ERROR-PRONE DNA REPAIR IN THE AFRICAN SWINE FEVER VIRUS: CHARACTERIZATION OF SIX ABASIC SITE PROCESSING ACTIVITIES AND EVIDENCE FOR A MUTAGENIC FUNCTION
DISSERTATION
Presented in Partial Fulfillment of the Requirements for
the Degree Doctor of Philosophy in the Graduate
School of The Ohio State University
By
Brandon James Lamarche, B.S.
* * * * *
The Ohio State University 2005
Dissertation Committee: Approved by Professor Ming-Daw Tsai, Adviser
Professor Dehua Pei ______Professor Ross Dalbey Adviser Graduate Program in Chemistry
ABSTRACT
The African Swine Fever Virus (ASFV) is a complex (∼150 genes), cytoplasmic, double-stranded DNA virus that causes a potentially lethal disease in domestic pigs.
Antigenic differences among field isolates suggest that ASFV exists as a diverse population of serotypes in some regions of Africa. Moreover, restriction fragment length polymorphisms – in the absence of major genome rearrangements – suggest that the genetic diversity of ASFV may arise from point mutations or small insertions/deletions.
Consistent with its intracellular location, ASFV encodes three DNA repair proteins: a repair polymerase, an AP endonuclease, and a DNA ligase. The ASFV repair polymerase, Pol X, is extremely error prone during single nucleotide gap filling, leading others to hypothesize that it might contribute to the genetic variability of ASFV.
In order for the error-proneness of Pol X to be biologically relevant it would need to function within a repair system in which each of the components tolerated and/or utilized the mismatched intermediates and products being formed. The work described here was undertaken in order to assess whether such a system exists.
Herein we demonstrate for the first time that ASFV gene E296R is an AP endonuclease, a 3’-phosphodiesterase, and a 3’→5 exonuclease. Pol X and ASFV DNA ligase are both shown to contribute lyase activity towards 5’-2-deoxyribose-5-phosphate.
ii With this complement of activities we demonstrate that ASFV is capable of effecting repair of abasic sites without the need to recruit host factors.
Having established a complete abasic site repair system, we assessed its capacity for mutagenesis. The catalytic efficiency of nick sealing by ASFV DNA ligase was determined for substrates containing all 16 possible base pair combinations at the 3’ side of a nick. Our results indicate this enzyme to be the lowest fidelity DNA ligase ever reported – capable of ligating a 3’ mismatched nick more efficiently than nicks containing Watson-Crick base pairs. Comparison of the mismatch specificity of Pol X with that of ASFV DNA ligase suggests that the latter may have evolved towards low fidelity for the purpose of generating the broadest possible spectrum of sealed mismatches. In competition experiments where mismatched nicks were incubated with both ASFV DNA ligase and ASFV AP endonuclease, the ligation activity competed very effectively with the 3’→5’ exonuclease editing activity – supporting the hypothesis that error-prone repair of the ASFV genome may facilitate its diversification.
iii
Dedicated to every member of my family, each of whom
has supported me in their own unique way
iv ACKNOWLEDGMENTS
My advisor, Ming-Daw Tsai, supported me patiently during the course of this work and allowed me the freedom to pursue my own interests. Alex Showalter initiated studies of ASFV Pol X and DNA ligase, developed the assays for determining the active concentration of DNA ligases, and helped train me during the early phase of my research.
Working with me, Yu (Kathy) Wang and Sandeep Kumar studied mismatch ligation by human DNA ligase IV and translesion synthesis by Pol X, respectively.
v VITA
October 30, 1976...... Born - Bellflower, California
1995-1999 ...... B.S. Chemistry, Westmont College
1999-present...... Graduate Teaching and Research Associate, NIH CBIP Predoctoral Fellow, The Ohio State University
PUBLICATIONS
1. Lamarche, B.J.; Showalter, A.K.; Tsai, M.–D. “An Error-Prone Viral DNA Ligase” Biochemistry ASAP Article; DOI: 10.1021/bi047706g (2005).
2. Bakhtina, M.; Lee, S.; Wang, Y.; Dunlap, C.A.; Lamarche, B.J.; Tsai, M.-D. “Use of Viscogens, dNTPαS and Rhodium(III) as Probes in Stopped-Flow Experiments to Obtain New Evidence for the Mechanism of Catalysis by DNA Polymerase β” Biochemistry 44, 5177-5187 (2005).
3. Tsai, M.-D.; Zhao, L.; Lamarche, B.J. “Enzyme Reaction Mechanisms: Stereochemistry” in Encyclopedia of Biological Chemistry, Vol. 2, (eds. Lennarz, W.J. & Lane, M.D.) 45-50 (Elsevier, Boston, 2004).
FIELDS OF STUDY
Major Field: Chemistry
vi TABLE OF CONTENTS
Page Abstract ...... ii Dedication ...... iv Acknowledgments...... v Vita...... vi List of Figures ...... xi
Chapters:
1. INTRODUCTION...... 1 1.1 Genetic Diversity and Evasion of Environmental Challenge...... 1 1.2 The African Swine Fever Virus...... 2 1.3 ASFV DNA Polymerase X and the Mutagenic Repair Hypothesis ...... 3 1.4 The Questions to be Addressed...... 4 1.5 Introduction to Base Excision Repair...... 5
2. ASFV DNA LIGASE...... 7 2.1 Background on DNA Ligation ...... 8 2.2 Identification and Previous Characterization of ASFV DNA Ligase.... 9 2.3 Subcloning, Protein Expression and Purification...... 11 2.4 DNA Ligation Assays...... 15 2.4.1 Preparation of Substrates...... 15 2.4.2 Manual Quench Assay Protocol ...... 17 2.4.3 Assay Buffer Optimization...... 17 2.5 Quantitation of Active Enzyme Concentration ...... 20
7 2.5.1 Forcing T4 DNA Ligase to “Burst” Using a 3’-NH2- Containing Nick ...... 20 2.5.2 Ligation of a Pre-Adenylylated Nick by ASFV DNA Ligase ... 22 2.5.3 Comparison of the Two Ligase Quantitation Assays ...... 26 2.6 Analysis of DNA Ligation Fidelity ...... 27 2.6.1 Previous Studies ...... 28 2.6.2 ASFV and T4 DNA Ligases...... 29 2.6.2.1 Experimental Details and Primary Data...... 29 2.6.2.2 Fidelity Profiles and Catalytic Parameters ...... 31 2.6.2.3 Assessment of Data Reproducibility...... 34 2.6.2.4 Data Analysis ...... 34 2.6.2.5 Importance of Ionic Strength...... 36 2.6.2.6 An Active Site Point Mutation...... 37 2.6.3 Synopsis of the Mechanism of DNA Ligation Fidelity with Consideration of Protein Structure...... 39 2.7 ASFV DNA Ligase vs. Pol X: Comparison of Specificity Profiles and Implications for the Mutagenic Repair Hypothesis ...... 43 2.8 Low Fidelity DNA Ligation in Other Systems...... 45
3. ASFV AP ENDONUCLEASE...... 46 3.1 AP Endonucleases in DNA Repair...... 46 3.2 ASFV Gene E296R: Identification and Preliminary Analysis ...... 48 3.3 Cloning, Protein Expression and Purification...... 49 3.4 Physical Characterization...... 55 3.5 AP Endonuclease Substrates and Assay Details ...... 56 3.6 Metal Cofactor Usage...... 57 3.7 Quantitation of Active Enzyme Concentration ...... 61 8 3.8 Demonstration of Three Repair Activities ...... 62 3.8.1 AP Endonuclease...... 63 3.8.2 3’-Phosphodiesterase...... 64 3.8.3 3’→5’ Exonuclease...... 66 3.9 Mismatch Editing: Implications for the Fidelity of AP Site Repair...... 67 3.9.1 5’-2-Deoxyribose-5-phosphate Influences Editing Efficiency.. 68 3.9.2 Editing Efficiency as a Function of 3’ Base Pair...... 70 3.9.3 Mismatch Ligation vs. Mismatch Editing: Competition Assays...... 76
4. 5’-2-DEOXYRIBOSE-5-PHOSPHATE REMOVAL: IDENTIFICATION AND CHARACTERIZATION OF A “MISSING” ACTIVITY IN ASFV AP SITE REPAIR ...... 79 4.1 An Apparent Bottleneck in ASFV APE-Initiated AP Site Repair ...... 79 4.2 Consideration of Potential Substrate Scenarios ...... 80 4.3 Assay Details...... 81 4.4 5’-dRP Removal from a Single Nucleotide Gap...... 83 4.5 5’-dRP Removal from a Nick...... 88 4.6 AP Site Incision by ASFV Proteins ...... 92 4.7 Summary of ASFV Proteins’ Activity Against the Three Abasic Lesions ...... 100 4.8 Reductive Trapping Assay: Discerning Between Lyase and Hydrolase Mechanisms ...... 100 4.9 Adenylylation of ASFV DNA Ligase Enhances its Activity Against Nicked 5’-dRP ...... 110
9 5. DNA REPAIR IN ASFV...... 112 5.1 Chapter Overview...... 112 5.2 Completeness of the ASFV DNA Repair System ...... 112 5.2.1 Reconstitution of AP Site Repair ...... 112 5.2.2 A Missing DNA Glycosylase ...... 115 5.2.3 Searching for NIR Activity ...... 117 5.2.4 Potential Recruitment of Host-derived Repair Factors by the ASFV-encoded PCNA Homologue...... 118 5.3 Further Characterization of Pol X ...... 119 5.3.1 Influence of 5’-2-Deoxyribose-5-phosphate on Gap Filling ....119 5.3.2 Translesion Synthesis Capabilities...... 122 5.4 Reassessing the Mutagenic DNA Repair Hypothesis ...... 127
BIBLIOGRAPHY ...... 130
10 LIST OF FIGURES
Figure Page
1.1 The two routes of short patch base excision repair ...... 7
2.1 Simplified ATP-dependent DNA ligase mechanism...... 9
2.2 Elution of ASFV DNA ligase from an S-100 column...... 13
2.3 Silver-stained SDS polyacrylamide gel showing elution of ASFV DNA ligase from hydroxyapatite by 320 mM sodium phosphate...... 14
2.4 Nicked DNA substrates...... 16
2.5 Optimization of DNA ligation assay buffer ...... 19
2.6 Burst assay to determine the total concentration of active T4 DNA ligase...... 22
2.7 Three assays to determine the concentration of active ASFV DNA ligase in both the adenylylated and unadenylylated forms ...... 23
2.8 Empirical determination of the optimal nicked substrate concentration for deadenylylating DNA ligase prior to incubation with adenylylated substrate ... 26
2.9 Comparison of active T4 DNA ligase concentrations determined by the two different ligase quantitation assays...... 27
2.10 Primary data and saturation curves for sealing nicked C:G and C:T by ASFV DNA ligase...... 30
2.11 Catalytic efficiency of nick sealing as a function of 3’ base pair for T4 DNA ligase and ASFV DNA ligase...... 32
2.12 Kinetic parameters for nick ligation by T4 DNA ligase as a function of 3’ base pair identity...... 33
xi 2.13 Kinetic parameters for nick ligation by ASFV DNA ligase as a function of 3’ base pair identity...... 33
2.14 Alignment of motif I from the nucleotidyltransferase domain of representative NAD+- and ATP-dependent DNA ligases ...... 38
2.15 Kinetic parameters for nick ligation by WT and N153D ASFV DNA ligase ..... 38
2.16 Summary of DNA ligase structures solved to date ...... 41
2.17 Comparison of the catalytic efficiencies of Pol X (for mismatch synthesis) and ASFV DNA ligase (for mismatch ligation)...... 44
3.1 Properties of E. coli exonuclease III and endonuclease IV...... 47
3.2 Phylogenetic analysis of representative AP endonucleases...... 48
3.3 Analysis of ASFV APE expression in Rosetta(DE3)pLysS cells ...... 50
3.4 ASFV APE purification buffers ...... 51
3.5 Silver-stained gels summarizing the purification of ASFV APE on cobalt resin ...... 52
3.6 Silver-stained gel of ASFV APE purified from BW528(DE3)pRAREpLysS cells in the absence of EDTA...... 54
3.7 Sequence of His-tagged ASFV APE, with fragments identified by MALDI-TOF mass spectrometry highlighted ...... 56
3.8 Inherent instability of an abasic site ...... 57
3.9 Percentage similarity/identity of various APEs...... 57
3.10 Activity of ASFV APE – purified in the absence of metal chelators – against A:THF...... 59
3.11 Activity of apo-ASFV APE – purified in the presence of EDTA – against A:THF ...... 61
xii 3.12 Confirmation that ASFV gene E296R encodes an AP endonuclease...... 63
3.13 ASFV APE is also a 3’-phosphodiesterase ...... 66
3.14 3’→5’ exonuclease activity of ASFV APE...... 67
3.15 Influence of a 5’-dRP analog (5’-Pi-THF) on the 3’→5’ mismatched nick editing activity of ASFV APE...... 69
3.16 Time courses for ASFV APE 3’→5’ exonuclease activity against all 16 possible base pair combinations at the 3’-OH side of a nick ...... 70
3.17 Comparison of ASFV APE’s mismatched nick editing activity with the catalytic efficiencies of Pol X and ASFV DNA ligase...... 71
3.18 Analysis of mismatched nick editing by ASFV APE...... 73
3.19 E. coli endonuclease IV complexed with abasic (THF) DNA ...... 74
3.20 A hypothetical base flipping mechanism for 3’→5’ exonuclease activity in the endonuclease IV family of APEs ...... 75
3.21 Mismatch editing vs. mismatch ligation: competition assays between ASFV APE and DNA ligase ...... 77
4.1 In situ generation of substrates containing 5’-dRP ...... 81
4.2 5’-dRP removal from a single nucleotide gap, hot gel...... 84
4.3 5’-dRP removal from a single nucleotide gap, bar plot ...... 85
4.4 5’-dRP removal from a nick, hot gel...... 88
4.5 5’-dRP removal from a nick, bar plot ...... 89
4.6 AP site incision, bar plot ...... 93
4.7 AP site incision, hot gel...... 95
xiii 4.8 Different routes for AP site processing that give products consistent with those generated by Pol X and ASFV DNA ligase...... 97
4.9 Summary of ASFV proteins’ activity against three different abasic lesions.....100
4.10 Reductive trapping to probe for a lyase mechanism ...... 101
4.11 Reductive trapping of ASFV APE on abasic site variants ...... 104
4.12 Reductive trapping of Pol X on abasic site variants...... 106
4.13 Reductive trapping of ASFV DNA ligase on abasic site variants...... 108
4.14 Reductive trapping of Pol β on abasic site variants ...... 109
4.15 Adenylylation enhances the efficiency with which ASFV DNA ligase is trapped on nicked 5’-dRP...... 111
xiv CHAPTER 1
INTRODUCTION
1.1 Genetic Diversity and Evasion of Environmental Challenge
The long-term fitness of a species is dependent both on its ability to faithfully replicate/repair its genome and on the plasticity of its genome in the face of environmental challenge. Without rigorous genetic maintenance, genes previously optimized for specific protein structures/activities are mutated – usually with deleterious effects. However, in the face of environmental challenge, when conditions no longer match those to which an organism is well suited, rigorous genome maintenance becomes a hindrance to genetic diversification/adaptation and therefore to survival. New conditions require new proteins or protein variants, and hence, new genes or gene variants. It is now clear that rather than relying on the occasional genetic changes resulting from imperfect “housekeeping”, nature often employs enzyme systems capable of potent, and rapid genetic diversification (1, 2). A detailed knowledge of the mechanisms responsible for diversification is critical for: i) understanding the extent and rate of biological evolution past and present, and ii) designing therapeutics effective against rapidly evolving pathogens.
