacteria and plating on medium containing a colorimetric substrate.

Plasmids with a frameshift mutation generated by BER of the Tg lesion gave rise to white colonies. The ratio of blue colonies to white colonies was used to determine the frequency of TLS, and the blue colonies were sequenced to determine the accuracy of TLS. This method would not have detected deletions (frameshift mutations) of 1 or 2 nucletoides

116 generated by Pol θ-mediated bypass of Tg unless the authors were specifically searching for them, so it is possible that Pol θ generates frameshift mutations during Tg bypass, but these mutations have remained undetected to date. Regardless of whether Pol θ generates frameshifts mutations in vivo, its propensity to do so in vitro makes the choice of this polymerase for cellular Tg bypass somewhat surprising.

The low efficiency of extension past Tg by Pol θ is also surprising. Pol θ is unique among TLS polymerases in that it conducts both nucleotide insertion and extension for a number of lesions.176–179 Typically TLS polymerases do not conduct both steps, and if they do, they do so for very few (often one) lesions. For instance, Pol η and Pol ν are capable of nucleotide insertion and extension past Tg.220,249 However, these polymerases do not contribute to Tg bypass in cells.177 Steady-state kinetic analysis indicates that Pol η inserts

-1 -1 nucleotides equally efficiently (kcat/Km: 21 μM /min ) opposite T and Tg, making it 30- to

220 50-fold more efficient at nucleotide insertion than Pol θ. Additionally, Pol η (kcat/Km:

0.54 μM-1/min-1) extends a primer past Tg at least 100 times more efficiently than Pol θ

-1 -1 -1 -1 220 (0.005 μM /min on 53 and 0.0014 μM /min on 55). Interestingly, Pol η is much more error prone than Pol θ during nucleotide insertion opposite Tg with Fins of 0.12 for dG, 0.05 for dC, and 0.02 for T.220 The relative efficiencies of Pol θ and Pol η argue against a lesion bypass model where TLS polymerases are randomly recruited to the replication fork until one of the polymerases bypasses the lesion. After all, Pol η is more efficient than Pol θ at both steps of TLS and would therefore be expected to make a significant contribution to

Tg bypass in human cells. Instead, the selection of a less efficient TLS polymerase for Tg bypass suggests a lesion-specific model for TLS, where specific polymerases are recruited to bypass specific lesions. Still, it is unclear why Pol θ is recruited instead of Pol η. Based

117 solely upon nucleotide insertion opposite Tg, this selection would appear reasonable: Pol

η is highly efficient at insertion opposite Tg, yet it is error-prone; Pol θ is less efficient, although still relatively efficient, and it is quite accurate at insertion opposite Tg. However, once extension past the lesion is considered, selection of Pol θ is perplexing. Pol θ is both less efficient than Pol η at extension past Tg, and it is also highly prone to inducing frameshift mutations on appropriate sequences. It is possible that accessory proteins and other replication factors suppress the formation of frameshift mutations during Pol θ bypass of Tg. In this case, the selection of Pol θ over Pol η would appear more reasonable because

Pol θ is considerably more accurate than Pol η during nucleotide insertion. It is also possible that Pol η forms frameshift mutations during bypass of Tg, although this has not been examined. The previous report on Tg bypass utilized a sequence which would not give rise to frameshift mutations.220 Talk about sequence and whether deletions could be formed Taken together, these results raise the possibility that Pol θ forms frameshift mutations in vivo, or that unknown regulatory or accessory mechanisms prevent formation of these mutations.

3.3 Pol θ Lyase Activity on 5’-dRP, DOB, and pC4-AP

Figure 41. Generation of 5’-dRP, DOB, and pC4-AP from respective photochemical precursors.

118

Chart 10. Ternary complexes used for lyase activity of Pol θ on 5’-dRP, DOB, and pC4-AP.

3.3.1 Pol θ Lyase Active Site Titration

To study the lyase activity of Pol θ in vitro, oligonucleotide ternary complexes 58-

60 were prepared such that a single strand break is present 5’ to the lesion (Chart 10).

Abasic lesions were generated from the respective photochemical precursors reported previously (Figure 41).14,26,324 An active site titration was employed for Pol θ lyase using

58 as a substrate. For many BER enzymes, the rate-limiting step occurs after the chemical step, causing the steady-state rate to be slower than the pre-steady state rate. Pol θ was hypothesized to exhibit similar pre-steady state burst kinetics, and as such, the active fraction of Pol θ lyase was determined in a similar fashion to that reported for other lyases.333,334 The concentration of 58 (100 nM) was kept constant while the concentration of Pol θ was increased from 10 nM to 40 nM (Figure 42A). As the concentration of Pol θ increased, the amplitude of the pre-steady state burst phase increased, as did the y-intercept.

The y-intercept was plotted as a function of Pol θ concentration, and the slope of the resulting line corresponded to the active fraction (20.1%) of Pol θ (Figure 42B).

119

Figure 42. Active site titration of Pol θ with 5’-dRP. (58). A) Pol θ (10, 20, 30, or 40 nM) was incubated with 58 (100 nM) at 37 °C. The concentration of Pol θ is indicated for each reaction. B) The y-intercept was determined for each reaction in A and plotted against [Pol θ]. The slope of this line (0.201) is the active fraction (20.1%) of enzyme. 3.3.2 Removal of 5’-dRP, DOB, and pC4-AP by Pol θ Lyase

The lyase activity of Pol θ has been demonstrated on 5’-dRP; however, the activity on pC4-AP and DOB was previously unknown. Single turnover rate constants were measured for Pol θ lyase by employing an excess of enzyme (1 μM) over DNA substrate

(50 nM) (Table 12). Under these conditions, Pol θ excised the oxidized abasic sites more rapidly than 5’-dRP, with the rate constants following the order pC4-AP > DOB > 5’-dRP.

This reactivity trend differs from Pol β and Pol λ, both of which exhibit reduced activity on pC4-AP and DOB relative to 5’-dRP.13,15,122 Pol θ lyase activity on 5’-dRP is also

-1 approximately 1000-fold weaker than Pol β (kobs: 120-360 min ) and approximately 25-

-1 13,15,334 fold weaker than Pol λ (kobs: 4.5 min ).

120

Table 12. Removal of abasic and oxidized abasic lesions by Pol θ under single turnover conditions.

-1 Lesion kobs (min ) 5’-dRP (58) 0.17 ± 0.02 DOB (59) 0.32 ± 0.01 pC4-AP (60) 0.91 ± 0.11 Data are the average ± std. dev. of two experiments, each consisting of 3 replicates.

3.3.3 Inactivation of Pol θ Lyase by pC4-AP

Under multiple turnover conditions, Pol θ lyase activity exhibited considerably different properties than under single turnover conditions (Figure 43). Removal of pC4-AP was initially rapid but quickly ceased within 5-10 minutes, after Pol θ had conducted approximately 4 turnovers. Removal of DOB and 5’-dRP continued at a linear rate for the entirety of the experiment (1 h), with Pol θ conducting approximately 20 turnovers for

DOB and 6-7 turnovers for 5’-dRP. The fact that Pol θ abruptly lost activity on pC4-AP yet retained activity on 5’-dRP and DOB under the same assay conditions suggested that

Pol θ lyase activity was inactivated by pC4-AP but not DOB. Interestingly, previous reports showed that both DOB and pC4-AP inactivate Pol β and Pol λ, with DOB being a more potent inactivator of these enzymes.13–15

121

Figure 43. Removal of 5’-dRP, DOB, and pC4-AP by Pol θ under multiple turnover conditions. Pol θ (2.5 nM) was incubated with 5’-dRP, DOB, and pC4-AP (58-60, 100 nM) at 37 °C. Error bars represent the standard deviation for three replicates.

Figure 44. Stepwise inhibition of Pol θ by pC4-AP. Pol θ (5 nM, 350 frmol) was incubated with pC4-AP (60, 100 nM) at 25 °C. Aliquots (1 eq, 350 fmol) of Pol θ were added at 20 min and 40 min (indicated by arrows). Error bars represent the standard deviation for three replicates.

Further evidence for inactivation of Pol θ by pC4-AP was gained by analyzing product formation as additional aliquots of Pol θ were added to the reaction mixture (Figure

122

44). A rapid cessation of product formation was observed after 3-4 turnovers of pC4-AP.

When an additional aliquot of enzyme was added, a burst of 2-3 turnovers was observed, and then product formation ceased again. Another burst of 3-4 turnovers was observed when another aliquot of enzyme was added. It is important to note that this experiment was conducted at 25 °C, while the experiment shown in Figure 43 was conducted at 37 °C. This change was made to reduce the background cleavage of pC4-AP which goes nearly to completion in 1 h at 37 °C. By comparison, the background reaction proceeds to <40% completion at 25 °C over the same time period.

The background reaction of pC4-AP generates a diffusible electrophile which could theoretically inactivate the enzyme by alkylation. The low concentration of both the enzyme and DNA employed in these experiments would argue against such rapid inactivation by a bimolecular reaction, but evidence against such a mechanism was still acquired. Addition of ethanolamine (10 μM) in 2,000-fold excess over Pol θ (5 nM) did not prevent enzyme inhibition (Figure 45). If a diffusible electrophile was inactivating the

Figure 45. Ethanolamine does not prevent inactivation of Pol θ by pC4-AP. Pol θ (5 nM) was incubated with pC4-AP (60, 100 nM) at 25 °C in the presence of ethanolamine (10 μM). Error bars represent the standard deviation for two experiments performed in triplicate. The background reaction was subtracted from the enzymatic reaction at each time point.

123 enzyme, ethanolamine would be expected to stop such a process, suggesting that inhibition of Pol θ by pC4-AP occurs via an electrophile generated within the enzyme active site.

Biochemical experiments gave inferential evidence for irreversible inactivation of Pol

θ lyase by pC4-AP. Previous results showed that when other lyase enzymes excise pC4=AP, they are covalently modified, giving rise to lactam 61.13,15 This also occurs on histone proteins when C4-AP is present within nucleosome core particles.335 We attempted to gain further evidence for formation of this modification by utilizing a fluorescent thiol probe (62) previously shown to be reactive towards 61 on histones.335 These experiments were unsuccessful, likely due to difficulty in moving the excess fluorescent probe. Pol θ

(400 nM) was reacted with pC4-AP (60, 2 μM) and then incubated with a large excess (10 mM) of 62. Dialysis or acetone precipitation was used to remove excess probe, and the reaction was analyzed by SDS-PAGE. Fluorescent imaging showed considerable fluorescent signal in both control and pC4-AP reactions (not shown), indicating that excess fluorescent probe had not been removed. The large excess of fluorescent probe that was still present likely washed out any signal resulting from reaction of 62 with

61. Additionally, it is unclear whether 61, generated by reaction of Pol θ with pC4-AP (60) would react with 62 under these reaction conditions. When 61 is generated on histones tails, it is presumably exposed to solvent and is relatively accessible to 62. However, it is unknown whether 61 generated within the Pol θ lyase active site would be accessible to the bulky fluorescent probe. Denaturants

(0.1% SDS or 2 M urea) were added to the reaction mixture to denature the protein and render the putative lactam

124 modification accessible to the probe. Unfortunately, this was ineffective as 62 was insoluble in the reaction mixture once urea or SDS were added. These attempts to detect lactam were abandoned in favor of direct evidence by UPLC-MS/MS.

3.3.4 Attempted Identification of the Catalytic Nucleophile of Pol θ Lyase by

Reductive Trapping and LC-MS/MS

Scheme 21. Reductive trapping of Schiff base intermediate formed during lyase reaction. Reductant= NaBH4 or NaCNBH3.

Pol θ excises 5’-dRP from DNA by proceeding through a transient DPC involving a

Schiff base. The nucleophile responsible for this activity is presumed to be a lysine, as lysine is the only side chain with an amino group. Although the N-terminus contains a free amine that could act as a nucleophile in the lyase reaction, the lyase activity of Pol θ was localized to an internal position, precluding the N-terminal amine from this function.17 The catalytic lysine residue in the Pol θ lyase active site can be identified by reductively trapping the Schiff base intermediate formed during excision of an abasic lesion (Scheme

21), followed by DNA digestion, protease treatment, and UPLC-MS/MS analysis. The reducing agent, NaBH4, trapped the Schiff base for 5’-dRP (58), DOB (59), and pC4-AP

(60), giving a stabilized DPC (Figure 46). However, NaBH4 reduces the aldehyde moiety of each abasic lesion as well as the Schiff base intermediate, so that no more than 5-10%

125

Figure 46. Trapping of Schiff base intermediate for 5’-dRP, DOB, and pC4-AP. Pol θ (250 nM) was incubated with 5’-dRP (58), DOB (59), or pC4-AP (60) (50 nM) and NaBH4 (2.5 mM) at 37 °C for 1 h and analyzed by SDS-PAGE. of the protein could be crosslinked to the DNA in a typical experiment. Increasing the DNA concentration would not be expected to have any effect on the low trapping yield because

NaBH4 would reduce the lesion on the excess DNA before Pol θ could form a Schiff base.

NaCNBH3 is more selective for the Schiff base over the aldehyde starting material.

Therefore, the DNA substrate can be used in excess over the polymerase when NaCNBH3 is used as a reductant. However, NaCNBH3 only trapped the Schiff base intermediate formed during excision of DOB, but not 5’-dRP or pC4-AP (not shown). We speculated that the major catalytic nucleophile is the same for each lesion, so treatment of Pol θ and

DOB with NaCNBH3 was utilized in attempts to determine the catalytic nucleophile of Pol

θ lyase. Treatment with NaCNBH3 (50 mM) with DOB present in five-fold excess gave essentially complete conversion of Pol θ to a reduced DPC (Figure 47). This determination is based on the yield of reduced DPC (20%) compared to the maximum yield (20%) under these conditions.

126

Figure 47. Trapping of Schiff base intermediate for DOB by NaCNBH3. Pol θ (50 nM) was incubated with DOB (59) and NaCNBH3 at the indicated concentrations (50, 100, or 250 nM) at 37 °C. Two bands are visible for free DNA, presumably due to partial melting of the ternary complex.

After trapping with NaCNBH3, reactions were concentrated, and the reduced DPC was separated from free protein by SDS-PAGE. Curiously, no free protein was observed in these experiments even though the active fraction of Pol θ was only 20% (Figure 48).

Although it is formally possible that catalytically inactive protein still formed the Schiff base intermediate and was trapped as the reduced

DPC, it is unclear if this is feasible. Inactive Pol θ did not interfere with removal of 5’-dRP under single turnover conditions as long as active Pol θ was in sufficient excess. This likely suggests that catalytically inactive Pol θ does not bind DNA or participate in lesion removal, precluding the Figure 48. Isolation of reduced DPC by formation of a Schiff base intermediate by SDS-PAGE. Pol θ (100 pmol) was trapped as a reduced DPC with DOB (59, 1 nmol) by inactive polymerase. Gel slices for DPC and NaCNBH3 (50 mM). DPC and free protein were separated by 10% SDS-PAGE with 5% free protein (control) were excised from the gel, stacking.

127 crushed, and resuspended in buffer, and the DNA was digested in-gel by treatment with

Dnase I, phosphodiesterase I, phosphodiesterase II, and alkaline phosphatase. Following

DNA digestion, Pol θ was treated with dithiothreitol to reduce disulfides, cysteines were alkylated with iodoacetamide, and the protein was digested in-gel with trypsin. The digests of the reduced DPC and the free protein (control) were analyzed by UPLC-MS/MS (Figure

49).

Figure 49. Coverage map for in-gel trypsin digestion of unmodified Pol θ (control) and Pol θ- reduced DPC (analyte). The putative lyase region is surrounded by a red line. Peptides detected in control and analyte are highlighted green. Those only in the control are blue. Those only in analyte are yellow. Sequence coverage was reasonably good (68% for Pol θ-reduced DPC, 75% for unmodified control), but a peptide-DNA adduct was not detected, so the catalytic nucleophile was not identified using this method (Figure 49). Sequence coverage in the

128 lyase region of the protein (circled in red, beginning with

Gly2307 and extending until approximately Arg2520) was quite good, covering 11 out of 13 lysines. Most importantly, Lys2383, identified by later experiments as the major nucleophile in lyase activity, was covered in these experiments by detection of the peptide Q2380QAK2383. This peptide was generated by cleavage at Arg2379 and Lys2383. As lysine modification would lead to missed trypsin cleavage at that site, modification of Lys2383 would produce a peptide spanning Glu2380 to

Lys2396, the same length peptide observed in later experiments which detected modification of Lys2383 by pC4-AP (Section 3.3.6). However, a peptide spanning Glu2380 to Lys2396 was not detected in this method, perhaps due to cleavage to Lys2383 (Figure 49). Importantly, cleavage at Lys2396 was observed in both control and DPC samples, suggesting that modification of Lys2383 could reasonably have been detected using this method. There are several possible explanations for why Lys2383 modification was not detected in these experiments. Incomplete digestion of the DNA component of the reduced DPC would likely generate a heterogenous mixture of peptide-DNA adducts. This would reduce sensitivity and could possibly have hindered detection of Lys2383 modification.

Additionally, the reduced DPC formed by incubation of Pol θ and DOB with NaCNBH3 is unstable, with a half-life of approximately 20 h at 37 °C. As in-gel DNA digestion was conducted for 12 h followed by protease digestion for 24 h, the isolated peptide-DNA adduct was unlikely to have persisted long enough for UPLC-MS/MS analysis. This method was therefore abandoned in favor of UPLC-MS/MS analysis of Pol θ following reaction with pC4-AP.

129

Biochemical experiments suggested that Pol θ was irreversibly inactivated during excision of pC4-AP from DNA. Previous reports showed that lysine residues on Pol β and

Pol λ are modified following excision of pC4-AP, giving 61.13,15 Detection of such a covalent modification on Pol θ would serve a dual purpose: both providing a mechanistic basis for inactivation of Pol θ lyase by pC4-AP and potentially identifying the major nucleophile for Pol θ lyase activity. Lysine modification during excision of pC4-AP occurs with elimination of the sugar fragment from DNA, so digestion of the DNA is unnecessary, eliminating a step which proved challenging in earlier attempts to identify the nucleophile in Pol θ lyase activity. Furthermore, the lactam modification is quite stable at 37 °C and is unlikely to be degraded on the timescale of a typical experiment.336 Previous UPLC-

MS/MS analysis of Pol β following excision of pC4-AP identified covalent modification of Lys84, but modification of Lys72, the major nucleophile in Pol β lyase activity, was not

14 detected. The failure to detect Lys72 modification resulted from a lack of sequence coverage for this region of the protein in both control digests and in digests of Pol β incubated with pC4-AP-containing DNA. Therefore, Lys72 may be modified by pC4-AP— and could even be the major target of pC4-AP modification—but the modification would have escaped detection in the previous report. To determine whether Pol θ could reasonably be expected to be modified at the major catalytic nucleophile, mapping of Pol β modification by pC4-AP was carried out, as Pol β is known to be inactivated by pC4-AP and the major catalytic nucleophile for Pol β lyase was previously identified.13,155,337

3.3.5 Modification of Pol β Lys72 by pC4-AP

130

A B

Figure 50. Comparison of observed and predicted spectra for modified peptide 63. A) Observed spectrum. B) Predicted spectrum. z=2.

To determine whether Lys72 of Pol β is modified following excision of pC4-AP (60),

Pol β (168 pmol) was incubated with pC4-AP (1 nmol) and subjected to digestion with trypsin/Lys-C followed by UPLC-MS/MS analysis. Sequence coverage was good (>70%), but most importantly, the entirety of the 8 kDa lyase domain was covered in this analysis, including Lys72

(not shown). A single modified peptide, 63

(calculated mass: 1511.7872; observed mass: 1511.7876) was detected (Figure 50). This peptide contains two lysines, but only Lys72 is present at an internal position. Lys81 is located at the C-terminal end of this peptide and is unlikely to be modified, as lysine modification generally prevents trypsin cleavage.338 Furthermore, fragmentation is consistent with modification of Lys72 (Figure 51, Table 13).

131

Figure 51. MS/MS of modified peptide 63.

Table 13. Fragmentation of modified peptide 63.

b found calc y found Calc 1 13 1512.8097 1512.7939 2 185.1280 185.1285 12 3 11 1328.6708 1328.6727 4 520.2723 520.2761 10 5 633.3408 633.3601 9 993.5055 993.5252 6 748.3754 748.3871 8 880.4296 880.4411 7 7 765.4038 765.4142 8 6 636.3660 636.3716 9 5 489.3104 489.3031 10 1208.6106 1208.6198 4 376.2214 376.2191 11 3 305.1800 305.1820 12 2 204.1358 204.1343 13 1

132

The observation of only a single modified lysine (Lys72) in this analysis suggests that the previous investigation, which found only Lys84 modification, may have detected only a minor modification. The crystal structure of Pol β bound to a model BER substrate shows

156 that Lys84 is well-positioned to act as a secondary nucleophile in the lyase reaction.

Therefore, it appears that Lys72-catalyzed excision of pC4-AP primarily leads to modification of Lys72; however, modification of Lys84 by 64, the diffusible species released by pC4-AP excision, constitutes a minor pathway (Scheme 22). Having established that incubation of Pol β with pC4-AP-containing DNA (60) leads to modification of the major nucleophilic residue for lyase activity, Lys72, this method was used to identify the major nucleophilic residue for Pol θ lyase and to gain direct support for irreversible inactivation of Pol θ by pC4-AP (60).

Scheme 22. Proposed mechanism for modification of Pol β by pC4-AP.

3.3.6 Identification of Lys2383 as the Major Nucleophile for Pol θ Lyase Activity

133

Figure 52. Coverage maps for Pol θ digestion by trypsin or Glu-C. Peptides highlighted in blue were detected by LC-MS/MS while unlighted regions not detected. The lyase region is circled in red. A) trypsin digest (61% sequence coverage). A) Glu-C digest (39% sequence coverage).

134

To map the location of lactam following pC4-AP (60) excision by Pol θ, protease digestion followed by UPLC-MS/MS was utilized. A previous report showed that the lyase domain of Pol θ is present within a structured 24 kDa region within the C-terminal

17 polymerase domain. This region begins with Gly2307 and extends until approximately

Arg2520. Therefore, it was important that digestion of Pol θ not only gave good overall sequence coverage of the protein, but also gave good coverage of this region of the protein.

Initial experiments where Pol θ was digested with trypsin in solution gave reasonable sequence coverage (60-70%) but were missing several peptides within the putative lyase region of the protein, most notably any peptides containing Lys2383 (Figure 52a). This result differed from in-gel trypsin digestion of Pol θ, where the lyase region of the protein was covered well, including the residue Lys2383. This suggested that structured elements of the protein within the lyase region were refractory to trypsin digestion. In-gel digestion does not suffer from this drawback, as SDS-PAGE fully denatures the protein, making previously structured elements accessible. Although SDS-PAGE followed by in-gel digestion could have been used for these analyses, this method is more time-consuming than solution phase digestion. Furthermore, electrophoresis, staining and destaining of the gel with acidic buffer, and overnight in-gel digestion could provide more opportunity for modifications to decompose. Therefore, solution phase digestion of Pol θ was the preferred method, and several attempts were made to improve digestion efficiency.

Denaturation of highly structured proteins with urea often improves protein digestion; however, denaturation of Pol θ with urea (8 M) prior to addition of protease gave poor digestion and reduced sequence coverage further (not shown). Glu-C digestion gave poor sequence coverage and did not improve coverage in the lyase region over digestion with

135 trypsin (Figure 52b). Denaturation of Pol θ with urea prior to Glu-C digestion reduced sequence coverage even further (not shown). It is possible that rigorous optimization of denaturation and digestion conditions may have been possible, although this was not pursued, as a mixture of trypsin and Lys-C considerably improved digestion efficiency, giving reproducible coverage of approximately 80%. Most importantly, nearly the entirety of the lyase region of the protein identified by Prasad was covered in these digests (Figure

17 53). Only Lys2435 was absent in trypsin/Lys-C digests; however, this lysine was covered by Glu-C digest, and incubation of Pol θ with pC4-AP (60) followed by Glu-C digestion did not lead to modification at this position (not shown). Therefore, the lack of coverage of Lys2435 by trypsin/Lys-C was not concerning.

Figure 53. Coverage map for Pol θ digestion by trypsin Lys-C. Peptides highlighted in blue were detected by LC-MS/MS while those which are white were not detected. The lyase region is surrounded by a red line.

136

Initial attempts to detect lactam 61 using trypsin/Lys-C digestion were unsuccessful. In these experiments, Pol θ (1 nmol) was incubated with pC4-AP (5 nmol) in a 1 mL reaction.

The reaction mixture was concentrated by Amicon centrifugal filter to a volume of 50-100

μL and digested by trypsin and Lys-C (2 μg). Concentration of the sample was necessary for digestion and to analyze the maximum amount of the sample by UPLC-MS. Use of a centrifugal filter was advantageous as it allowed for buffer exchange to remove residual

Pol θ storage buffer. As the storage buffer contains β-mercaptoethanol, removal of this component helped to avoid undesired reaction with 61. However, considerable protein loss occurred during concentration on the Amicon centrifugal filter, likely due to protein binding to the hydrophobic surface of the membrane. Initially, concentration with a Speed vac concentrator was used to concentrate reaction mixtures, thereby avoiding sample loss on the centrifugal filter. This was unsuitable for protease digestion; however, as concentration by a factor of 10 or 20 was necessary for the digestion reaction, and this gave rise to high concentrations of buffer, salt, and glycerol, which were inhibitory to the proteases. This problem was overcome by blocking the membrane with unreacted Pol θ, which minimized sample loss and allowed for effective desalting on the centrifugal filter.

137

Figure 54. MS/MS of peptide 65 containing Lys2383 modification. Table 14. Comparison of predicted and observed fragments for peptide 65.

b Found Predicted y Found Predicted 1 17 1949.944 1949.9613 2 16 3 15 4 14 5 13 6 12 7 11 8 10 9 1098.4962 1098.5038 9 909.4584 909.4863 10 1211.5826 1211.5878 8 11 1324.643 1324.6719 7 739.3467 739.3808 12 6 626.2722 626.2967 13 5 463.2239 463.2334 14 4 15 3 16 2 17 1931.9026 1931.9507 1

138

Analysis of the protein digest by UPLC-MS/MS found a single modified peptide, 65 (Figure 54,

Table 14). This peptide contains two lysines, although only Lys2383 is present at an internal position. Fragmentation of this peptide was consistent with modification of Lys2383, although fragmentation was not as extensive as for the modified peptide found for Pol β

(Figure 51, Table 13). Lys2383 is present within the lyase domain identified by reductive trapping followed by limited proteolysis.17 Interestingly, this residue is highly conserved in A-family polymerases and coordinates the γ-phosphate of the incoming dNTP during

276 DNA synthesis. The detection of the lactam modification on Lys2383 suggested that this residue is the catalytic nucleophile of Pol θ lyase activity. To test this hypothesis, Lys2383 was mutated to either alanine or arginine, and Pol θ K2383A and K2383R were expressed and purified in the same fashion as the wild type protein, giving similar yields and purities

(Figure 55).

Figure 55. SDS-PAGE analysis of Pol θ Lys2383 variants.

139

Predictably, mutation of Lys2383 had a profound effect on the polymerase activity.

Qualitative experiments showed that K2383A inserted only a few nucleotides during synthesis on native primer-template 48, even with a high concentration (500 μM) of each dNTP (Figure 56). Mutation of Lys2383 to Arg (K2383R), a more conservative mutation as arginine is also positively charged and could conceivably coordinate the dNTP, rescued polymerase activity slightly, although even with extended reaction times of 30 min, this mutant was incapable of extending a primer by 16 nucleotides to the full-length product

(Figure 56). Steady-state kinetic experiments showed that the catalytic efficiency of

K2383A is reduced by 28,000-fold relative to the wild type enzyme (Table 15). This is due to both a large increase in the Km as well as a considerable reduction in the kcat. K2383R

Figure 56. Polymerase activity of Pol θ Lys2383 variants on native DNA. Pol θ variants (50 nM) were incubated with 48 (100 nM) and dNTPs (500 μM) at 25 °C for the indicated time.

140

Pol θ is nearly 10-fold more efficient than K2383A, making K2383R 3,000-fold less efficient than the wild type enzyme. Active site titrations were attempted for both K2383A and K2383R in the same fashion as the wild type protein to ensure that the preparations of mutant protein were sufficiently active. However, the mutant proteins did not exhibit pre-steady state burst kinetics, presumably because the drastic change in reactivity led to a change in the rate- limiting step (not shown). A pre-steady state burst phase is only evident if the rate-limiting step occurs after the chemical reaction which is being observed. The drastic reduction in the rate for the mutant proteins likely changed the rate-limiting step to the nucleotidyl transfer reaction, eliminating the pre-steady state burst observed for the wild type enzyme.

Table 15. Comparison of steady-state rate constants for Pol θ variants on 48.

Pol θ -1 kcat/Km kcat (min ) Km (μM) Efficiency variant (min-1•μM-1) Wild Type 65.1 ± 11.7 4.8 ± 0.6 13.6 1 K2383A 0.5 ± 0.1 1094 ± 37 4.8 x 10-4 3.5 x 10-5 K2383R 1.4 ± 0.2 311 ± 24 4.3 x 10-3 3.2 x 10-4 Data are the average ± std. dev. of two experiments, each consisting of 3 replicates.

141

Table 16. Comparison of lyase activity (on 53) and DNA binding (on 66) of Lys2383 variants.

-1 Pol θ variant kobs (min ) Kd (nM) Wild Type 0.165 ± 0.021 6.6 ± 0.1 K2383A 0.014 ± 0.002 2.8 ± 0.5 K2383R 0.034 ± 0.002 8.8 ± 1.0 Data are the average ± std. dev. of three experiments.

Removal of 5’-dRP (58) by K2383A and K2383R was measured under single-turnover conditions and compared to wild type Pol θ (Table 16). By employing a large excess of Pol

θ (1 μM) over substrate (50 nM), the effect of any differences in specific activity of the protein preparations upon kobs was minimized or eliminated. K2383A Pol θ excised 5’-dRP

-1 with a rate constant (kobs) of 0.014 ± 0.002 min , a greater than 90%% reduction in activity

-1 relative to the wild type (0.165± 0.021 min ). Mutation of Lys2383 to Arg rescued some activity relative to K2383A; however, this mutant still excised DNA at a five-fold slower rate than the wild type protein. Predictably, Lys2383 mutants also showed a diminished ability to excise oxidized abasic lesions DOB (59) and pC4-AP (60) (Figure 57). Both

K2383A and K2383R Pol θ variants showed the same reactivity trend as the wild type enzyme (pC4-AP >

DOB > 5’-dRP), indicating that even in the absence of the primary nucleophile, pC4-AP and DOB are removed more efficiently than 5’- dRP, likely due to their greater Figure 57. Comparison of single turnover rate constants for lyase activity of Pol θ Lys2383 variants on 58-60. intrinsic reactivity.

142

As Lys2383 is responsible for the majority of lyase activity, this residue is likely the major catalytic nucleophile for the lyase reaction. Multiple experimental approaches were used to validate this observation and rule out the possibility that Lys2383 mutants exhibited reduced lyase activity due to a low fraction of active protein or a DNA binding defect.

Dissociation constants were measured by fluorescence anisotropy using 66, which contains

5’-P-F, a model of 5’-dRP. It was necessary to use this model substrate, as Pol θ would excise 5’-dRP if a substrate such as 58 was used, and dissociation constants would then be measured for the product of the lyase reaction instead of the substrate. Both Pol θ variants were proficient in DNA binding to 66 (Table 16), with the Kd for the K2383R variant nearly identical to that of the wild type. K2383A, meanwhile, bound DNA more strongly than the wild type. The reduced lyase activity observed upon mutation of Lys2383 to Ala or Arg is therefore not due to a DNA binding defect for the mutants

Lys2383 variants were expressed and purified a second time to ensure that the reduced activity of K2383A and K2383R was not due to a technical mistake during protein purification. Single-turnover rate constants (1 replicate) were measured at two different enzyme concentrations side-by-side for the second batch of each variant (Table 17). These rate constants compared favorably to those measured previously (Table 16). Additionally, this experiment validated that the reduced activity of K2383A and K2383R is not due to a low fraction of active protein. If the specific activity of the protein preparation was low and the active protein was not in sufficient excess over the substrate, reducing the protein concentration would reduce the reaction rate. Additionally, if the substrate was in excess over the enzyme due to a low active fraction of enzyme, biphasic, burst-phase kinetics would be observed. This was not the case, as the reaction rate remained essentially the

143 same whether the protein was present at 1 μM or 500 nM, indicating that the enzyme was in sufficient excess.

Table 17. Comparison of lyase activity on 58 of Lys2383 variants from a second, independent preparation of each variant. A single measurement was conducted for each in the same experiment.

-1 -1 Pol θ kobs (min ) kobs (min ) Variant 500 nM Pol θ 1 μM Pol θ K2383A 0.013 0.014 K2383R 0.026 0.030

Additional experiments were undertaken to support the role of Lys2383 as the nucleophile in the lyase reaction. Reductive trapping of the Schiff base intermediate formed during 5’-dRP removal was conducted for wild type, K2383A, and K2383R Pol θ using

NaBH4 (Scheme 21). In these experiments, the transient DPC formed during the lyase reaction is trapped by reductant, which stabilizes the DPC. Mutation of the major nucleophile would be expected to reduce the yield of the Schiff base intermediate.

Curiously, the yield of the trapping reaction was similar for wild type Pol θ, K2383A, and

K2383R (Figure 58). As both K2383A and K2383R mutants showed residual lyase activity, it is unsurprising that some Schiff base trapping was observed. However, the similar yields between wild type and Lys2383 mutants suggested two possibilities. 1) Lys2383 may not be the nucleophile in the lyase reaction and may instead play a different role, although what this role could be is unclear. 2) Lys2383 is the major nucleophile, yet other lysines compensate in the absence of this residue, forming the Schiff base with a comparable rate but being hindered at the β-elimination step. The latter possibility was most likely, as Lys2383 is modified by pC4-AP and its mutation drastically reduces lyase

144

Figure 58. Trapping yield of Pol θ Lys2383 variants relative to wild type. Pol θ variants (250 nM) were incubated with 5’-dRP (58, 50 nM) and NaBH4 (2.5 mM). Reactions were analyzed by SDS- PAGE and the yield of reduced crosslink for K2383A and K2383R was divided by the yield for wild type Pol θ. Error bars represent the std. dev. of five experiments.

activity; however, neither possibility could be definitively excluded. Therefore, additional

mutagenesis experiments were undertaken. There is no crystal structure for Pol θ bound to

a BER substrate; however, the crystal structure for Pol θ in complex with a primer template

and incoming ddGTP is available (PDB: 4X0Q).276 Analysis of this structure suggests that

Lys2575 and Lys2577 are nearby Lys2383 in the enzyme active site, and therefore may

substitute for Lys2383 in the lyase reaction when it is mutated (Figure 59). The

Figure 59. Crystal structure of Pol θ in a ternary complex with a primer-template and incoming ddGTP. Lys2383 coordinates the γ-phosphate of ddGTP. The distance from the phosphorous of the γ phosphate is shown for Lys2383 as well as nearby lysines Lys2575 and Lys2577. PDB ID: 4X0Q.

145

K2383A/K2575A/K2577A mutant and K2575A/K2577A mutant forms of Pol θ were therefore expressed and purified in the same fashion as the wild type (Figure 60).

Figure 60. SDS-PAGE analysis of all Pol θ variants.

Figure 61. Trapping yield of all Pol θ variants relative to wild type. *Double= K2575A/K2577A **Triple= K2383A/K2575A/K2577A. Pol θ variants (250 nM) were incubated with 5’-dRP (58, 50 nM) and NaBH4 (2.5 mM). Reactions were analyzed by SDS-PAGE and the yield of reduced crosslink was divided by the yield for wild type Pol θ. Error bars represent the std. dev. of at least three experiments.

146

Mutation of both Lys2575 and Lys2577 in addition to Lys2383 (K2383A/K2575A/K2577A) reduced lyase activity on 58 to 3% of the wild type enzyme (Table 18). Furthermore, trapping of the Schiff base was reduced by 80% (Figure 61). This suggests that Lys2575 and/or Lys2577 may act as nucleophiles in the absence of Lys2383. However, fluorescence anisotropy measurements found that the triple mutant exhibited reduced binding to 66 (4- fold higher Kd) relative to wild type (Table 18). Therefore, the observed loss in lyase activity and Schiff base trapping in the triple mutant could potentially result from a structural change and/or reduced DNA binding of this protein. Still, it was possible that

Lys2575 and/or Lys2577 could act as a nucleophile in the lyase reaction. Therefore, the

K2575A/K2577A mutant was analyzed. Lyase activity on 58 was reduced by only 50% relative to the wild type enzyme, yet DNA binding remained essentially identical to the wild type (Table 18). The comparatively minor reduction in lyase activity relative to the

K2383A mutant indicates that neither Lys2575 nor Lys2577 is the major nucleophile in the lyase reaction. Curiously, the yield of Schiff base trapping was increased by nearly two- fold for the K2575A/K2577A mutant, suggesting that one or both residues may participate in a step after formation of the Schiff base, perhaps in deprotonation of the C2’ position to catalyze β-elimination (Scheme 23). Although carboxylate side chains are usually implicated in the β-elimination step of BER, the identity of the base which catalyzes β- elimination is not known for all BER polymerases.

147

Table 18. Comparison of Lyase Activity (on 53) and DNA binding (on 66) of Pol θ variants.

Pol θ -1 kobs (min ) Kd (nM) Variant Wild Type 0.165 ± 0.021 6.7 ± 0.5 K2383A 0.014 ± 0.002 2.8 ± 0.5 K2383R 0.034 ± 0.002 8.8 ± 0.4 Double* 0.081 ± 0.001 5.7 ± 0.9 Triple** 0.006 ± 0.001 26.9 ±1.6 Data are the average ± std. dev. of three experiments. *Double: K2575A/K2577A **Triple: K2383A/K2575A/K2577A

Scheme 23. Proposed function of Lys2383 and Lys2575/Lys2577 in 5’-dRP excision by Pol θ.

Taken together, these data indicate that Lys2383 is the major nucleophile in the lyase reaction, accounting for more than 90% of activity during removal of 5’-dRP. The location of this residue is consistent with previous experiments where limited proteolysis mapped the nucleophile in the lyase reaction to a 23 kDa region within the C-terminal polymerase

17 domain. This region includes Lys2383 but does not include the final 8 kDa in the polymerase domain, where nearby Lys2575 and Lys2577 reside, further suggesting that

Lys2575 and Lys2577 are not candidates for the lyase reaction. Aside from these two residues, two additional lysines are 10-11 Å away from the central phosphorous atom of the γ-

148 phosphate of the dNTP (used as a reference point for lysines near Lys2383). These two residues, Lys2571 and Lys2586, which are already considerably removed from Lys2383 are also in the extreme C-terminus of the protein and are outside of the lyase domain mapped by limited proteolysis.17

Lys2383 is also essential for efficient polymerase activity, with mutation of Lys2383 to Ala reducing activity approximately 30,000-fold and mutation to Arg reducing activity

3,000-fold. This is consistent with the crystal structure of a ternary complex comprised of

Pol θ bound to a primer-template and an incoming dNTP, where Lys2383 coordinates the γ-

276 phosphate of the dNTP. The apparent dual role for Lys2383 in lyase and polymerase reactions is surprising and, to our knowledge, has not been reported for a BER polymerase previously. A previous report showed that polymerase and lyase activities of Pol θ reside

17 in the same domain. In addition, mutation of Asp2540 and Glu2541 eliminated polymerase activity but not lyase activity, indicating that these two residues are not involved in both

17 activities. Lys2383 is thus unique, as it is a major contributor to lyase and polymerase activities. This observation may have additional ramifications. Pol θ is an attractive drug

283 target for cancer therapy. The identification that a single residue, Lys2383, is essential for polymerase and lyase activity suggests that targeting of Lys2383 with covalent inhibitors may provide a path to inactivate both polymerase and lyase activities, potentially proving therapeutically useful.339

3.4 Expanded Substrate Scope of Pol θ and Pol β Lyase

3.4.1 Lyase Activity of Pol θ on Abasic Lesions Near DSB Termini

149

The substrate scope of Pol θ lyase was analyzed to determine if the enzyme acts on substrates other than BER intermediates. The impetus for this analysis was provided by the relatively weak lyase activity of Pol θ on BER intermediates, along with the role for Pol θ in DSB repair. Abasic and oxidized abasic sites are generated at DSB termini by ionizing radiation.301 Previous reports indicate that Ku70/80 excises these lesions, while Pol β is unable to do so.16 Ku has relatively weak lyase activity; however, causing us to speculate that Pol θ may remove abasic and oxidized abasic sites at DSB termini. Excision of 5’-dRP

(67), DOB (68), and pC4-AP (69) in a 5’-overhang was analyzed under single turnover conditions. Interestingly, Pol θ showed proficient lyase activity on these substrates, with excision of these lesions following the same trend as observed for BER intermediates 58-

60 (pC4-AP > DOB > 5’-dRP). Interestingly, single turnover rate constants for each lesion were very similar between the two substrates (BER intermediate or DSB), indicating that

Pol θ lyase shows little preference for a ternary complex or a 5’ overhang (Table 19).

Table 19. Comparison of single turnover rate constants for Pol θ lyase on ternary complexes (58-60) and DSBs (67-69).

-1 -1 kobs (min ) BER kobs (min ) DSB Lesion intermediate (58-60) (67-69) 5’-dRP 0.17 ± 0.02 0.12 ± 0.01 DOB 0.32 ± 0.01 0.32 ± 0.02 pC4-AP 0.91 ± 0.11 0.86 ± 0.04 Data are the average ± std. dev. of two experiments, each consisting of 3 replicates.

150

Multiple turnover experiments gave similar results to those obtained using ternary complexes. Pol θ removed DOB efficiently, conducting approximately 15 turnovers and continuing to remove the lesion at a linear rate over the course of 1 hr (Figure 62). Removal of 5’-dRP was less efficient, with Pol θ conducting 4-5 turnovers over 1 hr but predictably doing so at a linear rate. Although single turnover rate constants showed that pC4-AP is the best substrate for Pol θ, removal of this lesion rapidly ceased after a burst of approximately 6 turnovers. Addition of multiple aliquots of Pol θ to a solution of pC4-AP gave only short bursts of product formation, as the enzyme was apparently inactivated following excision of pC4-AP (Figure 63). These data suggest that Pol θ acts upon abasic and oxidized abasic sites at DSB termini with very similar properties to a ternary complex.

Figure 62. Inhibition of Pol θ lyase by pC4-AP in a 5’ overhang. Pol θ (2.5 nM) was incubated with 5’-dRP, DOB, and pC4-AP (67-69, 100 nM) at 37 °C. Error bars represent the standard deviation for 3 replicates.

151

Figure 63. Stepwise inhibition of Pol θ lyase by pC4-AP in a 5’ overhang. Pol θ (2.5 nM) was incubated with pC4-AP (69, 100 nM) at 25 °C. Pol θ (2.5 nM) was added at 20 min and 40 min (indicated by arrows) Error bars represent the standard deviation for 3 replicates.

3.4.2 Lyase Activity of Pol β on Abasic Lesions Near DSB Termini

A previous report showing that Pol β has very slow lyase activity on DSB termini suggested an interesting difference between Pol θ and Pol β, with Pol θ acting upon both ternary complexes and DSB termini and Pol β acting only upon BER intermediates.16

However, analysis of Pol β lyase on 67-69 revealed that Pol β is quite proficient at removing abasic and oxidized abasic lesions at DSB termini (Table 20). Indeed, Pol β removes 5’-dRP (67) 71-fold more efficiently than Pol θ and removes pC4-AP (69) about

4-fold more efficiently than Pol θ. Interestingly, both polymerases exhibit comparable activity on DOB (68). These results highlight several important differences between Pol θ and Pol β lyase activity. Pol β shows a preference for 5’-dRP over oxidized abasic sites, following the trend 5’-dRP > pC4-AP >DOB. Pol θ; however, exhibits a preference for oxidized abasic sites over 5’-dRP, following the trend pC4-AP > DOB > 5’-dRP. The two polymerases also differ in their relative preference for BER intermediates over DSB

152 termini. Comparison to previous reports for Pol β lyase activity suggest that Pol β is at least

60-fold more active on ternary complexes than on DSB termini,.13,122 Pol θ, on the other hand, exhibits comparable activity on BER intermediates and DSB termini. Fluorescence anisotropy measurements on 70 (Chart 11) suggest that this preference for a BER substrate over a DSB does not stem from poor binding of the DSB by Pol β (Table 21). Indeed, both

Pol β and Pol θ bound more tightly to a DSB than to a ternary complex.

Table 20. Single turnover rate constants for Pol β and Pol θ lyase on abasic lesions at a DSB terminus.

-1 -1 Lesion Pol β kobs (min ) Pol θ kobs (min ) 5’-dRP (67) 8.5 ± 1.5 0.12 ± 0.01 DOB (68) 0.32 ± 0.06 0.32 ± 0.02 pC4-AP (69) 3.1 ± 0.3 0.86 ± 0.04 Data are the average ± std. dev. of two experiments, each consisting of 3 replicates.

Chart 11. Substrates used for measuring dissociation constants for Pol β and Pol θ. Fluorophore= dichloro diphenyl fluorescein

153

Table 21. Dissociation constants for Pol θ and Pol β with 66 and 70.

Substrate Kd (nM) Pol θ Kd (nM) Pol β Ternary complex (66) 6.7 ± 0.5 2.7 ± 1.2 DSB (70) 0.4 ± 0.1 1.3 ± 0.1 Data are the average ± std. dev. of three experiments.

Although Pol β shows a clear preference for a BER intermediate over an abasic lesion at a DSB terminus, it is still quite proficient at removing abasic and oxidized abasic lesions from both substrates. These data are in stark contrast with those reported by Strande which showed that Pol β is not proficient at cleaving 5’-dRP at a DSB terminus and Ku is the only enzyme capable of doing so.16 Pol β, in fact, removes 5’-dRP from a DSB terminus with a single turnover rate constant (8.5 min-1) that is 33-fold greater than the value reported for Ku (0.26 min-1).16 Furthermore, the rate constant for excision of 5’-dRP by Pol θ (0.17 min-1) is comparable to that reported for Ku, suggesting that robust lyase activity on DSBs may be a general phenomenon for BER polymerases. The biological implications of this activity are unclear, however. Ku is rapidly recruited to laser-induced DSBs in cells, with

Ku foci appearing within seconds and saturating within 5 min.92 Therefore, Ku likely binds to a DSB before Pol θ or Pol β. Although Ku removes 5’-dRP more slowly than Pol β does, unless Pol β or Pol θ are rapidly recruited to the DSB, Ku would likely remove the lesion before either polymerase is able to do so. Indeed, Ku appears to remove the majority of AP and 5’-dRP from DSBs in human cell extracts, suggesting that the lyase activity of Pol β or Pol θ is unnecessary at DSB termini.16 However, some evidence suggests a function for

Pol β in the repair of DSBs.340 Therefore, Pol β lyase activity may function in DSB repair, even if this has not been directly demonstrated. Taken together, these data suggest that both

Pol β and Pol θ are proficient at removing abasic lesions from DSB termini, although the

154 preference for the type of lesion and the substrate structure differ for each polymerase, with

Pol θ showing increased activity on oxidized abasic sites and Pol β showing decreased activity on oxidized abasic sites. Additionally, the catalytic activity of Pol θ is comparable on a ternary complex and a DSB, while Pol β shows a strong decrease in activity on a DSB.

3.4.3 Probing the Scope of Pol θ Lyase on Various AP Substrates

Chart 12. Oligonucleotide complexes used for studying the scope of Pol θ lyase.

Several substrates were prepared containing AP in different structural contexts

(Chart 12). Pol θ had no lyase activity on AP within duplex DNA (71) and minimal activity

(<10% reaction in 1 h) on AP recessed by several nucleotides in a 5’-overhang (72). Kinetic analysis was not conducted on this substrate as AP excision was so slow, indicating that

Pol θ activity was reduced by at least several orders of magnitude relative to a BER intermediate. A similar result was obtained for a substrate containing AP recessed within a 3’-overhang (73). When the abasic lesion was present opposite a two-nucleotide gap on the other strand (74), Pol θ exhibited weak lyase activity (0.014 ± 0.002 min-1), a reduction

155 of approximately 10-fold relative to a ternary complex (58) (Table 22). Although 74 is unlikely to be biologically relevant, it was intriguing that Pol θ exhibited lyase activity on this substrate because AP is present in the center of an intact strand. The only difference between 74 and 71 (a substrate upon which Pol θ has no activity) is the presence of a two- nucleotide gap on the opposing strand. This raised an intriguing possibility that Pol θ may have lyase activity on clustered lesions, specifically on those containing an abasic lesion on one strand and a nearby gap on the other. Clustered lesions, which are defined as two or more lesions within two helical turns, are formed by ionizing radiation as well as some chemotherapeutics and are considered an important contributor to the cytotoxicity of these therapies.341 The potential for human enzymes to act upon clustered lesions is of interest because attempted repair of a clustered lesion can convert it into a more deleterious

DSB.342,343

Table 22. Lyase activity of Pol θ on alternative substrates 71-74.

-1 Substrate kobs (min ) 71 No activity 72 Negligible activity (<10% in 1 hr) 73 Negligible activity (<10% in 1 hr) 74 0.014 ± 0.002 Data are the average ± std. dev. of two experiments, each consisting of 3 replicates

3.4.4 Lyase Activity of Pol θ and Pol β on Clustered Lesions

156

Chart 13. Clustered lesion substrates used for studying Pol θ and Pol β lyase activity. An arrow indicates the location of a single nucleotide gap. This gap contains 5’-phosphate and 3’-hydroxyl termini. P=phosphate. Fluorophore=dichloro diphenyl fluorescein

Several substrates (75-78) were prepared containing a single nucleotide gap on the strand opposite AP (75a-78a), C4-AP (75b-78b), or F (75c-78c), a stable analogue of AP used for binding experiments (Chart 13). This gap contains 5’-phosphate and 3’-hydroxyl termini, as would be generated by BER. The position of this single nucleotide gap was chosen so that the substrate scope could be analyzed as a function of clustered lesion

157 polarity. This polarity refers to the distance between two lesions in the plus or minus direction, where lesions 3’-to one another are denoted minus polarity, and lesions 5’-to one another are denoted positive polarity. BER enzymes often exhibit a polarity preference during repair of clustered lesions, so the position of the single nucleotide gap had to be chosen such that it could reasonably have been generated by BER. Examples of clustered lesions which could give rise to 75a-78a are presented in Chart 14.

Chart 14. Possible biologically relevant clustered lesions which are processed by BER to give a clustered lesion with an AP site on the bottom strand and a single nucleotide gap on the top strand.

Substrates 75a-78a could be generated by a full cycle of BER on the top strand of a clustered lesion (such as 79-82), giving a single nucleotide gap on that strand.

Glycosylase activity on a damaged nucleobase on the bottom strand would then form 75a-

78a. 75b-78b could be formed in a similar fashion, except formation of C4-AP on the bottom strand would proceed via hydrogen atom abstraction from the sugar, so glycosylase

158 activity on the bottom strand would not be required. There is literature precedent for activity of glycosylases and Ape1 upon clustered lesions of defined polarity. Multiple oxidized bases, including Tg and 8-oxodG, are substrates for their cognate glycosylases in yeast and bacteria when present in the -1 orientation, suggesting that BER of a clustered lesion containing two oxidized bases could give rise to AP on the top strand.73,344 Ape1 has robust activity when AP sites are present in the -1 orientation and could cleave the top strand to trigger repair by a BER polymerase, giving 75a.76 Substrate 76a could be generated by BER of the top strand of 80. Ape1 shows reduced activity when two AP sites are present in the +1 orientation shown in 80, but its activity is greatest on the top strand of this substrate and nearly nonexistent on the bottom strand, suggesting that 80 could give rise to 76a.76 Clustered lesions containing 77a could be generated by OGG1 activity on 81.

A previous investigation found that yOGG1 activity was only slightly inhibited when a single nucleotide gap was present at the +5 position to 8-oxoG.73 Furthermore, Ape1 cleaves AP in the +5 position indicating that BER can act upon 81 to give 77a.76 Substrate

78a could be generated by BER of 82, initiated by the action of a glycosylase, likely NEIL1 but possibly SMUG1.345 Transfection of a similar substrate into yeast gave rise to DSBs, indicating that BER of 82 likely generates 78a as an intermediate.346

Substrates 75b-78b were also prepared where AP was replaced with C4-AP. The degree to which C4-AP inhibits glycosylase or endonuclease activity on the opposite strand is unknown. However, if C4-AP affects BER of adjacent lesions to a similar degree as does

AP, then Ape1 would be expected to function in BER to generate 75b-78b. Ape1 has normal or nearly normal activity on bistranded AP substrates in the negative polarity (75 and 78) as well as when two AP sites are present in the +5-polarity (77). Additionally,

159

Ape1 would be expected to initiate BER to generate 76b, albeit at a slower rate than on the other damage clusters.75 OGG1 would be expected to function normally in BER to generate

75b, 77b¸ and 78b if 8-oxo-dG was present in a clustered lesion containing C4-AP in the -

1, -4, or +5 polarity.73 yOGG1 is inhibited by more than 6-fold when an AP site is present in the +1 position, suggesting that formation of 76b would be less efficient than the others, assuming comparable inhibition of yOGG1 by AP and C4-AP. Bacterial Nth shows a drastic reduction in activity on Tg when AP is present in the +1 or -1 position.344 However,

Nei glycosylase shows robust activity regardless of the position of AP, suggesting that 75b-

78b could also be formed from clusters containing C4-AP with nearby Tg lesions.344

Having established that substrates 75-78 are biologically relevant, the lyase activity of Pol θ and Pol β were analyzed on 75a-78a. Both polymerases exhibited distinct polarity preferences on these clustered lesions. Pol θ showed activity only on substrates with the single nucleotide gap located in the negative polarity (-1 and -4; 75a and 77a), with no activity on positive polarity substrates (+1 and +5; 76a and 77a) (Table 23). Pol β showed the opposite polarity preference, efficiently removing AP on positive polarity substrates

76a (+1) and 77a (+5) but exhibiting minimal lyase activity on the negative polarity substrates (75a and 78a) (Table 23). Pol β lyase activity on negative polarity substrates was essentially identical to its activity on duplex DNA (71) (Table 23). This suggests that the presence of a single nucleotide gap on the opposite strand only stimulates lyase activity when in a preferred polarity (negative for Pol θ and positive for Pol β).

160

Table 23. Lyase Activity of Pol θ and Pol β on clustered lesions containing AP.

-1 -1 Substrate kobs (min ) Pol θ kobs (min ) Pol β Duplex (71) No activity 0.012 ± 0.007 Cluster - 1 (75a) 0.07 ± 0.01 0.02 ± 0.01 Cluster + 1 (76a) no activity 0.42 ± 0.04 Cluster + 5 (77a) no activity 0.60 ± 0.04 Cluster - 4 (78a) 0.12 ± 0.03 0.013 ± 0.005 Data are the average ± std. dev. of two experiments, each consisting of 3 replicates.

Table 24. Lyase activity of Pol θ and Pol β on clustered lesions containing C4-AP.

-1 -1 Substrate kobs (min ) Pol θ kobs (min ) Pol β Cluster - 1 (75b) 0.19 ± 0.05 0.04 ± 0.01 Cluster + 1 (76b) 0.10 ± 0.01 0.63 ± 0.01 Cluster + 5 (77b) 0.09 ± 0.01 0.96 ± 0.06 Cluster - 4 (78b) 0.26 ± 0.02 0.09 ± 0.03 Data are the average ± std. dev. of two experiments, each consisting of 3 replicates.

C4-AP is more labile than AP, with a half-life of approximately 100 hours in duplex

DNA compared to a half-life of approximately 1000 hours for AP.26,335 Additionally, Pol

θ excises pC4-AP 6-fold more rapidly than 5’-dRP in a ternary complex and 9-fold more rapidly than 5’-dRP in a DSB (Table 19 and Table 20). Therefore, 75b-78b were predicted to be cleaved more rapidly by Pol θ lyase than 75a-78a, and indeed this was the case (Table

24). The preference for negative polarity substrates was also evident on substrates containing C4-AP, consistent with observations for AP-containing clusters. However, the rate of strand cleavage for negative polarity C4-AP-containing clusters was accelerated by only about 2-fold relative to the respective AP-containing cluster (compare 75b to 75a and

161

78b to 78a). Positive polarity clusters (76 and 77) on the other hand underwent a substantial rate increase when the lesion was changed from AP to C4-AP. Interestingly, the rate constant for cleavage of 76b and 77b was only 2- to 3-fold less than for negative polarity clusters 75b and 78b. A similar phenomenon was observed for Pol β, where changing the lesion from AP to C4-AP resulted in a modest increase (~1.5-fold) in rate constant for the preferred polarity substrates (75b and 78b) and a larger increase (2-fold and 8-fold) for the non-preferred polarity (76b and 77b) (Table 24).

Table 25. Dissociation constants for Pol θ and Pol β on model clustered lesions.

Substrate Kd (nM) Pol θ Kd (nM) Pol β Cluster - 1 (75c) 1.1 ± 0.3 3.6 ± 1.1 Cluster + 1 (76c) 0.9 ± 0.4 2.0 ± 0.5 Cluster + 5 (77c) 0.7 ± 0.2 2.6 ± 1.4 Cluster - 4 (78c) 0.9 ± 0.03 2.9 ± 1.2 Data are the average ± std. dev. of three experiments.

The opposite polarity preferences exhibited by Pol θ and Pol β were intriguing; however, it was conceivable that these preferences were due to more favorable binding on the preferred substrates over the clusters that were not substrates. Therefore, dissociation constants were measured by fluorescence anisotropy for model clustered lesions 75c-78c containing the AP analogue, F. Dissociation constants for Pol θ were within experimental error of each other for 75c-78c, indicating that Pol θ does not show a polarity preference for binding to clustered lesions (Table 25). The same phenomenon was observed for Pol β, although Pol β binding was approximately two- to three-fold weaker than that of Pol θ.

Interestingly, Pol θ bound the clustered lesions approximately 5-fold more tightly than it bound the model BER substrate 66, while Pol β showed similar binding to 66 and 75c-78c

162

(Table 25 and Table 21). As each polymerase binds 75-

78 with similar affinity, it is possible that the polymerases bind to the same structural element for each substrate, perhaps the single nucleotide gap on the strand opposing the abasic lesion. In such a scenario, the polarity preference observed for catalysis could be explained by the resulting positioning of the abasic lesion.

Inferential support for this theory was gained from fluorescence anisotropy studies of Pol θ binding to 83. Pol θ binding to duplex DNA containing F (83) is four- to five-fold weaker (Kd: 4.0 ± 0.2 nM) than for gapped DNA containing F (Kd: 0.7-1.1 nM for 75c-

78c). This indicates that the single nucleotide gap in 75-78 is important for substrate binding. Furthermore, Pol θ showed no lyase activity on duplex DNA (71, Table 22), despite binding 83 with a low nanomolar dissociation constant. It is therefore possible that

Pol θ binds the single nucleotide gap on the strand opposite AP, and the reactivity of each clustered lesion is dictated by the resulting positioning of the lyase active site relative to the lesion. However, it is also possible that Pol θ binds directly to the lesion in each substrate, with the single nucleotide gap making the lesion more accessible, and thus explaining the tight binding of Pol θ to 75-78 than 83. If this is true; however, it is unclear why Pol θ binds to 76 and 77 but does not show lyase activity on these substrates.

163

Figure 64. Crystal structure of Pol β with gapped DNA (PDB ID: 1BPX). The 8-kDa lyase domain is colored cyan while the polymerase domain is colored green. The location of relevant nucleotides (relative to the single nucleotide gap) is indicated. Relevant lysines in the 8-kDa lyase domain are colored by element (C: green, N: blue).

The crystal structure of Pol β with gapped DNA may give insight into the polarity preference exhibited by this polymerase on clustered lesions 75-78 (PDB ID: 1BPX).347

The DNA substrate in this structure is a good approximation of clustered lesions 75-78 because it is a ternary complex containing a single nucleotide gap on one strand. Although the abasic site is absent in this substrate, it is possible that Pol β binds the single nucleotide gap of 75-78, and the gapped DNA substrate is therefore bound in a similar orientation to these clustered lesions. In this structure, the DNA is kinked 90 ° so that the 5’-phosphate adjacent to the gap is positioned within the lyase active site (Figure 64). Positive polarity clustered lesions 76a, b and 77a, b would appear to position AP (or C4-AP) nearby to the lyase active site, while negative polarity clusters would have AP (or C4-AP) far removed

(75a, b and 78a, b) from the lyase active site (Figure 64). Therefore, if Pol β does bind to

164 the single nucleotide gap, it is reasonable that positive polarity clusters are considerably better substrates for Pol β than those of negative polarity. However, the +1 and +5 positions are still quite far from Lys72, the major nucleophile involved in Pol β lyase activity (16.3 Å and 21.3 Å, respectively). It is therefore unclear how abasic lesions in these positions are excised. Lys35 is closer to the +1 position (10.3 Å) than Lys72, while Lys60 is closer than

Lys72 to the +5 position (17 Å), but these distances are still quite large (Figure 65).

Interestingly, in the crystal structure of Pol β bound to nicked DNA containing the F analogue of 5’-dRP (similar to 66), Lys72 is positioned more than 10 Å from the C1’ of

F.156 This would necessitate a rotation of the substrate to position the C1’ for attack by

Lys72. It is therefore possible that there is sufficient substrate flexibility to position abasic lesions at the +1 and +5 positions nearby to Lys72. Duplex destabilization by AP could contribute to the flexibility necessary for this to occur. The dependence upon a large conformational change for catalysis could explain the greater than 200-fold reduction in rate constant for 76a and 77a relative to a BER substrate.13,334 It is also possible that such a conformational change in the substrate is not possible, and the abasic lesion cannot be positioned near Lys72. In this case, lysines other than Lys72 could initiate catalysis, possibly less efficiently, consistent with the large reduction in activity for 76-77 relative to a prototypical Pol β substrate such as 66.

165

Figure 65. Crystal structure of lyase domain of Pol β with gapped DNA with relevant lysines indicated (PDB ID: 1BPX). The 8-kDa lyase domain is colored cyan while the polymerase domain is colored green.

Interestingly, Pol β exhibited lyase activity on negative polarity substrates 75a and

78a, even though AP is expected to reside in the polymerase domain, far from the lyase active site in these substrates (Figure 64). It is possible that lysine residues outside of the

Pol β lyase domain catalyze strand cleavage of positive polarity clusters. Interestingly, the rate constants for excision of 75a and 78a by Pol β are on the same order as those reported for excision of AP by nucleosome core particles, which occurs via reaction of AP with lysine residues of histones.26 The comparable rate constants for Pol β and nucleosome core particles along with the predicted location of AP suggests that a similar reaction may cause

AP cleavage in negative polarity clusters (75a and 78a). Consistent with this proposed mechanism, Lys234 is 6.4 Å from the nucleotide at the -4 position of gapped DNA, suggesting that this residue may act as a nucleophile in the reaction of Pol β with 75a

(Figure 66). There is no obvious candidate lysine near the -1 position, so it is unclear how 166

AP is cleaved in this position. A similar phenomenon could be at play for negative polarity clusters containing C4-AP (76b and 78b). In this case, the greater lability of C4-AP explains the increased rate of cleavage of 76b and 78b relative to 76a and 78a (Table 23 and Table 24). Although the polarity preferences exhibited by Pol β on clustered lesions

75-78 are generally consistent with binding of the single nucleotide gap, there is no direct evidence for this proposal. Pol β could also directly recognize the abasic lesion in each substrate. Interestingly, Pol β lyase activity on negative polarity clusters (the non-preferred polarity) was similar to that on duplex 71 (Table 23). This result could be consistent with binding of Pol β directly to the abasic lesion instead of the single nucleotide gap. However, there is no direct evidence for this binding model.

Figure 66. Crystal structure of Pol β bound to gapped DNA with Lys234 indicated. The nucleotide at the -4-polarity is circled in black. The polymerase domain is shown in green and the lyase domain in blue.

167

A B

Figure 67. Crystal structure of Pol θ bound to a primer template with the O-helix highlighted. A) The O-helix is shown in cyan, the primer is red, and the template is blue. Residues 2355-2374 have been deleted for clarity. B) Pol θ is shown in a surface representation and colored gray. The O-helix is shown as a cartoon and is colored green with Lys2383 shown in stick representation.

The structural basis for the polarity preference of Pol θ on 75-78 is even less clear.

The only crystal structures of Pol θ show it bound to a primer-template in the closed conformation.276 In these structures, the O-helix of the fingers subdomain occupies the position which would likely be occupied by the flanking DNA strand present in structures

75-78 (Figure 67). Therefore, it is difficult to speculate about how Pol θ would bind to 75-

78 and how this would affect the positioning of the O-helix. The positioning of the O-helix is likely quite important for lyase activity, because the major nucleophile in lyase activity,

Lys2383, is located in this helix (Figure 67b). To investigate the role of Lys2383 in cleavage of negative polarity AP substrates (75a and 78a), the activity of the K2383R variant of Pol

θ was measured on each substrate. Interestingly, the lyase activity was reduced on each substrate, despite the varying location of AP in each substrate (Table 26). K2383R Pol θ showed a 60% reduction in activity on the -1-polarity substrate (75a) and a 78% reduction in activity on the -4-polarity substrate (78a). This suggests that clustered lesions are

168 accommodated in the lyase active site, allowing for attack and strand cleavage catalyzed by the major nucleophile, Lys2383. It is possible that the flexibility of the O-helix, which can move from an open conformation to a closed conformation, allows Pol θ to utilize the same residue (Lys2383) for nucleophilic attack on AP present in variable locations relative to a single nucleotide gap.

Table 26. Comparison of single turnover rate constants for wild type (WT) and K2383R Pol θ on negative polarity clustered lesions.

-1 -1 kobs (min ) Pol θ kobs (min ) Pol θ Substrate WT K2383R Cluster - 1 (75a) 0.07 ± 0.01 0.028 ± 0.007 Cluster - 4 (78a) 0.12 ± 0.03 0.033 ± 0.001 Data are the average ± std. dev. of two experiments, each consisting of 3 replicates.

Pol θ and Pol β both exhibit lyase activity on clustered lesions containing a single nucleotide gap on the opposite strand from AP or C4-AP. This is the first report of lyase activity of either polymerase on these types of clustered lesions and suggests that Pol θ and

Pol β can convert some non-DSB clusters to DSBs independently of Ape1 activity. This activity was relatively surprising, especially for Pol β, which has minimal activity on unincised AP.348 Interestingly, both polymerases exhibit distinct polarity preferences for lyase activity on clustered lesions. The crystal structure of Pol β with gapped DNA suggests that Pol β may bind the single nucleotide gap, causing positive polarity AP to occupy the lyase active site. Unfortunately, there is no crystal structure for Pol θ bound to gapped

DNA, so the structural basis for the polarity preference for Pol θ is even less clear.

Although both enzymes convert clustered lesions to DSBs in vitro, it is unclear if they do so in vivo. Clustered lesions 79-82 would require activity of a glycosylase, Ape1, and a

BER polymerase for their formation (converting 79-82 to 75-78). These studies 169 demonstrate that action of Pol θ or Pol β on 75-78 could convert these clustered lesions into DSBs by cleavage of the bottom strand. However, Ape1 could fulfill this role in vivo instead of Pol θ or Pol β.

To our knowledge, the activity of Ape1 on clusters containing a single nucleotide gap on the opposing strand has not been reported. Ape1 does; however, convert bistranded lesions containing AP or F into DSBs, meaning that Ape1 has activity on AP or F with a single strand break (SSB) on the opposing strand.75,76 Assuming similar activity on clusters containing a SSB or a single nucleotide gap, Ape1 would be expected to cleave the bottom strand of 75 and 78 much more rapidly than Pol θ or Pol β. Previous investigations on clustered lesions did not report rate constants for Ape1 activity, so a direct comparison cannot be made between Ape1 and Pol θ or Pol β. However, Ape1 conducts hundreds of turnovers within 1 h on negative polarity substrates containing, generating DSBs.75

Therefore, Ape1 would be expected to cleave 75 and 78 much more rapidly than Pol θ or

Pol β, meaning that these polymerases may not cleave 75 or 78 in vivo. Ape1 activity on

77 would likely be similarly robust, precluding the need for Pol θ or Pol β lyase activity to generate a DSB. However, Ape1 activity is strongly inhibited in vitro when two AP sites are present in the +1 orientation.75,76 Ape1 slowly incises one strand of 80, generating 76b, but it does not incise the other strand to generate a DSB.75,76 The observation that Pol β efficiently removes AP from 76b to generate a DSB suggests that this clustered lesion can be converted to a DSB without the activity of Ape1. These results expand the scope of clustered lesions which may be converted to DSBs by aberrant BER.

3.5 Lyase Activity of Ku on Abasic and Oxidized Abasic Lesions

3.5.1 Purification of Ku

170

Figure 68. Representation of construct used for Ku expression.349 rbs= ribosome binding site.

Ku was expressed in Rosetta DE3 pLysS E. coli and purified according to the method developed by Les Hanakahi with some changes.349 Ku is expressed from a bicistronic pET14b construct which encodes hexaHis tagged Ku80 and untagged Ku70 (Figure 68).