1 Though genome diversification can be effected by both genetic shift (the large scale rearrangements associated with recombination) and genetic drift (the accumulation of point mutations and small insertions/deletions), this report focuses solely on the latter.
Conceptually, a genetically encoded mechanism (as opposed to an exogenous one) for promoting genetic drift might take two forms. In the first, mitigation or ablation of a
DNA repair system could give rise to increased mutation rates; we will refer to this as
“passive mutagenesis”. This is well exemplified by down-regulation of the mismatch repair system in E. coli (3). Mitigation of this post-replication editing system results in a
100- to 1,000-fold increase in mutation frequency, with point mutations being prevalent
(3-5). A second means of effecting genetic drift may involve an enzyme or a group of enzymes that actively introduce mutations into the genome; we will refer to this as
“active mutagenesis”. This functionality is recognized in the reverse transcriptase of
HIV, which facilitates viral hypermutation by its very low fidelity genome replication (6-
8).
The work described herein was undertaken to ascertain whether active mutagenesis might be achieved via an alternative mechanism. In particular, we asked:
“Can genetic drift be induced by a low fidelity DNA repair system?”
1.2 African Swine Fever Virus
When domestic pigs were introduced in Kenya in the early 1900’s they were found to contract a severe disease, now known to be caused by the African swine fever virus
(ASFV), associated with internal organ hemorrhages and nearly 100% lethality (9).
ASFV has since spread to different parts of Africa, the Iberian Peninsula, the Caribbean,
2 and South America (9). With its spread, less virulent strains have emerged. To date there are no effective therapeutics or vaccines against ASFV and the safest mechanism for quelling an outbreak is quarantine and wholesale slaughter of infected herds (10).
Owing to its classification as one of the three most deadly pig pathogens, ASFV is considered a “high-consequence pathogen” by the US Department of Agriculture and research on the intact virus is restricted to just a few laboratories worldwide – one of these being Plum Island.
ASFV is now known to be a large (∼180 kb), complicated (∼150 genes), unique
(only member of the new virus family Asfarviridae), double-stranded DNA virus that principally targets porcine macrophage cells (11) where it displays a distinct phase of
DNA synthesis in the cytoplasm prior to virion assembly (12, 13). Antigenic differences among field isolates suggest that ASFV exists as a diverse population of serotypes in some regions of Africa (14). Additionally, restriction fragment length polymorphisms – in the absence of major genome rearrangements – are consistent with genetic diversity in
ASFV arising from point mutations or small insertions/deletions (15).
Consistent with its intracellular location, ASFV encodes its own replicative DNA polymerase and three canonical base excision repair (BER) enzymes: a putative class II
AP endonuclease (APE), a DNA repair polymerase, and an ATP-dependent DNA ligase
(16).
1.3 ASFV DNA Polymerase X and the Mutagenic Repair Hypothesis
Sequencing of the ASFV genome in 1995 revealed the presence of a gene encoding a 20 kD protein with homology to the eukaryotic DNA repair polymerase Pol β (16). In 1997
3 analysis of this ASFV protein, named Pol X, revealed it to be a functional homologue of
Pol β: Pol X is a highly distributive DNA-directed DNA polymerase, lacks a 3’→5’ exonuclease activity, and shows a preference for gapped substrates containing 5’- phosphate (17). On the basis of these findings, Pol X was suggested to play a role in repair of the viral genome. Being that ASFV also encodes a DNA ligase (which had at this time already been characterized), and a putative AP endonuclease, a repair pathway resembling BER was hypothesized. Noting that Pol X is expressed late during ASFV infection, and that ASFV is known to inhibit host cell apoptosis, it was further surmised that expression of this repair system might be timed to cope with damaged sustained by the viral genome in a late cellular response to ASFV infection (17).
In 2001 Showalter and Tsai demonstrated that during single-nucleotide gap filling
Pol X is very error-prone and, in particular, catalyzes the formation of G:G mismatches with an efficiency comparable to that of the correct, Watson-Crick G:C base pair (18).
Moreover, Pol X was found to be more error-prone on gapped DNA (the preferred substrate) than on template-primer. In light of this data, it was suggested that Pol X may have evolved to function in a mutagenic DNA “repair” pathway capable of conferring diversity to the viral genome. The low catalytic efficiency of Pol X was deemed to be an important characteristic of the enzyme, preventing its mutagenic capacity from being so potent as to be lethal (18).
1.4 The Questions to be Addressed
In order for the error-proneness of Pol X to be biologically relevant, it would need to function within the context of a system capable both of tolerating and utilizing DNA
4 nicks containing 3’ mismatched base pairs. A preliminary analysis of the ASFV DNA ligase demonstrated that it is indeed more capable of sealing a 3’ G:G mismatched nick than is the DNA ligase from bacteriophage T4 (19).
The work described herein was undertaken in order to assess, as fully as possible, whether error-prone DNA repair might be contributing to the diversity observed in
ASFV. The critical questions being asked are: i) Is a complete DNA repair system, including ASFV-encoded and host-derived activities, available for converting damaged sites in the ASFV genome into contiguous duplex product?, and if such a system exists ii)
Are each of the components of this system able to tolerate and/or utilize mismatched repair intermediates and mismatched duplex products?
1.5 Introduction to Base Excision Repair
To set the stage for the work described in this report, an overview of BER is at this point necessary. A large percentage of DNA lesions involve damage to the nitrogenous bases, and the primary means for repairing these is BER. Though the details of BER are generally conserved among the different kingdoms of life, a number of variations have been documented. As a general overview, Figure 1.1 outlines “short patch” (involving replacement of just the damaged nucleotide) BER as it occurs in mammals. The pathway is initiated upon excision of the damaged base by a damage specific DNA glycosylase to generate an apurinic/apyrimidinic (AP) site. AP sites are usually then hydrolyzed, 5’ to the lesion, by an AP endonuclease (APE) – generating a single nucleotide gap flanked by a 3’-hydroxyl and 5’-2-deoxyribose-5-phosphate (5’-dRP). Subsequently, gap filling is
5 effected by a DNA polymerase, which also has the capability of removing 5’-dRP via a lyase mechanism. Finally, the nick so generated is sealed by a DNA ligase.
In an alternative route of short patch BER, an AP lyase incises 3’ to the abasic ribose to generate a 5’-phosphate and the 3’ polymerase-blocking group 4-hydroxy-2- pentenal-5-phosphate (Figure 1.1). The latter can be excised by a 3’-phosphodiesterase, which is an activity typically found in the APE mentioned above, to generate a single nucleotide gap with a 3’-OH. Gap filling and ligation then proceed as above.
A recent addition to the BER pathway is the 3’→5’ exonuclease activity of the multifunctional APE (Figure 1.1). With a preference for mismatched nicks, in humans this activity appears to compensate for the editing activity that is absent in the repair polymerase Pol β.
6 gap filling 5’-dRP lyase
(polymerase) (polymerase)
OH O O O O P nick sealing O P O O OH O O (DNA ligase) AP hydrolase O O O P O O O HO (APE) OH OH O O P O O O O O O P O O P O O O O P O O O O O O O O P O P O 3’→5’ exonuclease O O OH → O (APE) 3’ 5’ exonuclease (APE) O O O O O P O P O O
O O OH O P O O O O P OH O AP lyase O O H (polymerase) OH nick sealing O O O O O O (glycosylase) P P O O O P (DNA ligase) O O O
3’-phosphodiesterase gap filling (APE) (polymerase)
Figure 1.1 The two routes of short patch base excision repair. The initiating step, glycosylase-mediated removal of a-damaged base to generate the abasic site shown at the far left, is not shown. See text for details. 7 CHAPTER 2
ASFV DNA LIGASE
2.1 Background on DNA Ligation
DNA ligation was inherent in early models of DNA recombination and repair long before any such activity was experimentally verified (20). Upon demonstration that (i) genetic recombination is effected by the breakage and rejoining of DNA (1961) (21), and that (ii) linear bacteriophage λ DNA is converted to covalently closed duplex circles after infection of a host bacterium (1964) (22), the search for a “DNA ligase” began in earnest.
As pointed out by I. R. Lehman (20), five independent laboratories reported discovery of
DNA ligases nearly simultaneously in 1967 (23-27) – providing an indication of the extreme interest in this novel class of enzyme.
Playing essential roles in replication (Okazaki fragment processing), repair, and recombination, DNA ligases are now known to be ubiquitous in all three kingdoms of life, in addition to being encoded within the genomes of a wide variety of viruses.
Despite their diversity in size, sequence, and cofactor usage (either ATP or NAD+), all
DNA ligases characterized to date catalyze metal-dependent phosphodiester bond formation between adjacent 3’-hydroxyl and 5’-phosphoryl termini in duplex DNA via a
8
Figure 2.1 Simplified ATP-dependent DNA ligase mechanism.
similar mechanism. In the first step of this reaction the ε-NH2 group of a lysine residue attacks the α-phosphate of ATP (or the pyrophosphate linkage of NAD+), eliminating pyrophosphate (or NMN+) and forming an enzyme-AMP intermediate (20). This
“charging” process appears to effect structural rearrangements which facilitate binding of nicked DNA (28). The AMP moiety (also referred to as an adenylyl group) is subsequently transferred to the 5’-phosphate of a DNA nick, thereby activating the 5’- phosphate for nucleophilic attack by the adjacent 3’-hydroxyl (20). Nick sealing coincides with elimination of AMP and release of the sealed duplex. This reaction mechanism is summarized for an ATP-dependent DNA ligase in Figure 2.1.
2.2 Identification and Previous Characterization of ASFV DNA Ligase
Due to the fact that at least one phase of ASFV replication occurs in the cytoplasm, and since host DNA replication and repair factors are predominantly sequestered within the nucleus, it was predicted that the ASFV genome would encode its own DNA ligase (29).
In 1992 and 1993 sequencing efforts by independent groups revealed an ASFV gene, present in each of two different virus isolates, encoding a 48 kD polypetide with significant homology to known ATP-dependent DNA ligases (30, 31). Incubation of
ASFV infected cell extracts with [α-32P]ATP resulted in a unique, labeled band migrating
9 with an apparent MW of 45-49.5 kD on an SDS polyacrylamide gel (30, 31). Since formation of a covalent protein-AMP adduct is not a common reaction, this was taken as strong evidence for the existence of an ASFV ATP-dependent DNA ligase. This conclusion was further supported by the following facts: i) inclusion of either nicked
DNA or pyrophosphate in the reaction mixture resulted in loss of radioactivity from the band of interest (consistent with transfer of AMP from the protein to the nick or pyrophosphate, respectively) (30, 31), and ii) incubation of [α-32P]ATP with extracts of
E. coli that had been transformed with the putative ASFV DNA ligase gene gave rise to a labeled band of MW similar to that observed in ASFV-infected cell extracts (30, 31).
Collectively, this work unequivocally established the existence of an ASFV-encoded
ATP-dependent DNA ligase.
To our knowledge, nothing else was published on ASFV DNA ligase in the subsequent eight years. In 2001, using a semi-quantitative assay, Showalter and coworkers demonstrated that the ASFV DNA ligase is much more tolerant of a 3’ G:G mismatched nick than is the DNA ligase from bacteriophage T4 (19). This finding suggested that DNA repair in ASFV may indeed be mutagenic, with Pol X generating mismatched nicks and the ASFV DNA ligase sealing these to give mismatched duplex products. The studies described in the remaining sections of this chapter were therefore undertaken in order to assess, more quantitatively, the fidelity (i.e. error-tolerance) of
ASFV DNA ligase.