Ku solubility is dependent upon heterodimerization, so purified Ku consists of equimolar

Ku70 and Ku80, even though only Ku80 has an affinity tag.349 In a previous report by

Hanakahi, considerable amounts of Ku80 were observed in the insoluble fraction, suggesting that heterodimerization is inefficient and limits the yield.349 In this report, the protein was purified by Co2+ affinity chromatography followed by Heparin chromatography. The hexaHis tag binds tightly to Co2+ allowing for affinity purification of Ku. Heparin is a highly sulfated polysaccharide which has affinity for a range of proteins for protein purification.350 In Hanakahi’s report, this method yielded ~ 1 mg Ku/L of cell culture, approximately 5 to 10-fold less than reported for Pol θ catalytic core.271,319

Initial attempts to purify Ku utilized a method reported by Dynan.351 This method deviated slightly from Hanakahi’s purification method. Co2+ affinity chromatography was used for the first step, followed by anion exchange chromatography instead of Heparin chromatography. Size exclusion chromatography, which was not utilized by Hanakahi, was added for the final step. Notably, Co2+ affinity and anion exchange chromatography did not utilize elution gradients during purification. Instead lysate was applied to the Co2+ column, washed with lysis buffer, and immediately eluted with high imidazole

171 concentration. The eluent from the Co2+ column was applied to the anion exchange column, washed, and eluted directly with high salt. Typically, elution gradients are utilized to remove contaminating proteins, and without the elution gradient, the purity of Ku was quite poor in my hands (not shown). Anion exchange chromatography did little to improve the purity after Co2+ affinity chromatography. Size exclusion chromatography considerably improved the purity, although in most preparations the purity was still unacceptable using this method (not shown). Furthermore, the yield of Ku was relatively low. Therefore, the expression and purification of Ku were optimized in several ways (see Appendix for chromatograms).

Ku was expressed in Fernbach flasks using Terrific Broth medium instead of the standard Luria Broth (LB). Terrific Broth supports denser bacterial cultures, increasing the yield of bacterial cells harvested from the culture. For Ku expression, this led to a corresponding increase in the yield of Ku per liter of culture. The purity of Ku was greatly improved by utilizing a shallow, linear imidazole gradient for elution from the Co2+ column. Further improvements were made by replacing anion exchange medium with

Heparin medium, as employed by

Hanakahi.349 Similarly to the Co2+ column, a shallow elution gradient was utilized, considerably increasing the purity.

Following size exclusion chromatography, the purity of Ku was quite high (Figure 69).

Using this method, the yield of Ku varied Figure 69. Analysis of Ku purity by SDS- from 1-5 mg/L of culture. PAGE.

172

3.5.2 Lyase Activity of Ku on AP, C4-AP, and L in the Penultimate Position

Figure 70. Generation of abasic and oxidized abasic lesions by photolysis.

Chart 15. Substrates used for studying Ku lyase activity on abasic lesions at the penultimate position. P=phosphate.

The lyase activity of Ku on AP is most efficient on DSB substrates containing a 5’- overhang.16 Therefore, substrates 84-86 were prepared to compare the removal of oxidized abasic lesions C4-AP and L with that of AP (Chart 15). These substrates contain a biotin modification distal to the end containing the abasic lesion. Addition of streptavidin (500 nM) to the reaction mixture blocks this end, rendering it inaccessible to Ku and ensuring that Ku only binds to the end containing the abasic lesion.16 Removal of AP (84) was markedly less efficient than previously reported, with a single turnover rate constant (kobs)

173

Figure 71. Ku excision of C4-AP, AP, and L. Ku (10 nM) was incubated with 3’-32P-labeled AP (84), C4-AP (85), or L (86) (2 nM) at 37 °C. Aliquots were removed at the indicated times, quenched with NaBH4 (100 mM) and analyzed by denaturing PAGE. Error bars represent the standard deviation of three replicates. of 0.012 ± 0.004 min-1 (Figure 71). This was independent of whether AP was generated by

UDG treatment of a dU-containing substrate or by photochemical deprotection of the o- nitroveratryl-protected precursor. Excision of C4-AP (85) in the same position was

-1 considerably more efficient (kobs: 0.16 ± 0.02 min ). This is consistent with the greater lability of C4-AP.335 The C1’-oxidized abasic lesion, L (86), was not a substrate for Ku lyase activity (<3% reaction in 1 h), presumably because this lesion cannot form a Schiff base (Figure 71). Previous results showed that DSBs containing reduced AP or the AP analogue F, neither of which can be removed by Ku lyase activity, are refractory to repair by NHEJ.16 The observation that Ku efficiently removes C4-AP but shows no activity on

L suggests that the former presents no barrier to NHEJ while the latter presents a significant barrier.

174

3.5.3 Lyase Activity of Ku on DOB, pC4-AP, and 5’-dRP in the Terminal

Position

Substrate 87 was synthesized to examine the removal of DOB within the same sequence context. However, as DOB is necessarily present at the 5’ terminus (instead of the penultimate position like AP and C4-AP in this substrate), the starting material and product were inseparable by denaturing PAGE (not shown). As this substrate was a 42mer, we reasoned that it was too long to give efficient separation between starting material and product. A shorter substrate 88, which was still of sufficient length for Ku binding, was prepared. The product and starting material were separable by 20% denaturing PAGE, so additional substrates 88-90 were prepared to compare the excision of 5’-dRP, DOB, and pC4-AP at the 5’-terminus (Chart 16).

Chart 16. Substrates used to study excision of abasic lesions at the 5’ terminal position.

-1 Excision of DOB (88) was rapid, occurring slightly faster (kobs: 0.21 ± 0.02 min ) than internal C4-AP (85) (Figure 72). When pC4-AP (89) was present at the 5’-position,

-1 the reaction was considerably faster (kobs:1.09 ± 0.13 min ) than both DOB and penultimate C4-AP. 5’-dRP (90) was also removed more rapidly from the terminal position

175

Figure 72. Ku excision of pC4-AP, DOB, and 5’-dRP. (10 nM) was incubated with pC4-AP (88), DOB (89), or 5’-dRP (90) (2 nM) at 37 °C. Aliquots were removed at the indicated times, quenched with NaBH4 (100 mM) and analyzed by denaturing PAGE. Error bars represent the standard deviation of three replicates. than AP was from the penultimate position, with a single turnover rate constant (kobs) of

-1 0.042 ± 0.007 min . These data indicate that DOB, pC4-AP, and 5’-dRP are all removed from DSB termini by Ku; however, the oxidized abasic sites are removed most efficiently.

Removal of penultimate AP and terminal 5’-dRP was markedly slower than reported previously. It is unclear why this is the case; however, the previous report utilized baculovirus expression of Ku in insect cells.16 It is therefore possible that expression of Ku in E. coli yields protein that is less active than Ku purified from insect cells. Full-length

Pol θ cannot be expressed in E. coli, perhaps due to poor folding of the large protein.319 It is possible that a similar phenomenon makes Ku difficult to express in E. coli.

176

Figure 73. Lyase activity of Ku as a function of concentration. Ku (1, 2, or 4nM) with C4-AP (85, 2 nM). Aliquots were removed at the indicated time, quenched with NaBH4 and products were visualized by denaturing PAGE.

Figure 74. Effect of pre-incubation on Ku lyase activity. Ku (10 nM) was incubated for 30 min or 0 min at 37 °C in reaction buffer with 100 ng/μL BSA prior to addition of C4-AP (85, 2 nM). Aliquots were removed at the indicated time, quenched with NaBH4 and products were visualized by denaturing PAGE.

Biochemical experiments were undertaken to determine whether Ku lyase is inactivated by oxidized abasic lesions. These experiments were unsuccessful, largely because it is unclear whether Ku lyase conducts more than one turnover. When the ratio of

Ku:C4-AP was varied, excision ceased after one turnover or less (Figure 73). It is unclear

177 whether this cessation of product formation is due to inactivation of Ku lyase by C4-AP or by an intrinsic inability of Ku to do multiple turnovers. Incubation of Ku with AP or 5’- dRP could have addressed the latter possibility as neither lesion could conceivably inactivate Ku lyase; however, the rates of removal for both lesions were far too slow for this (Figure 71 and 72). Pre-incubation experiments were envisioned as a means to assay for lyase inactivation by C4-AP or DOB. Ideally, Ku would be pre-incubated with unlabeled C4-AP or DOB for a fixed time and added to AP. This would be predicted to cause a concentration-dependent and pre-incubation time-dependent decrease in lyase activity. However, pre-incubation of Ku in reaction buffer led to a complete loss of activity, even with C4-AP as the substrate instead of AP (Figure 74). This effect was rescued somewhat by addition of bovine serum albumin; however, Ku still lost a substantial amount of activity under these conditions. As removal of AP is already markedly slower than that of C4-AP, pre-incubation experiments were deemed impractical as inactivation of lyase activity by DOB or C4-AP could not be distinguished from intrinsic loss of lyase activity.

A previous report showed that lactam (61) generated by reaction of histones with C4-AP is reactive towards a fluorescent thiol probe.335 This probe was utilized to attempt detection of a putative lactam on Ku following removal of C4-AP. Although this probe was unsuccessful for detection of 61 on Pol θ, the crystal structure of Pol θ with a primer-template suggests that the active site of the enzyme may not be particularly accessible to a large hydrophobic molecule such as 62.276 The major nucleophile for Ku lyase, Lys31, is not resolved in the

178 crystal structure of Ku bound to DNA, likely because this region is predicted to be

89 flexible. The most N-terminal amino acid resolved in the crystal structure is Glu34.

Appending additional amino acids to Ku70 and modeling the structure of Ku bound to

303 DNA suggests that Lys31 is relatively exposed to solvent. Therefore, it is plausible that

62 would react with 61 on Lys31 of Ku70, even though a similar experiment was unsuccessful for Pol θ.

Attempts to detect 61 by reaction with 62 were unsuccessful for Ku. When Ku (100 nM) was incubated with 80 (50 nM) followed by addition of a large excess of 62, removal of free probe, and SDS-PAGE analysis, there was no increase in fluorescent signal for Ku reacted with C4-AP relative to Ku alone (not shown). We speculated that excess 62 may have led to a large background signal for both free Ku and Ku+C4-AP, and this may have obscured the signal coming from reaction of 62 with 61. Dialysis was unsuccessful at removing 62, perhaps due to non-covalent interaction between 62 and the membrane.

Acetone precipitation was successful in removing a large amount of the probe, but this was still insufficient to eliminate considerable background signal during gel scanning. The most successful method involved acetone precipitation followed by treatment with SDS-PAGE

Sample Prep Columns (Pierce). The protein binds to the column, which is then washed with a solution of 50% DMSO in water and then eluted. Although this method removed the most 62, considerable background fluorescent signal was still observed. It is possible that the background fluorescent signal was high enough to obscure signal due to reaction of 61 with 62. It is also possible DTT present in the Ku storage buffer reacted with 61.

3.5.4 Attempted Detection of DNA-Protein Crosslinks Between Ku and

Oxidized Abasic Lesions

179

In addition to covalent modification (such as 61) detected during excision of pC4-

AP by BER polymerases, oxidized abasic lesions also form DPCs with lyase enzymes.

Although Pol β does not excise L from DNA, attempted excision leads to formation of a

DPC in vitro.352 Pol β and Pol λ form DPCs during excision of DOB, inactivating the lyase activity of the enzymes.14,15 Treatment of cells with 1,10-copper-ortho-phenanthroline, which primarily generates L, forms Pol β-DPCs suggesting that DPC formation by oxidized abasic sites may be a biologically relevant phenomenon.353 Crosslinking of Ku to a DSB containing an oxidized abasic lesion would almost certainly prevent NHEJ and would therefore be highly cytotoxic. Therefore, SDS-PGE was used to analyze crosslinking of Ku to DNA by the oxidized abasic lesions C4-AP, L, and DOB.

Figure 75. Assaying for Ku crosslinking to C4-AP or L. Ku (10 nM) was incubated with C4-AP (85) or L (86) (2 nM). Where indicated, NaBH4 (2.5 mM) was added 30 s after the reaction was initiated. Samples were analyzed by 10% SDS-PAGE.

Ku (10 nm) was incubated with C4-AP (85), L (86), or DOB (88) (2 nM) and analyzed by SDS-PAGE. When NaBH4 (2.5 mM) was present in the reaction with C4-AP

(85), a reduced Ku-DPC (5-10% yield) was observed (Figure 75). This DPC presumably results from reduction of the Schiff base formed during C4-AP excision by Ku. A DPC

180 was not detected in the absence of NaBH4, indicating that C4-AP does not covalently trap

Ku at DSBs containing this lesion (Figure 75). Similarly, no DPC was detected when Ku was incubated with L (Figure 75). As Ku does not form a Schiff base with L, NaBH4 was not included in any reactions between Ku and L. When Ku was incubated with DOB, the intermediate could be trapped with NaBH4 (Figure 76). Interestingly, a DPC was also observed in the absence of NaBH4, suggesting that Ku is trapped at DSBs containing this lesion (Figure 76). This DPC was quite stable and persisted even after heating at 95 °C for

5 min. However, trapping of Ku in the absence of NaBH4 was inefficient and typically only

1-2% of DNA formed a DPC, even under conditions where all the DNA underwent reaction with Ku.

Figure 76. Assaying for Ku crosslinking to C4-AP or DOB. Ku (10 nM) was incubated with C4- AP (85) or DOB (88) (2 nM). Where indicated, NaBH4 (2.5 mM) was added 30 s after the reaction was initiated. Samples were analyzed by 10% SDS-PAGE.

3.5.5 Attempted Detection of Ku Modification by LC-MS/MS

181

Chart 17. Substrates used for LC-MS/MS experiments.

UPLC-MS/MS successfully detected covalent modification (61) of Pol θ and Pol β formed during excision of pC4-AP and was therefore used to detect modification of Ku.

Lacking biochemical evidence to suggest which lesions inactivate Ku lyase, both C4-AP and DOB were examined as possible candidates for covalent modification. Ku binds both ends of duplex DNA, so a biotin modification was included distal to the abasic lesion in substrates 84-90 which were used for kinetic experiments. Incubation with streptavidin blocked Ku from binding to this end of the duplex, allowing for binding only to the end containing the lesion. However, streptavidin could not be used in MS experiments as the protease would digest streptavidin as well as Ku, complicating analysis of the Ku digest.

Substrates 91 and 92 were synthesized, containing a stem loop structure at the end distal to the lesion, blocking Ku binding at this site (Chart 17). The design of this substrate was inspired by the substrate used for determining the crystal structure of Ku bound to duplex

DNA, where it was also necessary to block Ku binding to one end of the duplex without

182 addition of streptavidin.89 Excision of C4-AP from 91 was similar to that of 85, so 2 equivalents of Ku were needed to drive the reaction to completion (not shown). Excision of DOB from 92 was not measured because the product is very difficult to separate from the starting material on substrates longer than 26 bp. However, C4-AP and DOB are removed with similar efficiency on substrates 85 and 88, suggesting that removal of DOB will also proceed to near completion on 92.

Figure 77. Coverage map for Ku digestion by trypsin. Peptides highlighted in blue were detected by LC-MS/MS while those which are white were not detected. 183

Figure 78. Coverage map for digestion of Ku by trypsin/Lys-C. Peptides highlighted in blue were detected by LC-MS/MS while those which are white were not detected.

UPLC-MS/MS experiments were originally hindered by difficulty in digesting Ku to completion. Digestion with trypsin gave only 40% sequence coverage, although the

303 proposed major nucleophile, Lys31 of Ku70, was covered in trypsin digests (Figure 77).

Digestion with Glu-C worsened sequence coverage to 14% (not shown). Digestion with trypsin/Lys-C mix greatly improved sequence coverage (>70%), similar to observations

184 for Pol θ digestion (Figure 78). Although Lys31 was consistently covered in these digests, incubation of Ku with DOB or C4-AP did not give rise to detectable modification (not shown). Previous reports showed that additional residues on Ku70 (Lys160) and Ku80

(Lys543-545, Lys565, Lys566, and Lys568) participate in removal of AP in the absence of

303 Lys31. We therefore speculated that C4-AP and/or DOB covalently modified these residues instead of Lys31.

Of the six compensatory lysines present on Ku80, five were covered by trypsin/Lys-

C digest and none of these residues were found to be modified by DOB or C4-AP. Lys568 on Ku80 and Lys160 on Ku70 were absent from trypsin/Lys-C digests. Biochemical evidence indicates that Lys160 is the secondary nucleophile when Lys31 is mutated to alanine, so this residue was regarded as most likely (other than Lys31 which was found not to be modified) to be modified by C4-AP or DOB.303 Digestion with trypsin/Lys-C followed by Glu-C digestion gave coverage of Lys568 of Ku80, although modifications were not detected (Figure 78). Similarly to digestion with trypsin/Lys-C alone, Lys31 modification was not detected when Ku was incubated with C4-AP (91) or DOB (92) nor was modification of any other lysines (Figure 79). Importantly, Lys160 was not detected by this method, indicating that another method was needed to cover this residue.

185

186

Figure 79. Coverage map for Ku incubated with C4-AP (91) or DOB (92) and digested by trypsin/Lys-C followed by chymotrypsin. A) C4-AP B) DOB. Peptides detected in control and analyte are highlighted green. Those only in the control are blue. Those only in analyte are yellow.

Analysis of the primary sequence of Ku70 suggested that chymotrypsin was a good candidate protease to cover Lys160, as cleavage at nearby Phe159 and Phe169 would generate

187 a suitably sized peptide for UPLC-MS/MS analysis. Digestion of Ku with chymotrypsin alone gave poor overall sequence coverage (30%) and did not cover Lys160 (not shown); however, sequential digestion of Ku with trypsin/Lys-C followed by chymotrypsin gave greatly improved sequence coverage (Figure 78). Lys31 and Lys160 of Ku70 as well as all six candidate lysines (Lys543-545, Lys565, Lys566, and Lys568) on Ku 80 were covered by this method, yet modification was not found at any position. DNA was isolated from the reaction between Ku and 91 and the reaction was confirmed to have proceeded to completion.

It is unclear why Ku modification was not detected following excision of C4-AP or

DOB. This may indicate that Ku is not modified by either C4-AP or DOB. However, it is possible that Ku is modified by one or both of these lesions, but the modification was not detectable in these experiments. This could stem from multiple factors including inefficient modification as well as modification of multiple lysines. For example, Ku modification may be heterogeneous. Modification of multiple residues (Lys312, Lys324, and Lys273) was

15 detected when Pol λ excised pC4-AP. Ku possesses several lysine residues (Lys160 of

Ku70 as well as Lys543-545, Lys565, Lys566, and Lys568 of Ku80) which can compensate for

303 removal of AP in the absence of the major nucleophile, Lys31. Therefore, Ku modification may be distributed over a number of different lysine residues, reducing the detectable signal of modified Ku peptides. Additionally, it is reasonable to speculate that modification of Ku by C4-AP or DOB may be inefficient. Pol β is inactivated by DOB after approximately 4 turnovers and pC4-AP after approximately 8 turnovers, while Pol λ is inactivated by both lesions after 4 turnovers.13–15 Pol θ is inactivated by only pC4-AP, and this occurs after approximately 4 turnovers on a BER substrate and 6 turnovers on a

188

DSB. It is therefore reasonable to speculate that modification of Ku may be inefficient if it does occur, with only a fraction of the enzyme modified after a single turnover. Ku does not appear to conduct multiple turnovers, so it was not possible to maximize the fraction of modified Ku by incubation with excess DNA, as for Pol θ. Instead, a two-fold excess of

Ku over DNA was needed to ensure complete reaction, meaning that only 50% of Ku could be modified even if modification was quantitative.

Conclusions

The results presented here reveal interesting biochemical phenomena during the interactions of polymerases and repair enzymes with oxidative DNA damage. Pol θ inserts nucleotides less efficiently opposite C4-AP and L than it does for AP.322 Pol θ also has a strong propensity to form frameshift mutations during lesion bypass, for both abasic and oxidized abasic sites as well as the oxidized nucleobase lesion, Tg.323 The propensity to induce frameshift mutations also varies for each lesion, following the trend C4-AP > AP >

L on the same local sequence. Pol θ had an even greater tendency to induce frameshift mutations during bypass of Tg, albeit on a different local sequence. It remains to be seen whether Pol θ forms such frameshift mutations during in vivo lesion bypass. If so, this could contribute to genomic instability and genetic diversification in tumors, potentially driving poor cancer prognosis or response to treatment for tumors which upregulate Pol θ.

Similar variability between lesions is also apparent for Pol θ lyase activity, where

DOB is removed more efficiently than 5’-dRP, but pC4-AP irreversibly inactivates Pol θ lyase. Interestingly, Pol θ promotes cellular resistance to bleomycin, which produces C4-

AP along with a smaller proportion of DSBs.20,50 The inactivation of Pol θ by pC4-AP

189 suggests that the lyase activity of Pol θ may not be important for promoting bleomycin resistance. Instead, the repair of DSBs produced by bleomycin is more likely the mechanism by which Pol θ promotes cellular resistance to DSBs.20

Reaction of Pol θ with pC4-AP followed by trypsin/Lys-C digestion and UPLC-

MS/MS analysis found that Lys2383 was covalently modified by pC4-AP providing a chemical basis for the inactivation of Pol θ lyase activity by pC4-AP. This modification directed further studies into the Pol θ lyase reaction. Mutation of Lys2383 to Ala or Arg considerably reduced the lyase activity, indicating that this residue is likely the major nucleophile in the lyase reaction. Interestingly, Lys2383 is also essential for efficient polymerase activity, consistent with its role in coordinating the incoming dNTP during

DNA synthesis.276 This is the first demonstration that a BER polymerase utilizes a single residue in lyase and polymerase activities. Mutation of Lys207 in Pol ι reduces lyase and

354 polymerase activity, but the mechanism for this phenomenon is unclear. Lys207 makes direct contacts with the phosphodiester backbone of DNA and may be important for DNA binding, so this residue may not be directly involved in either lyase or polymerase activity

232,354 of Pol ι. By contrast, mutation of Lys2383 does not impact binding of Pol θ to DNA but dramatically reduces lyase and polymerase activity. The dual role for Lys2383 may inform the design of inhibitors towards Pol θ, which is currently an appealing drug target for cancer therapy.283 As the polymerase activity of Pol θ essential for DSB repair by alt-

20 EJ, inactivation of polymerase activity by covalent modification of Lys2383 would be expected to inhibit DSB repair by Pol θ.

The interactions of Ku with oxidized abasic lesions were distinct from that of abasic sites AP and 5’-dRP. The oxidized abasic lesions C4-AP and DOB were excised from DNA

190 considerably faster than AP or 5’-dRP. Reaction of Ku with DOB or C4-AP followed by

UPLC-MS/MS analysis did not detect covalent modification of the protein. However, oxidized abasic sites did provide some unique challenges for Ku. The C1’ oxidized abasic lesion, L, was not a substrate for Ku lyase activity, indicating that this lesion may persist at DSB termini and impede end joining. Additionally, a small amount of DNA-protein crosslink was detected following excision of DOB from DNA. These results are consistent with previous observations showing that oxidized abasic lesions can give rise to more deleterious outcomes than AP.13–15,22

4. Future Experiments

The results presented here suggest several interesting experiments. Pol θ forms frameshift mutations during in vitro bypass of AP, C4-AP, L, and Tg. It is unknown whether Pol θ contributes to the bypass of the former three lesions in cells. This question could be addressed by assessing the replication of heteroduplex plasmids containing AP,

C4-AP, or L in cells, similar to previous reports utilizing other lesions.177–179 A site-specific lesion is inserted into the lacZ gene, and the opposite strand contains a bulge with a restriction site. If the lesion is repaired by BER, replication of the opposite strand will result in insertion of an additional nucleotide, shifting the reading frame of the lacZ gene. Plasmid isolation and transformation into E. coli plated on the appropriate medium gives blue colonies for in-frame replication products and white colonies for out-of-frame products.

Frameshift mutations resulting from template slippage during lesion bypass can be distinguished from frameshifts resulting from BER, because the latter products can be digested with the specific restriction enzyme. The percentage of white colonies resulting from template slippage can therefore be determined by comparing colony ratios in the

191 presence or absence of restriction digest. Alternatively, plasmids could be isolated from white colonies and checked for digestion by the restriction enzyme or sequenced.

The contribution of Pol θ to cellular BER is unclear. When key residues in the lyase domain of Pol β (K72A/K35A/K68A) are mutated, the enzyme loses lyase enzyme but possesses normal polymerase activity.155,355 Expression of K72A/K35A/K68A Pol β in cells established that the lyase activity of Pol β is highly important for the cellular response to DNA alkylation.355 A similar method cannot be used for Pol θ because mutation of

Lys2383, the primary nucleophile for lyase activity identified here, dramatically reduces polymerase activity. Therefore, if a phenotype resulted from expression of K2383A Pol θ in cells, it could not be conclusively assigned to a defect in lyase activity, as the polymerase activity would also be deficient. It may be possible to reduce lyase activity by mutating other residues which are not essential for polymerase activity. Without a crystal structure for Pol θ bound to a lyase substrate, it is unclear which residues to mutate; however, crystallographic data could aid in this search. If other residues were identified as important for lyase activity but not polymerase activity, the importance of Pol θ lyase activity in cellular BER could be assessed. Even without such information, it may be possible to assess BER by Pol θ using reporter plasmids. As mentioned previously, heteroduplex plasmids containing site-specific lesions can distinguish BER from TLS and could also be used to measure the contribution of Pol θ to BER.

5. Experimental

General Methods Oligonucleotides were synthesized via automated DNA synthesis on an

Applied Biosystems model 394 instrument. Commercially available DNA synthesis

192 reagents including 5’-phosphorylation reagent (Solid CPR II), thymidine glycol CE phosphoramidite, SIMA HEX (dichloro diphenyl fluorescein) phosphoramidite, 5’-biotin

TEG phosphoramidite, and THF abasic site analogue (dSpacer) were purchased from Glen

Research. C4-AP, DOB, L, and AP photochemical precursor phosphoramidites were synthesized as described previously.14,26,324,325

T4 polynucleotide kinase, terminal deoxynucleotidyl transferase, Acc65I, EcoRI, uracil DNA glycosylase, Dpn 1, Phusion polymerase, dNTPs, Quick Ligase Kit, NEB buffer 3.1 were from New England Biolabs. Radionuclides were from Perkin Elmer. C18

Sep Pak cartridges were from Waters. Protease inhibitor cocktail (EDTA-free) was from

Roche. HisTrap column, Heparin column, HiTrap Q column, HiTrap SP column, and Talon column were from GE Healthcare. Nucleotide Removal Kit, Maxi Prep and Mini Prep Kits were from Qiagen. SDS-PAGE Sample Prep Columns were from Pierce. BL21 DE3,

Rosetta pLysS, and Rosetta 2 pLysS E. coli were from Millipore. Poly-Prep columns were from Bio-Rad. DH5α E. coli, DynadBeads M-270 Streptavidin, and SYPRO Ruby protein stain were from ThermoFisher. SUMO protease 2 was from LifeSensors. Tryptone and yeast extract were from BD Biosciences. NP-40 substitute (BioXtra) and α-lactose were from Sigma Aldrich. Imidazole was from Acros. Dextrose and glycerol were from Fisher.

Fluorescence anisotropy measurements were conducted using an AVIV Biomedical Model

ATF 107 spectrofluorometer at the Center for Molecular Biophysics at Johns Hopkins

University. Photolyses were carried out in a Rayonet photoreactor fitted with 16 lamps having a maximum output at 350 nm. Sonication was conducted using a Branson SFX-150 sonifier. Protein purification was conducted using an AKTA FPLC.

Synthesis, Purification, and Characterization of Modified Oligonucleotides

193

Oligonucleotides containing Tg precursor were synthesized using UltraMild phosphoramidites from Glen Research. For UltraMild syntheses, the standard Cap A solution was replaced with cap mix containing THF (distilled over Na), trimethylacetic anhydride (distilled over P2O5), and 2, 6-lutidine (distilled over CaH2) in a 8:1:1 ratio by volume.356 All other oligonucleotides were synthesized using standard reagents according to the manufacturer. Oligonucleotides containing Tg precursor were deprotected using concentrated ammonium hydroxide (2 h, room temperature) followed by desilylation using a solution (300 µL) containing triethylamine trihydrofluoride, triethylamine, and anhydrous N-methylpyrrolidinone (in a 4:3:6 ratio by volume) at 65 °C for 1.5 h.330,331 The desilylation solution was removed by ethanol precipitation (30 μL of 300 mM sodium acetate pH 5.6, 1 mL ethanol, -80 °C 1 h) followed by centrifugation (16,000 g, 30 min, 4

°C) and removal of the supernatant. Oligonucleotides containing SIMA HEX were deprotected using concentrated ammonium hydroxide (12 h, 55 °C). Oligonucleotides containing biotin-TEG modification were deprotected with a solution (1 mL) of concentrated ammonium hydroxide (500 µL) and 40 % methylamine in H2O (500 µL) at

65 °C for 1 h followed by drying in a Speed Vac concentrator. The oligonucleotide was manually detritylated with 300 μL of 80% acetic acid, followed by ethanol precipitation

(75 μL of 3M sodium acetate pH 5.6, 900 μL ethanol, incubated on dry ice 30 min), followed by centrifugation (16,000 g, 30 min, 4 °C) and removal of the supernatant.

All other oligonucleotides (including THF abasic site analogue, 5’-phosphate, C4-

AP, DOB, AP, or L precursors were deprotected using a solution (1 mL) of concentrated ammonium hydroxide (500 µL) and 40 % methylamine in H2O (500 µL) at 65 °C for 1 h.

Following deprotection, all oligonucleotides were dried in a Speed Vac concentrator,

194 resuspended in loading buffer (95% formamide with 10 mM EDTA, 150 µL) and purified by 20% denaturing polyacrylamide gel electrophoresis (PAGE) and visualized using a TLC plate and UV lamp. Oligonucleotides containing photochemical precursors were exposed to UV light for the shortest time possible and were kept covered from ambient light at all times. The band was excised from the gel, crushed into small pieces, and eluted from the gel matrix with elution buffer (10 mL, 100 mM NaCl, 1 mM EDTA) for 12 h with shaking.

The solution was filtered using a Poly-Prep column and desalted using C-18-Sep-Pak cartridges. Modified oligonucleotides were characterized by MALDI-TOF MS, ESI-MS, or UPLC-MS. For MALDI-TOF MS, oligonucleotides (100 pmol) were desalted by Zip-

Tip and resuspended in matrix (1 μL) containing 3-hydroxypicolinic acid (10 mg/mL) and ammonium citrate (30 mg/mL). For ESI-MS, oligonucleotides (3 nmol) were dissolved in

H2O (120 μL) and precipitated by addition of NH4OAc (5M, 50 μL, pH 5.6) followed by cold ethanol (600 μL). The mixture was incubated on dry ice (20 min), centrifuged (16,000 g, 30 min) and the supernatant was removed. For UPLC-MS, 10 pmol of oligonucleotide was used without desalting.

Radiolabeling and Preparation of Oligonucleotide Complexes

Oligonucleotides were 5'- 32P-labeled by T4 polynucleotide kinase and γ- 32P-ATP.

Radiolabeling of the 3'-termini was accomplished using - 32P-cordycepin triphosphate and terminal deoxynucleotidyl transferase. Ternary complexes were prepared by mixing

3'-32P-labeled oligonucleotide with the appropriate template and flanking strand in a 1:1.5:3 ratio in phosphate buffered saline (10 mM sodium phosphate, 100 mM NaCl, pH 7.2), heating to 95 °C, and slowly cooling to 25 °C. Primer-template complexes were prepared in the same fashion, but the 5'-32P-labeled primer strand and the template were mixed in a

195

1:1.5 ratio. Ternary complexes containing fluorophore-labeled oligonucleotides were prepared by annealing the fluorophore-labeled template with the other two strands in a

1:1.5:1.5 ratio.

Oligonucleotides containing AP were generated by treatment of dU-containing oligonucleotides with uracil DNA glycosylase (5 units, 37 °C, 15 min) or by photolysis of the photochemical precursor (350 nm, 10 min), the synthesis of which was reported previously.26 Oligonucleotides containing 5’-dRP, C4-AP, DOB, and L were generated by photolysis of the oligonucleotide containing the appropriate precursor (350 nm, 10 min for

5’-dRP, C4-AP, and DOB, 20 min for L).14,261,325

Expression of Pol θ QM (residues 1792-2590) fragment An aliquot (20 μL) of E. coli

(Rosetta 2 DE3 pLysS) was transformed with 10 ng of plasmid (received from Dr. Sylvie

Doublié) encoding the QM fragment (residues 1792-2590) of Pol θ with an N-terminal

SUMO tag and a hexaHis tag.271 Cells were plated on LB + ampicillin (100 μg/mL) and chloramphenicol (34 μg/mL) plates and grown at 37 °C for 16 h. Five colonies were suspended in LB medium (20 mL) and a portion (1 mL) was added to 1 L of autoinduction media (12 g/L tryptone, 24 g/L yeast extract, 4 mL glycerol, 0.05% (w/v) dextrose, 0.2%

(w/v) α-lactose, 0.017 M KH2PO4, 0.072 M K2HPO4, 100 μg/mL ampicillin, 34 μg/mL chloramphenicol) in a 2.8 L baffled Fernbach flask. Cultures were grown with 220 rpm shaking for 60-64 h at 20 °C at which point saturation of the culture was confirmed by measuring the OD595, which was typically 10-12. Measurement of the OD595 was accomplished by diluting 50 µL of culture in 950 µL of fresh media and back-calculating the optical density. The cells were harvested by centrifugation (4800 g, 5 min, 4 °C), the

196 media was poured off, and the cells were frozen with liquid nitrogen, and stored at -80 °C giving ~23 g of cells/L of culture.

Purification of Pol θ QM fragment The entirety of the purification was conducted at 4

°C. The following buffers were prepared, all of which were adjusted to pH 7 with concentrated HCl: buffer A: 20 mM Tris-HCl, 300 mM NaCl, 10 mM imidazole, 5 mM

BME, 10% glycerol, 0.01% NP-40 substitute (BioXtra); buffer B: 20 mM Tris-HCl, 300 mM NaCl, 500 mM imidazole, 5 mM BME, 10% glycerol, 0.01% NP-40 substitute

(BioXtra); buffer C: 20 mM Tris-HCl, 300 mM NaCl, 5 mM BME, 10% glycerol, 0.01%

NP-40 substitute (BioXtra); buffer D: 20 mM Tris-HCl, 1 M NaCl, 5 mM BME, 10% glycerol, 0.01% NP-40 substitute. Cells (10 g) were resuspended in buffer A (100 mL) containing protease inhibitor cocktail without EDTA (1 tablet per 100 mL). Cells were lysed by 5 cycles (30 s on, 60 s off, 50% amplitude) of sonication in an ice-water bath. The temperature was monitored with a thermometer and sonication was paused if the temperature exceeded 4 °C, which typically did not occur. The lysate was cleared by two rounds of centrifugation (15000 g, 25 min, 4 °C) and applied (flowrate 1 mL/min) to a Ni2+ column (5 mL HisTrap) which had been equilibrated with 5-10 column volumes of buffer

A. The protein was eluted by applying a linear gradient to buffer B (0-100% buffer B over

100 min with a flowrate of 1 mL/min). Pol θ typically eluted as a broad peak beginning at approximately 30% buffer B. Fractions were analyzed by SDS-PAGE based on analysis of the UV chromatogram. Pol θ-containing fractions (~15-25 mL) were pooled and applied to Heparin column (5 mL HiTrap Heparin) equilibrated with 10 column volumes of buffer

C) at a flowrate of 1 mL/min. The protein was eluted by applying a linear gradient to buffer

D (0-100% over 100 min at flowrate 1 mL/min). Pol θ typically eluted beginning at 55%

197 buffer D in a volume of 6-8 mL. SDS-PAGE analysis was typically unnecessary at this step because the Pol θ-containing peak was the largest and sharpest peak observed during elution; however, analysis by 10% SDS-PAGE could still be employed. Pol θ-containing fractions (~6-8 mL) were pooled. An aliquot (40 µL) was saved while the remainder of the protein was cleaved by SUMO protease 2 (5 U) to remove the SUMO tag. The cleavage reaction was aliquoted into 2 mL tubes and gently rotated overnight in a tube rotator in a cold room at 4 °C for at least 4 h (typically overnight). The tubes could also be placed into a large ice bath, covered, and incubated overnight on a rotary shaker at approximately 60 rpm. Comparison of uncleaved and cleaved proteins by SDS-PAGE showed that cleavage of the SUMO tag was complete. Cleavage was typically only 50% if the tubes were not rotated. The cleaved protein was applied to a Ni2+ column (5 mL HisTrap, GE Healthcare, equilibrated with 10 column volumes of buffer A). SUMO protease 2 and SUMO (both hexaHis tagged) bound tightly to the column, while Pol θ bound weakly and eluted after washing with ~5 column volumes of buffer A. In some cases, two elution peaks were observed, with an initial peak at one column volume and the second, Pol θ-containing peak at approximately 5 column volumes. The initial peak at 1 column volume typically contained mostly low molecular weight impurities and only a small amount of Pol θ.