10 2.3 Subcloning, Protein Expression and Purification
Expression of ASFV DNA ligase from pET-17b in BL21(DE3)pLysS cells, as described previously (19), gave low yields of the protein. With the hope that alternative expression conditions might improve yields, different combinations of expression plasmids (pET-
17b, pET-21b, pBAD/HisB), expression hosts (BL21, BL21(DE3)pLysS,
BL21(DE3)pLysE, CodonPlus, Rosetta(DE3)pLysS), and expression conditions (media type, and temperature and length of induction) were examined. This particular body of work must be qualified by stating that is was conducted during what was primarily a learning phase, when techniques were still being absorbed and a research “mentality” developed. Accordingly, in hindsight, these efforts at expression optimization were not as focused, systematic, thorough, or efficient as they might have been were they carried out by someone more experienced. Critical features of the protocol that was eventually selected for expression and purification are described below.
The ASFV DNA ligase gene was subcloned from pET 17b-ASFV ligase (19) into the
NcoI and KpnI sites of pBAD/HisB, which contains a tightly regulated arabinose- inducible E. coli RNA polymerase promoter. This construct does not include the N- terminal His tag. Use of the NcoI site resulted in the conservative leucine to valine mutation at amino acid #2. With this plasmid, protein expression is expected to be most efficient in TOP10 cells (or an isogenic variant), which are WT in arabinose import but deficient in arabinose metabolism (resulting in intracellular accumulation of the inducer).
Despite this advantage of TOP10 cells, BL21 (which are WT in arabinose metabolism) might be rationalized as a more optimal expression host by the following chain of reasoning: i) even when using a broad spectrum of protease inhibitors during purification,
11 a truncated fragment of ASFV DNA ligase was consistently observed after expression in a variety of host types – suggesting that this protein contains a protease sensitive site, ii)
TOP10 cells are WT for the E. coli proteases, and iii) BL21 cells – and their variants – are deficient in the major proteases Lon and Omp. Though a quantitative analysis of the yield of full length ASFV DNA ligase was never attempted using TOP10 cells, the protein was found to express with moderate efficiency in BL21 cells so these were used in the work described below.
After reaching mid-log phase in SOB medium at 37 °C, the temperature was decreased to 30 °C, L-arabinose added to a concentration of 0.13% (w/v), and shaking continued for 6 hours before harvesting by centrifugation. In hindsight, protein yields might be enhanced for this expression system by periodically titrating in additional arabinose to compensate for metabolic loss.
Protein purification was at 4 °C in a buffer consisting of 50 mM Tris-HCl, 400 mM
KCl, 10% glycerol (v/v), 1 mM EDTA, pH → 8.0 at 4 °C with KOH, and 10 mM DTT.
Cells were lysed by sonication in this buffer supplemented with 100 µM PMSF, 100 µM
AEBSF, and Complete Protease Inhibitor Cocktail according to the manufacturer’s protocol. After pelleting cellular debris, the clarified lysate was applied to a DEAE column (8 cm x 5 cm i.d.) that had been equilibrated with the same buffer. The flow through, including 600 mL of wash (using the same buffer), was applied to a P-11 column that was then washed with 1 L of the same buffer. Elution was in purification buffer using a linear KCl gradient (0.4-1.4 M; 350 mL total volume) at a flow rate of ∼0.8 mL/min., collecting 5.6 mL fractions. ASFV DNA ligase-containing fractions (which
12 eluted between 500 and 800 mM KCl) were pooled, concentrated in an Amicon
centrifugal filter device (10 kD MW limit), and loaded onto an S-100 column (200 cm x
3 cm i.d.) equilibrated with 150 mM KCl purification buffer. Active, full length ASFV
ligase eluted from this column in the void volume along with an active, truncated form
and a high MW contaminant (Figure 2.2). It was found that by rerunning the sample on
an S-200 column in the presence of 4% Tween 20 (v/v) the two proteins could be
resolved. Ligase-containing S-200 eluate was pooled and dialyzed into 10 mM sodium
A B
C
Figure 2.2 Elution of ASFV DNA ligase from an S-100 column. (A) S-100 elution profile. The expected elution volume for ASFV DNA ligase is indicated; this volume was calculated using a standard curve that had been generated under similar running conditions. (B) Silver-stained SDS polyacrylamide gel of S-100 fractions. Note the near perfect overlap of the 48 and ∼80 kD bands. (C) Checking S-100 fractions for adenylylation activity. Fractions were incubated with Mg2+ and [α-32P]ATP and then resolved on a 12% SDS polyacrylamide gel prior to phosphor screen autoradiography. The arrow indicates faint bands which were consistently observed at this phase of the purification. We attribute these to an active, truncated form of ASFV DNA ligase.
13 phosphate, 10% glycerol (v/v), 50 µM EDTA, pH 7.3 at 4 °C, 10 mM DTT, and 0.5%
Tween 20 (v/v) before being loaded onto a hydroxyapatite column (1.2 cm i.d.; bed volume was ∼10 mL) that had been equilibrated with the same buffer. For washing/elution a step gradient was employed, consisting of 20, 40, 80, 160, and 320 mM sodium phosphate. Tween 20 was excluded from the latter two steps. ASFV DNA ligase eluted, at >95% purity, late in the 320 mM step (Figure 2.3). Ligase containing fractions were pooled and dialyzed against the following buffer: 50 mM Tris-borate, 100 mM KCl,
15% glycerol (v/v), 10 mM DTT, pH → 8.0 at 4 °C with KOH. After concentration, the sample was supplemented to a final concentration of 50% glycerol (v/v), flash frozen in
N2 (l) and stored at –80 °C. Subsequently, working aliquots were kept at –20 °C.
The T4 DNA ligase used in kinetic assays was from New England Biolabs and was used without further purification.
Figure 2.3 Silver-stained SDS polyacrylamide gel showing elution of ASFV DNA ligase from hydroxyapatite by 320 mM sodium phosphate.
14 2.4 DNA Ligation Assays
2.4.1 Preparation of Substrates
Oligonucleotides were purified by denaturing polyacrylamide gel electrophoresis under standard conditions. After gel extraction, oligos were desalted on Sep-Pak C18 columns, dried in a speed-vac, and resuspended in TE buffer (10 mM Tris-HCl, 1 mM EDTA, pH
7.5). Concentrations were determined by UV absorbance at 260 nm using extinction coefficients calculated by the oligo analyzer tool on IDT’s website.
The appropriate oligonucleotides were labeled using T4 polynucleotide kinase
(PNK) and [γ-32P]ATP. After heat inactivating PNK, free ATP was removed on a
Microspin G-25 column. Nicked substrates (Figure 2.4) were assembled at room temperature by combining upstream 26mer with 45mer template and 5’-phosphorylated downstream 19mer at a ratio of 1:1.2:1.44, respectively; heating and slow cooling the oligonucleotide mixture in order to effect complete annealing was found to be unnecessary. Substrates were predominantly constructed with the 32P-label at the 5’ terminus of the upstream 26mer. However, this occasionally allowed for blunt end ligation to occur (for those base pairs displaying a very high KM and therefore requiring
32 high [DNAnicked] to saturate the enzyme). When this was the case the P-label was placed at the 5’ terminus of the downstream 19mer using PNK to catalyze the exchange reaction. On average, only 2% of the substrate molecules were labeled. Prior to storage at 4 °C, substrate was diluted to a concentration of 5-30 µM in the following buffer: 50 mM Tris-borate, 100 mM KCl, 15% glycerol (v/v), pH→7.8 with KOH while at 37 °C.
15
Figure 2.4 Nicked DNA substrates. Substrates are named according to the base pair that is 3’ (upstream) to the nick. This base pair is described as X:Y, where “X” denotes the templating nucleotide and “Y” represents the nucleotide that would have been incorporated by a polymerase in the preceding gap filling step of repair. “OH” is the 3’- hydroxyl and “P” is the 5’-phosphate.
Herein, the base pair identity at the 3’-OH side of a nick is described as X:Y, where “X” represents the templating nucleotide and “Y” is the nucleotide that would have been inserted by a polymerase in the preceding gap filling step. Note that this notation is the reverse of what has occasionally been used in the DNA ligase literature.
To facilitate comparison of mismatch formation/utilization by Pol X and ASFV DNA ligase, the 45mer templates used here are identical to those used in previous kinetic analyses of Pol X (18). The presence of two slightly different classes of ligation substrates in Figure 2.4 is rooted in DNA polymerase methodology. When monitoring polymerase catalyzed single nucleotide incorporation on a template/primer substrate, the nascent templating nucleotide must be different than the templating nucleotide 5’ to it in order to prevent multiple incorporation events. It is for this reason, along with our intention to make comparisons between Pol X and ASFV DNA ligase, that two slightly different templating sequences were employed in these studies.
16 2.4.2 Manual Quench Assay Protocol
Unless noted otherwise, all DNA ligase assays for determining active enzyme concentration and for fidelity analyses were performed manually at 37 °C in optimized assay buffer (vide infra) as follows. After a 5 minute pre-incubation at 37 °C, a solution containing DNA ligase, MgCl2, ATP, DTT, and BSA at twice the intended concentrations was added to an equal volume of nicked DNA substrate, also at twice the intended concentration. Generally, 10 µL aliquots were removed at the appropriate time points and quenched in 10 µL of formamide containing xylene cyanol and bromophenol blue.
Reaction products were resolved on 15% (or 19% for the T4 ligase burst assays) denaturing polyacrylamide gels and visualized by phosphor screen autoradiography using a STORM840 scanner from Molecular Dynamics. Band intensity quantitation and data plotting were with ImageQuant and SigmaPlot, respectively.
2.4.3 Assay Buffer Optimization
ASFV DNA ligase was initially assayed in buffer similar to that which had been used in studies of Pol X (18): 50 mM Tris-borate, 50 mM KCl, 3% glycerol (v/v), pH→9.0 with
KOH at room temperature, supplemented with 10 mM MgCl2, 1 mM ATP, 100 µM DTT, and 100 µg/mL BSA. However, under these conditions the activity of ASFV DNA ligase was low and data was difficult to reproduce. Subsequently, a cursory pH titration demonstrated the enzyme to have optimal activity in the pH range 7.5-8.5, with activity decreasing dramatically above pH 8.75 (data not shown). Accordingly, pH was fixed at
7.8 and individual buffer components were titrated as described below.
17 “Base buffer” consisted of: 50 mM Tris-borate, 50 mM KCl, 3% glycerol (v/v), pH → 7.8 with KOH while at 37 °C, supplemented with 1 mM ATP, 10 mM MgCl2, 100
µM DTT, and 100 µg/mL BSA. Ligation of nicked G:C substrate, present at a concentration of 100 nM, was monitored at 37 °C in the titrations described below. The one exception was that in the KCl titration nicked C:G was used instead of nicked G:C.
Magnesium titration: base buffer but with MgCl2 at a total concentration of 5, 7.5, 10,
13.3, 16.6, or 20 mM. BSA titration: base buffer but with BSA at a concentration of 0,
0.05, 0.1, 0.2, 0.4, 0.8, 1.6, 4, 8, 16, 32, or 62.5 mg/mL. KCl titration: base buffer but with KCl at 25, 45, 65, 85, 105, 125, 145, 175, 215, 265, or 325 mM. Glycerol titration: base buffer but with glycerol at 0, 0.5, 1, 2, 4, 8, 16, or 32% (v/v), and 2 mg/mL BSA and
15 mM MgCl2. DTT titration: base buffer but with DTT at a concentration of 0, 0.05,
0.1, 0.3, 0.6, 2, 10, or 20 mM, and 2 mg/mL BSA and 15 mM MgCl2. Data for each of these buffer optimization titrations is shown below in Figure 2.5.
On the basis of these titrations, the following optimal buffer was chosen: 50 mM
Tris-borate, 100 mM KCl, 15% glycerol (v/v), pH→7.8 with KOH while at 37 °C, plus
15 mM MgCl2, 1 mM ATP, 300 µM DTT, and 1.5 mg/mL BSA. In addition to approximating physiological pH and potassium concentration, these conditions also afforded increased activity for ASFV DNA ligase and excellent reproducibility of kinetic data. T4 DNA ligase showed good activity in the optimized ASFV DNA ligase assay buffer, allowing the two enzymes to be studied under identical conditions.
18
A B 100 100
80 80
60 60
40 40 relative activity (%) activity relative (%) activity relative 20 20
0 0 0 5 10 15 20 0 102030405060
[MgCl2] (mM) [BSA] (mg/mL)
C D 100 100
80 80
60 60
40 40 relative activity (%) activity relative (%) activity relative 20 20
0 0 0 50 100 150 200 250 300 0 5 10 15 20 25 30
[KCl] (mM) % glycerol (v/v)
E
100
80
relative activity (%) activity relative 60
40 0 5 10 15 20
[DTT] (mM)
Figure 2.5 Optimization of DNA ligation assay buffer. ASFV DNA ligase was assayed for its ability to seal 100 nM nicked substrate as a function of [MgCl2] (A), [BSA] (B), [KCl] (C), % glycerol (D), and [DTT] (E). Details for each of these titrations are listed in the text.
19 2.5 Quantitation of Active Enzyme Concentration
It is often the case for recombinant enzymes that the UV-determined protein concentration is considerably higher than that determined by kinetic analyses, indicating that not all of the purified protein is active and/or that contaminant protein is present.
This technical difficulty is compounded for DNA ligases since the active (natively folded, undamaged) form of these enzymes exists as a mixture of a catalytically incompetent form (lacking the cofactor) and a catalytically competent form (with the
AMP cofactor covalently bound) (Figure 2.1). Upon overexpression and subsequent purification, the ratio of these forms can vary dramatically depending on the identity of the particular enzyme, the expression host, the method of purification, etc.
Most previous studies of DNA ligases haven’t required an accurate knowledge of the concentration of active enzyme (in either of its forms) and have simply used UV approximated values. To ensure that kcat values for DNA ligation could be reported as accurately as possible, A. Showalter (of the Tsai lab) previously developed novel assays for determining active enzyme concentrations. These assays, along with subsequent refinements, are described below. To our knowledge these represent the most rigorous, and also the most practical, methods available for determining the concentration of a
DNA ligase.
2.5.1 Forcing T4 DNA Ligase to “Burst” Using a 3’-NH2-Containing Nick
One approach for determining the total concentration of active DNA ligase would be to quantitatively pre-charge the enzyme and then incubate it with nicked DNA under conditions that either result in a single turnover or that force the enzyme to show burst
20 kinetic behavior. As shown in parts A and B of Figure 2.6, when incubating T4 DNA ligase with a 3’ amino-terminated nick (at 1180 nM), the second step of nick sealing
(DNA adenylylation) proceeds with high efficiency while the third step (nick sealing) is blocked altogether. This gives rise to biphasic kinetic behavior where the steady state phase is limited by dissociation of the enzyme•AMP-DNA complex or a conformational change associated with this. This data was fit to the burst equation: [product] = A[1-exp(- kobst)] + ct, where A is the burst amplitude, kobs is the observed single exponential rate constant, and c is the steady-state rate of transferring AMP from the AMP-ligase to nicked DNA. The steady state rate constant (kss) was obtained by dividing c by A. For
-1 -1 the plot in Figure 2.6, A = 60.6 nM, kobs = 1.63min , and kss = 0.016 min . Since the enzyme was preincubated with ATP prior to mixing with DNA (allowing all active protein molecules to become charged), and since 1180 nM 3’-amino-terminated substrate was empirically demonstrated to be sufficient for saturating T4 ligase (data not shown), the burst amplitude corresponds to the total concentration of active enzyme in this assay.
Multiplying the burst amplitude by the enzyme dilution factor for this assay (200-fold dilution) indicates that for this particular stock of T4 DNA ligase [E]active, total = 12.12 µM.
The 3’-amino-terminated oligonucleotide described above was synthesized by Pol
β-catalyzed incorporation of 2’-deoxy-3’-amino-cytidinetriphosophate into the appropriate gapped substrate. This 3’-amino-terminated 26mer was gel purified, quantitated, and then assembled into a nicked substrate as outlined in Section 2.4.1; it was similar to the G:C substrate shown in Figure 2.4 except that the 5’-32P label was placed on the downstream 19mer using PNK to catalyze the exchange reaction.
21
Figure 2.6 Burst assay to determine the total concentration of active T4 DNA ligase. (A) When using a 3’ amino-terminated nick, T4 DNA ligase still catalyzes transfer of the adenylyl group from the enzyme’s catalytic lysine to the 5’-phosphate of the nick. However, the final nick-sealing step is blocked – allowing for the adenylylated DNA intermediate to accumulate. (B) Biphasic behavior of T4 DNA ligase acting on a 3’ amino-terminated nick (present at 1180 nM). Data (●) were fit to the burst equation; kinetic parameters, included the empirically determined enzyme concentration, are listed in the text. The gel displayed in the inset shows accumulation of the adenylylated DNA intermediate as a function of time.
2.5.2 Ligation of a Pre-Adenylylated Nick by ASFV DNA Ligase
In contrast to T4 DNA ligase, ASFV DNA ligase adenylylated the 3’ amino-terminated substrate extremely inefficiently; these reactions typically proceeding to only ∼5% completion, and did not show burst behavior. Accordingly, an alternative set of three assays were employed for determining the total concentration of active enzyme. We say
22 that the total concentration of active (natively folded, undamaged) enzyme is equal to the sum of the active adenylylated and active unadenylylated forms:
[E]active, total = [E]active, adenylylated + [E]active, unadenylylated (equation 1)
The concentration of active, adenylylated enzyme was determined by the amplitude of a single turnover assay in which enzyme was incubated with 100 nM nicked G:C in the absence of exogenous ATP (Figure 2.7A). Data points were fit to a single exponential:
[product] = A[1-exp(-kobst)], where A is the single turnover amplitude (14.4 nM) and kobs
18 A 140 B 16 preincubation without DNA 120 nicked 14 preincubation with DNAnicked 12 100
10 80 8 60 [45mer] (nM) (nM) [45mer] 6 40 4 20 2
0 0 0 5 10 15 20 25 30 35 0246810 time (minutes) time (minutes)
Figure 2.7 Three assays to determine the concentration of active ASFV DNA ligase in both the adenylylated and unadenylylated forms. (A) Determination of the concentration of adenylylated ASFV DNA ligase. A single turnover was performed by incubating ASFV DNA ligase with 100 nM nicked G:C in the absence of ATP. The data (!) was fit to a single exponential; kinetic parameters, included the empirically determined enzyme concentration, are listed in the text. (B) Determination of the concentration of unadenylylated ASFV DNA ligase. Steady state ligation of 590 nM adenylylated DNA by 240 pM ASFV DNA ligase after a preincubation with (○; v0 total = 13.94 nM/min.) or without (●; v0 unadenylylated = 2.33 nM/min.) 450 nM unlabeled, nicked DNA. As outlined in the text, these initial velocities were used, along with the single turnover amplitude from part A, to determine the concentration of active, unadenylylated enzyme.
23 is the observed rate constant (1.99 min.-1). Since the enzyme was diluted 75-fold in this particular assay, this gives [E]active, adenylylated = 1.08 µM.
[E]active, unadenylylated was determined indirectly using synthetic, nicked DNA in which the 5’-phosphate was linked to the phosphate moiety of AMP. This substrate represents the DNA ligase reaction intermediate formed immediately prior to nick sealing
(Figure 2.1) and will be referred to as adenylylated DNA or adenylylated substrate. With the exception of the extra-helical 5’-adenylyl group, this substrate was similar to the G:C substrate shown in Figure 2.4. Parallel reactions with adenylylated DNA were performed in assay buffer lacking ATP and had the following format. ASFV DNA ligase was preincubated for 15 minutes in the absence (assay #1) or presence (assay #2) of 450 nM unlabeled, nicked DNA and then mixed with the 32P-labeled, adenylylated substrate. The final concentrations of enzyme and adenylylated DNA were 240 pM and 590 nM, respectively. In assay #1, both the adenylylated and the unadenylylated forms of the ligase persist after the preincubation step. Upon addition of adenylylated substrate, only the unadenylylated form of the enzyme shows turnover (the 5’ adenylyl moiety on the
DNA precludes binding by adenylylated ligase). From this assay we obtained an initial velocity, v0 unadenylylated = 2.33 nM/min (Figure 2.7B). In assay #2, the nicked DNA present during the preincubation effects deadenylylation of the ligase. Accordingly, upon addition of adenylylated substrate, all active ligase molecules show turnover, allowing determination of the initial velocity v0 total = 13.94 nM/min (Figure 2.7B). The ratio v0 total/v0 unadenylylated is equivalent to the ratio [E]active, total/[E]active, unadenylylated, allowing us to write the expression [E]active, total = n[E]active, unadenylylated, where n is the empirically determined proportionality constant. We have, v0 total/v0 unadenylylated = 5.98 = [E]active,
24 total/[E]active, unadenylylated. Substitution into equation 1, along with the [E]active, adenylylated value determined above, gives [E]active, unadenylylated = 217 nM. Accordingly, 83% of the enzyme existed in the adenylylated form after expression in E. coli and subsequent purification.
The concentration of unlabeled, nicked DNA used in the preincubation step described above was not arbitrarily chosen to be 450 nM. If present at too low a concentration this substrate would not effect complete deadenylylation of the ligase. If present at too high a concentration, the residual nicked molecules from the preincubation step might subsequently inhibit ligation of the labeled, adenylylated substrate. In order to determine what substrate concentration in the preincubation step would optimally
“activate” ASFV DNA ligase, the ratio v0 total/v0 unadenylylated was monitored for a series of reactions in which the concentration of unlabeled, nicked substrate in the preincubation step was varied between 0.6 and 1200 nM. Figure 2.8A suggests that when the substrate concentration during the 15 minute preincubation is below 450 nM, ASFV DNA ligase is not quantitatively deadenylylated. The decrease in v0 total/v0 unadenylylated on going from 450 to 1200 nM substrate is consistent with ligation of labeled, adenylylated substrate being inhibited by the unlabeled, nicked substrate molecules left over from the preincubation.
Accordingly, it appears that unadenylylated ASFV DNA ligase does interact with nicks, albeit with a reduced affinity.
The optimal concentration of unlabeled, nicked DNA to be used in the preincubation step is expected to vary from one DNA ligase to the next, being a function of the enzyme’s affinity for the different DNA species. To test this, an identical set of
25 A B 6 2.6
5 2.2
4
0 unadenylylated 1.8 0 unadenylylated V / V 3 / 0 total 0
0 total 1.4 V 2 V 1 1.0 0 200 400 600 800 1000 1200 0 200 400 600 800 1000 1200 [unlabeled, nicked DNA] during preincubation (nM) [unlabeled, nicked DNA] during preincubation (nM)
Figure 2.8 Empirical determination of the optimal nicked substrate concentration for deadenylylating DNA ligase prior to incubation with adenylylated substrate. Similar to the assay shown in Figure 2.7B, the initial velocity (v0) of ligating adenylylated DNA was monitored after enzyme was preincubated with unlabeled, nicked DNA (giving v0 total) or after enzyme was preincubated without unlabeled, nicked DNA (giving v0 unadenylylated). See text for details. The ratio v0 total/v0 unadenylylated is plotted as a function of the concentration of unlabeled, nicked substrate used during the preincubation step for ASFV DNA ligase (A) and T4 DNA ligase (B).
assays were performed with T4 DNA ligase (Figure 2.8B). Though the curves in parts A and B of Figure 2.8 have similar shapes, the T4 enzyme is optimally activated by 300 nM substrate during the preincubation (vs. 450 nM for the ASFV enzyme).
2.5.3 Comparison of the Two Ligase Quantitation Assays
In order to assess the accuracy of the two different ligase quantitation assays described above, concentration values for T4 DNA ligase are compared in Figure 2.9. The active concentration of T4 DNA ligase determined by these two methods differs by less than
2%. Such a high level of convergence from two independent methodologies suggests that both provide an accurate measure of active DNA ligase concentration.
26
[E] [E] [E] assay method active, adenylylated active, unadenylylated active, total (µM) (µM) (µM)
burst ------12.12
ligation of preadenylated 7.30 4.61 11.91 DNA
Figure 2.9 Comparison of active T4 DNA ligase concentrations determined by the two different ligase quantitation assays.
While it is well known that DNA ligases can bind to and seal adenylylated DNA
(32, 33), herein we took advantage of this property for the purpose of determining the concentration of active enzyme existing in both the adenylylated and the unadenylylated states. Because the adenylylated DNA substrate is a universal intermediate in the DNA ligation reaction mechanism, this particular assay is expected to be broadly applicable for determining the concentration of both ATP- and NAD+-dependent DNA ligases. In contrast, as suggested by the differences observed between ASFV and T4 DNA ligases,
3’ amino-terminated nicked substrates may give rise to burst kinetic behavior among only a subset of DNA ligases.
2.6 Analysis of DNA Ligation Fidelity
All work described up to this point was preparative, geared towards being able to evaluate in a quantitative manner whether the ASFV DNA ligase is capable of sealing 3’ mismatched nicks – which in vitro assays suggest are generated with high frequency during single nucleotide gap filling by Pol X. The DNA ligation assays that we conducted (vide infra) are the most exhaustive ever reported and, accordingly, not only
27 enable us to assesses the credence of the ASFV mutagenic DNA repair hypothesis, but also afford us the opportunity to begin evaluating the mechanisms by which DNA ligases discriminate between matched vs. mismatched substrates. In Section 2.6 we: i) summarize previous analyses of DNA mismatch ligation, ii) provide experimental details pertaining to our ligation fidelity studies, iii) present fidelity profiles and catalytic parameters for both ASFV and T4 DNA ligases, and iv) discuss the apparent determinants of DNA ligation fidelity.
2.6.1 Previous Studies
Enzymatic fidelity is generally defined as (kcat/KM)correct/(kcat/KM)incorrect, where “correct” and
“incorrect” denote the canonical substrate and a non-canonical substrate, respectively. When applied to DNA ligation, the fidelity parameter therefore assesses an enzyme’s ability to discriminate between nicks containing matched (Watson-Crick) vs. mismatched (non Watson-
Crick) base pairs.
Though the biological relevance of DNA ligation fidelity has until recently been questionable, the efficiency of ligating mismatched nicks has been studied semi-quantitatively for a large number of DNA ligases in the last ∼15 years. These studies of mismatch ligation have been rooted in two major interests: (i) in the in vitro diagnostic techniques of ligase detection reaction/ligase chain reaction, single nucleotide polymorphisms are detected based on the ability/inability of a DNA ligase to seal mismatched nicks located at the site of a suspected point mutation (the sensitivity of this technique is determined by the ability of the DNA ligase to discriminate between matches and mismatches either 3’, 5’, or both 3’ and 5’ to the nick) (34,
28 35); (ii) analysis of mismatch tolerance has proven useful for differentiating between the different human enzymes (36, 37).
To our knowledge, none of the previous analyses of DNA ligation fidelity have employed catalytic parameters; instead they have primarily compared the extent of ligation – for match vs. mismatch – after a given length of incubation. While this methodology is sufficient for the purposes mentioned above, it often dramatically overestimates the efficiency of mismatch ligation; at early, unmonitored time points the ratio [ligated match]/[ligated mismatch] may be very high, but given a long enough incubation period it will usually approach unity. In order to evaluate DNA ligation fidelity as accurately as possible, in this study we therefore used the steady state catalytic parameters kcat and KM – which are also advantageous in that they provide information about how the enzyme discerns match from mismatch.