Fractions from both peaks were analyzed by SDS-PAGE. Concentration by Amicon Ultra centrifugal filter (30 K cutoff) was typically conducted at this step; however, a considerable portion of protein bound to the membrane and was lost. Therefore, fractions containing the most Pol θ (as determined by SDS-PAGE) were subjected to Bradford assay and if the concentration was sufficiently high, (typically 5-10 µM), these fractions were not concentrated and were instead aliquoted into small portions and frozen with liquid nitrogen

198 prior to storage at -80 °C. Less concentrated fractions were subjected to concentration by

Amicon Ultra (30 K cutoff) by centrifugation (10 min, 5000 g, 4 °C) followed by gentle pipetting to mix the concentrate. Centrifugation was repeated the same way, and the concentrate (now ~ 1.5 mL) was mixed by pipetting and then aliquoted into microcentrifuge tubes which were flash frozen with liquid nitrogen and stored at – 80 °C.

Bradford assay using bovine serum albumin as a standard was used to determine the concentration. Protein yield was ~5-10 mg/L of culture when the protein was not concentrated; however, the yield was reduced by approximately 50% by concentration.

Generation of Pol θ mutants Pol θ variants were generated by site-directed mutagenesis of the Pol θ plasmid reported previously.271 PCR was conducted with the following primers: primers for the K2383A mutation were 5'-d(CTG AGG CAG CAG GCA GCA

CAG ATT TGC TAT GGG ATC A) and 5'-d(TGA TCC CAT AGC AAA TCT GTG CTG

CCT GCT GCC TCA G). Primers for the K2383R mutation were 5'-d(CTG AGG CAG

CAG GCA AGA CAG ATT TGC TAT GGG ATC A) and 5'-d(TGA TCC CAT AGC

AAA TCT GTC TTG CCT GCT GCC TCA G). Primers for the K2575A/K2577A mutation were 5’-d(GTC TGT GAA ATT GAA AGT GGC AAT AGG CGC CAG CTG GGG AGA

GC) and 5’-d(GCT CTC CCC AGC TGG CGC CTA TTG CCA CTG CCA CTT TCA

ATT TCA CAG AC). The K2575A/K2577A mutant was generated by PCR amplification of the wild type plasmid using primers for the K2575A/K2577A mutant. The

K2383A/K2575A/K2577A mutant plasmid was generated by PCR amplification of the

K2383A mutant plasmid using primers for the K2575A/K2577A mutant.

PCR was conducted using 50 ng of plasmid and 15 pmol of each primer in a 50 μL reaction using Phusion polymerase according to manufacturer protocol. Briefly, the

199 reaction contained the provided high fidelity buffer (10 μL), dNTPs (1 μL of 10 mM solution), both primers (1 μL for each, 15 μM solution), the parental plasmid (50 ng, 1 μL),

Phusion polymerase (1 unit, 0.5 μL), and H2O (35.5 μL). PCR was conducted with 95 °C initial denaturing (30 s), followed by 18 cycles of 95 °C denaturing (30 s), 55 °C annealing

(1 min), 72 °C extension (5 min). Parental plasmid was digested by addition of Dpn1 (1

μL, 20 units) at 37 °C for 4 h. An aliquot (2 μL) of this mixture was transformed into DH5-

α cells (40 μL) by incubation on ice (30 min), heat shock at 42 °C (30 s), and incubation on ice (5 min). SOC medium (200 μL) was added and outgrowth was conducted at 37 °C for 1 h with 250 rpm shaking. A portion (100 μL) of the transformed cells was plated on an LB plate with ampicillin (100 μg/mL). Plates were grown for 16 h at 37 °C. A single colony was picked and resuspended in LB media (5 mL) with ampicillin (100 μg/mL) and grown for 16 h at 37 °C. Plasmids were isolated by Mini prep according to the manufacturer protocol and sequenced to confirm mutagenesis. The mutant proteins were expressed and purified in the same fashion as the wild type.271,319

Sanger Sequencing Sanger sequencing was conducted by the Johns Hopkins Genetic

Resources Core Facility. Samples (12 µL) contained plasmid (100 ng/µL) and a sequencing primer (500 nM). The following primers were used for sequencing of Pol θ QM plasmid:

5’-d(CTT CTT TCT TGT GGC ATC TCC TTG); 5’-d(AAC ATA TAA TCG AAG CCA

AGC); 5’-d(GAA GGG TTT CAG CGT GAA TCC); 5’-d(GCG TAG ATA TTT GCC

AGG).

Expression of hexa-His tagged Ku70/80 Terrific Broth (2 L) was prepared according to the standard recipe (for 2 L of broth: 24 g tryptone, 48 g yeast extract, 8 mL glycerol, H2O diluted to 1.8 L with H2O and this solution was autoclaved. Rosetta DE3 pLysS E. coli

200

(EMD Millipore) were transformed with Ku vector (10 ng) according to the manufacturer’s protocol and grown on LB plates supplemented with ampicillin (100 μg/mL) and chloramphenicol (34 μg/mL). A single colony was picked and grown in LB media (50 mL) containing ampicillin (100 μg/mL) and chloramphenicol (34 μg/mL) at 37 °C until visually confluent (12 h, 250 rpm). The next morning, the overnight culture (20 mL) was poured into the Terrific Broth along with an autoclaved solution (200 mL) of potassium phosphate buffer (0.17M KH2PO4 and 0.72M K2HPO4). Ampicillin (100 mg/mL, 2 mL) was added to a final concentration of 100 μg/mL, and chloramphenicol was added (34 mg/mL, 2 mL) to a final concentration of 34 μg/mL. The culture was divided into 2.8 L baffled Fernbach flasks (1L of culture in each) and was grown at 37 °C. After 5 h, when the optical density

(OD595) reached 1.0, the culture was immersed in an ice bath until the temperature reached

18 °C and a solution of IPTG in water was added to a final concentration of 0.25 mM.

Induction was carried out for 18 h at 18 °C with 240 rpm shaking. Cells were harvested by centrifugation (4200 g, 10 min, 4 °C), frozen with liquid nitrogen, and stored at -80 °C yielding 22 g of cells/L of culture.

Purification of Ku70/80 The composition of buffers (all pH 8.0) used during Ku purification is listed below. Protease inhibitor cocktail, BME, DTT, and PMSF were added to buffers immediately before use. Lysis buffer: 50 mM Tris-HCl, 1 M NaCl, 0.4 M

NH4OAc, 2 mM β-mercaptoethanol (BME), and 1× Complete protease inhibitor cocktail.

Buffer E: 50 mM Tris-HCl, 50 mM NaCl, 5% glycerol, 5 mM imidazole, 1 mM BME, and

100 μM phenylmethane sulfonyl fluoride (PMSF). Buffer F: 50 mM NaCl, 5% glycerol,

250 mM imidazole, 1mM BME, and 100 μM PMSF. Buffer G: 50 mM Tris-HCl, 50 mM

NaCl, 500 μM EDTA, 1 mM DTT, 5% glycerol, and 100 μM PMSF. Buffer H: 50 mM

201

Tris-HCl, 1 M NaCl, 500 μM EDTA, 1 mM DTT, 5% glycerol, and 100 μM PMSF. Ku storage buffer: 25 mM Tris-HCl, 2 mM DTT, 150 mM KCl, 10% glycerol.

Cells (15 g) were resuspended in lysis buffer (150 mL) and treated with lysozyme

(2 mg/mL) on ice for 30 min, followed by four cycles of sonication in an ice-water bath

(30 s on, 60 s off, 50 % amplitude). Purification was conducted at 4 °C. The cell lysate was cleared by two rounds of centrifugation (12000 g, 20 min, 4 °C) and applied to a Talon column (5 mL, GE healthcare, equilibrated with 50 mL lysis buffer before application of lysate) at a flow rate of 1 mL/min. The column was washed with lysis buffer (1 mL/min flowrate) until the UV trace reached baseline (approximately 15 column volumes) and then washed with two column volumes of buffer E (1 mL/min). Elution was conducted using a linear gradient of 0-100% buffer F over 50 min at a flow rate of 1 mL/min. Ku usually eluted at 75-100 mM imidazole. Fractions (1 mL) were analyzed by SDS-PAGE to determine purity. Ku-containing fractions (usually 10-15 mL) were pooled and directly applied to Heparin column (two 1 mL columns connected in tandem) which had been equilibrated with 15 column volumes of buffer G. The flow rate for injection was 0.4 mL/min. Protein was eluted with a gradient to 100% buffer H at a flowrate of 0.5 mL/min over 40 min. Ku protein usually eluted at 50% buffer B. Fractions (1 mL) were analyzed by SDS-PAGE and Ku-containing fractions (3) were concentrated using an Amicon spin column (30 K molecular weight cutoff) according to the manufacturer’s protocol to a volume of 500 μL. Size exclusion column (Superdex 200, 30/100 GL, GE Healthcare) was equilibrated with two column volumes of Ku storage buffer. The concentrated eluent from the Heparin column was injected onto the Superdex at a flow rate of 0.5 mL/min and purification was conducted at the same flow rate. Ku protein usually eluted at

202 approximately 8 mL after injection (the column volume is 24 mL). Fractions (0.5 mL) were analyzed by SDS-PAGE to determine purity. Two fractions (500 μL) contained pure

Ku70/80. These were aliquoted, frozen on liquid nitrogen, and stored at -80 °C. Protein concentration was determined by Bradford Assay. The yield was typically 1-2 mg/L of culture.

Expression of Polymerase Beta BL21 DE3 E. coli (40 µL) were transformed with a plasmid encoding human DNA polymerase β (10 ng) by incubation on ice (30 min), heat shock (42 °C, 20 s), and incubation on ice (5 min). SOC media (400 µL) was added and the cells were grown with shaking (37 °C, 250 rpm, 1 h). An aliquot of transformed cells

(20 µL) was mixed with SOC media (30 µL) and plated on LB plates containing amipicillin

(100 µg/mL) followed by incubation at 37 °C for 16 h. A single colony was resuspended in LB media (100 mL) containing ampicillin (100 µg/mL) and incubated 12 h at 37 °C with shaking. A portion (20 mL) was centrifuged (4200 g, 5 min) and the spent media was poured off. The pellet was resuspended in fresh LB media and 10 mL was added to 1 L LB media containing ampicillin (100 µg/mL). The cells were grown at 37 °C with 250 rpm shaking for 4 h until the OD595 was 0.7. The incubator was cooled to 20 °C and cells were incubated with shaking for 45 min followed by induction with IPTG (1 mL of 1 M IPTG added to each culture) followed by growth at 20 °C for 18 h. Cells were harvested by centrifugation (4200 g, 10 min). The pellet was washed with 5 mM sodium phosphate, 50 mM NaCl pH 7.2 and frozen on liquid nitrogen prior to storage at -80 °C.

Pol Beta Purification Cells harvested from 1 L of culture were resuspended in 70 mL of lysis buffer (50 mM Tris HCl pH 8.0, 0.1 mM EDTA, 400 mM KCl, 5 % glycerol, 1 mM

DTT, 1 tablet of complete protease inhibitor) and lysed by sonication in an ice water bath

203

(10 s on, 10 s off, 6 min, 50 % amplitude). The lysate was clarified by centrifugation (15000 g, 20 min) two times and loaded (1 mL/min) onto a 5 mL HiTrap Q column which had been equilibrated in lysis buffer. The flowthrough, which contained Pol Beta, was collected and diluted with an equal volume of buffer E (50 mM Tris HCl pH 8.0, 0.1 mM EDTA, 5

% glycerol, 1 mM DTT) to reduce the KCl concentration to 200 mM. This solution was loaded (1 mL/min) onto a 5 mL HiTrap SP column which had been equilibrated with buffer

F (50 mM Tris HCl pH 8.0, 0.1 mM EDTA, 200 mM KCl, 5 % glycerol, 1 mM DTT). The column was washed with buffer E until UV absorbance reached baseline and then Pol β was eluted using a gradient (1 mL/min, 0-100% over 50 min) to buffer F (50 mM Tris HCl pH 8.0, 0.1 mM EDTA, 800 mM KCl, 5 % glycerol, 1 mM DTT). Pol β eluted at approximately 400 mM KCl. Fractions were pooled, the concentration was determined by

Bradford assay, and the solution was aliquoted, frozen on dry ice, and stored at -80 °C.

Protein yield was approximately 15 mg/L of culture.

Bradford Assay Bradford assay was used to determine the concentration of purified proteins. A standard curve was generated by adding Bradford reagent (900 μL, Sigma

Aldrich) to solutions (100 μL) containing 100, 200, or 400 pmol of bovine serum albumin.

After 5 min incubation at room temperature, the absorbance (595 nm) of each standard was measured using a spectrophotometer giving values of 0.078, 0.159, 0.352, or respectively.

Absorbance was plotted against amount of BSA and fit to straight line giving the equation y=0.000921x – 0.0185 where y is the absorbance and x is the amount of protein in pmol.

Pol θ Polymerase Active Site Titration The active fraction of Pol θ polymerase active site was determined by pre-steady state kinetic analysis using a KinTek RQF-3. A solution of Pol θ (50 nM) and primer-template 27 (100 nM) was mixed with dGTP (1 mM) and

204 quenched at defined time points with a solution of 80% formamide and 100 mM EDTA.

In a typical experiment, a solution (150 μL) of Pol θ (100 nM) and primer-template (200 nM) was prepared in reaction buffer (10 mM Tris•HCl pH 8.0, 25 mM KCl, 1 mM BME,

10 mM MgCl2). A solution of dGTP (1 mL, 1 mM) was prepared in the same reaction buffer. The drive syringes of the RQF-3 were filled with reaction buffer while the quench syringe was filled with a solution of 80% formamide and 100 mM EDTA. The solutions of dGTP and Pol θ-primer-template were drawn into disposable syringes (1 mL) and the sample loops of the RQF-3 instrument were filled with either solution. The reaction time

(3, 4, 5, 6, 10, 25, 50, 100, 250, 500, 1000, 2500, 5000 ms) was set on the computer controller of the RQF-3 and the quench volume was NOT set as constant. Reaction initiation and quenching were automatically conducted by the instrument as soon as the

“start” button was pressed. Reactions were dispelled from the instrument through the exit loop and collected in a microcentrifuge tube (1.6 mL) containing a solution (50 μL) of 500 mM EDTA with trace bromophenol blue and xylene cyanol. Radiation in each sample was normalized and an aliquot (5 μL) of each sample was loaded on a 20% denaturing PAGE, which was run at 55 W. The fraction of extended primer was plotted as a function of time

-kt and this was fit to the equation P=A(1-e ) + ksst where P is fraction of extended primer, A is the burst amplitude, k is the burst phase rate constant, kss is the steady-state rate constant, and t is time. The fraction of active enzyme was determined by dividing the experimental burst amplitude by the theoretical burst amplitude (50%) if all the enzyme were active.

This experiment was carried out three times and the active fraction was determined to be

47.5 ± 6.6%.

205

Pol θ Full-Length Bypass Experiments for Abasic and Oxidized Abasic Lesions. Pol θ

(23.8 nM) was incubated with 28-32 (100 nM) at room temperature with all four dNTPs

(500 μM) in reaction buffer (20 mM Tris•HCl pH 8.0, 25 mM KCl, 10 mM MgCl2, 1 mM

BME). In a typical experiment, a 5  solution of primer-template complex (4 μL) was added to 10  reaction buffer (200 mM Tris•HCl, 250 mM KCl, 100 mM MgCl2, 10 mM

BME, 2 μL), H2O (10 μL) and 10  dNTP mix (5 mM, 2 μL). A 10  solution of Pol θ in storage buffer (238 nM, 2 L) was added and aliquots (4 μL) were removed at the indicated times, usually 1 min and 30 min, and quenched with a solution of 95% formamide with 25 mM EDTA (6 μL) containing trace bromophenol blue and xylene cyanol. Samples were heated (95 °C, 1 min) and then immediately placed on ice prior to loading (5 μL) on 20% denaturing PAGE run at 55 W for approximately 3 h followed by exposure to a phosphor storage cassette and phosphorimaging.

Pol θ Full-Length Bypass Experiments for Tg. Pol θ (20 nM) was incubated with 48-51

(50 nM) at room temperature with all four dNTPs (500 μM) in reaction buffer (20 mM

Tris•HCl pH 8.0, 25 mM KCl, 10 mM MgCl2, 1 mM BME). In a typical experiment, a 10

 solution of primer-template complex (2 μL) was added to 10  reaction buffer (200 mM

Tris•HCl, 250 mM KCl, 100 mM MgCl2, 10 mM BME, 2 μL), H2O (12 μL) and 10  dNTP mix (5 mM, 2 μL). A 10  solution of Pol θ in storage buffer (200 nM, 2 L) was added and aliquots (4 μL) were removed at 1, 10, 30 min, and quenched with a solution of 95% formamide with 25 mM EDTA (6 μL) containing trace bromophenol blue and xylene cyanol. Samples were heated (95 °C, 1 min) and then immediately placed on ice prior to loading (5 μL) on 20% denaturing PAGE run at 55 W for approximately 3 h followed by exposure to a phosphor storage cassette and phosphorimaging.

206

Steady-State Kinetic Analysis for Nucleotide Insertion Opposite Abasic Lesions.

Polymerase reactions were conducted with Pol  ( pM), primer-template complex (28-

3) (50 nM), and various concentrations of the indicated dNTP at room temperature in a reaction buffer (20 mM Tris•HCl pH 7.5, 25 mM KCl, 10 mM MgCl2, 1 mM BME). In a typical experiment, a solution of primer-template (500 nM, 30 µL) was mixed with 10  reaction buffer (200 mM Tris•HCl, 250 mM KCl, 100 mM MgCl2, 10 mM BME, 30 µL), and H2O (60 µL). A solution of Pol θ (4.75 nM, 30 µL) was added to bring the final volume of the 2  DNA-enzyme solution to 150 μL. The 2  DNA-enzyme solution (3 µL) was added to the appropriate 2  dNTP solution (3 µL). The reaction was conducted for a fixed time at room temperature and then quenched with 95% formamide loading buffer containing 25 mM EDTA (8 µL). Samples were heated (95 °C, 1 min) and then immediately placed on ice prior to loading (5 μL) on a 20% denaturing PAGE. The gel was run at 55 W for approximately 3 h followed by exposure to a phosphor storage cassette and phosphorimaging analysis. The reaction velocity was plotted as a function of dNTP concentration, and the data were fit to the Michaelis−Menten equation. The kcat was determined by dividing Vmax (in nM/min) by the active concentration of the enzyme (in nM) determined by active site titration. The dNTP concentration range and reaction time varied for each primer-template complex.

The following specific reaction conditions were used. For 28: dATP: 2.5, 5, 10 µM:

2 min; 15, 20, 25, 50 µM: 1 min. dGTP: 10, 20, 30 μM: 4 min; 50, 75 µM: 2 min; 100, 150,

200 µM: 1 min. For 29: dATP: 10, 20, 40 μM: 5 min; 60, 100 µM: 2 min; 200, 300, 500

µM: 1 min. dGTP: 150, 300, 400, 500 μM: 4 min; 750, 1000 µM: 2 min; 1500 µM: 1 min.

For 30: dATP: 5, 10, 20, 40 µM: 5 min; 60, 100, 150 µM: 2.5 min. dGTP: 20, 50, 75, 100,

207

150 µM: 5 min, 250, 500 µM: 3 min. For 31: dATP: 2.5, 5, 10 μM: 2 min; 20, 30, 40, 60,

100 μM: 1 min. dGTP: 10, 20 μM: 4 min; 30, 50 μM: 3 min, 100, 150, 200, 500 μM: 2 min. For 32: 0.5, 1, 2.5, 5, 10, 25, 50, 100 μM: 20s.

Steady-State Kinetic Analysis of Extension Past Abasic and Oxidized Abasic Lesions.

Polymerase extension reactions were conducted with Pol  (475 pM for reactions with native templates and  nM for reactions with lesion-containing templates), primer- template complexes 33-37 and 39-43 (50 nM for native templates and 25 nM for lesion- containing templates), and various concentrations of the indicated dNTP at room temperature in a reaction buffer (20 mM Tris•HCl pH 7.5, 25 mM KCl, 10 mM MgCl2, 1 mM BME). Experimental setup was the same as for translesion synthesis kinetics (above section) with the exception of the difference in Pol θ and primer-template concentrations.

The concentration range of dNTP and the reaction time varied for each lesion

The following specific reaction conditions were used. For 33a: dCTP: 0.25, 0.5, 1

μM: 1 min; 2.5, 5, 10, 25, 50 μM: 30 s. For 34a: dCTP: 25, 50, 75. 100 µM: 10 min; 200,

300 µM: 5 min; 500, 1000 µM: 3 min. dTTP: 50, 100, 150 µM: 6 min; 200, 300 µM: 4 min; 400, 500, 750 µM: 3 min. For 35a: dCTP: 50, 100, 150, 200, 250, 300, 400, 500 µM:

15 min. dTTP: 100, 200, 400, 500, 750, 1000, 1500, 2000 µM: 5 min. For 36a: dCTP: 25,

50, 75, 100 µM: 3 min; 200, 400, 700, 1000 µM: 1 min. dTTP: 25, 50, 75, 100, 200, 400,

700, 1000 µM: 5 min. For 37a: Identical to 4a. For 39a: dATP: 0.5, 1, 2.5, 5, 10, 25, 50,

100 μM: 20s. For 39b: dATP: 2.5, 5, 10 µM: 5 min; 25, 50 µM: 3 min; 100, 200, 300 µM:

1.5 min. dTTP: 2.5, 5, 10 µM: 5 min; 25, 50 µM: 2.5 min; 100, 200, 300 µM: 1 min. 39c: dATP: 5, 10, 20, 50, 100, 250, 500, 1000 µM. dTTP: 25, 50, 75, 100 µM: 5 min; 250, 500,

1000, 2000 µM; 1 min. For 40a: dATP: 0.5, 1, 2.5, 5, 10, 25, 50, 100 μM: 30s. For 40b:

208 dATP: 50, 100, 200, 300 µM: 10 min; 400, 500, 1000, 2000 µM: 5 min. dTTP: 25, 50 µM:

5 min; 75, 100, 250 µM: 2.5 min; 500, 1000, 2000 µM: 1 min. For 41a: dGTP: 0.5, 1, 2.5,

5, 10, 25, 50, 100 μM: 15s. 41b: dATP 5, 10 µM: 4 min; 25, 50 µM: 2 min; 100, 250, 500,

1000 µM: 45 s. For 41c: dATP: 10, 25, 50 µM: 4 min; 50, 100, 250 µM: 2 min; 500, 1000,

2000 µM: 1 min. For 43a: dATP: 0.5, 1, 2.5, 5, 10, 25, 50, 100 μM: 20s. 43b: dTTP: 10,

25, 50 µM: 5 min; 100, 200, 300, 500, 1000 µM: 2 min. For 43c: dTTP: 5, 10, 25 µM: 2 min; 50, 100 µM: 1 min; 500, 1000, 2000 µM: 45 s.

Steady-state kinetic analysis for Pol θ on primer-templates containing Tg and control

T Polymerase reactions were conducted with Pol θ (432 pM for 48−51 except for dGTP and dTTP insertion, for which a concentration of 4.32 nM was used; 4.32 nM for 52−55, except for dGTP insertion on 53, for which a concentration of 2 nM was used), primer−template complexes (50 nM), and various concentrations of the indicated dNTP at

25 °C in reaction buffer [10 mM Tris-HCl (pH 8), 25 mM KCl, 10 mM MgCl2, and 1 mM

BME]. The concentration range of dNTP and the reaction time were selected such that reactions did not proceed past 20% completion. In a typical experiment, a 2×

DNA−enzyme solution was prepared by mixing the primer−template complex (1 μM, 10

μL), 10× reaction buffer (20 μL), 10× Pol θ (4.32 nM, 20 μL) in storage buffer [20 mM

Tris-HCl (pH 7), 300 mM NaCl, 10% glycerol, and 5 mM BME], and H2O (50 μL). The

2× DNA−enzyme solution (3 μL) was mixed with the appropriate 2× dNTP solution (3 μL) to initiate the reaction, which was quenched after a fixed time with 95% formamide loading buffer containing 25 mM EDTA (8 μL). An aliquot (4 μL) was loaded on a 20% denaturing

PAGE gel, which was run at 55 W for approximately 3.5 h. The gel was analyzed by phosphorimaging. The reaction velocity was plotted as a function of dNTP concentration,

209 and the data were fit to the Michaelis−Menten equation. The kcat was determined by dividing Vmax (in nM) by the active concentration of the enzyme (in nM) determined by active site titration.

Specific reaction times and dNTP concentrations were as follows: For 48: [dATP]:

0.5, 1, 2.5, 5, 10, 25, 100, 500 µM: 20 s. [dTTP]: 50, 100, 250, 500, 1000, 2000, 3000,

5000 µM: 5 min. [dGTP]: same as dTTP. For 49: [dATP]: 2.5, 5 µM: 5 min; 10, 25, 50

µM: 2 min; 100, 500, 2000 µM: 30 s. [dTTP]: 50, 100, 250, 500, 1000, 2000, 3000, 5000

µM: 5 min. [dGTP]: same as dTTP. For 50: [dATP]: 0.5, 1, 2.5, 5, 10, 25, 100, 500 µM:

20 s. [dTTP]: 50, 100, 250, 500, 1000, 2000, 3000, 5000 µM: 5 min. [dGTP]: 25, 50, 100,

250 µM: 5 min; 500, 1000, 2000, 3000 µM: 1.5 min. For 51: same as 2. For 52: [dATP]:

0.5, 1, 2.5, 5, 10, 25, 100, 500 µM: 20 s. For 53: [dATP]: 10, 25, 50, 100, 500 µM: 5 min;

1000, 2000, 3000 µM: 2 min. [dTTP]: 25, 50 µM: 2 min; 100, 250, 500 µM: 1 min; 1000,

2000, 3000 µM: 30 s. For 54: [dTTP]: 25, 50, 100 µM, 5 min; 250, 500, 1000, 2000, 3000

µM: 20s. For 55: [dTTP]: 25, 50, 100, 250, 500, 1000, 2000, 3000 µM: 5 min. [dGTP]: 5,

10, 25 µM: 3 min; 50, 100 µM: 1 min; 500, 1000, 2000 µM: 30 s.

Sequencing of Pol θ Bypass Products of AP and Tg Pol θ (95 nM) was incubated with primer-template complex 46a or 46b (dU or AP, respectively), 56 or 57 (T or Tg, respectively) (400 nM) at room temperature with all four dNTPs (100 µM) in reaction buffer (20 mM Tris•HCl pH 8.0, 25 mM KCl, 10 mM MgCl2, 1 mM BME) at a reaction volume of 20 µL for 30 min. The solution was mixed with Dynabeads M-270 Streptavidin

(20 µg, 20 µL) and incubated at room temperature for 30 min with mixing by pipette every

10 min. Magnetic beads were concentrated using a magnetic particle concentrator. The supernatant was removed and the beads were washed five times with washing buffer (5

210 mM Tris•HCl pH 7.5, 0.5 mM EDTA, 1 M NaCl, 50 µL). The biotinylated strand was separated from the complementary, lesion-containing strand by incubation with NaOH (0.1

M, 50 µL) at room temperature for 5 min. The beads were washed once with NaOH (0.1

M, 50 µL), twice with Tris•EDTA buffer (10 mM Tris•HCl, 1 mM EDTA, pH 7.0, 50 µL) and twice with H2O (50 µL). The beads were resuspended in H2O (40 µL). An aliquot (5

µL) was amplified by PCR using Phusion polymerase according to the manufacturer’s protocol. This entailed 30 PCR cycles (98 °C for 10 s, 59 °C for 30 s, 72 °C for 30s) with

PCR primers 5’-d(AGA TGG AAT TCG TTC GAC C) and 5’-d(GTA GGT ACC GAT

TAA TCA CAG C). The PCR product was purified by 16% non-denaturing polyacrylamide gel electrophoresis. The gel was placed into a container with ethidium bromide (10 μg/mL, 100 mL) in water and stained for 20 min with gentle shaking. The stain was poured off and the gel was incubated in water for 10 min with gentle shaking.

The product band was visualized by UV (312 nm), excised from the gel, crushed manually, and eluted in buffer (100 mM NaCl, 1 mM EDTA, 1 mL) by shaking for 12 h at room temperature. Gel pieces were removed by Poly-Prep column and a Nucleotide Removal

Kit was used to remove ethidium bromide from the sample. The purified PCR product was digested with Acc65I (1 µL, 10 units) and EcoRI (1 µL, 20 units) in NEB Buffer 3.1 at 37

°C for 4 h. The reaction was purified using a Nucleotide Removal Kit. pBlueScript SK- plasmid (5 µg) was digested with Acc65I and EcoRI (identical conditions as for the PCR product) and purified by 1% agarose gel electrophoresis. The linearized plasmid was excised from the gel and purified using a Gel Extraction Kit. The digested PCR product

(75 fmol) was ligated into linearized pBlueScript SK- plasmid (25 fmol) using a Quick

Ligase Kit. This entailed mixing of the digested PCR product (2 μL, 37.5 nM) with

211 linearized plasmid (1 μL, 25 nM) and diluting with H2O (7 μL). The provided 2× Quick

Ligase Buffer (10 μL) was added followed by Quick Ligase (1 μL) and mixing of the solution by gently pipetting. The reaction was incubated at 25 °C for 5 min and an aliquot

(3 µL) of the ligation reaction was immediately transformed into DH5 α cells by incubation on ice (30 min), heat shock (42 °C, 30 s), incubation on ice (5 min) and addition of SOC medium (200 μL) followed by outgrowth at 37 °C for 1 h with 250 rpm shaking. The majority (200 μL) of the culture was plated on LB media containing ampicillin (100 µg/ mL). After growth at 37 °C for 16 h, the plates were sent to GeneWiz for colony sequencing using the Genewiz primer M13(-47). All commercially available kits used in this procedure were used according to the manufacturer’s protocol.

Pol θ Lyase Active Site Titration Pol θ (10, 20, 30, or 40 nM in protein concentration) was incubated with 5’-dRP (58, 100 nM) at 37 °C in a reaction buffer consisting of 50 mM

HEPES pH 7.5, 20 mM KCl, 1 mM EDTA, 1 mM β-mercaptoethanol. Reactions (30 μL) were prepared by adding H2O (21 μL), a 10× solution of 58 (3 μL), 10× reaction buffer (3

μL), and a 10× solution (3 μL) of Pol θ in storage buffer (20 mM Tris HCl pH 7, 300 mM

NaCl, 10% glycerol, 5 mM BME) to a microcentrifuge tube and incubating at 37 °C.

Aliquots (4 μL) were removed at the indicated time points (20, 30, 45, 60 min) and frozen on dry ice. Upon completion of the experiment, aliquots were quenched with a solution of

NaBH4 (1 μL, 500 mM) and incubated at room temperature for 1.5 h. An equal volume (5

μL) of formamide loading buffer (95% formamide, 10 mM EDTA, trace bromophenol blue and xylene cyanol) was added to each sample. Samples were heated (95 °C, 1 min) and then separated by 20% denaturing PAGE. The gel was exposed to a phosphor storage cassette and analyzed by phosphorimaging. Product formation (in nM) was plotted as a

212 function of time for each protein concentration and the data were fit to a straight line. The y-intercept of each line was then plotted as a function of the protein concentration. The slope of this line is the active fraction of protein (20.1%).

Measurement of Pol θ single turnover lyase kinetics DNA substrates (50 nM) were incubated with an excess of Pol θ (at least 250 nM) at 37 °C in a reaction buffer consisting of 50 mM HEPES pH 7.5, 20 mM KCl, 1 mM EDTA, 1 mM β-mercaptoethanol. Notably, as long as Pol θ was in sufficient excess, increasing the concentration of Pol θ did not change the rate constant. In a typical experiment, a 10× solution of DNA (500 nM) in 1× phosphate buffered saline (10 mM sodium phosphate, 100 mM NaCl, pH 7.2) was prepared. The abasic lesion was generated by photolysis (350 nm, 10 min) for 58-60 and

67-69, while UDG treatment (0.5 μL, 5 U, 37 °C, 15 min in 1 × phosphate buffered saline) was used for 71-78. The 10× DNA solution (3 μL) was added to a solution of H2O (21 μL),

10× reaction buffer (3 μL). A 10× solution of Pol θ in storage buffer (20 mM Tris HCl pH

7, 300 mM NaCl, 10% glycerol, 5 mM BME) was added and the reaction was incubated at

37 °C. Aliquots (4 μL) were removed at the indicates times and frozen on dry ice. At the end of the experiment, a 500 mM solution of NaBH4 was prepared in H2O and immediately added (1 μL) to each aliquot to quench the reaction. The reduction of the substrate was nearly instantaneous, but residual NaBH4 sometimes interfered with mobility during gel electrophoresis, so the reactions were incubated at room temperature for 1.5 h with occasional centrifugation on a bench-top centrifuge to allow for complete reaction of residual NaBH4. Alternatively, ethanol precipitation could be used to remove excess

NaBH4 without the necessity of a prolonged incubation, although this was typically unnecessary. Following quenching of the reactions, samples were mixed with an equal

213 volume (5 μL) of 95% formamide containing 10 mM EDTA and trace bromophenol blue and xylene cyanol and subjected to 20% denaturing polyacrylamide gel electrophoresis at

55W. For 58-60 and 67-69, electrophoresis was conducted for approximately 4 h For 75-

78 it was conducted for 45 min. The gel was exposed to a phosphor storage cassette and analyzed by phosphorimaging. The fraction of product was plotted against time and fit to

-kobs*x the equation y =ymax*(1-e ).