2.6.2 ASFV and T4 DNA Ligases
2.6.2.1 Experimental Details and Primary Data
Saturation curves were obtained for each base pair by plotting the initial velocity (v0) of
DNA ligation as a function of nicked DNA substrate concentration and fitting the data to the Michaelis-Menten equation: v0 = Vmax[S]/(Km+[S]). kcat was obtained by dividing
Vmax by the enzyme concentration. Each saturation curve was repeated at least twice and the reported data represents the single best independent trial. Each saturation curve contained v0’s for seven different substrate concentrations, as shown in Figure 2.10. For most of the 32 saturation curves generated (16 for each enzyme) the highest substrate concentration used was greater than 5 x Km. Enzyme concentrations were varied
29
160 A C 140 100
120 80 100 60 80
60 [45mer] (nM) [45mer] (nM) 40 40 20 20
0 0 0 10203040 0 5 10 15 20 25 30 time (minutes) time (minutes)
B D 4 4
3 3
2 2 (nM/minute) (nM/minute) 0 0 V V 1 1
0 0 0 200 400 600 800 1000 1200 0 100 200 300 400 500 600 [nicked C:G] (nM) [nicked C:T] (nM)
Figure 2.10 Primary data and saturation curves for sealing nicked C:G and C:T by ASFV DNA ligase. (A) Time course for sealing nicked C:G. Substrate concentrations are as follows: 40 nM (●), 80 nM (○), 120 nM (■), 250 nM (□), 400 nM (!), 800 nM (") and 1,200 nM (▼). Lines are linear fits forced through the origin. (B) Initial velocities from (A) plotted as a function of substrate concentration. Data points (●) were fit to the Michaelis-Menten equation as outlined in the text. (C) Time course for sealing nicked C:T. Substrate concentrations are as follows: 40 nM (●), 60 nM (○), 80 nM (■), 120 nM (□), 240 nM (!), 460 nM (") and 680 nM (▼). Lines are linear fits forced through the origin. (D) Initial velocities from (C) plotted as a function of substrate concentration. Data points (●) were fit to the Michaelis-Menten equation as outlined in the text.
30 (between 52 pM and 19 nM for T4 ligase, and between 64 pM and 31 nM for ASFV
DNA ligase) for each substrate so that time points, corresponding to less than 30% turnover, could be taken in the range of 10 seconds to 1 hour.
2.6.2.2 Fidelity Profiles and Catalytic Parameters
As described in the previous section, ASFV DNA ligase was assayed in the steady state to generate saturation curves for nicked substrates containing all 16 possible base pair combinations at the 3’-OH side of the nick. Having previously been described as a low- fidelity DNA ligase (38) – on the basis of its limited discrimination against a 3’ G:T mismatch – T4 DNA ligase was also assayed to serve as a reference. For both of these enzymes, catalytic efficiency is plotted as a function of base pair in parts A and B of
Figure 2.11. kcat and KM values for each enzyme are also listed in Figures 2.12 and 2.13.
31
6e+7 A 5e+7 )
-1 4e+7 s -1
(M 3e+7 M K /
cat 2e+7 k
1e+7
0
T T T T A C G A C G A C A G C G : : : : : : : : : : : : T A T C G: T T G: C A A A C C G: G: base pair
B 1e+7
) 8e+6 -1 s -1 6e+6 (M M K /
cat 4e+6 k
2e+6
0
T T T T A C G A C G A C A G C G : : : : : : : : : : : : : : T A T C G: T T G: C A A A C C G G base pair
Figure 2.11 Catalytic efficiency of nick sealing as a function of 3’ base pair for T4 DNA ligase (A) and ASFV DNA ligase (B). Note the different scales for the y-axes.
32 a -1 -1 -1 b base pair kcat (s ) KM (nM) kcat/KM (M s ) fidelity A:T 3.5 ± 0.29 74 ± 15 4.7 x 107 ---- A:C 0.23 ± 0.011 280 ± 47 8.2 x 105 57 A:A 0.043 ± 0.0011 680 ± 62 6.3 x 104 750 A:Gc 0.049 ± 0.053 290000 ± 320000 1.7 x 102 280000
T:A 6.0 ± 0.22 170 ± 24 3.5 x 107 ---- T:T 0.26 ± 0.0066 150 ± 12 1.7 x 106 21 T:C 0.51 ± 0.030 1300 ± 240 3.9 x 105 90 T:G 0.22 ± 0.0090 1100 ± 140 2.0 x 105 180
G:C 4.9 ± 0.23 160 ± 22 3.1 x 107 ---- G:T 2.9 ± 0.11 640 ± 62 4.5 x 106 6.9 G:G 0.46 ± 0.018 1600 ± 140 2.9 x 105 110 G:A 0.015 ± 0.0014 2600 ± 570 5.8 x 103 5300
C:G 7.2 ± 0.10 220 ± 8.7 3.3 x 107 ---- C:T 3.2 ± 0.075 940 ± 59 3.4 x 106 9.7 C:A 0.51 ± 0.035 750 ± 140 6.8 x 105 49 C:C 0.048 ± 0.0056 3500 ± 860 1.4 x 104 2400
Figure 2.12 Kinetic parameters for nick ligation by T4 DNA ligase as a function of 3’ base pair identity. a In the base pair notation X:Y, X refers to the templating nucleotide, and Y is the nucleotide that would have been inserted by a polymerase in the preceding step of BER (see Figure 2.4 for structures). b Fidelity c is defined as: (kcat/KM)correct /(kcat/KM)incorrect. T4 DNA ligase seals the A:G mismatch extremely inefficiently. At 10 µM A:G substrate, the enzyme was nowhere close to being saturated. Thus, the kcat and KM values reported for this substrate were obtained by considerable extrapolation.
a -1 -1 -1 b base pair kcat (s ) KM (nM) kcat/KM (M s ) fidelity A:T 0.80 ± 0.089 180 ± 50 4.4 x 106 ---- A:C 0.28 ± 0.0090 290 ± 33 9.7 x 105 4.5 A:A 0.041 ± 0.00065 2100 ± 77 2.0 x 104 220 A:G 0.00093 ± 0.000050 1600 ± 270 5.8 x 102 7600
T:A 0.81 ± 0.058 130 ± 27 6.2 x 106 ---- T:T 0.27 ± 0.023 580 ± 160 4.7 x 105 13 T:C 0.37 ± 0.025 240 ± 53 1.5 x 106 4.1 T:G 0.27 ± 0.014 510 ± 110 5.3 x 105 12
G:C 1.0 ± 0.04 200 ± 30 5.0 x 106 ---- G:T 0.35 ± 0.028 1000 ± 200 3.5 x 105 14 G:G 0.24 ± 0.0047 2100 ± 83 1.1 x 105 45 G:A 0.014 ± 0.00068 1100 ± 210 1.3 x 104 380
C:G 0.67 ± 0.017 180 ± 14 3.7 x 106 ---- C:T 1.2 ± 0.043 120 ± 11 1.0 x 107 0.37 C:A 0.16 ± 0.011 1000 ± 230 1.6 x 105 23 C:C 0.059 ± 0.0015 1400 ± 99 4.2 x 104 88
Figure 2.13 Kinetic parameters for nick ligation by ASFV DNA ligase as a function of 3’ base pair identity. a, b Base pair notation and fidelity are the same as in Figure 2.12.
33 2.6.2.3 Assessment of Data Reproducibility
As discussed below, the fidelity values for ASFV DNA ligase are extraordinary. In order to assess the precision of these values, the activity of ASFV DNA ligase against nicked
C:G and C:T was reexamined using substrates assembled from different preparations of the component oligonucleotides (identical sequences, but the oligonucleotide stocks were different than those used in generating data for Figure 2.13). From this set of assays the fidelity of C:T ligation by ASFV DNA ligase was determined to be 0.41. This compares very well with the value of 0.37 in Figure 2.13, confirming the C:T preference of ASFV
DNA ligase and suggesting a precision of roughly ± 10% for the reported fidelity values.
2.6.2.4 Data Analysis
When compared with T4 DNA ligase (Figure 2.12), ASFV DNA ligase (Figure 2.13) displays lower fidelity for sealing 11 of the 12 possible mismatched nicks. Especially salient, the ASFV enzyme ligates the C:T mismatch three-fold more efficiently than the corresponding Watson-Crick base pair, C:G (Figures 2.11B and 2.13). This preference for C:T results from both a higher kcat and a lower KM, relative to C:G. This is the first reported example of a DNA ligase that preferentially seals a mismatched nick.
Comparison with previous, more qualitative, examinations of DNA ligase fidelity (36,
38) indicates that ASFV DNA ligase has, by far, the lowest fidelity of any DNA ligase reported to date.
A few noteworthy trends emerging from the data in Figures 2.11-2.13 are as follows. First, T4 DNA ligase seals correctly matched nicks between 6- and 11-fold more efficiently than ASFV DNA ligase; this difference is largely attributable to the
34 higher kcat values of the T4 enzyme. Second, base pair size/geometry appears to be an important determinant of ligation efficiency. Consistent with what has been published for a diverse set of DNA ligases (37, 39-41), ASFV and T4 DNA ligases seal the bulky 3’ purine:purine mismatches (A:A, G:G, G:A and A:G) very inefficiently. In contrast, the purine:pyrimidine, pyrimidine:purine, and pyrimidine:pyrimidine mismatches tend to be better tolerated. Though the literature shows that the preferred 3’ mismatch varies from one enzyme to the next, G:T, T:G, C:T, and T:C have generally proven to be well tolerated (37, 39, 40), and this is also the case for the two enzymes reported on here.
Note, however, that the efficiency of sealing the two permutations of a given base pair, such as the mismatched C:T vs. T:C, can vary considerably – indicating that simple base pair size/geometry is not the sole determinant of ligation efficiency.
On the basis of our kinetic data, it is clear that discrimination against mismatches occurs both at the level of nick binding as well as during the chemical steps (one or both of the adenylyl transfers involving the DNA nick; we cannot discriminate between these since for most assays the 32P label was placed on the upstream oligo – preventing detection of the adenylylated DNA intermediate). This is particularly evident among the four purine:purine mismatches and C:C – all of which display extremely high KM’s and low kcat’s with both T4 and ASFV DNA ligases.
Mismatch discrimination at the level of DNA binding has recently been examined using nucleotide analogs with Tth DNA ligase. Liu and coworkers demonstrated that, similar to DNA polymerases, the relative positioning of minor groove hydrogen-bond acceptors (the N3 nitrogen of the purines and the C2 carbonyl of the pyrimidines) is a critical factor in deciphering between match vs. mismatch at the 3’ base pair of a nick
35 (42). It will be interesting to see whether this is a universal fidelity determinant among
DNA ligases, and if so, how the ASFV enzyme has redirected its specificity in order to accommodate 3’ mismatches while still sealing the canonical Watson-Crick base pairs.
Inversion of stereochemistry during the third step (nick-sealing) of DNA ligation by T4 ligase (43) is consistent with an associative in-line attack of the 5’-phosphate by the 3’-oxygen; it is assumed that all ligases operate by a similar mechanism. The relative positions of the 5’-phosphate and the 3’-hydroxyl are expected to fluctuate with the identity of the 3’ base pair. Since ASFV DNA ligase displays an enhanced kcat with the pyrimidine:pyrimidine mismatch C:T (relative to C:G), it would appear that this enzyme repositions these reactive groups on the annealed, nicked substrate in order to facilitate chemistry. The mechanism by which this is accomplished is of particular interest.
2.6.2.5 Importance of Ionic Strength
Lindahl and coworkers demonstrated that the concentration of KCl used in their assay buffer dramatically influenced the fidelity of DNA ligation for human DNA ligases I and
III (36). We have found this to also be true for ASFV DNA ligase; the fidelity for sealing the C:T mismatch is higher at 150 mM KCl than it is at 100 mM KCl. However, even at
150 mM KCl, C:T is still sealed more efficiently than C:G (data not shown). Though a number of examples of 3’ mismatch tolerance by a DNA ligase have been described in the literature, these studies have been conducted at low, non-physiological ionic strength
(37, 39, 40, 44, 45). Despite the fact that these artificial conditions often enhance mismatch ligation efficiency, in no case did the ligase being studied actually seal a
36 mismatch preferentially – highlighting the uniqueness of the ASFV enzyme described in this report.
2.6.2.6 An Active Site Point Mutation
Phylogenetic analysis of known DNA ligases demonstrates a distinct bifurcation between the ATP- and the NAD+-dependent enzymes. Despite their low overall sequence homology both of these enzyme families belong to the covalent nucleotidyltransferase superfamily and share a conserved active site primary and tertiary structure known as the nucleotidyltransferase domain (46). This domain provides many of the residues involved in catalysis while a C-terminal oligonucleotide binding (OB) domain contributes to both
DNA binding as well as catalysis [reviewed in (46)].
Comparison of the nucleotidyltransferase domain of ASFV DNA ligase with that of other DNA ligases reveals a striking difference. As noted by the Dixon lab in 1992
(30), the ASFV enzyme contains an asparagine residue at a position in nucleotidyltransferase motif I that is conserved as aspartate in other DNA ligases (Figure
2.14); we have been unable to identify another ligase containing this point mutation. This divergence in the ASFV ligase primary structure begs the question of whether this particular residue contributes to the enzyme’s markedly low fidelity. Accordingly, the
ASFV DNA ligase N153D mutant was constructed, purified, and assayed as described for
WT protein.