Aliquots were removed at the following time points. For 5’-dRP (58): 1, 2, 5, 10,

15, 20 min for the wild type; 2, 5, 10, 20, and 30 min for K2383R mutant; 2, 5, 10, 15, and

20 min for K2575A/K2577A and 5, 10, 15, 30, and 60 min for K2383A and

K2383A/K2575A/K2577A. For DOB (59): 45 s, 90 s, 2 min, 5 min, 7.5 min, and 10 min for the wild type; 1, 2, 5, 10, 20 min for K2383R; 2, 5, 10, 20, 30 for K2383A. For pC4-

AP (60): 20 s, 40s, 1 min, 2 min, 5 min, and 10 min for the wild type; 0.5, 1, 2, 5, 10 min for K2383R; 1, 2, 5, 10, 15 min for K2383A. For 67: 2.5, 5, 7.5, 10, 20, 30 min; 68: 1, 2,

3, 4, 6, 10 min; 69: 0.5, 1, 2, 4, 6, 10 min. For 71-74: 10, 20, 30, 45, 60 min. For 75a and

78a: 2, 5, 10, 20, 30, 60 min; 75b-78b: 1, 2.5, 5, 10, 20, 30 min.

Measurement of Pol β single turnover lyase kinetics Pol β single turnover kinetics were conducted in the same fashion as Pol θ single turnover kinetics. Reaction times were as follows: For 67: 5, 10, 20, 30, 60, 300 s; 68: 1, 2, 4, 6, 10, 20 min; 69: 10, 20, 40, 60, 120,

300 s. 75a, 76a, 78a: 1, 2.5, 5, 10, 20, 30 min; 77a: 0.5, 1, 2.5, 5, 10, 30 min. For 75b and

78b: 2.5, 5, 10, 20, 30, 60 min; 76b: 1, 2.5, 5, 10, 20, 30 min; 77b: 0.5, 1, 2.5, 5, 10, 30 min.

Analysis of Pol θ lyase activity under multiple turnover conditions. Ternary complexes

3'-32P-58-60 or 67-69 (100 nM) were incubated with Pol θ (2.5 nM) at 37 °C in a reaction 214 buffer consisting of 50 mM HEPES pH 7.5, 20 mM KCl, 1 mM EDTA, 1 mM β- mercaptoethanol. In a typical experiment, a 10 × solution of DNA (500 nM) in 1 × phosphate buffered saline (10 mM sodium phosphate 100 mM NaCl, pH 7.2) was prepared and photolyzed (350 nm, 10 min) to generate the appropriate abasic site (dRP, DOB, or pC4-AP). The 10 × DNA solution (3 μL) was added to a solution of H2O (21 μL) and 10 × reaction buffer (3 μL). A 10 × solution of Pol θ (3 μL) in storage buffer (20 mM Tris HCl pH 7, 300 mM NaCl, 10% glycerol, 5 mM BME) was added and the reaction was incubated at 37 °C. To account for the background reaction in the absence of enzyme, a control sample was treated in the exact same fashion except instead of adding Pol θ to the reaction, the same volume of Pol θ storage buffer (3 μL) was added. Aliquots (4 μL) were removed from the reaction at 2.5, 5, 7.5, 10, 20, and 30 min for 60 and 69 or 10, 20, 30, 45, and 60 min for 58, 59, 67, and 68 and frozen on dry ice. At the end of the experiment, a 500 mM solution of NaBH4 was prepared in H2O and immediately added (1 μL) to each aliquot to quench the reaction. Samples were incubated at room temperature for 1.5 h and analyzed by 20% denaturing PAGE followed by phosphorimaging as described for measurement of

Pol θ single turnover lyase kinetics. Reactions were conducted in triplicate for each experiment and each experiment was conducted at least twice.

Stepwise Inhibition of Pol θ Lyase Activity by pC4-AP Pol θ (5 nM) was incubated with pC4-AP (60 or 69, 100 nM) in reaction buffer buffer (50 mM HEPES pH 7.5, 20 mM KCl,

2 mM DTT, 1 mM EDTA) at 25 °C. A 10 × solution of 60 or 69 was prepared by photolysis

(10 min, 350 nm) and 7 μL of this solution was added to a microcentrifuge tube containing

H2O (49 μL) and 10 × reaction buffer (7 μL). The reaction was initiated by the addition of a 10 × solution of Pol θ in buffer C (7 μL, 350 fmol) bringing the volume of the reaction

215 to 70 μL. A control reaction was setup to measure the background reaction without enzyme and an equal volume of buffer C was added to this reaction whenever Pol θ was added to enzymatic reactions. Aliquots (4 μL) were removed from the reactions at 5, 10, 15, and 20 min and frozen on dry ice. Immediately after removing the aliquot at 20 min, 1 eq of Pol θ in buffer C (350 fmol, 1 μL) was added to the reaction. Aliquots were removed at 25, 30,

35, and 40 min and frozen on dry ice. Another aliquot of Pol θ in buffer C (350 fmol, 1 μL) was added immediately after removing the aliquot at 40 min. Aliquots were removed from the reaction at 45, 50, 55, and 60 min and frozen on dry ice. At the end of the experiment, all aliquots were quenched with a solution of NaBH4 (1 μL, 500 mM) and incubated at room temperature for 1.5 h followed by 20% denaturing PAGE and analysis by phosphorimaging.

Steady-state polymerase kinetics for Pol θ K2383A and K2383R mutants Polymerase reactions were conducted with Pol θ K2383A and K2383R variants (5 nM), primer- template complex 48 (50 nM), and a range of dATP concentrations (62.5, 125, 250, 500,

1000, 2000, 3000, 5000 μM) at 25 °C in reaction buffer (10 mM Tris HCl pH 8, 25 mM

KCl, 10 mM MgCl2, 1 mM BME) for a fixed time (1.5 min for K2383R and 4 min for

K2383A). The concentration range of dATP and the reaction time were selected such that reactions did not proceed past 20% completion (single-hit conditions). In a typical experiment, a 2× DNA-polymerase solution was prepared by mixing primer-template (1

µM, 9 μL), 10 × reaction buffer (18 μL), 10 × Pol θ (4.32 nM or 50 nM, 18 μL) in storage buffer (20 mM Tris HCl pH 7, 300 mM NaCl, 10% glycerol, 5 mM BME), and H2O (45

μL). The 2× DNA-enzyme solution (3 μL) was mixed with the appropriate 2× dNTP solution (3 μL) to initiate the reaction, which was quenched after a fixed time with 95%

216 formamide loading buffer containing 25 mM EDTA (6 μL). An aliquot (4 µL) was analyzed by 20% denaturing PAGE run at 55 W for approximately 3.5 h. The gel was analyzed by phosphorimaging, and the data were fit to the Michaelis-Menten equation. The kcat was determined by dividing Vmax (in units of nM/min) by the active concentration of enzyme (in nM). Active site titration of the wild type enzyme was reported previously.

Mutant proteins could not be subject to active site titration due to the extremely slow rate of reaction, so the active fraction was assumed to be equal for mutant and wild type proteins.

Fluorescence anisotropy measurements. Anisotropy measurements were conducted using a solution of dichloro diphenyl fluorescein-labeled DNA (66, 70, 75c-78c, 83) (250 pM) and Pol θ (varying concentrations) in reaction buffer (50 mM HEPES pH 7.5, 20 mM

KCl, 1 mM EDTA, and 1 mM β-mercaptoethanol). Samples also contained 10 % Pol θ storage buffer (20 mM Tris HCl pH 7, 300 mM NaCl, 10% glycerol, 5 mM BME) by volume. In a typical experiment, a sample (300 μL) was prepared by mixing Pol θ (30 μL,

2 μM) in storage buffer with 10 × reaction buffer (30 μL), DNA substrate (30 μL, 2.5 nM), and H2O (210 μL). The concentration of Pol θ in this solution, termed solution A, is 200 nM. Samples containing varying concentrations of Pol θ were prepared by serial dilution with solution B. Solution B (10 mL) was prepared by mixing H2O (7.95 mL) with 10 × reaction buffer (1 mL), Pol θ storage buffer (1 mL), and DNA (50 nM, 50 μL). By mixing solution A (150 μL) with solution B (150 μL), the concentration of Pol θ was decreased to

100 nM, while the concentration of DNA, reaction buffer, and storage buffer remained unchanged. An aliquot (150 μL) of this new solution was then mixed with solution B (150

μL) to prepare a new solution containing 50 nM Pol θ. Serial dilutions were repeated in

217 this fashion to vary the concentration of Pol θ (listed below). Samples were incubated at

25 °C for 1 h and fluorescence anisotropy (A) was measured using a portion (125 μL) of each sample with PMT voltage of 800 mV, 8 nm slit width, and 535 nm excitation and 556 nm emission. Fluorescence anisotropy was measured in the absence of polymerase (A0), and the change in anisotropy (A-A0) was calculated for each sample and plotted against the

n n concentration of Pol θ. The data were fit to the Hill equation A=Amax([enzyme] /Kd +n) where n is the Hill coefficient using Origin 7.0.

For 66, serial dilutions were repeated such that samples contained Pol θ concentrations of 200, 100, 50, 25, 12.5, 6.25, 3.13, 1.57 nM for wild type Pol θ; 300, 150,

75, 37.5, 18.75, 9.38, 4.69, 2.34, 1.17 nM for K2383A and K2383R; 100, 50, 25, 12.5,

6.25, 3.13, 1.57 nM for K2575A/K2577A; and 200, 100, 75, 50, 37.5, 25, 12.5, 6.25 nM for K2575A/K2577A/K2383A. For 70, and 75c-78c, and 83 the Pol θ concentration was

100, 50, 25, 12.5, 6.25, 3.13, 1.56, 0.78, 0.39 nM.

The experiment was conducted in the same fashion for Pol β, for which the concentrations were 100, 50, 25, 12.5, 6.25, 3.13, 1.56, 0.78, 0.39 nM for all substrates.

UPLC-MS/MS analysis of Pol θ modification by pC4-AP. A solution of Pol θ (100 μL,

10 μM) was mixed with H2O (700 μL), 10× reaction buffer (100 μL, 500 mM HEPES pH

7.5, 200 mM KCl, 10 mM EDTA), and 60 (100 μL, 100 μM) and incubated at room temperature for 30 min. The reaction mixture was concentrated by centrifugation using an

Amicon 10K centrifugal filter. To prevent loss of protein, the centrifugal filter was blocked with Pol θ prior to addition of the sample. Blocking was conducted by adding Pol θ (500

μL, 1 μM,) followed by centrifugation (13,000 g, 5 min, 4 °C) and removal of the supernatant. Following blocking of the membrane filter, half of the sample (500 μL) was 218 added to the Amicon centrifugal filter, and centrifugation was carried out (13,000 g, 5 min,

4 °C). The remainder of the sample was added and centrifugation was repeated. The sample was then washed twice with 400 μL of reaction buffer (50 mM HEPES pH 7.5, 20 mM

KCl, 1 mM EDTA) and concentrated by centrifugation in the Amicon centrifugal filter to

100 μL. Trypsin/Lys-C mix was reconstituted in the resuspension buffer provided by the manufacturer (20 μL) and added (2 μg, 2 μL) to the sample, which was incubated at 37 °C for 4 h. The digestion mixture was concentrated in a Speed Vac concentrator to 40 μL and a portion (10 μL) was analyzed by UPLC-MS/MS using an ACQUITY UPLC HSS T3

Column (100 Å, 1.8 µm, 2.1 mm × 100 mm). The flow rate was 0.3 mL/min running a gradient from 85:5:10 water: acetonitrile:1% formic acid to 50:40:10 water: acetonitrile:1% formic acid over 35 min. Analysis was conducted using BioPharmaLynx with tolerance set at 20 ppm and allowing for 3 missed cleavages.

UPLC-MS/MS analysis of Pol β modification by pC4-AP. A solution of Pol β (20 μL,

8.4 μM) was mixed with H2O (150 μL), 10× reaction buffer (20 μL, 500 mM HEPES pH

7.5, 200 mM KCl, 10 mM EDTA), and 60 (10 μL, 100 μM) and incubated at room temperature for 30 min. The reaction mixture was concentrated by Speed vac rotary concentrator to 60 μL. Digestion buffer (5 μL, 500 mM Tris HCl pH 8.0) was added.

Trypsin/Lys-C mix was reconstituted in the resuspension buffer provided by the manufacturer (20 μL) and added (1.5 μg, 1.5 μL) to the sample, which was incubated at 37

°C for 5 h. A portion (10 μL) of the digestion mixture was analyzed by UPLC-MS/MS using an ACQUITY UPLC HSS T3 Column (100 Å, 1.8 µm, 2.1 mm × 100 mm). The flow rate was 0.3 mL/min running a gradient from 85:5:10 water: acetonitrile:1% formic acid to

219

50:40:10 water: acetonitrile:1% formic acid over 35 min. Analysis was conducted using

BioPharmaLynx with tolerance set at 20 ppm and allowing for 3 missed cleavages.

Kinetic analysis of Ku lyase activity Reactions were conducted with Ku (10 nM unless otherwise noted), DNA substrates 84-86 or 88-90 (2 nM), streptavidin (500 nM) and 1×

Ku reaction buffer (25 mM sodium phosphate buffer pH 7.4, 100 μM EDTA, 150 mM KCl,

1 mM DTT). In a typical experiment, a solution of DNA, Ku reaction buffer, and streptavidin was prepared such that the concentration of each was 1.25× compared to the concentration of that component in the reaction. This was accomplished by adding a 10 nM solution of DNA (25 μL), a 5 μM solution of streptavidin (12.5 μL), 10× Ku reaction buffer

(12.5 μL) and H2O (50 μL) to a clear Eppendorf tube. This solution was either photolyzed

(350 nm, 10 min) or incubated with UDG (1 unit at 37°C for 15 min). The 1.25× DNA solution (24 μL) was then added to an Eppendorf tube containing H2O (3 μL) and Ku (3

μL of 100 nM solution) to achieve a final volume of 30 μL and the final concentrations of each component listed above. Reactions were incubated at 37°C and aliquots (4 μL) were removed at a fixed time and immediately frozen on dry ice. At the end of the time course experiment, aliquots were treated with 200 mM NaBH4, centrifuged briefly, vortexed, and centrifuged briefly again before 1.5 h incubation at 4°C. Following incubation, calf thymus

DNA (1 μg, 1 μL) was added to each sample along with NaOAc (2 μL, 3 M, pH 5.2) and

H2O to 20 μL. Three volumes (60 μL) of cold ethanol were added and samples were incubated on dry ice for 20 min, spun at 16,000 rcf for 20 min, and the supernatant was removed. Both the samples and the supernatant were counted by a liquid scintillation counter to ensure complete or nearly complete recovery. Residual ethanol was removed in a Speed Vac concentrator and samples were dissolved in 5 μL loading buffer (90%

220 formamide, 10 mM EDTA) with dye, heated for 1 minute at 95° C, and separated by 12% denaturing PAGE at constant power of 55W (approximately 3 hours). The gel was exposed to a phosphor storage cassette and imaged by phosphorimaging. The fraction of product

-kobs*x was plotted against time and fit to the equation y =ymax*(1-e ).

Reaction times were as follows: For 84: 5, 10, 20, 30, 45, 60 min; For 85 and 88: 1, 2, 5,

10, 20, 30 min; For 86: 10, 30, 60 min; For 89: 0.25, 0.5, 0.75, 1, 2, 5 min; For 90: 5, 10,

15, 20, 30, 45, 60 min.

Attempted detection of lactam modification on Ku and Pol θ using fluorescent probe

Ku or Pol θ was incubated with C4-AP-containg DNA to form the putative lactam modification in a 250 μL reaction. For Ku reactions, 25 pmol of Ku (25 μL, 1 μM) and 25 pmol of 85 (5 μL, 5 μM) were used. For Pol θ, 100 pmol of Pol θ (16.66 μL, 6.6 μM) and

500 pmol of 60 (10 μL, 50 μM) were used. The remainder of the reaction solution was composed of 10× reaction buffer (25 μL) and water (to 250 μL). The final concentrations were as follows: 100 nM Ku, 50 nM 85, 25 mM sodium phosphate buffer pH 7.4, 100 μM

EDTA, 150 mM KCl; 400 nM Pol θ, 2 μM 60, 50 mM HEPES pH 7.5, 20 mM KCl, 1 mM

EDTA. Reactions were incubated at 37 °C for 1 h. Fluorescent probe 62 (synthesized as reported335,357) was added (50 μL, 15 mM) and the solution was concentrated to 20 µL using a Speed vac concentrator. Ar was bubbled through the solution, and it was incubated overnight at 37 °C. The solution was diluted with H2O to 100 μL and pipetted into a dialysis cap and dialyzed against 500 mL of 1× phosphate buffered saline (10 mM sodium phosphate, 100 mM NaCl, pH 7.2), 30 mM KCl, 5 mM BME, 100 μM PMSF for 8 h. The buffer was changed to a two-fold dilution of the previous dialysis buffer and dialyzed overnight. The buffer was changed to H2O with 100 μM PMSF and dialyzed an additional

221

12 hours. The sample was removed from the dialysis button by puncturing the membrane with a pipette tip. The solution was frozen, lyophilized, and re-suspended in H2O (15 μL), to which 4× SDS loading buffer (5 μL) was added. Samples were loaded on an SDS-PAGE

(4% stacking, 12% resolving) and run for 2 h at 200 V. The gel was scanned using a

Typhoon phosphorimager (488 nM ex 522 em). SYPRO Ruby was used for total protein staining according to the manufacturer’s instructions and then imaged with a Typhoon phosphorimager using the SYPRO Ruby setting.

Attempted detection of lactam modification on Ku by UPLC-MS

A solution of Ku (100 μL, 8 μM) was mixed with H2O (700 μL), 10 × reaction buffer (100 μL, 250 mM sodium phosphate buffer pH 7.4, 1 mM EDTA, 1.5 M KCl), and

91 or 92 (100 μL, 4 μM) and incubated at 37 °C for 1 h. An aliquot (10 μL) was removed from the reaction and quenched by addition of NaBH4 (1 μL, 1 M). The reaction mixture was concentrated by centrifugation using an Amicon 10K centrifugal filter. The sample was then washed twice with 400 μL of reaction buffer (25 mM Sodium phosphate buffer pH 7.4, 100 μM EDTA, 150 mM KCl) and concentrated by centrifugation in the Amicon centrifugal filter to 100 μL. Trypsin/Lys-C mix was reconstituted in the resuspension buffer provided by the manufacturer (20 μL) and added (2 μg, 2 μL) to the sample, which was incubated at 37 °C for 3 h. A solution (10 μL) of 10 × chymotrypsin buffer (1 M Tris HCl,

100 mM CaCl2 pH 8.0) was added. Chymotrypsin (25 ug) was reconstituted in 1 mM HCl

(25 μL) and added (2 μg, 2 μL) and the reaction was incubated at 37 °C for 4 h. The digestion mixture was concentrated in a Speed Vac concentrator to 40 μL and a portion (10

μL) was analyzed by UPLC-MS/MS using an ACQUITY UPLC HSS T3 Column (100 Å,

1.8 µm, 2.1 mm × 100 mm). The flow rate was 0.3 mL/min running a gradient from 85:5:10

222 water: acetonitrile:1% formic acid to 50:40:10 water: acetonitrile:1% formic acid over 35 min. Analysis was conducted using BioPharmaLynx with tolerance set at 20 ppm and allowing for 3 missed cleavages.

The aliquot which was removed from the reaction was incubated at room temperature (1 h) to allow complete reaction of NaBH4. Sodium acetate (3 M, 2 μL) was added along with H2O (8 μL) to a final volume of 20 μL. Cold ethanol (60 μL) was added and the sample was incubated on dry ice (30 min) followed by centrifugation (16,000 g, 30 min, 4 °C). The supernatant was removed, and the sample was dried by a Speed vac concentrator. The DNA was resuspended in H2O (10 μL) and heated to 95 °C (5 min) to denature the duplex and Ku. The DNA was 3’-end labeled with cordycepin and terminal transferase followed by Sephadex G-25 to remove unincorporated cordycepin. A portion

(5 μL) of the 3’-labeled DNA was visualized by 12% denaturing PAGE and phosphorimaging analysis.

Attempted detection of reduced DPC between Pol θ and DOB by UPLC-MS

Pol θ (3 μM active protein, 40 μL) in buffer C was added to a microcentrifuge tube containing H2O (260 μL), DOB (59, 20 μL, 100 μM), and 10 × reaction buffer (40 μL, 500 mM HEPES pH 7.5, 200 mM KCl, 10 mM EDTA). A solution of NaCNBH3 (500 mM) was prepared and added (40 μL) to the reaction. A control reaction which was identical except for the omission of 59 was also prepared. The reactions were incubated at 37 °C

(1.5 hr), frozen on liquid nitrogen, and concentrated to ~30 μL by vacuum centrifugation.

The samples contained 13.3% glycerol from the Pol θ storage buffer, so instead of SDS- loading buffer, a solution (8 μL) of SDS (10%) DTT (100 mM), and bromophenol blue

(trace) was added and the samples were heated at 95 C for 5° min. Samples were loaded

223 on 10% SDS-PAGE with 5% stacking and run at 175 V for 2.5 hr. Protein was visualized by staining with Coommassie blue stain (0.025% solution in 5:4:1 H2O:MeOH:acetic acid) on a rotary shaker for 1 hr at room temperature and then destaining for 6 hr in destaining buffer (5:4:1 H2O:MeOH:acetic acid) on a rotary shaker. The gel was washed with H2O (3 times, 10 min wash) on a rotary shaker, and proteins were excised from the gel with a razor blade and transferred to a 1.6 mL microcentrifuge tube. Gel pieces were crushed and a solution of NH4HCO3 (100 μL, 25 mM) was added, followed by addition of a solution of

DTT (10 μL, 300 mM) and incubation at 50 °C (20 min) to reduce disulfides. Cysteines were alkylated by addition of saturated iodoacetamide (10 μL) and incubation at room temperature in the dark for 20 min. Samples were briefly spun in a benchtop centrifuge and the solution was discarded. Gel pieces were washed three times by addition of 50% MeCN in 25 mM NH4HCO3 (150 μL), incubation for 10 min at room temperature, and then discarding of the liquid. Gel pieces were dehydrated by addition of acetonitrile (100 μL), incubation for 10 min at room temperature, and discarding of the liquid. Gel pieces were resuspended (10 mM Tris-HCl, 15 mM MgCl2, pH 7.0, 200 μL). Phosphodiesterase I and

II (250 mU each), alkaline phosphatase (6 U), and Dnase I (200 mU) were added and the samples were incubated at 37 °C for 24 hr. Samples were dried in a Speed vac concentrator.

Modified trypsin was reconstituted in 25 mM NH4HCO3 at a concentration of 50 ng/μL and 100 μL of this solution was added to each sample. Samples were incubated at 37 °C for 20 hr. The solution was transferred to a clean microcentrifuge tube for each sample.

Peptides were extracted by addition of 0.1% formic acid in 60% MeCN (100 μL) followed by incubation at room temperature for 15 min. The solution was transferred to a new tube.

Peptide extraction was repeated in the same fashion and the solution was combined with

224 the previous peptide solution. Samples were concentrated to dryness by vacuum centrifugation and resuspended in 0.1% formic acid (50 μL). Digests were analyzed by

UPLC-MS (10 μL injection).

225

6. Appendix

a) b)

c) d)

Figure S1. Sample kinetic plots for TLS by Pol θ. a-c) insertion opposite AP (28) and d, e) insertion opposite L (29). a) dA insertion b) dG insertion c) T insertion d) dA insertion e) dG insertion. Data are the average ± standard deviation of three replicates. 226

a) b)

c) d)

Figure S2. Sample kinetic plots for a-c) insertion opposite C4-AP (30). d, e) insertion opposite F (31). f) insertion opposite dU (32). Plots a, d, f) dA insertion; Plots b, e) dG insertion; Plots c) T insertion. Data are the average ± standard deviation of three replicates. 227

a) b)

c) d)

e) f)

Figure S3. Sample kinetic plots for a, b) extension past dA:AP (34a). c, d) extension past dA:C4-AP (35a). e, f) extension past dA:L (36a) Plots. a, c, e: dC insertion; Plots b, d, f: T insertion. Data are the average ± standard deviation of three replicates. 228

a) b)

c) d)

e) f)

Figure S4. Sample kinetic plots for a, b) extension past dA:F (37a). c, d) extension past dA:AP (39b). e, f) extension past dA:C4-AP (39c). a: dC insertion. b, d, f: T insertion. c, e: dA insertion. Data are the average ± standard deviation of three replicates. 229

a) b)

c) d)

e) f)

Figure S5. Sample kinetic plots for a, b) extension past dA:AP (40b). c) extension past dA:AP (41b). d) extension past dA:C4-AP (41c). e) extension past dA:AP (43b). f) extension past dA:C4-AP (43c) a, c d: dA insertion. b, e, f: T insertion. Data are the average ± standard deviation of three replicates. 230

a) b)

c) d)

Figure S6. Sample kinetic plots for a) dC insertion past dA:dU (33a). b) dA insertion past dA:dU (39a). c) dA insertion past dA:dU (40a). d) dG insertion past dA:dU (41a). e) dA insertion past dA:dU (43a). Data are the average ± standard deviation of three replicates. 231

a) b)

c) d)

e) f)

Figure S7. Sample kinetic plots for insertion opposite Tg a-c) 49; d-f) 51. a, d) dA insertion. b, e) dG insertion c, f) T insertion. Data are the average ± standard deviation of three replicates.

232

a) b)

c) d)

e) f)

Figure S8. Sample kinetic plots for insertion opposite T a-c) 48; d-f) 50. a, d) dA insertion. b, e) dG insertion c, f) T insertion. Data are the average ± standard deviation of three replicates.

233

a) b)

c) d)

e) f)

Figure S9. Sample kinetic plots for extension past dA:Tg or dA:T. a, b) extension past dA:Tg (52) c, d) extension past dA:Tg (54). e) extension past dA:T (48) f) extension past dA:T (50). a, c, f) T insertion b, e) dA insertion. d) dG insertion. Data are the average ± standard deviation of three replicates.

234

a) b)

c) d)

Figure S10. Sample kinetic plots for excision of dRP (3'-32P-58) by Pol θ variants. a) wild type b) K2383A c) K2383R d) K2575A/K2577A e) K2383A/K2575A/K2577A. Data are the average ± standard deviation of three replicates.

235

a) b)

c) d)

e) f) f)

Figure S11. Sample kinetic plots for excision of a-c) DOB (3'-32P-59) or d-f) pC4-AP (3'- 32 P-60) by Pol θ Lys2383 variants. a) wild type b) K2383A c) K2383R d) wild type e) K2383A f) K2383R. Data are the average ± standard deviation of three replicates.

236

a) b)

c) d)

Figure S11. Plots for fluorescence anisotropy experiments with Pol θ variants on 66. a) wild type b) K2383A c) K2383R d) K2575A/K2577A e) K2383A/K2575A/K2577A.

237

a) b)

Figure S12. Sample kinetic plots for dA insertion opposite T (48) by Pol θ variants. a) wild type b) K2383A c) K2383R.

238

a) b)

c) d)

e) f)

Figure S13. Sample kinetic plots for Pol θ (a-c) and Pol β (d-f) on abasic lesions at a DSB. a, d) 5’-dRP (67) b, e) DOB (68) c, f) pC4-AP (69).

239

a) b)

c) d)

e) f)

Figure S14. Sample kinetic plots for Pol θ with clustered lesions. a) AP -4 (78a) b) AP -1 (75a) c) C4-AP -4 (78b) d) C4-AP +5 (77b) e) C4-AP -1 (75b) f) C4-AP +1 (76b)

240

a) b)

c) d)

Figure S15. Sample kinetic plots for Pol β with clustered lesions containing AP. a) AP -4 (78a) b) AP +5 (77a) c) AP -1 (75a) d) AP +1 (76a).

241

a) b)

d) c)

Figure S16. Sample kinetic plots for Pol β with clustered lesions containing AP. a) C4-AP -4 (78b) b) C4-AP +5 (77b) c) C4-AP -1 (75b) d) C4-AP +1 (76b).

242

a) b)

c) d)

e) f)

Figure S17. Plots for fluorescence anisotropy experiments with Pol β. a) ternary complex 66 b) DSB 70 c) cluster -4 78c d) cluster +5 77c e) cluster -1 75c f) cluster +1 76c.

243

a) b)

c) d)

e) f)

Figure S18. Plots for fluorescence anisotropy experiments with Pol θ. a) ternary complex (66) b) DSB (70) c) cluster -4 (78c) d) cluster +5 (77c) e) cluster -1 (75c) f) cluster +1 (76c). 244

Figure S19. A) Plots for fluorescence anisotropy experiments measuring Pol θ binding to 83. B) Sample kinetic plot for Pol θ K2383R lyase activity on 75a and c) 78a.

245

5’-d(CGA CCG GCT CGT ATG UTG TGT GGA GCT GTG G) Calculated mass: 9616.2

Observed mass: 9616.0

Figure S20. ESI-MS of template used for 32 and 33. Used to prepare 28 and 34.

246

5’-d(CGA CCG GCT CGT ATG XTG TGT GGA GCT GTG G) Calculated mass: 9667.1 Observed mass: 9667.0

Figure S21. ESI-MS of precursor to template used in 29, 36.

247

5’-d(CGA CCG GCT CGT ATG FTG TGT GGA GCT GTG G) Calculated mass: 9506.1 Observed mass 9507.0

Figure S22. ESI-MS of precursor to template used in 31, 37.

248

5’-d(CGA CCG GCT CGT AAG UTG TGT GGA GCT GTG G) Calculated mass: 9625.3 Observed mass: 9627.0

Figure S23. ESI-MS of precursor to template used in 38.

249

5’-d(CGA CCG GCT CGT ATT UTG TGT GGA GCT GTG G) Calculated mass: 9591.2 Observed mass: 9591.0

Figure S24. ESI-MS of precursor to template used in 39a, b and 75a-78a. m/z= 9630 corresponds to M+K+.

250

5’-d(CGA CCG GCT CGT ATG XTG TGT GGA GCT GTG G) Calculated mass: 9903.6 Observed mass: 9904.0

Figure S25. ESI-MS of precursor to template used in 39c.

251

5’-d(CGA CCG GCT CGT AAT UTG TGT GGA GCT GTG G) Calculated mass: 9600.2 Observed mass: 9601.0

Figure S26. ESI-MS of precursor to template used for 40a, b.

252

5’-d(CGA CCG GCT CGT ATC UTG TGT GGA GCT GTG G) Calculated mass:9576.2 Observed mass:9576.0

Figure S27. ESI-MS of precursor to template used for 41b.

253

5’-d(CGA CCG GCT CGT ATC XTG TGT GGA GCT GTG G) Calculated mass: 9888.6 Observed mass: 9888.0

Figure S28. ESI-MS of precursor to template used for 41c.

254

5’-d(CGA CCG GCT CGT ACT UTG TGT GGA GCT GTG G) Calculated mass: 9576.2 Observed mass: 9576.0

Figure S29. ESI-MS of precursor to template used for 43b.

255

5’-d(CGA CCG GCT CGT ACT XTG TGT GGA GCT GTG G) Calculated mass: 9888.6 Observed mass: 9889.0

Figure S30. ESI-MS of precursor to template used in 43c.

256

5’-d(TAT GXT GAC CTA CAC GCT AGA AC) Calculated mass: 7087.6 Observed mass: 7094.0

Figure S31. ESI-MS of AP precursor oligonucleotide used to prepare 44b.

257

5’-d(TAT GUT GAC CTA CAC GCT AGA AC) Calculated mass: 6993.6 Observed mass: 6994.0

Figure S32. ESI-MS of oligonucleotide used to prepare 45a, b.

258

5’-d(BGT AGG TAC CAG TTA ATC ACA GCT CCA CAC A) B: Biotin-TEG Calculated mass: 9713.6 Observed mass: 9714.0

Figure S33. ESI-MS of primer used in 46.

259

5’-d(AGA TGG AAT TCG TTC GAC CGG CTC GTA TTU TGT GTG GAG CTG TGG TTA) Mass calculated: 14892.7 Mass observed: 14895.0

Figure S34. ESI-MS of template used in 46.

260

5’-d(AGG CCT CAC ATG CAT TgTA GGA CCT GCA GCT) Mass (calculated): 9201.0 Mass (observed): 9202.6

1200 Intens. [a.u.] 9202.579

1000

800

600

400

200

0 6000 7000 8000 9000 10000 11000 m/z

Figure S35. MALDI-TOF MS of Tg-containing template used for 49, 53.

261

5’-d(AGG CCT CAC ATG CTA TgTA GGA CCT GCA GCT) Mass (calculated): 9201.0 Mass (observed): 9200.6

1200 Intens. [a.u.] 9200.596

1000

800

600

400

200

6000 6500 7000 7500 8000 8500 9000 9500 10000 10500 11000 m/z Figure S36. MALDI-TOF MS of Tg-containing template used for 50, 54.

262

5’-d(AGA TGG AAT TCG TTC GAC CGG CTC GTA TTTg TGT GTG GAG CTG TGG TTA) Mass (calculated): 14940.7 Mass (observed): 14942.7

2000 Intens. [a.u.] 14942.693

1500

1000

500

0 10000 11000 12000 13000 14000 15000 16000 17000 18000 19000 m/z

Figure S37. MALDI-TOF MS of Tg-containing template used for 57

263

5’-d(XC CGT AAT GCA GTC T) Mass (calc): 4811.2 Mass observed: 4810.0

Figure S38. ESI-MS of DOB precursor oligonucleotide used for 59. m/z = 4832 corresponds to M + Na.

264

5’-d(PXC CGT AAT GCA GTC T) P: phosphate Mass (calc): 4920.3 Mass observed: 4921.0

Figure S39. ESI-MS of pC4-AP precursor oligonucleotide used for 60.

265

5’-d(PFC CGT AAT GCA GTC T) P: phosphate Mass (calc): 4498.9 Mass observed: 4499.0

Figure S40. ESI-MS of F-containing oligonucleotide used for 66.

266

5’- d(XAG ACT GCA TTA CGG AAA GCG TTA GCC ATT A) Mass (calc): 9998.6 Mass observed: 9996.1

9996.124 Intens. [a.u.] 800

600 X:

400

200

0 7000 8000 9000 10000 11000 12000 13000 14000 15000 m/z

Figure S41. MALDI TOF-MS of dichloro diphenyl fluorescein-containing oligonucleotide used for 66.

267

5'-d(PXG GAA ATC AAA TGT AAG TAG AGG TCA) P: phosphate Calculated mass: 8253.438 Observed mass: 8253.554

S42. LC-MS spectrum for 5’-dRP precursor used to prepare 67 and 90.

268

5'-d(XG GAA ATC AAA TGT AAG TAG AGG TCA) Calculated mass: 8358.6 Observed mass: 8361.0

S43. ESI-MS spectrum for DOB precursor used to prepare 68 and 88.

269

5'-d(PXG GAA ATC AAA TGT AAG TAG AGG TCA) P: phosphate Calculated mass: 8467.6 Observed mass: 8470.0

S44. ESI-MS spectrum for pC4-AP precursor used to prepare 69 and 89.

270

5’-d(XTT GAC CTC TAC TTA CAT TTG ATT TC) Calculated mass: 8307.5 Observed mass: 8307.4

6000 8307.412 Intens. [a.u.]

5000

4000 X:

3000

2000

1000

0 5000 6000 7000 8000 9000 10000 11000 12000 13000 14000 15000 m/z

S45. MALDI-MS for dichloro diphenyl fluorescein template used to prepare 70.

271

5'-d(XG GAA ATC AAA TGT AAG TAG AGG TCA) Calculated mass: 8047.2 Observed mass: 8044.05

x104 8044.050 Intens. [a.u.] 1.25

1.00

0.75

0.50

0.25

0.00 5000 6000 7000 8000 9000 10000 11000 12000 13000 14000 15000 m/z

S46. MALDI-MS for F-containing oligonucleotide used to prepare 70.

272

5’-d(PGC TGG CCG AGC ATA AG) P=phosphate Calculated mass: 5010.3 Observed mass: 5011.0

S47. ESI MS of oligonucleotide used to prepare 75

273

5’-d(PGC TGG CCG AG) P=phosphate Calculated mass: 3148.0 Observed mass: 3150.0

S48. ESI MS of oligonucleotide used to prepare 76

274

5’-d(PGC TGG CCG AGC ATA) P=phosphate Calculated mass: 4367.8 Observed mass: 4370.0

S49. ESI MS of oligonucleotide used to prepare 77

275

5’-d(PGC TGG CCG AGC ATA AGA CA) P=phosphate Calculated mass: 5925.9 Observed mass 5928.0

S50. ESI MS of oligonucleotide used to prepare 78

276

5'-d(Y-CGA CCG GCT CGT ATT XTG TGT GGA GCT GTG G) Calculated mass: 10240.5 Observed mass: 10240.4

10240.441 Intens. [a.u.] 8000

Y: 6000

4000

2000

0 6000 7000 8000 9000 10000 11000 12000 13000 14000 15000 m/z

S51. MALDI-MS for template containing F and dichloro diphenyl fluorescein used to prepare 75c-78c.

277

5’-d(PGU GGA AAT CAA ACG TAA GTA GAA TCC AAAGTC TCT TTC TTC CG) P=phosphate Calculated mass: 13274.6 Observed mass: 13278.0

S52. ESI-MS for dU precursor used to prepare 84.

278

5’-d(PGX GGA AAT CAA ACG TAA GTA GAA TCC AAAGTC TCT TTC TTC CG) P=phosphate Calculated mass: 13585.9 Observed mass: 13591.0

S53. ESI-MS for C4-AP precursor oligonucleotide used to prepare 85.

279

5’-d(PGX GGA AAT CAA ACG TAA GTA GAA TCC AAAGTC TCT TTC TTC CG) P=phosphate Calculated mass: 13320.2 Observed mass: 13318.9

S54. LC-MS for L precursor oligonucleotide used to prepare 86.

280

5’-d(PTX CTT TCT TAG AGT CTT TAG TTT ATT GGG CGC G) P=phosphate Calculated mass: 10511.9 Observed mass: 10510.0

S55. LC-MS for C4-AP precursor oligonucleotide used to prepare 91.

281

X CTT TCT TAG AGT CTT TAG TTT ATT GGG CGC G Calculated mass: 10430.9 Observed mass: 10430.0

S56. ESI-MS for DOB precursor oligonucleotide used to prepare 92.

282

S57. Purification of SUMO- and hexaHis-tagged Pol θ by HisTrap column. The blue line represents UV absorbance, the green line the concentration of elution buffer. The UV absorbance is high (1300 mAU) during injection of the lysate and it decreases as the column is washed with buffer A. The first large peak is observed at 50 mM imidazole and corresponds to impuritites. The broad peak that is outlined with a square is Pol θ.

283

S58. Purification of Pol θ by Heparin column. The blue line represents UV absorbance, the green line the concentration of elution buffer, and the brown line is conductivity. The dashed vertical line represents the point at which Pol θ from the HisTrap column is injected. The UV absorbance is high during injection and decreases rapidly as the column is washed with buffer C. Top chromatogram:. When a sharp elution gradient was used (0-100 buffer D over 40 min), a large, sharp peak eluted at approximately 50-60% buffer D. This contained a large amount of Pol θ and some impurities. Bottom chromatogram: When a linear gradient of 0-100% buffer D over 100 min was used, Pol θ (eluting in fractions B9- B7) eluted as a smaller, broader peak but was separated from more impurities.

284

S58. Purification of untagged Pol θ by HisTrap column following cleavage of SUMO tag (which contains the hexaHis tag also). The blue line represents UV absorbance, the green line the concentration of elution buffer. The dashed vertical line represents the point at which Pol is injected. The UV trace has some rapid spikes due to air bubbles. It can be difficult to avoid air bubbles when injecting a small volume onto the FPLC. The largest peak contains pure Pol θ.

285

S59. Elution of Ku from Talon column using a linear gradient of elution buffer. The blue line represents UV absorbance, the green line the concentration of elution buffer. The sharp peak at ~735 mL contained impurities which eluted during wash with wash buffer. The peak at ~750 mL contained Ku. The absorbance is negative once the buffer is changed from lysis buffer to wash buffer.

286

S60. Purification of Ku by Heparin column. The blue line represents UV absorbance, the green line the concentration of elution buffer. The first large peak is the flowthrough from the column, which contains impurities. The second large peak observed contains Ku. This eluted at 50% TEDGP buffer containing 1M NaCl.

287

S61. Purification of Ku by size exclusion column. The first large peak is Ku and it eluted after approximately 7 mL. Later peaks are low-molecular weight impurities.

288

S62. Analysis of C4-AP excision from 91 by Ku. In lanes 1-3, the top strand of 91 was 32P- labeled and loaded without photolysis (lane 1), or with photolysis (lanes 2 and 3) followed by either NaBH4 (lane 2) or NaOH (lane 3). In lanes 4 and 5, aliquots from a control reaction containing 91 (without Ku) were removed after incubation at 37 °C for 60 min (lane 4) or 0 min (lane 5) and 32P-labeled. In lane 6, an aliquot was removed from the enzymatic reaction of Ku with 91 after 60 min and 32P-labeled.

289

7. References

(1) Lindahl, T.; Barnes, D. E. Repair of Endogenous DNA Damage. Cold Spring Harb.

Symp. Quant. Biol. 2000, 65, 127–133.

(2) J.H., H. DNA Damage, Aging and Cancer. N. Engl. J. Med. 2009, 361, 1475–1485.

(3) Merrick, C. J.; Jackson, D.; Diffley, J. F. X. Visualization of Altered Replication

Dynamics after DNA Damage in Human Cells. J. Biol. Chem. 2004, 279, 20067–

20075.

(4) Ciccia, A.; Elledge, S. J. The DNA Damage Response: Making It Safe to Play with

Knives. Mol. Cell 2010, 40, 179–204.

(5) Wirtz, S.; Nagel, G.; Eshkind, L.; Neurath, M. F.; Samson, L. D.; Kaina, B. Both

Base Excision Repair and O6-Methylguanine-DNA Methyltransferase Protect

against Methylation-Induced Colon Carcinogenesis. Carcinogenesis 2010, 31,

2111–2117.

(6) Park, J.; Chen, L.; Tockman, M. S.; Elahi, A.; Lazarus, P. The Human 8-Oxoguanine

DNA N-Glycosylase 1 (hOGG1) DNA Repair Enzyme and Its Association with

Lung Cancer Risk. Pharmacogenetics 2004, 14, 103–109.

(7) Koboldt, D. C.; Fulton, R. S.; McLellan, M. D.; Schmidt, H.; Kalicki-Veizer, J.;

McMichael, J. F.; Fulton, L. L.; Dooling, D. J.; Ding, L.; Mardis, E. R.; et al.

Comprehensive Molecular Portraits of Human Breast Tumours. Nature 2012, 490,

61–70.

(8) Edge, L. Endogenous DNA Damage as a Source of Genomic Instability in Cancer.

290

Cell 2017, 168, 644–656.

(9) Povirk, L. F. DNA Damage and Mutagenesis by Radiomimetic DNA-Cleaving

Agents: Bleomycin, Neocarzinostatin and Other Enediynes. Mutat. Res. - Fundam.

Mol. Mech. Mutagen. 1996, 355, 71–89.

(10) Kozarich, J. W.; Stubbe, J. Mechanisms of Bleomycin-Induced DNA Degradation.

Chem. Rev. 1987, 87, 1107–1136.

(11) Sung, J. S.; Demple, B. Roles of Base Excision Repair Subpathways in Correcting

Oxidized Abasic Sites in DNA. FEBS J. 2006, 273, 1620–1629.

(12) Greenberg, M. M. Abasic and Oxidized Abasic Site Reactivity in DNA: Enzyme

Inhibition, Cross-Linking, and Nucleosome Catalyzed Reactions. Accounts Chem.

Res. 2014, 47, 646–655.

(13) Jacobs, A. C.; Kreller, C. R.; Greenberg, M. M. Long Patch Base Excision Repair

Compensates for DNA Polymerase Beta Inactivation by the C4-Oxidized Abasic

Site. Biochemistry 2011, 50, 136–143.

(14) Guan, L.; Greenberg, M. M. Irreversible Inhibition of DNA Polymerase Beta by an

Oxidized Abasic Lesion. J. Am. Chem. Soc. 2010, 132, 5004–5005.

(15) Stevens, A. J.; Guan, L.; Bebenek, K.; Kunkel, T. A.; Greenberg, M. M. DNA

Polymerase Lambda Inactivation by Oxidized Abasic Sites. Biochemistry 2013, 52,

975–983.

(16) Roberts, S. A; Strande, N.; Burkhalter, M. D.; Strom, C.; Havener, J. M.; Hasty, P.;

Ramsden, D. A.. Ku Is a 5’-dRP/AP Lyase That Excises Nucleotide Damage near

291

Broken Ends. Nature 2010, 464, 1214–1217.

(17) Prasad, R.; Longley, M. J.; Sharief, F. S.; Hou, E. W.; Copeland, W. C.; Wilson, S.

H. Human DNA Polymerase Theta Possesses 5’-dRP Lyase Activity and Functions

in Single-Nucleotide Base Excision Repair in Vitro. Nucleic Acids Res. 2009, 37,

1868–1877.

(18) Yousefzadeh, M. J.; Wood, R. D. DNA Polymerase POLQ and Cellular Defense

against DNA Damage. DNA Repair (Amst). 2013, 12, 1–9.

(19) Rouquette, I.; Lepage, B.; Oumouhou, N.; Walschaerts, M.; Leconte, E.; Schilling,

V.; Gordien, K.; Brouchet, L. DNA Replication Stress Response Involving PLK1 ,

CDC6 , POLQ , RAD51 and CLASPIN Upregulation Prognoses the Outcome of

Early / Mid-Stage Non-Small Cell Lung Cancer Patients. Oncogenesis 2012, 1, 1–

10.

(20) Yousefzadeh, M. J.; Wyatt, D. W.; Takata, K. I.; Mu, Y.; Hensley, S. C.; Tomida,

J.; Bylund, G. O.; Doublié, S.; Johansson, E.; Ramsden, D. A.; McBride, K. M.;

Wood, R. D. Mechanism of Suppression of Chromosomal Instability by DNA

Polymerase POLQ. PLoS Genet. 2014, 10, e1004654.

(21) Datta, K.; Neumann, R. D.; Winters, T. a. Characterization of Complex

Apurinic/apyrimidinic-Site Clustering Associated with an Authentic Site-Specific

Radiation-Induced DNA Double-Strand Break. Proc. Natl. Acad. Sci. U. S. A. 2005,

102, 10569–10574.

(22) Greenberg, M. M. Looking beneath the Surface to Determine What Makes DNA

Damage Deleterious. Curr. Opin. Chem. Biol. 2014, 21, 48–55.

292

(23) Lindahl, T.; Nyberg, B. Rate of Depurination of Native Deoxyribonucleic Acid.

Biochemistry 1972, 11, 3610–3618.

(24) Fu, D.; Calvo, J. A.; Samson, L. D. Balancing Repair and Tolerance of DNA

Damage Caused by Alkylating Agents. Nat. Rev. Cancer 2012, 12, 104–120.

(25) Wilde, J. A.; Bolton, P. H.; Mazumder, A.; Manoharan, M.; Gerlt, J. A.

Characterization of the Equilibrating Forms of the Aldehydic Abasic Site in Duplex

DNA by17O NMR. J. Am. Chem. Soc. 1989, 111, 1894–1896.

(26) Sczepanski, J. T.; Wong, R. S.; McKnight, J. N.; Bowman, G. D.; Greenberg, M. M.

Rapid DNA-Protein Cross-Linking and Strand Scission by an Abasic Site in a

Nucleosome Core Particle. Proc. Natl. Acad. Sci. 2010, 107, 22475–22480.

(27) Sabouri, N.; Johansson, E. Translesion Synthesis of Abasic Sites by Yeast DNA

Polymerase ε. J. Biol. Chem. 2009, 284, 31555–31563.

(28) Haracska, L.; Unk, I.; Johnson, R. E.; Johansson, E.; Burgers, P. M.; Prakash, S.;

Prakash, L. Roles of Yeast DNA Polymerases Delta and Zeta and of Rev1 in the

Bypass of Abasic Sites. Genes Dev. 2001, 15, 945–954.

(29) Loeb, L. A.; Preston, B. D. Mutagenesis By Apurinic/Apyrimidic Sites. Annu. Rev.

Genet. 1986, 20, 201–230.

(30) Kubo, K.; Ide, H.; Wallace, S. S.; Kow, Y. W. A Novel, Sensitive, and Specific

Assay for Abasic Sites, the Most Commonly Produced DNA Lesion. Biochemistry

1992, 31, 3703–3708.

(31) Wei, S.; Shalhout, S.; Ahn, Y.; Bhagwat, A. S. A Versatile New Tool to Quantify

293

Abasic Sites in DNA and Inhibit Base Excision Repair. DNA Repair (Amst). 2015,

27, 9–18.

(32) Mendez, F.; Goldman, J. D.; Bases, R. E. Abasic Sites in DNA of HeLa Cells

Induced by Lucanthone Abasic Sites in DNA of HeLa Cells Induced by Lucanthone.

Cancer Invest. 2002, 20, 983–991.

(33) Nakamura, J.; Swenberg, J. A. Endogenous Apurinic/Apyrimidinic Sites in

Genomic DNA of Mammalian Tissues. Cancer Res. 1999, 59, 2522–2526.

(34) Rahimo, R.; Kosmatchev, O.; Kirchner, A.; Pfa, T.; Spada, F.; Brantl, V.; Mu, M.;

Carell, T. 5 ‑Formyl- and 5 ‑Carboxydeoxycytidines Do Not Cause Accumulation

of Harmful Repair Intermediates in Stem Cells. J. Am. Chem. Soc. 2017, 139,

10359–10364.

(35) Murphy, M. P. How Mitochondria Produce Reactive Oxygen Species. Biochem. J.

2009, 417, 1–13.

(36) Tainer, J. A.; Getzoff, E. D.; Richardson, J. S.; Richardson, D. C. Structure and

Mechanism of Copper, Zinc Superoxide Dismutase. Nature 1983, 206, 284–287.

(37) Winterbourn, C. C. Toxicity of Iron and Hydrogen Peroxide: The Fenton Reaction.

Toxicol. Lett. 1995, 82–83, 969–974.

(38) LaVerne, J. A. OH Radicals and Oxidizing Products in the Gamma Radiolysis of

Water. Radiat. Res. 2000, 153, 196–200.

(39) von Sonntag, C. Free-Radical-Induced DNA Damage and Its Repair. Springer:

Berlin, 2006.

294

(40) Cooke, M. S.; Evans, M. D.; Dizdaroglu, M.; Lunec, J. Oxidative DNA Damage :

Mechanisms , Mutation , and Disease. FASEB J. 2003, 17, 1195–1214.

(41) Adelman, R.; Saul, R. L.; Ames, B. N. Oxidative Damage to DNA: Relation to

Species Metabolic Rate and Life Span. Proc. Natl. Acad. Sci. U. S. A. 1988, 85,

2706–2708.

(42) Cathcart, R.; Schwiers, E.; Saul, R. L.; Ames, B. N. Thymine Glycol and Thymidine

Glycol in Human and Rat Urine: A Possible Assay for Oxidative DNA Damage.

Proc. Natl. Acad. Sci. U. S. A. 1984, 81, 5633–5637.

(43) Breimer, L. H.; Lindahl, T. Thymine Lesions Produced by Ionizing Radiation in

Double-Stranded DNA. Biochemistry 1985, 24, 4018–4022.

(44) Teebor, G.; Cummings, A.; Frenkel, K.; Shaw, A.; Voituriez, L.; Cadet, J.

Quantitative Measurement of the Diastereoisomers of Cis Thymidine Glycol in

Gamma-Irradiated DNA. Free Radic. Res. Commun. 1987, 2, 303–309.

(45) Brown, K. L.; Adams, T.; Jasti, V. P.; Basu, A. K.; Stone, M. P. Interconversion of

the Cis -5 R , 6 S - and Trans -5 R , 6 R -Thymine Glycol Lesions in Duplex DNA.

J. Am. Chem. Soc. 2008, 130, 11701–11710.

(46) Jovanovic, S. V.; Simic, M. G. Mechanism of OH Radical Reactions with Thymine

and Uracil Derivatives. J. Am. Chem. Soc. 1986, 108, 5968–5972.

(47) Su, X.; Huang, Q.; Dang, B.; Wang, X.; Yu, Z. Spectroscopic Assessment of Argon

Gas Discharge Induced Radiolysis of Aqueous Adenine and Thymine. Radiat. Phys.

Chem. 2011, 80, 1343–1351.

295

(48) Cadet, J.; Wagner, J. R.; Shafirovich, V.; Geacintov, N. E. One-Electron Oxidation

Reactions of Purine and Pyrimidine Bases in Cellular DNA. Int J Radiat Biol 2014,

90, 423–432.

(49) Balasubramanian, B.; Pogozelski, W. K.; Tullius, T. D. DNA Strand Breaking by

the Hydroxyl Radical Is Governed by the Accessible Surface Areas of the Hydrogen

Atoms of the DNA Backbone. Proc. Natl. Acad. Sci. 1998, 95, 9738–9743.

(50) Chen, J.; Stubbe, J. Bleomycins: Towards Better Therapeutics. Nat. Rev. Cancer

2005, 5, 102–112.

(51) York, N.; Biol, N. P. J. M.; Sausville, E. A.; Peisach, J.; Horwitz, S. B. Effect of

Chelating Agents and Metal Ions on the Degradation of DNA by Bleomycin.

Biochemistry 1978, 17, 2740–2746.

(52) Burger, R. M.; Peisach, J. Activated Bleomycin: A Transient Complex of Drug, Iron,

and Oxygen That Degrades DNA. J. Biol. Chem. 1981, 256, 11636–11644.

(53) Burger, R. M.; Peisachg, J.; Horwitz, S. B. Effects of O2 on the Reactions of

Activated Bleomycin. J. Biol. Chem. 1982, 257, 3372–3375.

(54) Rabow, L. E.; Stubbe, J.; Kozarich, J. W. Identification and Quantitation of the

Lesion Accompanying Base Release in Bleomycin-Mediated DNA Degradation. J.

Am. Chem. Soc. 1990, 112, 3196–3203.

(55) Andrea, A. D. D.; Haseltine, W. A. Sequence Specific Cleavage of DNA by the

Antitumor Antibiotics Neocarzinostatin and Bleomycin. Proc. Natl. Acad. Sci.

1978, 75, 3608–3612.

296

(56) Chen, B.; Zhou, X.; Taghizadeh, K.; Chen, J.; Stubbe, J.; Dedon, P. C. GC/MS

Methods To Quantify the 2-Deoxypentos-4-Ulose and 3 ′-Phosphoglycolate

Pathways of 4′ Oxidation of 2-Deoxyribose in DNA: Application to DNA Damage

Produced by γ Radiation and Bleomycin. Chem. Res. Toxicol. 2007, 20, 1701–1708.

(57) Regulus, P.; Duroux, B.; Bayle, P.; Favier, A.; Cadet, J.; Ravanat, J. Oxidation of

the Sugar Moiety of DNA by Ionizing Radiation or Bleomycin Could Induce the

Formation of a Cluster DNA Lesion. Proc. Natl. Acad. Sci. 2007, 104, 14032–

14037.

(58) Chan, W.; Chen, B.; Wang, L.; Taghizadeh, K.; Demott, M. S.; Dedon, P. C.

Quantification of the 2-Deoxyribonolactone and Nucleoside 5 ′-Aldehyde Products

of 2-Deoxyribose Oxidation in DNA and Cells by Isotope-Dilution Gas

Chromatography Mass Spectrometry: Differential Effects of γ-Radiation and. J. Am.

Chem. Soc. 2010, 132, 6145–6153.

(59) Myers, A. G. Proposed Structure of the Neocarzinostatin Chromophore-Methyl

Thioglycolate Adduct; A Mechanism for the Nucleophilic Activation of

Neocarzinostatin. Tetrahedron Lett. 1987, 28, 4493–4496.

(60) Goldberg, I. H. Mechanism of Neocarzinostatin Action: Role of DNA

Microstructure in Determination of Chemistry of Bistranded Oxidative Damage.

Acc. Chem. Res. 1991, 24, 191–198.

(61) Kappen, L. S.; Goldberg, I. H.; Wu, S. H.; Stubbe, J. A.; Worth, L.; Kozarich, J. W.

Isotope Effects on the Sequence-Specific Cleavage of dC in d(AGC) Sequences by

Neocarzinostatin: Elucidation of Chemistry of Minor Lesions. J. Am. Chem. Soc.

297

1990, 112, 2797–2798.

(62) Emanuel, C. J.; Newcomb, M.; Ferreri, C.; Chatgilialoglu, C. Kinetics of 2’-

deoxyuridin-1’-yl Radical Reactions. J. Am. Chem. Soc. 1999, 121, 2927–2928.

(63) Chen, B. Z.; Bohnert, T.; Zhou, X. F.; Dedon, P. C. 5 ’-(2-Phosphoryl-1,4-

Dioxobutane) as a Product of 5 ’-Oxidation of Deoxyribose in DNA: Elimination as

Trans-1,4-Dioxo-2-Butene and Approaches to Analysis. Chem. Res. Toxicol. 2004,

17, 1406–1413.

(64) Kawabata, H.; Takeshita, H.; Fujiwara, T.; Sugiyama, H.; Matsuura, T.; Saito, I.

Chemistry of Neocarzinostation-Mediated Degradation of d(GCATGC).

Mechanism of Spontaneous Thymine Release. Tetrahedron Lett. 1989, 30, 4263–

4266.

(65) Kinner, A.; Wu, W.; Staudt, C.; Iliakis, G. Gamma-H2AX in Recognition and

Signaling of DNA Double-Strand Breaks in the Context of Chromatin. Nucleic

Acids Res. 2008, 36, 5678–5694.

(66) Vilenchik, M. M.; Knudson, A. G. Endogenous DNA Double-Strand Breaks:

Production, Fidelity of Repair, and Induction of Cancer. Proc. Natl. Acad. Sci. U. S.

A. 2003, 100, 12871–12876.

(67) Lieber, M. R. The Mechanism of Double-Strand DNA Break Repair by the

Nonhomologous DNA End-Joining Pathway. Annu. Rev. Biochem. 2010, 79, 181–

211.

(68) Shrivastav, M.; De Haro, L. P.; Nickoloff, J. a. Regulation of DNA Double-Strand

298

Break Repair Pathway Choice. Cell Res. 2008, 18, 134–147.

(69) Ward, J. F.; Blakely, W. F.; Joner, E. I. Mammalian Cells Are Not Killed by DNA

Single-Strand Breaks Caused by Hydroxyl Radicals from Hydrogen Peroxide.

Radiat. Res. 1985, 103, 383–392.

(70) Ward, J. F. The Complexity of DNA Damage: Relevance to Biological

Consequences. Int. J. Radiat. Biol. 1994, 66, 427–432.

(71) Pastwa, E.; Neumann, R. D.; Mezhevaya, K.; Winters, T. A. Repair of Radiation-

Induced DNA Double-Strand Breaks Is Dependent upon Radiation Quality and the

Structural Complexity of Double-Strand Breaks. Radiat. Res. 2003, 159, 251–261.

(72) Radford, I. R. DNA Lesion Complexity and Induction of Apoptosis by Ionizing

Radiation. Int. J. Radiat. Biol. 2002, 78, 457–466.

(73) David-Cordonnier, M. H.; Boiteux, S.; O’Neill, P. Excision of 8-Oxoguanine within

Clustered Damage by the Yeast OGG1 Protein. Nucleic Acids Res. 2001, 29, 1107–

1113.

(74) Yang, N.; Chaudhry, M. A.; Wallace, S. S. Base Excision Repair by hNTH1 and

hOGG1: A Two Edged Sword in the Processing of DNA Damage in γ-Irradiated

Human Cells. DNA Repair (Amst). 2006, 5, 43–51.

(75) Chaudhry, M. A.; Weinfeld, M. Reactivity of Human Apurinic/apyrimidinic

Endonuclease and Escherichia Coli Exonuclease III with Bistranded Abasic Sites in

DNA. J. Biol. Chem. 1997, 272, 15650–15655.

(76) Paap, B.; Wilson, D. M.; Sutherland, B. M. Human Abasic Endonuclease Action on

299

Multilesion Abasic Clusters: Implications for Radiation-Induced Biological

Damage. Nucleic Acids Res. 2008, 36, 2717–2727.

(77) Absalon, M. J.; Wu, W.; Kozarich, J. W.; Stubbe, J. Sequence-Specific Double-

Strand Cleavage of DNA by Fe-Bleomycin. 2. Mechanism and Dynamics.

Biochemistry 1995, 34, 2076–2086.

(78) Dedon, P. C.; Goldberg, I. H. Free-Radical Mechanisms Involved in the Formation

of Sequence-Dependent Bistranded DNA Lesions by the Antitumor Antibiotics

Bleomycin, Neocarzinostatin, and Calicheamicin. Chem. Res. Toxicol. 1992, 5,

311–332.

(79) Shao, R.-G. Pharmacology and Therapeutic Applications of Enediyne Antitumor

Antibiotics. Curr. Mol. Pharmacol. 2008, 1, 50–60.

(80) Wang, R.; Li, L.; Zhang, S.; Li, Y.; Wang, X.; Miao, Q.; Zhen, Y. A Novel

Enediyne-Integrated Antibody-Drug Conjugate Shows Promising Antitumor

Efficacy against CD30 + Lymphomas. Mol. Oncol. 2018, 12, 339–355.

(81) Sievers, E. L.; Appelbaum, F. R.; Spielberger, R. T.; Forman, S. J.; Flowers, D.;

Smith, F. O.; Shannon-Dorcy, K.; Berger, M. S.; Bernstein, I. D. Selective Ablation

of Acute Myeloid Leukemia Using Antibody-Targeted Chemotherapy: A Phase I

Study of an Anti-CD33 Calicheamicin Immunoconjugate. Blood 1999, 93, 3678–

3684.

(82) Jeggo, P. a; Löbrich, M. DNA Double-Strand Breaks: Their Cellular and Clinical

Impact? Oncogene 2007, 26, 7717–7719.

300

(83) Vermeulen, K.; Van Bockstaele, D. R.; Berneman, Z. N. The Cell Cycle: A Review

of Regulation,deregulation and Therapeutic Targets in Cancer. Cell Prolif. 2003, 36,

131–149.

(84) Rothkamm, K.; Krüger, I.; Thompson, L. H.; Kru, I.; Lo, M. Pathways of DNA

Double-Strand Break Repair during the Mammalian Cell Cycle. Mol. Cell. Biol.

2003, 23, 5706–5715.

(85) Branzei, D.; Foiani, M. Regulation of DNA Repair throughout the Cell Cycle. Nat.

Rev. Mol. Cell Biol. 2008, 9, 297–308.

(86) Truong, L. N.; Li, Y.; Shi, L. Z.; Hwang, P. Y.-H.; He, J.; Wang, H.; Razavian, N.;

Berns, M. W.; Wu, X. Microhomology-Mediated End Joining and Homologous

Recombination Share the Initial End Resection Step to Repair DNA Double-Strand

Breaks in Mammalian Cells. Proc. Natl. Acad. Sci. U. S. A. 2013, 110, 7720–7725.

(87) McVey, M.; Lee, S. E. MMEJ Repair of Double-Strand Breaks (Director’s Cut):

Deleted Sequences and Alternative Endings. Trends Genet. 2008, 24, 529–538.

(88) Ceccaldi, R.; Liu, J. C.; Amunugama, R.; Hajdu, I.; Primack, B.; Petalcorin, M. I.

R.; O’Connor, K. W.; Konstantinopoulos, P. A; Elledge, S. J.; Boulton, S. J.;

Yusufzai, T.; D'Andrea, A. D. Homologous-Recombination-Deficient Tumours Are

Dependent on Polθ-Mediated Repair. Nature 2015, 518, 258–262.

(89) Walker, J. R.; Corpina, R. a; Goldberg, J. Structure of the Ku Heterodimer Bound

to DNA and Its Implications for Double-Strand Break Repair. Nature 2001, 412,

607–614.

301

(90) Dynan, W. S.; Yoo, S. Interaction of Ku Protein and DNA-Dependent Protein

Kinase Catalytic Subunit with Nucleic Acids. Nucleic Acids Res. 1998, 26, 1551–

1559.

(91) Mari, P.-O.; Florea, B. I.; Persengiev, S. P.; Verkaik, N. S.; Brüggenwirth, H. T.;

Modesti, M.; Giglia-Mari, G.; Bezstarosti, K.; Demmers, J. A A; Luider, T. M.;

Houtsmuller, A. D.; van Gent, D. C. Dynamic Assembly of End-Joining Complexes

Requires Interaction between Ku70/80 and XRCC4. Proc. Natl. Acad. Sci. U. S. A.

2006, 103, 18597–18602.

(92) Britton, S.; Coates, J.; Jackson, S. P. A New Method for High-Resolution Imaging

of Ku Foci to Decipher Mechanisms of DNA Double-Strand Break Repair. J. Cell

Biol. 2013, 202, 579–595.

(93) Gottlieb, T. M.; Jackson, S. P. The DNA-Dependent Protein Kinase: Requirement

for DNA Ends and Association with Ku Antigen. Cell 1993, 72, 131–142.

(94) Uematsu, N.; Weterings, E.; Yano, K. I.; Morotomi-Yano, K.; Jakob, B.; Taucher-

Scholz, G.; Mari, P. O.; Van Gent, D. C.; Chen, B. P. C.; Chen, D. J.

Autophosphorylation of DNA-PKCS Regulates Its Dynamics at DNA Double-

Strand Breaks. J. Cell Biol. 2007, 177, 219–229.

(95) Jiang, W.; Crowe, J. L.; Liu, X.; Nakajima, S.; Wang, Y.; Li, C.; Lee, B. J.

Differential Phosphorylation of DNA-PKcs Regulates the Interplay between End-

Processing and End-Ligation during Nonhomologous End-Joining. Mol. Cell 2015,

58, 172–185.

(96) Ma, Y.; Schwarz, K.; Lieber, M. R. The Artemis:DNA-PKcs Endonuclease Cleaves

302

DNA Loops, Flaps, and Gaps. DNA Repair (Amst). 2005, 4, 845–851.

(97) Rasouli-Nia, A.; Karimi-Busheri, F.; Weinfeld, M. Stable down-Regulation of

Human Polynucleotide Kinase Enhances Spontaneous Mutation Frequency and

Sensitizes Cells to Genotoxic Agents. Proc. Natl. Acad. Sci. U. S. A. 2004, 101,

6905–6910.

(98) Nick McElhinny, S. A.; Ramsden, D. A.. Sibling Rivalry: Competition between Pol

X Family Members in V(D)J Recombination and General Double Strand Break

Repair. Immunol. Rev. 2004, 200, 156–164.

(99) Sibanda, B. L.; Critchlow, S. E.; Begun, J.; Pei, X. Y.; Jackson, S. P.; Blundell, T.

L.; Pellegrini, L. Crystal Structure of an Xrcc4-DNA Ligase IV Complex. Nat.

Struct. Biol. 2001, 8, 1015–1019.

(100) Graham, T. G. W.; Walter, J. C.; Loparo, J. J. Two-Stage Synapsis of DNA Ends

during Non-Homologous End Joining. Mol. Cell 2016, 61, 850–858.

(101) Yano, K. I.; Chen, D. J. Live Cell Imaging of XLF and XRCC4 Reveals a Novel

View of Protein Assembly in the Non-Homologous End-Joining Pathway. Cell

Cycle 2008, 7, 1321–1325.

(102) Reynolds, P.; Anderson, J. A.; Harper, J. V.; Hill, M. a.; Botchway, S. W.; Parker,

A. W.; O’Neill, P. The Dynamics of Ku70/80 and DNA-PKcs at DSBs Induced by

Ionizing Radiation Is Dependent on the Complexity of Damage. Nucleic Acids Res.

2012, 40, 10821–10831.

(103) Li, Y.; Reynolds, P.; O’Neill, P.; Cucinotta, F. A. Modeling Damage Complexity-

303

Dependent Non-Homologous End-Joining Repair Pathway. PLoS One 2014, 9.

e85816

(104) Bétermier, M.; Bertrand, P.; Lopez, B. S. Is Non-Homologous End-Joining Really

an Inherently Error-Prone Process? PLoS Genet. 2014, 10, e1004086

(105) Rodgers, K.; Mcvey, M. Error-Prone Repair of DNA Double-Strand Breaks. J. Cell.

Physiol. 2016, 231, 15–24.

(106) Schimmel, J.; Kool, H.; van Schendel, R.; Tijsterman, M. Mutational Signatures of

Non‐homologous and Polymerase Theta‐mediated End‐joining in Embryonic Stem

Cells. EMBO J. 2017, e201796948.

(107) Difilippantonio, M. J.; Zhu, J.; Chen, H. T.; Meffre, E.; Nussenzweig, M. C.; Max,

E. E.; Ried, T.; Nussenzweig, A. DNA Repair Protein Ku80 Suppresses

Chromosomal Aberrations and Malignant Transformation. Nature 2000, 404, 510–

514.

(108) Brandsma, I.; Gent, D. C. Pathway Choice in DNA Double Strand Break Repair:

Observations of a Balancing Act. Genome Integr. 2012, 3, 9.

(109) Mao, Z.; Bozzella, M.; Seluanov, A.; Gorbunova, V. Comparison of

Nonhomologous End Joining and Homologous Recombination in Human Cells.

DNA Repair (Amst). 2008, 7, 1765–1771.

(110) Heyer, W.-D.; Ehmsen, K. T.; Liu, J. Regulation of Homologous Recombination in

Eukaryotes. Annu. Rev. Genet. 2010, 44, 113–139.

(111) Chapman, J. R.; Taylor, M. R. G.; Boulton, S. J. Playing the End Game: DNA

304

Double-Strand Break Repair Pathway Choice. Mol. Cell 2012, 47, 497–510.

(112) Sharma, S.; Hicks, J. K.; Chute, C. L.; Brennan, J. R.; Ahn, J.; Glover, T. W.;

Canman, C. E. REV1 and Polymerase Zeta Facilitate Homologous Recombination

Repair. Nucleic Acids Res. 2012, 40, 682–691.

(113) McVey, M.; Khodaverdian, V. Y.; Meyer, D.; Cerqueira, P. G.; Heyer, W.-D.

Eukaryotic DNA Polymerases in Homologous Recombination. Annu. Rev. Genet.

2016, 50, 393–421.

(114) Sung, P.; Klein, H. Mechanism of Homologous Recombination: Mediators and

Helicases Take on Regulatory Functions. Nat. Rev. Mol. Cell Biol. 2006, 7, 739–

750.