In Figure 2.15 we juxtapose catalytic parameters for the WT and N153D proteins acting on nicked C:G and C:T substrates. An active enzyme concentration was not determined for the mutant, so Vmax is used in place of kcat; the same concentration of
37
Figure 2.14 Alignment of motif I from the nucleotidyltransferase domain of representative NAD+- and ATP-dependent DNA ligases. “N” followed by a number indicates the distance from the N-terminus to the shown sequence. The asterisk marks the conserved catalytic lysine. The residue two positions downstream of this lysine is aspartate in all known DNA ligases, except ASFV DNA ligase – in which it is asparagine.
a -1 -1 -1 b enzyme base pair kcat (s ) KM (nM) kcat/KM (M s ) fidelity WT C:G 0.67 ± 0.017 180 ± 14 3.7 x 106 ---- WT C:T 1.2 ± 0.043 120 ± 11 1.0 x 107 0.37
a -1 -1 c enzyme base pair Vmax (nM s ) KM (nM) Vmax/KM (s ) fidelity N153D C:G 3.27 ± 0.02 570 ± 11 5.7 x 10-3 ---- N153D C:T 6.76 ± 0.13 750 ± 40 9.1x 10-3 0.64
Figure 2.15 Kinetic parameters for nick ligation by WT and N153D ASFV DNA ligase. a In the base pair notation X:Y, X refers to the templating nucleotide, and Y is the nucleotide that would have been inserted by a polymerase in the preceding step of BER b c (see Figure 2.4 for structures). Fidelity is defined as: (kcat/KM)correct /(kcat/KM)incorrect. In this case only, we define fidelity as (Vmax/KM)correct /(Vmax/KM)incorrect.
38 mutant protein was used in assays against both C:G and C:T, and here we calculate fidelity as (Vmax/KM)correct/(Vmax/KM)incorrect. The N153D mutation reduces the enzyme’s affinity for both the matched nick (by 3-fold) and the mismatch nick (by 6-fold).
However, the ratio Vmax, correct/Vmax, incorrect for the mutant (0.48) is similar to the ratio kcat, correct/kcat, incorrect for the WT protein (0.56). The overall result is that the N153D mutant is in fact more “faithful” (i.e. shows higher discrimination against the C:T mismatch) than the WT protein. However, relative to DNA ligases from other organisms, the N153D mutant still displays extremely low fidelity – suggesting that this particular residue contributes minimally to the unique mismatch tolerance of ASFV DNA ligase.
2.6.3 Synopsis of the Mechanism of DNA Ligation Fidelity with Consideration of Protein Structure
Participation of an error-prone DNA polymerase in DNA repair would give rise to nicks containing 3’-OH mismatched base pairs (as opposed to 5’-Pi mismatches); accordingly we have focused our analyses on 3’ mismatch ligation exclusively. For a DNA ligase, discrimination against 3’ mismatched nicks might conceptually occur at three different stages of the ligation reaction. That nick sensing/binding would occur without any interaction between the protein and the 3’ base pair seems highly unlikely, so differentiation between match vs. mismatch might first take place at the level of DNA binding. Second, a ligase, once bound to a nick, might adenylylate the 5’-phosphate with different efficiencies depending on whether the 3’ base pair is matched or mismatched.
The reason for this, at first glance, is not immediately obvious since the positioning of the
5’ phosphate should be relatively immune to the identity of the 3’ base pair. However,
39 the 3’-OH of the nick appears to be a critical component of the active site architecture during adenylylation of the 5’-phosphate – evidenced by the fact that 3’ dideoxy and 3’ amino-terminated nicks have been found to be adenylylated with low efficiency (28, 47).
Third, once nick adenylylation has occurred, a ligase might discriminate between 3’ matched vs. mismatched base pairs during the final, nick sealing step. The position of the
3’-OH, relative to the adenylylated 5’-phosphate, is expected to vary as a function of 3’ base pair identity; accordingly, the efficiency of the final, nick sealing step is expected to vary with 3’ base pair identity – unless the enzyme is capable of repositioning these reactive moieties. As described in previous sections, and immediately below, it appears that DNA ligases discriminate against mismatches at all three stages of the reaction.
The structures listed in Figure 2.16 have enabled a molecular understanding of cofactor recognition and covalent adduct formation. Large conformational changes associated with enzyme “charging” have also been elucidated. However, information about specific interactions with the nicked DNA substrate, and therefore about the structural determinants of fidelity, have only become available in the past six months.
In the structure of human DNA ligase I bound to an adenylylated nick (48), the protein shows minimal interaction with the 3’ base pair. This suggests that the protein does not “actively” discriminate between match/mismatch during the third, nick closing step and that the efficiency of this step ought to depend largely on the relative positions of the 3’-OH and 5’-Pi effected by the local DNA structure. When assayed against the 3’
T:C and T:G mismatches, which ligase I seals with moderate efficiency, the adenylylated
DNA intermediate is not detected (37). Consistent with the conclusion drawn from the complex structure, this kinetic result suggests that, at least for these two mismatches,
40
organism protein Enzyme PDB # date details ref. type deposited (cofactor)
bacteriophage DNA ligase ATP 1A0I 12/1/1997 -non-covalent complex with ATP (49) T7
Bacillus DNA ligase NAD+ 1B04 11/16/1998 -free adenylylation domain only (50) stearothermo- adenylylation philus domain
Thermus DNA ligase NAD+ 1DGS 11/25/1999 -enzyme-AMP covalent adduct (51) filiformis -Se-Met substituted
Chlorella DNA ligase ATP 1FVI 9/20/2000 -enzyme-AMP covalent adduct (28) virus -a sulfate group is bound in a position that may reflect the 5’ phosphate of a nick; accordingly it is suggested that this reflects the structure after nick binding
Homo sapiens DNA ligase IV ATP 1IK9 5/3/2001 -just a fragment of ligase IV bound to an (52) XRCC4 dimer
Homo sapiens DNA Ligase ATP 1IMO 5/11/2001 -structure is only of the BRCT domain, (53) IIIα BRCT (s) which is known to bind the BRCT domain domain of XRCC1 in base excision repair
Homo sapiens DNA Ligase ATP 1IN1 5/11/2001 -this is the family of 20 structures, which (53) IIIα BRCT (s) were used to generate the minimized, domain average structure 1IMO
Thermus DNA ligase NAD+ 1L7B 3/14/2002 -BRCT domain only ----- thermophilus BRCT domain (s) -to date, no paper has been published
Chlorella DNA ligase ATP 1P8L 5/7/2003 -enzyme-AMP covalent adduct (54) virus -similar to 1FVI, but without sulfate – allowing for comparison of structures before and after nick binding
Enterococcus DNA ligase NAD+ 1TA8 5/19/2004 -major conformational changes are seen (55) faecalis during the synthesis of NAD+ from NMN and AMP within the crystal lattice -(open form)
Enterococcus DNA ligase NAD+ 1TAE 5/19/2004 -major conformational changes are seen (55) faecalis during the synthesis of NAD+ from NMN and AMP within the crystal lattice -(closed form)
Homo sapiens DNA Ligase ATP 1UW0 1/27/2004 -zinc-finger domain only (56) IIIα zinc finger (s) domain
Thermus DNA ligase NAD+ 1V9P 11/25/1999 -enzyme-AMP covalent adduct (51) filiformis -same as 1DGS, except this is not Se-Met substituted (in paper this is listed as 1DGT)
Homo sapiens DNA ligase I ATP 1X9N 8/23/2004 -protein bound to adenylylated nick (48)
Figure 2.16 Summary of DNA ligase structures solved to date. (s) denotes a solution structure; all other structures were determined by X-ray.
41 once adenylylation of the nick occurs, the subsequent nick closure step is rapid (i.e. discrimination does not occur during step 3). Since ligase I is a high fidelity enzyme
(36), it must sense match vs. mismatch at earlier steps of the reaction (i.e. DNA binding and/or nick adenylylation).
Since the above mentioned T:C and T:G assays did not accumulate the adenylylated DNA intermediate (37), and since ligase I does not interact extensively with the 3’ base pair during the nick sealing step, it would appear that the global distortion the protein imposes on the nicked duplex (48) results in the 3’-OH and the adenylylated 5’-Pi being in positions appropriate for chemistry even when the 3’ base pair is a mismatch.
Are there 3’ base pairs, such as the bulkier purine:purine mismatches, for which this is not the case? What would happen if this enzyme were assayed against a 3’ G:G mismatch containing a pre-adenylylated 5’-phosphate? If it were able to bind this substrate, would the final chemical step proceed efficiently?
In contrast to human ligase I, human ligase III does appear to attain to fidelity, at least in part, by discerning match from mismatch during the final, nick sealing step. This protein ligates the T:C and T:G mismatches with reduced efficiency relative to T:A, with concomitant accumulation of the adenylylated intermediate for the mismatches only (37); discrimination against mismatches during the nick sealing step has similarly been observed with the Vaccinia virus DNA ligase (40).
We see throughout Section 2.6 token examples of ligation fidelity being effected, at least in part, at three different stages of the nick sealing reaction. Tth, ASFV, and T4
DNA ligases can discriminate at the level of DNA binding. DNA ligase I does not discriminate strongly during the third step and therefore – by default – must discriminate
42 at either the DNA binding or the nick adenylylation steps. DNA ligase III discriminates, at least partially, during the third step. The details of fidelity enforcement are likely to vary from one enzyme to the next and from one base pair to the next. It is easy to imagine, for example, that a bulky purine:purine mismatch might be selected against at all three steps of the reaction, while a purine:pyrimidene mismatch might be selected against at just the DNA adenylylation step.
2.7 ASFV DNA Ligase vs. Pol X: Comparison of Specificity Profiles and Implications for the Mutagenic Repair Hypothesis
Nature’s selective pressure is ultimately exerted on enzyme systems. Since the genes encoding ASFV Pol X and DNA ligase have presumably coevolved, comparing the mismatch specificities of these two enzymes may provide clues about the selective pressure that has promoted low-fidelity in ASFV DNA ligase. If ASFV DNA ligase has evolved towards low-fidelity for the purpose of sealing the replisome-blocking lesions
(mismatched nicks) generated by Pol X, then its mismatch specificity is expected to be similar to that of Pol X. In this scenario, since persistently unligated mismatched nicks could block genome replication and be lethal to the virus, the ligase should seal most efficiently those mismatches being formed the most frequently by Pol X; as a corollary, mismatches synthesized infrequently by Pol X would be sealed less efficiently by the ligase. While this sort of system would give rise to the largest number of replicatable genomes (i.e. those lacking nicks), it wouldn’t generate broad sequence diversity since some sealed mismatches would be over represented and others underrepresented.
Alternatively, if ASFV DNA ligase has evolved towards low-fidelity for the purpose of
43 promoting broad genomic diversity, then the mismatch specificities of these two enzymes are expected to complement one another; those mismatches synthesized least efficiently by Pol X would be sealed very efficiently by ASFV DNA ligase since only in this manner would the entire spectrum of mismatched duplex products be well represented. Note that
ASFV does not encode a mismatch repair system or a mismatch-specific glycosylase (16) that could repair mismatches after they had been ligated.
Figure 2.17 demonstrates very clearly that the mismatch specificity of ASFV
DNA ligase does not mirror the mismatch specificity of Pol X. In contrast, a modest complementation in the mismatch specificities of these two enzymes is observed – with the ligase efficiently sealing the mismatches generated least efficiently by Pol X.
1e+7 Pol X 1000 ASFV DNA ligase ) -1
1e+6 ) s -1 -1 100 s -1 (M (M M Pol X 1e+5 K / d,app /K cat k ASFV DNA ligase DNA ASFV
pol 10 k 1e+4
1
T T T G G A G A A C C C : : : : : : : : : : : : T T G C C A T A G A G C base pair
Figure 2.17 Comparison of the catalytic efficiencies of Pol X (for mismatch synthesis) and ASFV DNA ligase (for mismatch ligation). Base pair are arranged, from left to right, in order of decreasing Pol X catalytic efficiency. Note the different scales for the y-axes.
44 With perhaps even greater bearing on the credence of the mutagenic DNA repair hypothesis is the fact that the catalytic efficiency of ASFV DNA ligase far exceeds that of Pol X – regardless of the base pair (note the different scales for the y-axes in Figure
2.17). On the basis of the in vitro data presented up to this point, it seems inevitable that base substitutions would occur at a much higher frequency during ASFV-mediated DNA repair than in any other canonical DNA repair system characterized to date.
2.8 Low Fidelity DNA Ligation in Other Systems
Our results raise the important question of whether low-fidelity DNA ligation is unique to
ASFV. A candidate process in which an aberrant DNA ligase might be useful or necessary is somatic hypermutation (SHM). While error-prone DNA polymerases are responsible for the generation of point mutations in SHM, the mechanistic details await clarification (57). Since BER and mismatch repair are likely routes for processing SHM intermediates (57), the generation of mismatched nicks is at least conceivable. If this is in fact the case, the protein or protein complex responsible for ligating these nicks has likely not yet been characterized since human DNA ligases I, III, and IV do not seal mismatched nicks efficiently (36) (Y. Wang, B. Lamarche, and M.-D. Tsai, unpublished results).
45 CHAPTER 3
ASFV AP ENDONUCLEASE
3.1 AP Endonucleases in DNA Repair
Despite the prophylactic affects of compartmentalization within the nucleus (58, 59) and packaging into protein-bound higher order structures (60, 61), DNA is still highly susceptible to damage by chemical [alkylating agents (62), for example] and physical
[UV light (63) and ionizing radiation (64)] assaults from the environment.
Notwithstanding these exogenous sources of damage, normal metabolism can give rise to reactive oxygen species (hydroxyl radical, superoxide, etc.) which are also capable of destroying the genetic blueprint (59).
As mentioned in Section 1.5, a large percentage of DNA lesions involve modification of the nitrogenous bases [deamination, oxidation, ring fragmentation, etc.
(59)] and are processed via base excision repair (BER) (65), the defining feature of which is removal of the chemically modified base by a glycosylase to generate an apurinic/apyrmidinic (AP) site (see Figure 1.1 for the chemical structure of an AP site).
In addition to being generated by glycosylases, AP sites also arise spontaneously, primarily via depurination. It is estimated that on average, the genome of a human cell
46 sustains up to 10,000 such depurination events per day (66). If they linger, AP sites can themselves be mutagenic and/or lethal; accordingly, cells devote considerable energy to processing them. In humans, for example, the primary AP site repair enzyme, Ape1, is present at a concentration of ∼300,000-7,000,000 copies/cell (67, 68).