(115) Sfeir, A.; Symington, L. S. Microhomology-Mediated End Joining: A Back-up

Survival Mechanism or Dedicated Pathway? Trends Biochem. Sci. 2015, 40, 701–

714.

(116) Boulton, S. J.; Jackson, S. P. Saccharomyces Cerevisiae Ku70 Potentiates

Illegitimate DNA Double-Strand Break Repair and Serves as a Barrier to Error-

Prone DNA Repair Pathways. EMBO J. 1996, 15, 5093–5103.

(117) Yan, C. T.; Boboila, C.; Souza, E. K.; Franco, S.; Hickernell, T. R.; Murphy, M.;

Gumaste, S.; Geyer, M.; Zarrin, A. A.; Manis, J. P.; Rajewsky, K.; Alt, F. W. IgH

Class Switching and Translocations Use a Robust Non-Classical End-Joining

Pathway. Nature 2007, 449, 478–482.

(118) Koole, W.; Schendel, R. Van; Karambelas, A. E.; Heteren, J. T. Van; Okihara, K.

305

L.; Tijsterman, M. A Polymerase Theta-Dependent Repair Pathway Suppresses

Extensive Genomic Instability at Endogenous G4 DNA Sites. Nat. Commun. 2014,

5, 1–10.

(119) Kent, T.; Chandramouly, G.; Mcdevitt, S. M.; Ozdemir, A. Y.; Pomerantz, R. T.

Mechanism of Microhomology-Mediated End-Joining Promoted by Human DNA

Polymerase θ. Nat. Struct. Mol. Biol. 2015, 22, 230–237.

(120) Lu, G.; Duan, J.; Shu, S.; Wang, X.; Gao, L.; Guo, J.; Zhang, Y. Ligase I and Ligase

III Mediate the DNA Double-Strand Break Ligation in Alternative End-Joining.

Proc. Natl. Acad. Sci. 2016, 113, 1256–1260.

(121) Krokan, H. E.; Bjørås, M. Base Excision Repair. Cold Spring Harb. Perspect. Biol.

2013. 5:a012583

(122) Guan, L.; Bebenek, K.; Kunkel, T. A.; Greenberg, M. M. Inhibition of Short Patch

and Long Patch Base Excision Repair by an Oxidized Abasic Site. Biochemistry

2010, 49, 9904–9910.

(123) Woodrick, J.; Gupta, S.; Camacho, S.; Parvathaneni, S.; Choudhury, S.; Cheema,

A.; Bai, Y.; Khatkar, P.; Erkizan, H. V.; Sami, F.; Su, Y.; Scharer, O. D.; Sharma,

S.; Roy, R. A. New Sub‐pathway of Long‐patch Base Excision Repair Involving 5′

Gap Formation. EMBO J. 2017, 36, 1605–1622.

(124) Lindahl, T. An N-Glycosidase from Escherichia Coli That Releases Free Uracil from

DNA Containing Deaminated Cytosine Residues. Proc. Natl. Acad. Sci. U. S. A.

1974, 71, 3649–3653.

306

(125) Breimer, L. H.; Lindahl, T. DNA Glycosylase Activities for Thymine Residues

Damaged by Ring Saturation, Fragmentation, or Ring Contraction Are Functions of

Endonuclease III in Escherichia Coli. J. Biol. Chem. 1984, 259, 5543–5548.

(126) Demple, B.; Herman, T.; Chen, D. S. Cloning and Expression of APE, the cDNA

Encoding the Major Human Apurinic Endonuclease : Definition of a Family of DNA

Repair Enzymes. Proc. Natl. Acad. Sci. U. S. A. 1991, 88, 11450–11454.

(127) Matsumoto, Y.; Kim, K. Excision of Deoxyribose Phosphate Residues by DNA

Polymerase Beta During DNA Repair. Science 1995, 269, 699–702.

(128) Sobol, R. W.; Horton, J. K.; Kuhn, R.; Gu, H.; Singhal, R. K.; Prasad, R.; Rajewsky,

K.; Wilson, S. H. Requirement of Mammalian DNA Polymerase Beta in Base

Excision Repair. Nature 1996.

(129) Prasad, R.; Singhal, R. K.; Srivastava, D. K.; Molina, J. T.; Tomkinson, A. E.;

Wilson, S. H.; Chem, J. B. Specific Interaction of DNA Polymerase Beta and DNA

Ligase I in a Multiprotein Base Excision Repair Complex from Bovine Testis. J.

Biol. Chem. 1996, 271, 16000–16007.

(130) Hilbert, T. P.; Chaung, W.; Boorstein, R. J.; Cunningham, R. P.; Teebor, G. W.;

Biochemistry, G. W. Cloning and Expression of the cDNA Encoding the Human

Homologue of the DNA Repair Enzyme, Escherichia Coli Endonuclease III. J. Biol.

Chem. 1997, 272, 6733–6740.

(131) Das, A.; Wiederhold, L.; Leppard, J. B.; Kedar, P.; Prasad, R.; Wang, H.; Boldogh,

I.; Karimi-busheri, F.; Weinfeld, M.; Tomkinson, A. E.; Wilson, S. H.; Mitra, S.;

Hazra, T. K. NEIL2-Initiated, APE-Independent Repair of Oxidized Bases in DNA :

307

Evidence for a Repair Complex in Human Cells. DNA Repair (Amst). 2006, 5, 1439–

1448.

(132) Sebastian, E.; Alseth, I.; Forsbring, M.; Høydal, I.; Morland, I.; Luna, L.; Bjørås,

M.; Dalhus, B. Biochemical Mapping of Human NEIL1 DNA Glycosylase and AP

Lyase Activities. DNA Repair (Amst). 2012, 11, 766–773.

(133) Dobson, C. J.; Allinson, S. L. The Phosphatase Activity of Mammalian

Polynucleotide Kinase Takes Precedence over Its Kinase Activity in Repair of

Single Strand Breaks. Nucleic Acids Res. 2006, 34, 2230–2237.

(134) Gu, H.; Marth, J. D.; Orban, P. C.; Mossmann, H.; Rajewsky, K. Deletion of a DNA

Polymerase β Gene Segment in T Cells Using Cell Type-Specific Gene Targeting.

Science 1994, 265, 103–106.

(135) Asagoshi, K.; Yamada, T.; Terato, H.; Ohyama, Y.; Monden, Y.; Arai, T.;

Nishimura, S.; Aburatani, H.; Lindahl, T.; Ide, H. Distinct Repair Activities of

Human 7, 8-Dihydro-8-Oxoguanine DNA Glycosylase and Formamidopyrimidine

DNA Glycosylase for Formamidopyrimidine and 7, 8-Dihydro-8-Oxoguanine. J.

Biol. Chem. 2000, 275, 4956–4964.

(136) Takao, M.; Kanno, S. I.; Shiromoto, T.; Hasegawa, R.; Ide, H.; Ikeda, S.; Sarker, A.

H.; Seki, S.; Xing, J. Z.; Le, X. C.; Weinfield, M.; Kobayashi, K.; Miyazaki, J.;

Muijtjens, M.; Hoeijmakers, J. H.; van der Horst, G.; Yasui, A Novel Nuclear and

Mitochondrial Glycosylases Revealed by Disruption of the Mouse Nth1 Gene

Encoding an Endonuclease III Homolog for Repair of Thymine Glycols. EMBO J.

2002, 21, 3486–3493.

308

(137) Tsai-Wu, J.-J.; Liu, H.; Lu, A. Escherichia Coli MutY Protein Has Both N-

Glycosylase and Apurinic/apyrimidinic Endonuclease Activities on A-C and A-G

Mispairs. Proc. Natl. Acad. Sci. U. S. A. 1992, 89, 8779–8783.

(138) Dianov, G. L.; Hübscher, U. Mammalian Base Excision Repair: The Forgotten

Archangel. Nucleic Acids Res. 2013, 41, 3483–3490.

(139) Banerjee, A.; Yang, W.; Karplus, M.; Verdine, G. L. Structure of a Repair Enzyme

Interrogating Undamaged DNA Elucidates Recognition of Damaged DNA. Nature

2005, 434, 612–618.

(140) Bruner, S. D.; Norman, D. P. G.; Verdine, G. L. Structural Basis for Recognition

and Repair of the Endogenous Mutagen 8-Oxoguanine in DNA. Nature 2000, 403,

859–866.

(141) McCullough, A. K.; Dodson, M. L.; Lloyd, R. S. Initiation of Base Excision Repair:

Glycosylase Mechanisms and Structures. Annu. Rev. Biochem. 1999, 68, 255–285.

(142) Mullins, E. A.; Shi, R.; Parsons, Z. D.; Yuen, P. K.; David, S. S.; Igarashi, Y.;

Eichman, B. F. The DNA Glycosylase AlkD Uses a Non-Base-Flipping Mechanism

to Excise Bulky Lesions. Nature 2015, 527, 254–258.

(143) Parsons, Z. D.; Bland, J. M.; Mullins, E. A.; Eichman, B. F. A Catalytic Role for

C−H/π Interactions in Base Excision Repair by Bacillus Cereus DNA Glycosylase

AlkD. J. Am. Chem. Soc. 2016, 138, 11485–11488.

(144) Krokan, H. E.; Standal, R.; Slupphaug, G. DNA Glycosylases in the Base Excision

Repair of DNA. Biochem. J. 1997, 325, 1–16.

309

(145) Fromme, J. C.; Verdine, G. L. Structure of a Trapped Endonuclease III-DNA

Covalent Intermediate. EMBO J. 2003, 22, 3461–3471.

(146) Xanthoudakis, S.; Smeyne, R. J.; Wallace, J. D.; Curran, T. The Redox/DNA Repair

Protein, Ref-1, Is Essential for Early Embryonic Development in Mice. Proc Natl

Acad Sci U S A 1996, 93, 8919–8923.

(147) Ludwig, D. L.; MacInnes, M. A.; Takiguchi, Y.; Purtymun, P. E.; Henrie, M.;

Flannery, M.; Meneses, J.; Pedersen, R. A.; Chen, D. J. A Murine AP-Endonuclease

Gene-Targeted Deficiency with Post-Implantation Embryonic Progression and

Ionizing Radiation Sensitivity. Mutat. Res. - DNA Repair 1998, 409, 17–29.

(148) Mol, C. D.; Izumi, T.; Mitra, S.; Talner, J. A. DNA-Bound Structures and Mutants

Reveal Abasic DNA Binding by APE1 DNA Repair and Coordination. Nature 2000,

403, 451–456.

(149) Maher, R. L.; Bloom, L. B. Pre-Steady-State Kinetic Characterization of the AP

Endonuclease Activity of Human AP Endonuclease 1. J. Biol. Chem. 2007, 282,

30577–30585.

(150) Xu, Y. J.; DeMott, M. S.; Hwang, J. T.; Greenberg, M. M.; Demple, B. Action of

Human Apurinic Endonuclease (Ape1) on C1′-Oxidized Deoxyribose Damage in

DNA. DNA Repair (Amst). 2003, 2, 175–185.

(151) Yamtich, J.; Sweasy, J. B. DNA Polymerase Family X: Function, Structure, and

Cellular Roles. Biochim. Biophys. Acta - Proteins Proteomics 2010, 1804, 1136–

1150.

310

(152) García-Díaz, M.; Domínguez, O.; López-Fernández, L. A.; De Lera, L. T.; Saníger,

M. L.; Ruiz, J. F.; Párraga, M.; García-Ortiz, M. J.; Kirchhoff, T.; Del Mazo, J.;

Bernad, A.; Blanco, L. DNA Polymerase Lambda (Pol Λ), a Novel Eukaryotic DNA

Polymerase with a Potential Role in Meiosis. J. Mol. Biol. 2000, 301, 851–867.

(153) Belousova, E. A.; Lavrik, O. I. DNA Polymerases β and λ and Their Roles in Cell.

DNA Repair (Amst). 2015, 29, 112–126.

(154) Piersen, C. E.; Prasad, R.; Wilson, S. H.; Lloyd, R. S. Evidence for an Imino

Intermediate in the DNA Polymerase β Deoxyribose Phosphate Excision Reaction.

J. Biol. Chem. 1996, 271, 17811–17815.

(155) Prasad, R.; Beard, W. A.; Chyan, J. Y.; Maciejewski, M. W.; Mullen, G. P.; Wilson,

S. H. Functional Analysis of the Amino-Terminal 8-kDa Domain of DNA

Polymerase Beta as Revealed by Site-Directed Mutagenesis. J. Biol. Chem. 1998,

273, 11121–11126.

(156) Prasad, R.; Batra, V. K.; Yang, X. P.; Krahn, J. M.; Pedersen, L. C.; Beard, W. A.;

Wilson, S. H. Structural Insight into the DNA Polymerase β Deoxyribose Phosphate

Lyase Mechanism. DNA Repair (Amst). 2005, 4, 1347–1357.

(157) Daskalova, S. M.; Bai, X.; Hecht, S. M. Study of the Lyase Activity of Human DNA

Polymerase β Using Analogues of the Intermediate Schi Ff Base Complex.

Biochemistry 2018.

(158) Garcı́a-Dıaz,́ M.; Bebenek, K.; Kunkel, T. A.; Blanco, L. Identification of an

Intrinsic 5′-Deoxyribose-5-Phosphate Lyase Activity in Human DNA Polymerase

λ. J. Biol. Chem. 2001, 276, 34659–34663.

311

(159) Yoshimura, M.; Kohzaki, M.; Nakamura, J.; Asagoshi, K.; Sonoda, E.; Hou, E.;

Prasad, R.; Wilson, S. H.; Tano, K.; Yasui, A.; Lan, L.; Seki, M.; Wood, R. D.;

Arakawa, H.; Buerstedde, J. M.; Hochegger, H.; Okada, T.; Hiraoka, M.;Takeda.

Vertebrate POLQ and POLβ Cooperate in Base Excision Repair of Oxidative DNA

Damage. Mol. Cell 2006, 24, 115–125.

(160) Bebenek, K.; Frank, E. G.; Mcdonald, J. P.; Prasad, R.; Wilson, S. H.; Woodgate,

R.; Kunkel, T. A. 5’-Deoxyribose Phosphate Lyase Activity of Human DNA

Polymerase Iota in Vitro. Science 2001, 291, 2156–2160.

(161) Prasad, R.; Bebenek, K.; Hou, E.; Shock, D. D.; Beard, W. A.; Woodgate, R.;

Kunkel, T. A.; Wilson, S. H. Localization of the Deoxyribose Phosphate Lyase

Active Site in Human DNA Polymerase Iota by Controlled Proteolysis. J. Biol.

Chem. 2003, 278, 29649–29654.

(162) Prasad, R.; Poltoratsky, V.; Hou, E. W.; Wilson, S. H. Rev1 Is a Base Excision

Repair Enzyme with 5 -Deoxyribose Phosphate Lyase Activity. Nucleic Acids Res.

2016, 44, 10824–10833.

(163) Longley, M. J.; Prasad, R.; Srivastava, D. K.; Wilson, S. H.; Copeland, W. C.

Identification of 5’-deoxyribose Phosphate Lyase Activity in Human DNA

Polymerase Gamma and Its Role in Mitochondrial Base Excision Repair in Vitro.

Proc. Natl. Acad. Sci. U. S. A. 1998, 95, 12244–12248.

(164) Petta, T. B.; Nakajima, S.; Zlatanou, A.; Couve-privat, S.; Ishchenko, A.; Sarasin,

A.; Yasui, A.; Kannouche, P. Human DNA Polymerase Iota Protects Cells against

Oxidative Stress. EMBO J. 2008, 27, 2883–2895.

312

(165) Vaisman, A.; Woodgate, R. Translesion DNA Polymerases in Eukaryotes: What

Makes Them Tick? Crit. Rev. Biochem. Mol. Biol. 2017, 52, 274–303.

(166) Limoli, C. L.; Giedzinski, E.; Bonner, W. M.; Cleaver, J. E. UV-Induced Replication

Arrest in the Xeroderma Pigmentosum Variant Leads to DNA Double- Strand

Breaks , Gamma -H2AX Formation , and Mre11 Relocalization. Proc. Natl. Acad.

Sci. U. S. A. 2002, 99, 2–7.

(167) Christine, R.; Maria, J. Frequent Chromosomal Translocations Induced by DNA

Double-Strand Breaks. Nature 2000, 405, 697–700.

(168) Izhar, L.; Ziv, O.; Cohen, I. S.; Geacintov, N. E.; Livneh, Z. Genomic Assay Reveals

Tolerance of DNA Damage by Both Translesion DNA Synthesis and Homology-

Dependent Repair in Mammalian Cells. Proc. Natl. Acad. Sci. U. S. A. 2013, 110,

E1462–E1469.

(169) Vanoli, F.; Fumasoni, M.; Szakal, B.; Maloisel, L.; Branzei, D. Replication and

Recombination Factors Contributing to Recombination-Dependent Bypass of DNA

Lesions by Template Switch. PLoS Genet. 2010, 6, e1001205

(170) Li, Y.; Dutta, S.; Doublié, S.; Moh, H.; Taylor, J.; Ellenberger, T. Nucleotide

Insertion Opposite a Cis - Syn Thymine Dimer by a Replicative DNA Polymerase

from Bacteriophage T7. Nat. Struct. Mol. Biol. 2004, 11, 784–790.

(171) Aller, P.; Rould, M. A.; Hogg, M.; Wallace, S. S.; Doublié, S. A Structural Rationale

for Stalling of a Replicative DNA Polymerase at the Most Common Oxidative

Thymine Lesion, Thymine Glycol. Proc. Natl. Acad. Sci. U. S. A. 2007, 104, 814–

818.

313

(172) Gregory, M.; Lee, J. Y.; Zhao, Y.; Kondo, Y.; Ramo, S. Structure and Mechanism

of Human DNA Polymerase Eta. Nature 2010, 465, 1044–1048.

(173) Shachar, S.; Ziv, O.; Avkin, S.; Adar, S.; Wittschieben, J.; Reißner, T.; Chaney, S.;

Friedberg, E. C.; Wang, Z.; Carell, T.; Geacintov, N; Livneh, Z. Two-Polymerase

Mechanisms Dictate Error-Free and Error-Prone Translesion DNA Synthesis in

Mammals. EMBO J. 2009, 28, 383–393.

(174) Yoon, J.-H.; Bhatia, G.; Prakash, S.; Prakash, L. Error-Free Replicative Bypass of

Thymine Glycol by the Combined Action of DNA Polymerases Kappa and Zeta in

Human Cells. Proc. Natl. Acad. Sci. U. S. A. 2010, 107, 14116–14121.

(175) Yoon, J.; Prakash, L.; Prakash, S. Error-Free Replicative Bypass of (6-4)

Photoproducts by DNA Polymerase Zeta in Mouse and Human Cells. Genes Dev.

2010, 24, 123–128.

(176) Seki, M.; Masutani, C.; Yang, L. W.; Schuffert, A; Iwai, S.; Bahar, I.; Wood, R. D.

High-Efficiency Bypass of DNA Damage by Human DNA Polymerase Q. Embo J

2004, 23, 4484–4494.

(177) Yoon, J. H.; Choudhury, J. R.; Park, J.; Prakash, S.; Prakash, L. A Role for DNA

Polymerase θ in Promoting Replication through Oxidative DNA Lesion, Thymine

Glycol, in Human Cells. J. Biol. Chem. 2014, 289, 13177–13185.

(178) Yoon, J.-H.; Roy Choudhury, J.; Park, J.; Prakash, S.; Prakash, L. Translesion

Synthesis DNA Polymerases Promote Error-Free Replication through the Minor-

Groove DNA Adduct 3-Deaza-3-Methyl Adenine. J. Biol. Chem. 2017,

jbc.M117.808659.

314

(179) Yoon, J.; Hodge, R. P.; Hackfeld, L. C.; Park, J.; Choudhury, J. R.; Prakash, S.;

Prakash, L. Genetic Control of Predominantly Error-Free Replication through an

Acrolein-Derived Minor-Groove DNA Adduct. J. Biol. Chem. 2018, 293, 2949–

2958.

(180) Waters, L. S.; Minesinger, B. K.; Wiltrout, M. E.; Souza, S. D.; Woodruff, R. V;

Walker, G. C. Eukaryotic Translesion Polymerases and Their Roles and Regulation

in DNA Damage Tolerance. Microbiol. Mol. Biol. Rev. 2009, 73, 134–154.

(181) Diamant, N.; Hendel, A.; Vered, I.; Carell, T.; Reissner, T.; Wind, N. De; Geacinov,

N.; Livneh, Z. DNA Damage Bypass Operates in the S and G2 Phases of the Cell

Cycle and Exhibits Differential Mutagenicity. Nucleic Acids Res. 2011, 40, 170–

180.

(182) Friedberg, E. C.; Lehmann, A. R.; Fuchs, R. P. P. Trading Places: How Do DNA

Polymerases Switch during Translesion DNA Synthesis? Mol. Cell 2005, 18, 499–

505.

(183) Sabbioneda, S.; Bortolomai, I.; Giannattasio, M.; Plevani, P.; Muzi-Falconi, M.

Yeast Rev1 Is Cell Cycle Regulated, Phosphorylated in Response to DNA Damage

and Its Binding to Chromosomes Is Dependent upon MEC1. DNA Repair (Amst).

2007, 6, 121–127.

(184) Quinet, A.; Martins, D. J.; Vessoni, A. T.; Biard, D.; Sarasin, A.; Stary, A.; Menck,

C. F. M. Translesion Synthesis Mechanisms Depend on the Nature of DNA Damage

in UV-Irradiated Human Cells. Nucleic Acids Res. 2016, 44, 5717–5731.

(185) Michel, B. After 30 Years of Study, the Bacterial SOS Response Still Surprises Us.

315

PLoS Biol. 2005, 3, e255

(186) Kato, T.; Shinoura, Y. Isolation and Characterization of Mutants of Escherichia Coli

Deficient in Induction of Mutations by Ultraviolet Light. Mol. Gen. Genet. 1977,

156, 121–131.

(187) Fuchs, R. P. P.; Fujii, S. Translesion DNA Synthesis and Mutagenesis in

Prokaryotes. Cold Spring Harb. Perspect. Biol. 2013, 5, 1–23.

(188) Nelson, J. R.; Lawrence, C. W.; Hinkle, D. C. Deoxycytidyl Transferase Activity of

Yeast REV1 Protein. Nature. 1996, pp 729–731.

(189) Tang, M.; Bruck, I.; Eritja, R.; Turner, J.; Frank, E. G.; Woodgate, R.; O’Donnell,

M.; Goodman, M. F. Biochemical Basis of SOS-Induced Mutagenesis in

Escherichia Coli: Reconstitution of in Vitro Lesion Bypass Dependent on the

UmuD’2C Mutagenic Complex and RecA Protein. Proc. Natl. Acad. Sci. U. S. A.

1998, 95, 9755–9760.

(190) Tang, M.; Shen, X.; Frank, E. G.; O’Donnell, M.; Woodgate, R.; Goodman, M. F.

UmuD’(2)C Is an Error-Prone DNA Polymerase, Escherichia Coli Pol V. Proc.

Natl. Acad. Sci. U. S. A. 1999, 96, 8919–8924.

(191) Gruz, P.; Kim, S.; Yamada, M.; Matsui, K.; Fuchs, R. P. P.; Nohmi, T.; Digestif, D.

A. The dinB Gene Encodes a Novel E. Coli DNA Polymerase, DNA Pol IV,

Involved in Mutagenesis. Mol. Cell 1999, 4, 281–286.

(192) Mcdonald, J. P.; Levine, A. S.; Woodgate, R. The Saccharomyces Cerevisiae

RAD30 Gene, a Homologue of Escherichia Coli DinB and umuC, Is DNA Damage

316

Inducible and Functions in a Novel Error-Free Postreplication Repair Mechanism.

Genetics 1997, 147, 1557–1568.

(193) Gerlach, V. L.; Aravind, L.; Gotway, G.; Schultz, R. A.; Koonin, E. V; Friedberg,

E. C. Human and Mouse Homologs of Escherichia Coli DinB (DNA Polymerase

IV), Members of the UmuC/DinB Superfamily. Proc. Natl. Acad. Sci. U. S. A. 1999,

96.

(194) Kawasaki, Y.; Sugino, A. Yeast Replicative DNA Polymerases and Their Role at

the Replication Fork. Mol. Cells 2001, 12, 277–285.

(195) Fijalkowska, I. J.; Schaaper, R. M.; Jonczyk, P. DNA Replication Fidelity in

Escherichia Coli: A Multi-DNA Polymerase Affair. FEMS Microbiol. Rev. 2012,

36, 1105–1121.

(196) Johansson, E.; Dixon, N. Replicative DNA Polymerases. Cold Spring Harb.

Perspect. Biol. 2013, 5:a012799.

(197) Steitz, T. DNA Polymerases: Structural Diversity and Common Mechanisms. J Biol

Chem 1999, 274, 17395–17398.

(198) Masutani, C.; Kusumoto, R.; Yamada, A. The XPV (Xeroderma Pigmentosum

Variant) Gene Encodes Human DNA Polymerase Eta. Nature 1999, 399, 700–704.

(199) Sonoda, E.; Okada, T.; Zhao, G. Y.; Tateishi, S.; Araki, K.; Yamaizumi, M.; Yagi,

T.; Verkaik, N. S.; Gent, D. C. Van; Takata, M. Multiple Roles of Rev3, the

Catalytic Subunit of Pol Zeta in Maintaining Genome Stability in Vertebrates.

EMBO J. 2003, 22, 3188–3197.

317

(200) Yang, W. An Overview of Y ‑ Family DNA Polymerases and a Case Study of

Human DNA Polymerase η. Biochemistry 2014, 53, 2793–2803.

(201) Washington, M. T.; Johnson, R. E.; Prakash, S.; Prakash, L. Fidelity and

Processivity of Saccharomyces Cerevisiae DNA Polymerase Eta. J. Biol. Chem.

1999, 274, 36835–36838.

(202) Guo, C.; Fischhaber, P. L.; Luk-paszyc, M. J.; Masuda, Y.; Zhou, J.; Kamiya, K.;

Kisker, C.; Friedberg, E. C. Mouse Rev1 Protein Interacts with Multiple DNA

Polymerases Involved in Translesion DNA Synthesis. EMBO J. 2003, 22, 6621–

6630.

(203) Bienko, M.; Green, C. M.; Sabbioneda, S.; Crosetto, N.; Matic, I.; Hibbert, R. G.;

Begovic, T.; Niimi, A.; Mann, M.; Lehmann, A. R.; Dikic, I. Regulation of

Translesion Synthesis DNA Polymerase H by Monoubiquitination. Mol. Cell 2010,

37, 396–407.

(204) Bomar, M. G.; Pai, M.; Tzeng, S.; Li, S. S.; Zhou, P. Structure of the Ubiquitin-

Binding Zinc Finger Domain of Human DNA Y-Polymerase Eta. EMBO Journal,

2007, 8, 247–251.

(205) Bomar, M. G.; Souza, S. D.; Bienko, M.; Dikic, I.; Walker, G. C.; Zhou, P.

Unconventional Ubiquitin Recognition by the Ubiquitin-Binding Motif within the

Y Family DNA Polymerases I and Rev1. Mol. Cell 2010, 37, 408–417.

(206) Kelman, Z. PCNA: Structure, Functions, and Interactions. Oncogene 1997, 14, 629–

640.

318

(207) Bauer, G. A.; Burgers, P. M. J. Molecular Cloning , Structure and Expression of the

Yeast Proliferating Cell Nuclear Antigen Gene. Nucleic Acids Res. 1990, 18, 261–

265.

(208) Haracska, L.; Kondratick, C. M.; Unk, I.; Prakash, S.; Prakash, L. Interaction with

PCNA Is Essential for Yeast DNA Polymerase Eta Function. Mol. Cell 2001, 8,

407–415.

(209) Haracska, L.; Johnson, R. E.; Unk, I.; Phillips, B. B.; Hurwitz, J.; Prakash, L.;

Prakash, S. Targeting of Human DNA Polymerase Iota to the Replication Machinery

via Interaction with PCNA. Proc. Natl. Acad. Sci. U. S. A. 2001, 98, 14256–14261.

(210) Nair, D. T.; Johnson, R. E.; Prakash, L.; Prakash, S.; Aggarwal, A. K. Rev1 Employs

a Novel Mechanism of DNA Synthesis Using a Protein Template. Science 2005,

309, 2219–2222.

(211) Kuang, L.; Kou, H.; Xie, Z.; Zhou, Y.; Feng, X.; Wang, L.; Wang, Z. A Non-

Catalytic Function of Rev1 in Translesion DNA Synthesis and Mutagenesis Is

Mediated by Its Stable Interaction with Rad5. DNA Repair (Amst). 2013, 12, 27–37.

(212) Ross, A.; Simpson, L. J.; Sale, J. E. Vertebrate DNA Damage Tolerance Requires

the C-Terminus but Not BRCT or Transferase Domains of REV1. Nucleic Acids

Res. 2005, 33, 1280–1289.

(213) Jansen, J. G.; Tsaalbi-shtylik, A.; Langerak, P.; Calle, F.; Meijers, C. M.; Jacobs,

H.; Wind, N. De. The BRCT Domain of Mammalian Rev1 Is Involved in Regulating

DNA Translesion Synthesis. Nucleic Acids Res. 2005, 33, 356–365.

319

(214) Nelson, J. R.; Gibbs, P. E. M.; Nowicka, A. M.; Hinkle, D. C.; Lawrence, C. W.

Evidence for a Second Function for Saccharomyces Cerevisiae Rev1p. Mol.

Microbiol. 2000, 37, 549–554.

(215) Haracska, L.; Unk, I.; Johnson, R. E.; Johansson, E.; Burgers, P. M. J.; Prakash, S.;

Prakash, L. Roles of Yeast DNA Polymerases Zeta and Delta and of Rev1 in the

Bypass of Abasic Sites. Genes Dev. 2001, 15, 945–954.

(216) Pryor, J. M.; Gakhar, L.; Washington, M. T. Structure and Functional Analysis of

the BRCT Domain of Translesion Synthesis DNA Polymerase Rev1. Biochemistry

2012, 52, 254–263.

(217) Washington, M. T.; Johnson, R. E.; Prakash, S.; Prakash, L. Accuracy of Thymine

– Thymine Dimer Bypass by Saccharomyces Cerevisiae DNA Polymerase Eta.

Proc. Natl. Acad. Sci. U. S. A. 2000, 97, 3094–3099.

(218) Yoon, J.; Prakash, L.; Prakash, S. Highly Error-Free Role of DNA Polymerase Eta

in the Replicative Bypass of UV-Induced Pyrimidine Dimers in Mouse and Human

Cells. Proc. Natl. Acad. Sci. U. S. A. 2009, 106, 18219–18224.

(219) Choi, J.; Pfeifer, G. P. The Role of DNA Polymerase Eta in UV Mutational Spectra.

DNA Repair (Amst). 2005, 4, 211–220.

(220) Kusumoto, R.; Masutani, C.; Iwai, S.; Hanaoka, F. Translesion Synthesis by Human

DNA Polymerase η across Thymine Glycol. Biochemistry 2002, 41, 6090–6099.

(221) Avkin, S.; Adar, S.; Blander, G.; Livneh, Z. Quantitative Measurement of

Translesion Replication in Human Cells : Evidence for Bypass of Abasic Sites by a

320

Replicative DNA Polymerase. Proc. Natl. Acad. Sci. U. S. A. 2002, 99, 3764–3769.

(222) Avkin, S.; Livneh, Z. Efficiency, Specificity, and DNA Polymerase-Dependence of

Translesion Replication across the Oxidative DNA Lesion 8-Oxoguanine in Human

Cells. Mutat. Res. 2002, 510, 81–90.

(223) Franklin, A.; Milburn, P. J.; Blanden, R. V; Steele, E. J. Human DNA Polymerase

Eta, an A-T Mutator in Somatic Hypermutation of Rearranged Immunoglobulin

Genes, Is a Reverse Transcriptase. Nat. Immunol. 2004, 82, 219–225.

(224) Johnson, R. E.; Prakash, S.; Prakash, L. The Human DINB1 Gene Encodes the DNA

Polymerase Pol Theta. Proc. Natl. Acad. Sci. U. S. A. 2000, 97, 3838–3843.

(225) Washington, M. T.; Johnson, R. E.; Prakash, L.; Prakash, S. Human DINB1-

Encoded DNA Polymerase Kappa Is a Promiscuous Extender of Mispaired Primer

Termini. Proc. Natl. Acad. Sci. U. S. A. 2001, 99, 1910–1914.

(226) Haracska, L.; Prakash, L.; Prakash, S. Role of Human DNA Polymerase Kappa as

an Extender in Translesion Synthesis. Proc. Natl. Acad. Sci. U. S. A. 2002, 99,

16000–1605.

(227) Lone, S.; Townson, S. A.; Uljon, S. N.; Johnson, R. E.; Brahma, A.; Nair, D. T.;

Prakash, S.; Prakash, L.; Aggarwal, A. K. Human DNA Polymerase Kappa

Encircles DNA: Implications for Mismatch Extension and Lesion Bypass. Mol. Cell

2007, 25, 601–614.

(228) Washington, M. T.; Minko, I. G.; Johnson, R. E.; Wolfle, W. T.; Harris, T. M.;

Lloyd, R. S.; Prakash, S.; Prakash, L. Efficient and Error-Free Replication Past a

321

Minor-Groove DNA Adduct by the Sequential Action of Human DNA Polymerases

Iota and Kappa. Mol. Cell. Biol. 2004, 24, 5687–5693.

(229) Williams, H. L.; Gottesman, M. E.; Gautier, J. Replication-Independent Repair of

DNA Interstrand Crosslinks. Mol. Cell 2012, 47, 140–147.

(230) Hile, S. E.; Wang, X.; Lee, M. Y. W. T.; Eckert, K. A. Beyond Translesion

Synthesis : Polymerase Kappa Fidelity as a Potential Determinant of Microsatellite

Stability. Nucleic Acids Res. 2012, 40, 1636–1647.

(231) Johnson, R. E.; Washington, M. T.; Haracska, L.; Prakash, S.; Prakash, L.

Eukaryotic Polymerases Iota and Zeta Act Sequentially to Bypass DNA Lesions.

Nature 2000, 406, 1015–1019.

(232) Nair, D. T.; Johnson, R. E.; Prakash, S.; Prakash, L.; Aggarwal, A. K. Replication

by Human DNA Polymerase Iota Occurs by Hoogsteen Base-Pairing. Nature 2004,

430, 377–380.

(233) Johnson, R. E.; Prakash, L.; Prakash, S. Biochemical Evidence for the Requirement

of Hoogsteen Base Pairing for Replication by Human DNA Polymerase Iota. Proc.

Natl. Acad. Sci. U. S. A. 2005, 102, 10466–10471.

(234) Ziv, O.; Geacintov, N.; Nakajima, S.; Yasui, A.; Livneh, Z. DNA Polymerase Zeta

Cooperates with Polymerases Kappa and Iota in Translesion DNA Synthesis across

Pyrimidine Photodimers in Cells from XPV Patients. Proc. Natl. Acad. Sci. U. S. A.

2009, 106, 11552–11557.