Though repair of AP sites can be initiated by both AP endonucleases (APEs) and
AP lyases (as detailed in Section 1.5 and Figure 1.1), the APE route likely predominates in vivo [reviewed in Chapter 4 of (69)]. Phylogenetic analysis of known APEs produces a distinct bifurcation between a set of enzymes resembling E. coli exonuclease III1 and a set of enzymes resembling E. coli endonuclease IV. Despite their small size (∼28-32 kD), members of each of these two protein families typically posses more than one repair activity. A succinct overview of exonuclease III and endonuclease IV is provided in
Figure 3.1.
enzyme (gene) exonuclease III (xth) endonuclease IV (nfo) MW ∼28 kD ∼31.5 kD known activities •AP endonuclease •AP endonuclease •3’→5’ exonuclease (robust) •3’→5’ exonuclease (at nicks) •3’-phosphatase •3’-phosphatase •3’-phosphodiesterase •3’-phosphodiesterase •RNase H •RNase H •Nucleotide Incision Repair cofactor •absolute requirement for Mg2+ •contains a tri-zinc center requirements •inhibited by EDTA •stimulated by select transition metals misc. •the major APE in E. coli •∼10% of the APE activity in E. coli •inducible by oxidative DNA damage homologues •human Ape1 • Saccharomyces cerevisiae APN1 •Saccharomyces cerevisiae APN2 •ASFV APE
Figure 3.1 Properties of E. coli exonuclease III and endonuclease IV.
1 This apparent misnomer is rooted in the fact that the 3’→5’ exonuclease activity of this enzyme was discovered before its endonuclease activity. 47 3.2 ASFV Gene E296R: Identification and Preliminary Analysis
Upon sequencing the entire genome of ASFV strain BA71V (an avirulent isolate adapted to grow in Vero cells) in 1995, Yanez and coworkers described a gene – which they designated E296R – encoding a 33.5 kD polypeptide with homology to AP endonucleases
(16). Note that this gene was also identified in the partial genome sequence of the virulent Malawi LIL20/1 ASFV isolate (70). Though the Salas group casually stated in
2003 that this gene is needed for ASFV viability when infecting porcine macrophages
(71), the protein product of this gene – which we will refer to simply as ASFV APE – has not been characterized to date.
Phylogenetic analysis of ASFV APE indicates that it is most similar to endonuclease IV-like enzymes (Figure 3.2). While this homology implicates specific
DNA repair activities, the nuances of enzymatic catalysis – particularly for a
Figure 3.2 Phylogenetic analysis of representative AP endonucleases. Sequence alignments and the subsequent dendrogram were generated using MultAlign (72), available at: http://prodes.toulouse.inra.fr/multalin/multalin.html.
48 multifunctional protein – can only be understood empirically. In order to evaluate the potential contribution of ASFV APE to viral DNA repair, the work described below was undertaken.
3.3 Cloning, Protein Expression and Purification
ASFV gene E296R was amplified from a mixture of BASE and BASH (two separate clones containing genomic fragments from ASFV isolate BA71V; these were obtained from Linda Dixon, Institute for Animal Health, Pirbright Laboratory, UK) and ligated into the NdeI and XhoI sites of pET30; both a His-tagged and a non-tagged construct were generated.
Since the ASFV APE gene contains multiple codons that are used with low frequency in E. coli, Rosetta(DE3)pLysS cells were used as the expression host. As a preliminary expression analysis, cells containing the non-tagged pET30-ASFV APE were grown in 500 mL LB at 37 °C to mid log phase, induced with 1 mM IPTG, and were further incubated at 37 °C, 250 rpm. At the indicated time points post induction, aliquots were removed and diluted 5-fold with SDS loading buffer. Without being boiled, samples were run on an SDS 4-20% polyacrylamide gradient gel and stained with
GelCode Blue (Figure 3.3). Despite the uninduced sample being under-loaded, it is clear that a band of ∼29.9 kD is induced; in Section 3.4 we demonstrate that this does in fact correspond to full length ASFV APE. Though the relative intensity of this band doesn’t appear to change dramatically between 2 and 8.25 hours post induction, a ∼24.8 kD band
49
Figure 3.3 Analysis of ASFV APE expression in Rosetta(DE3)pLysS cells. Crude cell lysate is shown as a function of time post induction. Staining was with GelCode Blue.
increases in intensity as a function of time – suggesting that ASFV APE may be prone to proteolysis. On the basis of this observation subsequent induction periods were conducted at 30 °C.
Non-tagged ASFV APE could be purified to >90% homogeneity using DEAE, P-
11, hydroxyapatite, and S-100 chromatographies, in that order. However, the His-tagged construct displayed similarly high levels of expression and could be purified more quickly and with higher yield and higher purity than the non-tagged construct; accordingly, His-tagged ASFV APE was used exclusively in all the experiments described herein. A protocol for purification of His-tagged ASFV APE is provided below.
All purification steps were conducted on ice or at 4 °C. Cell lysis was by sonication in buffer A (Figure 3.4) supplemented with lysozyme, ROCHE protease
50 Contents Buffer A 50 mM NaPi, 300 mM NaCl, 10% glycerol, pH 8.0, 5 mM imidazole, 1 mM BME Buffer B 50 mM NaPi, 300 mM NaCl, 10% glycerol, pH 8.0, 1% NP-40, 5 mM imidazole, 1 mM BME Buffer C 50 mM NaPi, 300 mM NaCl, 10% glycerol, pH 7.0, 1% NP-40, 10 mM imidazole, 1 mM BME Buffer D 50 mM NaPi, 1 M NaCl, 10% glycerol, pH 7.0, 1% NP-40, 10 mM imidazole, 1 mM BME Buffer E 50 mM NaPi, 1 M NaCl, 10% glycerol, pH 7.0, 10 mM imidazole, 1 mM BME Buffer F 50 mM NaPi, 300 mM NaCl, 10% glycerol, pH 7.0, 10 mM imidazole, 1 mM BME Buffer G 50 mM NaPi, 300 mM NaCl, 10% glycerol, pH 7.0, 80 mM imidazole, 1 mM BME Buffer H 50 mM NaPi, 300 mM NaCl, 10% glycerol, pH 7.0, 170 mM imidazole, 1 mM BME
Figure 3.4 ASFV APE purification buffers. Abbreviations are as follows: NaPi, NaH2PO4; BME, β-mercapto ethanol.
inhibitor cocktail, and 100 µM AEBSF and PMSF. After pelleting cellular debris the clarified lysate was “batched” with 20 mL of Talon cobalt resin slurry in a centrifuge bottle; this involved gentle rocking for 25 minutes. After pelleting the resin, supernatant was aspirated and the resin batched a second time with 250 mL of fresh buffer A in analogous manner. Single batching steps in 350 mL of buffer B, and then in buffer C followed. The resin was then transferred to a column already containing a few mL of fresh, settled resin. This column was washed with 250 mL each of buffers D, E, F, and
G, in that order. Finally, elution was with 250 mL buffer H, collecting 20-30 mL fractions. ASFV APE eluted in buffer H at >95% purity (Figure 3.5). Eluate was concentrated by ultrafiltration and then dialyzed into storage buffer: 50 mM Tris-borate,
100 mM KCl, 5% glycerol (v/v), 5 mM DTT, pH→8.0 with KOH. In previous purification efforts, ASFV APE was found to precipitate when being dialyzed directly from the cobalt elution buffer (buffer H) into storage buffer. Accordingly, this transition was made gradually through a series of six different buffers in which the buffer identity, salt concentration and identity, and pH were adjusted incrementally. This dramatically
51
Figure 3.5 Silver-stained gels summarizing the purification of ASFV APE on cobalt resin. See Figure 3.4 for the composition of wash and elution buffers.
reduced protein precipitation. Glycerol was added to 40% (v/v) before flash freezing aliquots in N2 (l) and storing at –80 °C. Subsequently, working aliquots were kept at –20
°C. Protein prepared by the above protocol was used for all assays described in this and subsequent chapters, except those focusing on metal cofactor usage (described in Section
3.6).
E. coli exonuclease III and endonuclease IV represent robust activities which might conceivably contaminate ASFV APE preparations derived from
Rosetta(DE3)pLysS cells. To ensure that repair activity could be unequivocally attributed to ASFV APE and not these potential contaminants, ASFV APE was also expressed in BW528 cells (which are deficient in both exonuclease III and endonuclease
IV), obtained from Bernard Weiss at Emory University. Since heterologous protein expression in this double mutant strain was expected to be low, two modifications were made: i) an IPTG inducible T7 RNA polymerase gene was inserted into the bacterial genome using Novagen’s λDE3 lysogenization kit (T7 RNA polymerase is roughly 5- fold more efficient than its E. coli counterpart and it only transcribes genes preceded by a
T7 specific promoter – meaning that its activity is focused solely on transcription of the
52 gene of interest); and ii) the BW528(DE3) cells were then transformed with the pRAREpLysS plasmid (which supplies the cell with tRNAs that are normally present in low copy number; this plasmid was isolated from Rosetta(DE3)pLysS cells). In order to express ASFV APE from BW528(DE3)pRAREpLysS – which are kanamycin resistant
(owing to insertional inactivation of endonuclease IV) and chloramphenicol resistant (due to the pRAREpLysS plasmid) – the His-tagged viral gene had to be subcloned from pET30 to pET21, again using the NdeI and XhoI restriction sites. This expression scenario ultimately resulted in yields of ASFV that were not markedly reduced relative to
Rosetta(DE3)pLysS cells. However, after running an initial cobalt column as described above, the protein consistently required further purification. These additional steps are described below.
ASFV APE-containing cobalt fractions were pooled and diluted ∼5-fold using P-
11 buffer: 50 mM Tris-HCl, 75 mM KCl, 10% glycerol (v/v), pH→8.0 with KOH at 4
°C, 10 mM DTT. The diluted sample was loaded onto a P-11 column (∼40 mL bed volume) that had been equilibrated with the same buffer, and was then washed with an additional 350 mL of buffer. Elution was in the same buffer using a linear KCl gradient:
300 mL of 75 mM KCl + 300 mL of 1 M KCl. The column flow rate was 0.7 mL/minute. ASFV APE-containing fractions, spanning roughly 200-380 mM KCl, were concentrated in 5 kD NMWL centrifugal filters and applied to an S-100 column equilibrated with 50 mM Tris-HCl, 150 mM KCl, 10% glycerol (v/v), pH→8.0 with
KOH at 4 °C, 10 mM DTT. The flow rate was 0.85 mL/minute, with ASFV APE eluting at an apparent MW of 25.9 kD (note that the theoretical MW for this His-tagged construct
53 is 34.5 kD). S-100 eluate was loaded directly onto a hydroxyapatite column (∼10 mL bed volume) that had been equilibrated with 20 mM NaH2PO4, 10% glycerol (v/v), pH→7.3 with NaOH while at 25 °C, and 10 mM DTT. Washing and elution steps were conducted in the same buffer, containing different concentrations of NaH2PO4 and detergent. Wash
#1: 100 mL of 80 mM NaH2PO4 with 1% TWEEN 20 (v/v). Wash #2: 100 mL of 80 mM
NaH2PO4 with 1% Triton X-100 (v/v). Wash #3: 100 mL of 80 mM NaH2PO4 without detergent. Finally, elution was with 175 mM NaH2PO4 – yielding ASFV APE at ∼90-
95% purity (Figure 3.6). The protein was dialyzed directly into 50 mM Tris-borate, 100 mM KCl, 5% glycerol (v/v), pH→7.3 with KOH at 15 °C, 5 mM DTT, and then concentrated in a 5 kD NMWL centrifugal filter, adjusted to 40% glycerol (v/v), flash frozen in N2 (l), and stored at -80 °C. Subsequently, working stocks were kept at –20 °C.
An analogous purification of His-tagged ASFV APE from BW528(DE3)pRAREpLysS cells was also conducted with 1 mM EDTA present at all stages after the cobalt column.
The major human AP endonuclease, Ape1, was assayed for comparative purposes; it was purchased from Trevigen and used without further purification.
Figure 3.6 Silver-stained gel of ASFV APE purified from BW528(DE3)pRAREpLysS cells in the absence of EDTA.
54 3.4 Physical Characterization
Regardless of the expression host used, all preparations of His-tagged ASFV APE were of uniform molecular weight; the apparent molecular weight was, however, lower than the theoretical MW of 34.5 kD. Since SDS PAGE and gel filtration inferred exceedingly low MWs of 28.6-29.9 and 22.7-25.9 kD, respectively, we attempted to characterize the protein’s size by two additional methods.
Not pure enough for ESI mass spectrometry, ASFV APE was resolved by SDS
PAGE and the protein band excised and subjected to tryptic digestion prior to analysis by
MALDI-TOF mass spectrometry. It was hoped that this method would yield sequence coverage sufficient to confirm the presence or absence of N-terminal, C-terminal, and internal segments of the protein. However, as shown in Figure 3.7, only two fragments were observed. MALDI identification of the fragment approaching the C-terminus, along with the fact that the protein bound to a cobalt column very tightly (consistent with the presence of the C-terminal His tag), suggests a fully intact C-terminus.
Since mass spec was unable to provide information about the N-terminus (even though the experiment was run two separate times), ASFV was subjected to Edman sequencing. This indicated the first five residues of the protein to be “MFGAF”. Since this sequence corresponds to the correct N-terminus (Figure 3.7) we believe that our preparation of ASFV APE is in fact the full-length protein. While we cannot rule out the possibility of an internal truncation, this seems extremely unlikely considering the fact that both of the expression plasmids used (pET30-ASFV APE and pET21-ASFV APE) have been sequenced to confirm accuracy. At present, we have no explanation for the anomalous behavior of this protein during both SDS PAGE and gel filtration.