(235) Gueranger, Q.; Stary, A.; Aoufouchi, S.; Faili, A.; Sarasin, A.; Reynaud, C.; Weill,

322

J. Role of DNA Polymerases Eta , Iota, and Zeta in UV Resistance and UV-Induced

Mutagenesis in a Human Cell Line. DNA Repair (Amst). 2008, 7, 1551–1562.

(236) Morrison, A.; Christensen, R. B.; Alley, J.; Beck, A. K.; Bernstine, E. G.; Lemontt,

J. F.; Lawrence, C. W. REV3, a Saccharomyces Cerevisiae Gene Whose Function

Is Required for Induced Mutagenesis, Is Predicted To Encode a Nonessential DNA

Polymerase. J. Bacteriol. 1989, 171, 5659–5667.

(237) Nelson, J. R.; Lawrence, C. W.; Hinkle, D. C. Thymine-Thymine Dimer Bypass by

Yeast DNA Polymerase ζ. Science 1996, 272, 1646–1649.

(238) Johnson, R. E.; Prakash, L.; Prakash, S. Pol31 and Pol32 Subunits of Yeast DNA

Polymerase δ Are Also Essential Subunits of DNA Polymerase ζ. Proc. Natl. Acad.

Sci. U. S. A. 2012, 109, 12455–12460.

(239) Makarova, A. V; Stodola, J. L.; Burgers, P. M. A Four-Subunit DNA Polymerase

Zeta Complex Containing Pol Delta Accessory Subunits Is Essential for PCNA-

Mediated Mutagenesis. Nucleic Acids Res. 2012, 40, 11618–11626.

(240) Lee, Y.; Gregory, M. T.; Yang, W. Human Pol ζ Purified with Accessory Subunits

Is Active in Translesion DNA Synthesis and Complements Pol η in Cisplatin

Bypass. Proc. Natl. Acad. Sci. U. S. A. 2014, 111, 2954–2959.

(241) Sharma, S.; Shah, N. A.; Joiner, A. M.; Roberts, K. H.; Canman, C. E. DNA

Polymerase Zeta Is a Major Determinant of Resistance to Platinum-Based

Chemotherapeutic Agents. Mol. Pharmacol. 2012, 81, 778–787.

(242) Wittschieben, J.; Shivji, M. K. K.; Lalani, E.; Jacobs, M. A.; Marini, F.; Gearhart,

323

P. J.; Rosewell, I.; Stamp, G.; Wood, R. D. Disruption of the Developmentally

Regulated Rev3l Gene Causes Embryonic Lethality. Curr. Biol. 2000, 10, 1217–

1220.

(243) Sloun, P. P. H. Van; Varlet, I.; Sonneveld, E.; Boei, J. J. W. A.; Romeijn, R. J.;

Eeken, J. C. J.; Wind, N. De. Involvement of Mouse Rev3 in Tolerance of

Endogenous and Exogenous DNA Damage. Mol. Cell. Biol. 2002, 22, 2159–2169.

(244) Zander, L.; Bemark, M. Immortalized Mouse Cell Lines That Lack a Functional

Rev3 Gene Are Hypersensitive to UV Irradiation and Cisplatin Treatment. DNA

Repair (Amst). 2004, 3, 743–752.

(245) Lehner, K.; Jinks-Robertson, S. The Mismatch Repair System Promotes DNA

Polymerase Zeta-Dependent Translesion Synthesis in Yeast. Proc. Natl. Acad. Sci.

U. S. A. 2009, 106, 5749–5754.

(246) Efrati, E.; Tocco, G.; Eritja, R.; Wilson, S. H.; Goodman, M. F. “Action-at-a-

Distance” Mutagenesis: 8-Oxo-7,8-Dihydro-Deoxyguanosine Causes Base

Substitution Errors At Neighboring Template Sites When Copied By DNA

Polymerase Beta. J. Biol. Chem. 1999, 274, 15920–15926.

(247) Efrati, E.; Tocco, G.; Eritja, R.; Wilson, S. H.; Goodman, M. F. Abasic Translesion

Synthesis by DNA Polymerase Beta Violates the A-rule. J. Biol. Chem. 1997, 272,

2559–2569.

(248) Markkanen, E.; Castrec, B.; Villani, G.; Hübscher, U. A Switch between DNA

Polymerases δ and λ Promotes Error-Free Bypass of 8-Oxo-G Lesions. Proc. Natl.

Acad. Sci. U. S. A. 2012, 109, 20401–20406.

324

(249) Takata, K.; Shimizu, T.; Iwai, S.; Wood, R. D. Human DNA Polymerase N (POLN)

Is a Low Fidelity Enzyme Capable of Error-Free Bypass of 5 S -Thymine Glycol.

Nucleic Acids Res. 2006, 281, 23445–23455.

(250) Yamanaka, K.; Minko, I. G.; Takata, K.; Kolbanovskiy, A.; Kozekov, I. D.; Wood,

R. D.; Rizzo, C. J.; Lloyd, R. S. Novel Enzymatic Function of DNA Polymerase ν

in Translesion DNA Synthesis Past Major Groove DNA - Peptide and DNA - DNA

Cross-Links. Chem. Res. Toxicol. 2010, 23, 689–695.

(251) Schaaper, R. M.; Glickman, B. W.; Loeb, L. A. Mutagenesis Resulting from

Depurination Is an SOS Process. Mutat. Res. - Fundam. Mol. Mech. Mutagen. 1982,

106, 1–9.

(252) Weerasooriya, S.; Jasti, V. P.; Basu, A. K. Replicative Bypass of Abasic Site in

Escherichia Coli and Human Cells : Similarities and Differences. PLoS One 2014,

9, e107915

(253) Tessman, I.; Kennedy, M. A. DNA Polymerase II of Escherichia Coli in the Bypass

of Abasic Sites in Vivo. Genetics 1994, 136, 439–448.

(254) Maor-Shoshani, A.; Hayashi, K.; Ohmori, H.; Livneh, Z. Analysis of Translesion

Replication across an Abasic Site by DNA Polymerase IV of Escherichia Coli. DNA

Repair (Amst). 2003, 2, 1227–1238.

(255) Wang, F.; Yang, W. Structural Insight Into Translesion Synthesis By DNA Pol II.

Cell 2009, 139, 1279–1289.

(256) Gibbs, P. E. M.; Lawrence, C. W. Novel Mutagenic Properties of Abasic Sites in

325

Saccharomyces Cerevisiae. J. Mol. Biol. 1995, 251, 229–236.

(257) Zhao, B.; Xie, Z.; Shen, H.; Wang, Z. Role of DNA Polymerase η in the Bypass of

Abasic Sites in Yeast Cells. Nucleic Acids Res. 2004, 32, 3984–3994.

(258) Haracska, L.; Washington, M. T.; Prakash, S.; Prakash, L. Inefficient Bypass of an

Abasic Site by DNA Polymerase Eta. J. Biol. Chem. 2001, 276, 6861–6866.

(259) Pagès, V.; Johnson, R. E.; Prakash, L.; Prakash, S. Mutational Specificity and

Genetic Control of Replicative Bypass of an Abasic Site in Yeast. Proc. Natl. Acad.

Sci. U. S. A. 2008, 105, 1170–1175.

(260) Kroeger, K. M.; Kim, J.; Goodman, M. F.; Greenberg, M. M. Replication of an

Oxidized Abasic Site in Escherichia Coli by a dNTP-Stabilized Misalignment

Mechanism That Reads Upstream and Downstream Nucleotides. Biochemistry

2006, 45, 5048–5056.

(261) Kroeger, K. M.; Kim, J.; Goodman, M. F.; Greenberg, M. M. Effects of the C4 ′ -

Oxidized Abasic Site on Replication in Escherichia Coli. An Unusually Large

Deletion Is Induced by a Small Lesion. Biochemistry 2004, 43, 13621–13627.

(262) Kroeger, K. M.; Jiang, Y. L.; Kow, Y. W.; Goodman, M. F.; Greenberg, M. M.

Mutagenic Effects of 2-Deoxyribonolactone in Escherichia Coli . An Abasic Lesion

That Disobeys the A-Rule. Biochemistry 2004, 43, 6723–6733.

(263) Kow, Y. W.; Bao, G.; Minesinger, B.; Jinks-Robertson, S.; Siede, W.; Jiang, Y. L.;

Greenberg, M. M. Mutagenic Effects of Abasic and Oxidized Abasic Lesions in

Saccharomyces Cerevisiae. Nucleic Acids Res. 2005, 33, 6196–6202.

326

(264) Evans, J.; Maccabee, M.; Hatahet, Z.; Courcelle, J.; Bockrath, R.; Ide, H.; Wallace,

S. Thymine Ring Saturation and Fragmentation Products: Lesion Bypass,

Misinsertion and Implications for Mutagenesis. Mutat. Res. Toxicol. 1993, 299,

147–156.

(265) Lin, X.; Howell, S. B. DNA Mismatch Repair and p53 Function Are Major

Determinants of the Rate of Development of Cisplatin Resistance. Mol. Cancer

Ther. 2006, 5, 1239–1248.

(266) Ogi, T.; Mimura, J.; Hikida, M.; Fujimoto, H.; Fujii-kuriyama, Y.; Ohmori, H.

Expression of Human and Mouse Genes Encoding Pol K: Testis-Specific

Developmental Regulation and AhR-Dependent Inducible Transcription. Genes to

Cells 2001, 6, 943–953.

(267) Hoege, C.; Pfander, B.; Moldovan, G.; Pyrowolakis, G.; Jentsch, S. RAD6-

Dependent DNA Repair Is Linked to Modification of PCNA by Ubiquitin and

SUMO. Nature 2002, 419, 135–141.

(268) Haracska, L.; Torres-ramos, C. A.; Johnson, R. E.; Prakash, S.; Prakash, L.

Opposing Effects of Ubiquitin Conjugation and SUMO Modification of PCNA on

Replicational Bypass of DNA Lesions in Saccharomyces Cerevisiae. Mol. Cell.

Biol. 2004, 24, 4267–4274.

(269) Hendel, A.; Krijger, P. H. L.; Diamant, N.; Goren, Z.; Langerak, P.; Kim, J.;

Reissner, T.; Lee, K.; Geacintov, N. E.; Carell, T.; Myung, K.; Tateishi, S.;

D'Andrea, A.; Jacobs, H.; Livneh, Z. PCNA Ubiquitination Is Important, But Not

Essential for Translesion DNA Synthesis in Mammalian Cells. PLoS Genet. 2011,

327

7, e1002262.

(270) Masutani, C. Mechanisms of Accurate Translesion Synthesis by Human DNA

Polymerase Eta. EMBO J. 2000, 19, 3100–3109.

(271) Hogg, M.; Seki, M.; Wood, R. D.; Doublié, S.; Wallace, S. S. Lesion Bypass

Activity of DNA Polymerase Theta (POLQ) Is an Intrinsic Property of the Pol

Domain and Depends on Unique Sequence Inserts. J. Mol. Biol. 2011, 405, 642–

652.

(272) Boyd, J. B.; Sakaguchi, K.; Harris, P. V. mus308 Mutants of Drosophila Exhibit

Hypersensitivity to DNA Cross-Linking Agents and Are Defective in a

Deoxyribonuclease. Genetics 1990, 125, 813–819.

(273) Harris, P. V; Mazina, O. M.; Leonhardt, E. A.; Case, R. B.; Boyd, J. B.; Burtis, K.

C. Molecular Cloning of Drosophila mus308, a Gene Involved in DNA Cross-Link

Repair with Homology to Prokaryotic DNA Polymerase I Genes. Mol. Cell. Biol.

1996, 16, 5764–5771.

(274) Seki, M.; Marini, F.; Wood, R. D. POLQ (Pol Θ), a DNA Polymerase and DNA-

Dependent ATPase in Human Cells. Nucleic Acids Res. 2003, 31, 6117–6126.

(275) Ozdemir, A. Y.; Rusanov, T.; Kent, T.; Siddique, L. A.; Pomerantz, R. T.

Polymerase θ-Helicase Efficiently Unwinds DNA and RNA-DNA Hybrids. J. Biol.

Chem. 2018, 293, 5259–5269.

(276) Zahn, K. E.; Averill, A. M.; Aller, P.; Wood, R. D.; Doublié, S. Human DNA

Polymerase θ Grasps the Primer Terminus to Mediate DNA Repair. Nat. Struct. Mol.

328

Biol. 2015, 22, 304–311.

(277) Shima, N.; Munroe, R. J.; Schimenti, J. C. The Mouse Genomic Instability Mutation

chaos1 Is an Allele of Polq That Exhibits Genetic Interaction with Atm. Mol. Cell.

Biol. 2004, 24, 10381–10389.

(278) Beagan, K.; Armstrong, R. L.; Witsell, A.; Roy, U.; Renedo, N.; Baker, A. E.;

Scharer, O.; McVey, M. Drosophila DNA Polymerase Theta Utilizes Both Helicase-

like and Polymerase Domains during Microhomology-Mediated End Joining and

Interstrand Crosslink Repair. PLoS Genet. 2017, 13, e1006813.

(279) Wyatt, D. W.; Feng, W.; Conlin, M. P.; Yousefzadeh, M. J.; Roberts, S. A.;

Mieczkowski, P.; Wood, R. D.; Gupta, G. P.; Ramsden, D. A.. Essential Roles for

Polymerase θ-Mediated End Joining in the Repair of Chromosome Breaks. Mol. Cell

2016, 63, 662–673.

(280) Seki, M.; Wood, R. D. DNA Polymerase Theta (POLQ) Can Extend from

Mismatches and from Bases Opposite a (6-4) Photoproduct. DNA Repair (Amst).

2008, 7, 119–127.

(281) Mateos-Gomez, P. A.; Kent, T.; Deng, S. K.; Mcdevitt, S.; Kashkina, E.; Hoang, T.

M.; Pomerantz, R. T.; Sfeir, A. The Helicase Domain of Polθ Counteracts RPA to

Promote Alt-NHEJ. Nat. Struct. Mol. Biol. 2017, 24, 1116–1123.

(282) Goff, J. P.; Shields, D. S.; Seki, M.; Choi, S.; Epperly, M. W.; Dixon, T.; Wang, H.;

Bakkenist, C. J.; Dertinger, S. D.; Torous, D. K.; Wittschieben, J.; Wood, R. D.;

Greenberger, J. S. Lack of DNA Polymerase θ (POLQ) Radiosensitizes Bone

Marrow Stromal Cells in Vitro and Increases Reticulocyte Micronuclei after Total-

329

Body Irradiation. Radiat. Res. 2009, 172, 165–174.

(283) Higgins, G. S.; Boulton, S. J. Beyond PARP—POLθ as an Anticancer Target.

Science 2018, 359, 1217–1218.

(284) Lemée, F.; Bergoglio, V.; Fernandez-Vidal, A.; Machado-Silva, A.; Pillaire, M.-J.;

Bieth, A.; Gentil, C.; Baker, L.; Martin, A.-L.; Leduc, C.; et al. DNA Polymerase

Theta up-Regulation Is Associated with Poor Survival in Breast Cancer, Perturbs

DNA Replication, and Promotes Genetic Instability. Proc. Natl. Acad. Sci. U. S. A.

2010, 107, 13390–13395.

(285) Wood, R. D.; Doublié, S. DNA Polymerase θ (POLQ), Double-Strand Break Repair,

and Cancer. DNA Repair (Amst). 2016, 44, 22–32.

(286) Nijman, S. M. B. Synthetic Lethality: General Principles, Utility and Detection

Using Genetic Screens in Human Cells. FEBS Lett. 2011, 585, 1–6.

(287) Fell, V. L.; Schild-poulter, C. The Ku Heterodimer: Function in DNA Repair and

beyond. Mutat. Res. Mutat. Res. 2015, 763, 15–29.

(288) Weller, G. R.; Kysela, B.; Roy, R.; Tonkin, L. M.; Scanlan, E.; Della, M.; Devine,

S. K.; Day, J. P.; Wilkinson, A.; Devine, K. M.; Bowater, R. P.; Jeggo, P. A.;

Jackson, S. P.; Doherty, A. J. Identification of a DNA Nonhomologous End-Joining

Complex in Bacteria. Science 2002, 297, 1686–1690.

(289) Mimori, T.; Akizuki, M.; Yamagata, H.; Inada, S. Characterization of a High

Molecular Weight Acidic Nuclear Protein Recognized by Autoantibodies in Sera

from Patients with Polymyositis-Scleroderma Overlap. J. Clincal Investig. 1981, 68,

330

611–620.

(290) Mimori, T.; Hardint, A.; Steitz, A. Characterization of the DNA-Binding Protein

Antigen Ku Recognized by Autoantibodies from Patients with Rheumatic Disorders.

J. Biol. Chem. 1986, 261, 2274–2278.

(291) Gu, Y.; Shengfang, J.; Gao, Y.; Weaver, D. T.; Alt, F. W. Ku70-Deficient

Embryonic Stem Cells Have Increased Ionizing Radiosensitivity , Defective DNA

End-Binding Activity and Inability to Support V(D)J Recombination. Proc. Natl.

Acad. Sci. 1997, 94, 8076–8081.

(292) Smider, V.; W Kimryn, R.; Lieber, M. R.; Chu, G. Restoration of X-Ray Resistance

and V(D)J Recombination in Mutant Cells by Ku cDNA. Science 1994, 266, 288–

291.

(293) Nussenzweig, A.; Sokol, K.; Burgman, P.; Li, L.; Li, G. C. Hypersensitivity of

Ku80-Deficient Cell Lines and Mice to DNA Damage: The Effects of Ionizing

Radiation on Growth, Survival, and Development. Proc. Natl. Acad. Sci. 1997, 94,

13588–13593.

(294) Fisher, T. S.; Zakian, V. A. Ku: A Multifunctional Protein Involved in Telomere

Maintenance. DNA Repair (Amst). 2005, 4, 1215–1226.

(295) Li, G.; Nelsen, C.; Hendrickson, E. A. Ku86 Is Essential in Human Somatic Cells.

Proc. Natl. Acad. Sci. 2002, 99, 832–837.

(296) Errami, A.; Smider, V.; Rathmell, W. K.; He, D. M.; Hendrickson, E. A.;

Zdzienicka, M. Z.; Chu, G. Ku86 Defines the Genetic Defect and Restores X-Ray

331

Resistance and V(D)J Recombination to Complementation Group 5 Hamster Cell

Mutants. Mol. Cell. Biol. 1996, 16, 1519–1526.

(297) Yoo, S.; Kimzey, A.; Dynan, W. S. Photocross-Linking of an Oriented DNA Repair

Complex. J. Biol. Chem. 1999, 274, 20034–20039.

(298) Blier, P. R.; Griffith, A. J.; Craft, J.; Hardin, A. Binding of Ku Protein to DNA. J.

Biol. Chem. 1993, 268, 7594–7601.

(299) Singleton, B. K.; Torres-Arzayus, M. I.; Rottinghaus, S. T.; Taccioli, G. E.; Jeggo,

P. A. The C Terminus of Ku80 Activates the DNA-Dependent Protein Kinase

Catalytic Subunit. Mol. Cell. Biol. 1999, 19, 3267–3277.

(300) Teixeira-Silva, A.; Ait Saada, A.; Hardy, J.; Iraqui, I.; Nocente, M. C.; Fréon, K.;

Lambert, S. A. E. The End-Joining Factor Ku Acts in the End-Resection of Double

Strand Break-Free Arrested Replication Forks. Nat. Commun. 2017, 8, 1982.

(301) Lomax, M. E.; Folkes, L. K.; O’Neill, P. Biological Consequences of Radiation-

Induced DNA Damage: Relevance to Radiotherapy. Clin. Oncol. 2013, 25, 578–

585.

(302) Strande, N.; Roberts, S. a.; Oh, S.; Hendrickson, E. a.; Ramsden, D. A.. Specificity

of the dRP/AP Lyase of Ku Promotes Nonhomologous End Joining (NHEJ) Fidelity

at Damaged Ends. J. Biol. Chem. 2012, 287, 13686–13693.

(303) Strande, N. T.; Carvajal-Garcia, J.; Hallett, R. A.; Waters, C. A.; Roberts, S. A.;

Strom, C.; Kuhlman, B.; Ramsden, D. A.. Requirements for 5’dRP/AP Lyase

Activity in Ku. Nucleic Acids Res. 2014, 42, 11136–11143.

332

(304) O’Sullivan, R. J.; Karlseder, J. Telomeres: Protecting Chromosomes against

Genome Instability. Nat. Rev. Mol. Cell Biol. 2010, 11, 171–181.

(305) Hsu, H.; Gilley, D.; Galande, S. A.; Hande, M. P.; Allen, B.; Kim, S.; Li, G. C.;

Campisi, J.; Kohwi-shigematsu, T.; Chen, D. J. Ku Acts in a Unique Way at the

Mammalian Telomere to Prevent End Joining. Genes Dev. 2000, 14, 2807–2812.

(306) Teo, S.; Jackson, S. P. Identification of Saccharomyces Cerevisiae DNA Ligase IV :

Involvement in DNA Double-Strand Break Repair. EMBO J. 1997, 16, 4788–4795.

(307) Wang, Y.; Ghosh, G.; Hendrickson, E. A. Ku86 Represses Lethal Telomere Deletion

Events in Human Somatic Cells. Proc. Natl. Acad. Sci. 2009, 106, 12430–12435.

(308) Greider, C. W.; Blackburn, E. H. Identification of a Specific Telomere Terminal

Transferase Activity in Tetrhymena Extracts. Cell 1985, 43, 405–413.

(309) Stellwagen, A. E.; Haimberger, Z. W.; Veatch, J. R.; Gottschling, D. E. Ku Interacts

with Telomerase RNA to Promote Telomere Addition at Native and Broken

Chromosome Ends. Genes Dev. 2003, 17, 2384–2395.

(310) Wu, J.; Churikov, D.; Ge, V.; Chen, H.; Xue, J.; Churikov, D.; Hass, E. P.; Shi, S.;

Lemon, L. D.; Luciano, P.; Bertuch, A. A.; Zappulla, D. C.; Geli, V.; Wu, J.; Lei,

M. Structural Insights into Yeast Telomerase Recruitment to Telomeres. Cell 2018,

172, 331–343.

(311) Williams, J. M.; Ouenzar, F.; Lemon, L. D.; Chartrand, P.; Bertuch, A. A. The

Principal Role of Ku in Telomere Length Maintenance Is Promotion of Est1

Association. Genetics 2014, 197, 1123–1136.

333

(312) Krishna, T. S. R.; Kong, X. P.; Gary, S.; Burgers, P. M.; Kuriyan, J. Crystal Structure

of the Eukaryotic DNA Polymerase Processivity Factor PCNA. Cell 1994, 79,

1233–1243.

(313) Hedglin, M.; Kumar, R.; Benkovic, S. J. Replication Clamps and Clamp Loaders.

Cold Spring Harb. Perspect. Biol. 2013, 5, 1–19.

(314) Kubota, T.; Katou, Y.; Nakato, R.; Shirahige, K.; Donaldson, A. D. Replication-

Coupled PCNA Unloading by the Elg1 Complex Occurs Genome-Wide and

Requires Okazaki Fragment Ligation. Cell Rep. 2015, 12, 774–787.

(315) Postow, L.; Ghenoiu, C.; Woo, E. M.; Krutchinsky, A. N.; Chait, B. T.; Funabiki,

H. Ku80 Removal from DNA through Double Strand Break – Induced

Ubiquitylation. J. Cell Biol. 2008, 182, 467–479.

(316) Stingele, J.; Bellelli, R.; Alte, F.; Hewitt, G.; Sarek, G.; Maslen, S. L.; Tsutakawa,

S. E.; Borg, A.; Kjær, S.; Tainer, J. A.; Skehel, J. M.; Groll, M.; Boulton, S. J.

Mechanism and Regulation of DNA-Protein Crosslink Repair by the DNA-

Dependent Metalloprotease SPRTN. Mol. Cell 2016, 64, 688–703.

(317) Postow, L. Destroying the Ring: Freeing DNA from Ku with Ubiquitin. FEBS Lett.

2011, 585, 2876–2882.

(318) Sites, K. D.; Brown, J. S.; Lukashchuk, N.; Galanty, Y.; Jackson, S. P.; Sage, C.

Neddylation Promotes Ubiquitylation and Release of Article Neddylation Promotes

Ubiquitylation and Release of Ku from DNA-Damage Sites. Cell Rep. 2015, 11,

704–714.

334

(319) Malaby, A. W.; Martin, S. K.; Wood, R. D.; Doublié, S. Expression and Structural

Analyses of Human DNA Polymerase θ (POLQ). Methods Enzymol. 2017, 592,

103–121.

(320) Johnson, K. A. Conformational Coupling in DNA Polymerase Fidelity. Annu. Rev.

Biochem. 1993, 62, 685–713.

(321) Su, Y.; Guengerich, F. P. Pre-Steady-State Kinetic Analysis of Single-Nucleotide

Incorporation by DNA Polymerases. Curr. Protoc. Nucleic Acid Chem. 2017, 65,

7.23.1-7.23.10.

(322) Laverty, D. J.; Averill, A. M.; Doublié, S.; Greenberg, M. M. The A-Rule and

Deletion Formation during Abasic and Oxidized Abasic Site Bypass by DNA

Polymerase θ. ACS Chem. Biol. 2017, 12, 1584–1592.

(323) Laverty, D. J.; Greenberg, M. M. In Vitro Bypass of Thymidine Glycol by DNA

Polymerase θ Forms Sequence-Dependent Frameshift Mutations. Biochemistry

2017, 56, 6726–6733.

(324) Kim, J.; Gil, J. M.; Greenberg, M. M. Synthesis and Characterization of

Oligonucleotides Containing the C4’-oxidized Abasic Site Produced by Bleomycin

and Other DNA Damaging Agents. Angew. Chemie, Int. Ed. 2003, 42, 5882–5885.

(325) Kotera, M.; Roupioz, Y.; Defrancq, E.; Bourdat, È.; Garcia, J.; Coulombeau, C.;

Lhomme, J. The 7-Nitroindole Nucleoside as a Photochemical Precursor of 2 â€TM

-Deoxyribonolactone : Access to DNA Fragments Containing This Oxidative

Abasic Lesion. Chem. Eur. J. 2000, 6, 4163–4169.

335

(326) Chowdhury, G.; Guengerich, F. P. Liquid Chromatography-Mass Spectrometry

Analysis of DNA Polymerase Reaction Products. Curr. Protoc. Nucleic Acid Chem.

2011, 47, 7.16.1-7.16.11.

(327) Dean, F.; Nelson, J.; Giesler, T.; Lasken, R. Rapid Amplification of Plasmid and

Phage DNA Using Phi29 Polymerase and a Multiply-Pimed Rolling Circle

Amplification. Genome Res. 2001, 11, 1095–1099

(328) Goullet de Rugy, T.; Bashkurov, M.; Datti, A.; Betous, R.; Guitton-Sert, L.; Cazaux,

C.; Durocher, D.; Hoffmann, J. S. Excess Polθ Functions in Response to Replicative

Stress in Homologous Recombination-Proficient Cancer Cells. Biol. Open 2016, 5,

1485–1492.

(329) Iwai, S. Synthesis of Thymine Glycol Containing Oligonucleotides from a Building

Block with the Oxidized Base. Angew. Chemie - Int. Ed. 2000, 39, 3874–3876.

(330) Wincott, F.; Direnzo, A.; Shaffer, C.; Grimm, S.; Tracz, D.; Workman, C.; Sweedler,

D.; Gonzalez, C.; Scaringe, S.; Usman, N. Synthesis, Deprotection, Analysis and

Purification of RNA and Ribosomes. Nucleic Acids Res. 1995, 23, 2677–2684.

(331) Huang, H.; Imoto, S.; Greenberg, M. M. The Mutagenicity of Thymidine Glycol in

Escherichia Coli Is Increased When It Is Part of a Tandem Lesion. Biochemistry

2009, 48, 7833–7841.

(332) Hogg, M.; Sauer-Eriksson, A. E.; Johansson, E. Promiscuous DNA Synthesis by

Human DNA Polymerase θ. Nucleic Acids Res. 2012, 40, 2611–2622.

(333) Müller, T. A.; Tobar, M. A.; Perian, M. N.; Hausinger, R. P. Biochemical

336

Characterization of AP Lyase and m6A Demethylase Activities of Human AlkB

Homologue 1 (ALKBH1). Biochemistry 2017, 56, 1899–1910.

(334) Prasad, R.; Shock, D. D.; Beard, W. A.; Wilson, S. H. Substrate Channeling in

Mammalian Base Excision Repair Pathways: Passing the Baton. J. Biol. Chem.

2010, 285, 40479–40488.

(335) Zhou, C.; Sczepanski, J. T.; Greenberg, M. M. Histone Modification via Rapid

Cleavage of C4’-oxidized Abasic Sites in Nucleosome Core Particles. J. Am. Chem.

Soc. 2013, 135, 5274–5277.

(336) Zhang, Y.; Zhou, X.; Xie, Y.; Greenberg, M. M.; Xi, Z.; Zhou, C. Thiol Specific

and Tracelessly Removable Bioconjugation via Michael Addition to 5 ‑ Methylene

Pyrrolones. J. Am. Chem. Soc. 2017.

(337) Deterding, L. J.; Prasad, R.; Mullen, G. P.; Wilson, S. H.; Tomer, K. B. Mapping of

the 5’-2-Deoxyribose-5-Phosphate Lyase Active Site in DNA Polymerase Beta by

Mass Spectrometry. J. Biol. Chem. 2000, 275, 10463–10471.

(338) Zee, B. M.; Garcia, B. A. Discovery of Lysine Post-Translational Modifications

through Mass Spectrometric Detection. Essays Biochem. 2012, 52, 147–163.

(339) Arian, D.; Hedayati, M.; Zhou, H.; Bilis, Z.; Chen, K.; Theodore, L.; Greenberg, M.

M. Irreversible Inhibition of DNA Polymerase β by Small Molecule Mimics of a

DNA Lesion . No. Scheme 1.

(340) Ray, S.; Breuer, G.; DeVeaux, M.; Zelterman, D.; Bindra, R.; Sweasy, J. B. DNA

Polymerase Beta Participates in DNA End-Joining. Nucleic Acids Res. 2018, 46,

337

242–255.

(341) Sage, E.; Shikazono, N. Radiation-Induced Clustered DNA Lesions: Repair and

Mutagenesis. Free Radic. Biol. Med. 2017, 107, 125–135.

(342) Blaisdell, J. O.; Wallace, S. S. Abortive Base-Excision Repair of Radiation-Induced

Clustered DNA Lesions in Escherichia Coli. Proc. Natl. Acad. Sci. 2001, 98, 7426–

7430.

(343) Gulston, M.; de Lara, C.; Jenner, T.; Davis, E.; O’Neill, P. Processing of Clustered

DNA Damage Generates Additional Double-Strand Breaks in Mammalian Cells

Post-Irradiation. Nucleic Acids Res. 2004, 32, 1602–1609.

(344) Bellon, S.; Shikazono, N.; Cunniffe, S.; Lomax, M.; O’Neill, P. Processing of

Thymine Glycol in a Clustered DNA Damage Site: Mutagenic or Cytotoxic. Nucleic

Acids Res. 2009, 37, 4430–4440.

(345) Parsons, J. L.; Kavli, B.; Slupphaug, G.; Dianov, G. L. NEIL1 Is the Major DNA

Glycosylase That Processes 5-Hydroxyuracil in the Proximity of a DNA Single-

Strand Break. Biochemistry 2007, 46, 4158–4163.

(346) Kozmin, S. G.; Sedletska, Y.; Reynaud-Angelin, A.; Gasparutto, D.; Sage, E. The

Formation of Double-Strand Breaks at Multiply Damaged Sites Is Driven by the

Kinetics of Excision/incision at Base Damage in Eukaryotic Cells. Nucleic Acids

Res. 2009, 37, 1767–1777.

(347) Sawaya, M. R.; Prasad, R.; Wilson, S. H.; Kraut, J.; Pelletier, H. Crystal Structures

of Human DNA Polymerase β Complexed with Gapped and Nicked DNA: Evidence

338

for an Induced Fit Mechanism. Biochemistry 1997, 36, 11205–11215.

(348) Piersen, C. E.; McCullough, A. K.; Lloyd, R. S. AP Lyases and dRPases:

Commonality of Mechanism. Mutat. Res. - DNA Repair 2000, 459, 43–53.

(349) Hanakahi, L. A. 2-Step Purification of the Ku DNA Repair Protein Expressed in

Escherichia Coli. Protein Expr. Purif. 2007, 52, 139–145.

(350) Farooqui, A. A. Purification of Enzymes by Heparin-Sepharose Affinity

Chromatography. J. Chromatogr. A 1980, 184, 335–345.

(351) Jaafar, L.; Li, Z.; Li, S.; Dynan, W. S. SFPQ•NONO and XLF Function Separately

and Together to Promote DNA Double-Strand Break Repair via Canonical

Nonhomologous End Joining. Nucleic Acids Res. 2017, 45, 1848–1859.

(352) Demott, M. S.; Beyret, E.; Wong, D.; Bales, B. C.; Hwang, J. T.; Greenberg, M. M.;

Demple, B. Covalent Trapping of Human DNA Polymerase Beta by the Oxidative

DNA Lesion 2-Deoxyribonolactone. J. Biol. Chem. 2002, 277, 7637–7640.

(353) Quiñones, J. L.; Thapar, U.; Yu, K.; Fang, Q.; Sobol, R. W.; Demple, B. Enzyme

Mechanism-Based, Oxidative DNA–protein Cross-Links Formed with DNA

Polymerase β in Vivo. Proc. Natl. Acad. Sci. 2015, 112, 8602–8607.

(354) Miropolskaya, N.; Petushkov, I.; Kulbachinskiy, A.; Makarova, A. V. Identification

of Amino Acid Residues Involved in the dRP-Lyase Activity of Human Pol ι. Sci.

Rep. 2017, No. June, 1–7.

(355) Sobol, R. W.; Wilson, S. H. Mammalian DNA β-Polymerase in Base Excision

Repair of Alkylation Damage. Prog. Nucleic Acid Res. Mol. Biol. 2001, 68, 57–74.

339

(356) Zhu, Q.; Delaney, M. O.; Greenberg, M. M. Observation and Elimination of N-

Acetylation of Oligonucleotides Prepared Using Fast-Deprotecting

Phosphoramidites and Ultra-Mild Deprotection. Bioorganic Med. Chem. Lett. 2001,

11, 1105–1107.

(357) Baumann, S.; Schoof, S.; Bolten, M.; Haering, C.; Takagi, M.; Shin-ya, K.; Arndt,

H.-D. Molecular Determinants of Microbial Resistance to Thiopeptide Antibiotics.

J. Am. Chem. Soc. 2010, 132, 6973–6981.

8. Curriculum Vitae

Daniel Laverty was born in Melrose, Massachusetts in 1991. He earned a bachelor’s

degree in chemistry and biology from Merrimack College in 2013. He began his

graduate studies at Johns Hopkins University in 2013 and joined the Greenberg lab

to investigate the effects of oxidized abasic lesions on DNA polymerase θ and

Ku70/80. These studies resulted in two first-author publications along with a third

submitted and a fourth in preparation. He will start a post-doctoral fellowship in

Zachary Nagel’s laboratory at Harvard School of Public Health in late June 2018.

340