55 1 11 21 31 41 51 | | | | | | 1 MFGAFVSHRL WSDSGCTTTC ITNSIANYVA FGEQIGFPFK SAQVFIAGPR KAVINIQEDD 60 61 KVELLKMIVK HNLWVVAHGT YLDVPWSRRS AFVTHFIQQE LLICKEVGIK GLVLHLGAVE 120 121 PELIVEGLKK IKPVEGVVIY LETPHNKHHT YKYSTMEQIK ELFLRIRNTR LKQIGLCIDT 180 181 AHIWSSGVNI SSYNDAGQWL RSLENIHSVI PPSHIMFHLN DAATECGSGI DRHASLFEGM 240 241 IWKSYSHKIK KSGLYCFVEY ITRHQCPAIL ERNLGSSMQL QTALTAEFTT LKSLLKLEHH 300 301 HHHH
Figure 3.7 Sequence of His-tagged ASFV APE, with fragments identified by MALDI- TOF mass spectrometry highlighted in red.
3.5 AP Endonuclease Substrates and Assay Details
All assays involving ASFV APE were performed manually in a manner similar to what was described for ASFV DNA ligase in Section 2.4.2. APE substrates were identical in sequence to those shown in Figure 2.4. However, they employed an intact 45mer as the upper strand (instead of the 26mer and 19mer used to generate nicks), and the base pair of interest contained either deoxyuridine (U) or tetrahydrofuran (THF) in the upper strand. Substrate nomenclature remains consistent with that presented in Section 2.4.1;
A:U or A:THF therefore denote a 45mer duplex containing uridine or THF opposite to a templating adenosine. Detailed information accompanies each of the assays presented in this chapter. In short, canonical AP sites were generated in situ by excising uracil with E. coli uracil DNA glycosylase (UDG). In other instances, the THF analog of an AP site was employed; since THF lacks the C1’-hydroxyl of an AP site, it cannot equilibrate with the open-chain aldehyde and is therefore not subject to spontaneous strand scission via β- elimination (Figure 3.8).
56 (AP site) + H Pi Pi OH O Pi OH O Pi OH O O OH
Pi H Pi H Pi Pi
(THF site) Pi O
Pi H
Figure 3.8 Inherent instability of an abasic site. The chemical stability of THF enables study of APEs without the need to generate an abasic site in situ.
3.6 Metal Cofactor Usage
As shown in Figure 3.1, the two families of AP endonucleases employ catalytic metal ions very differently. Exonuclease III uses readily dissociatable magnesium; the active site of endonuclease IV contains three tightly bound zinc ions [reviewed in (73)].
Though the phylogenetic analysis presented in Figure 3.2 places ASFV APE in the endonuclease IV-like family, detailed analyses of protein sequences indicates that ASFV
APE is not entirely dissimilar to exonuclease III (Figure 3.9).
In order to discern whether ASFV APE is more similar to endonuclease IV or exonuclease III in its metal cofactor usage, and to establish optimal assay conditions for subsequent kinetic analyses, we posed the following questions as experimental
proteins % similarity % identity ASFV APE vs. endonuclease IV 24 17 ASFV APE vs. exonuclease III 21 14
Figure 3.9 Percentage similarity/identity of various APEs.
57 objectives: i) Does the active site of ASFV APE retain a tightly bond metal cofactor after purification in the absence of a chelator?; ii) If a metal cofactor is bound tightly within the ASFV APE active site, what is its identity?; iii) If a metal cofactor is bound tightly within the ASFV APE active site, what is the stoichiometry of protein:metal ion(s)?; and iv) After purification in either the presence or absence of EDTA which, if any, metal ions can activate the enzyme?
We have only just begun pursuing answers to the above questions, for reasons that are now worth explaining. Noting that ASFV APE is essential for ASFV viability in cell culture, Garcia-Escudero et al. referred to this protein as a homologue of human Ape1
(71) – which is a magnesium-dependent, exonuclease III-like protein. With this comparison in mind, and in our haste to simply demonstrate its activity qualitatively,
ASFV APE was successfully assayed in the presence of 10 mM MgCl2 – confirming the activities previously only inferred from protein homology (each of these activities is described in Sections 3.8.1-3.8.3). This qualitative data for ASFV APE, along with qualitative data for ASFV-encoded AP lyase activity (presented in Chapter 4), was used to supplement our ASFV DNA ligase fidelity analyses (described in Chapter 2) in early manuscripts speculating on the role of DNA repair in ASFV. A common objection to these manuscripts was that the proposed concept of mutagenesis via deviant DNA repair would require biological evidence in order to gain credence. In light of these objections it seemed best to simply break the repair pathway down into its constituent enzymes and describe these individually – a change that necessitated investigation of the APE and AP lyase activities in greater detail. It was at this point, when performing homology analyses on our own, that we recognized ASFV APE is most similar to endonuclease IV-like
58 proteins – and may therefore employ metal ions other than magnesium preferentially. At present, only the limited experiments described below have been conducted to address this issue. Each of these experiments was performed with ASFV APE that had been expressed in BW528(DE3)pRAREpLysS cells and purified on cobalt, P-11, S-100, and hydroxyapatite columns – either in the presence or absence of EDTA.
As shown in (Figure 3.10), after being purified in the absence of EDTA, ASFV
APE: i) does not require exogenous metal ions for activity, and ii) is inhibited roughly
40% by 200 µM EDTA during a 30 minute reaction at 37 °C. These findings suggest that the enzyme does in fact retain one or more catalytic metal ions over the course of its lengthy purification. The identity and stoichiometry of the active site metal ion(s) will be probed via atomic absorption spectroscopy in the near future. We have not yet examined whether, while retaining its endogenous metal cofactor, ASFV APE can be hyper- activated by exogenous metals – as is the case for endonuclease IV (74).
100
80
60
40
relative activity (%) 20
0 without EDTA with EDTA
Figure 3.10 Activity of ASFV APE – purified in the absence of metal chelators – against A:THF. Assay buffer was 50 mM Tris-borate, 100 mM KCl, 15% glycerol (v/v), pH→7.8 with KOH while at 37°C, 300 µM DTT, 1.5 mg/mL BSA, with or without 200 µM EDTA. Substrate concentration was 100 nM, while protein concentration was unspecified.
59 After being purified in the presence of EDTA, which – given the length of time required for purification – presumably results in at least partial loss of active site metal ions, it was found that ASFV APE could be differentially activated by the addition of exogenous metal salts (Figure 3.11). Similar to what has been observed for apo- endonuclease IV (74), apo-ASFV APE is efficiently activated by manganese and cobalt, but not by zinc; being that zinc is the endogenous metal ion of endonuclease IV, its inability to reactivate that enzyme may be attributable to difficulty reassembling the tri- metal center within the active site. Unlike apo-endonuclease IV which employs Ni2+ very poorly (74), apo-ASFV APE is activated most efficiently by this ion (Figure 3.11).
While our understanding of metal usage by ASFV APE is still modest, the enzyme clearly appears to be more similar to endonuclease IV than to exonuclease III in its cofactor preference. How does this fact influence our interpretation of data, described in
Sections 3.8.1-3.8.3, which was generated using magnesium as the sole cofactor? It seems extremely unlikely that activities detected in the presence of Mg2+ will be abolished in the presence of Mn2+, Co2+, or Ni2+. If anything, the data in Figure 3.11 suggests that ASFV APE repair activities will be more robust with these transition metals than with magnesium.
60 100
80
60
40
20 [incised product](nM)
0 2+ 2+ 2+ 2+ 2+ 2+ 2+ 2+ g n n a o i u d M M Z C C N C C
Figure 3.11 Activity of apo-ASFV APE – purified in the presence of EDTA – against A:THF. Assay buffer was 50 mM Tris-borate, 100 mM KCl, 15% glycerol (v/v), pH→7.8 with KOH while at 37°C, 1 mM EDTA and 15 mM of the specified metal ion (each of these was the chloride salt, expect nickel and copper – which were both the sulfate salts). Note that DTT and BSA were excluded from this assay, since in their presence considerable precipitation was observed; this is expected to be less of a problem at lower metal concentrations. Substrate concentration was 100 nM, while protein concentration was unspecified. Reactions were run for three minutes at 37 °C.
3.7 Quantitation of Active Enzyme Concentration
The concentrations of active ASFV APE and human Ape1 were determined by single turnover experiments conducted at 37 °C in assay buffer consisting of 50 mM Tris- borate, 100 mM KCl, 15% glycerol (v/v), pH→7.8 with KOH while at 37 °C, plus 300
µM DTT, and 1.5 mg/mL BSA. Protein was preincubated with 500 nM THF-containing
45mer duplex DNA in the absence of free metal ions (4 mM EDTA). Reactions were initiated by the simultaneous addition of 21 mM MgCl2 and two trapping agents: heparin at 22.5 mg/mL and heat degradation product (HDP) at 12.6 µM. HDP, consisting of a
45mer duplex containing a single nucleotide gap flanked by both a 3’-phosphate and a 5’- phosphate, was prepared simply by annealing the three appropriate oligonucleotides. The mixture of heparin and HDP was demonstrated to be an effective trap on the time scale
61 required to run this assay. At appropriate time points post initiation, aliquots were quenched with formamide and analyzed as described elsewhere.
The amplitude of these single turnover experiments could underestimate the true active enzyme concentration if: i) the APE were not saturated with DNA; and ii) if some of the DNA-bound APE needed to dissociate from its substrate in order to bind magnesium – in which case the protein could be trapped before rebinding substrate in a productive manner. As a preliminary examination of these possibilities, a higher MW
32P-labled THF-containing duplex was included in the magnesium/trapping solution.
Under the conditions described above we observed little to no turnover (less than ∼1%) of the larger substrate – suggesting that the APE was in fact saturated with DNA, and that dissociation from DNA was not occurring upon the addition of magnesium.
The above single turnover assays were conducted prior to our realization that
ASFV APE uses Co2+, Mn2+, and Ni2+ preferentially to Mg2+ (Figure 3.11). The higher activity in the presence of transition metal ions may be attributed to higher catalytic efficiency and/or an increase in the concentration of catalytically competent enzyme molecules. If the latter possibility is indeed important, then the single turnovers conducted with magnesium as the sole cofactor will underestimate the concentration of active ASFV APE and Ape1.
3.8 Demonstration of Three Repair Activities
As listed in Figure 3.1, AP endonucleases catalyze a variety of repair reactions. In this section we demonstrate, for ASFV APE, those activities expected to be most relevant to
DNA repair in the African Swine Fever Virus.
62 3.8.1 AP Endonuclease
45mer oligonucleotide containing the THF abasic site analog at position 26 (see Figure
2.4 for oligonucleotide sequences) was labeled at either the 5’ terminus with polynucleotide kinase (using [γ-32P]ATP) or at the 3’ terminus with terminal transferase
(using [α-32P]ddATP). These differentially labeled THF-containing oligonucleotides were simultaneously annealed to a complementary 45mer containing adenine opposite to
THF. 3.5% of the duplex substrate contained the 5’ label and another 3.5% contained the
3’ label. This labeling scheme enabled us to monitor chemistry both up and down stream of the lesion without having to doubly label the oligo, which is technically more difficult.
Assay buffer consisted of 50 mM Tris-borate, 100 mM KCl, 15% glycerol (v/v), pH→7.8 with KOH while at 37 °C, supplemented with 15 mM MgCl2, 300 µM DTT, and 1.5 mg/mL BSA. At 37 °C, 200 nM A:THF was incubated with 3.5 nM enzyme (either
ASFV APE or human Ape1). At the indicated time points aliquots were quenched with
Figure 3.12 Confirmation that ASFV gene E296R encodes an AP endonuclease. (A) The THF-containing substrate. For simplicity the upper strand is shown here with 32P- label at both the 5’ and the 3’ termini. However, the duplex was constructed as a mixture of molecules containing label at either the 5’ terminus or the 3’ terminus. The red arrow indicates the site of cleavage. See Figure 1.1 for detailed chemical structures. (B) Monitoring strand incision as a function of time. Ape1 was used as a positive control.
63 formamide containing xylene cyanol and bromophenol blue and resolved on a 15% denaturing polyacrylamide gel under standard conditions. Note that when run on a 19% gel the 5’ labeled 45mer is readily resolved from the 3’ labeled 45mer. Though the
25mer and 5’-Pi-THF-19mer reaction products are necessarily produced at the same rate, in Figure 3.12 the bands for the former appear more intense since the activity of the 5’ label exceeded that of the 3’ label.
ASFV APE incised the THF substrate to give product bands identical in size to those generated by the well-characterized human Ape1; these product bands were also identical in size to synthetic standards (Figure 3.12). This confirms that the protein product of ASFV gene E296R is indeed an AP endonuclease.
3.8.2 3’-Phosphodiesterase
Recent data from the Salas group suggests that AP sites in ASFV will, at least occasionally, be processed by the lyase activity of Pol X (71). As shown in Figure 1.1, this activity gives rise to a single nucleotide gap containing the 3’ polymerase blocking- moiety 4-hydroxy-2-pentenal-5-phosphate. To see whether ASFV APE possesses a 3’- phospohodiesterase activity capable of removing this group, the following assay was performed.
A 5’ labeled 45mer oligonucleotide containing deoxyuridine at position 26 (see
Figure 2.4 for oligonucleotide sequences) was annealed to a complementary 45mer
(containing adenine opposite to uracil). 21.25 pmol of this duplex was incubated with
0.85 “units” of UDG in a buffer consisting of 50 mM HEPES, 100 mM KCl, 15% glycerol (v/v), pH→7.8 with KOH while at 37 °C, 1 mM DTT, 1 mM EDTA, and 1.5
64 mg/mL BSA (total volume was 39 µL). Incubation was at 37 °C for 60 minutes before adding a 10% excess of UDG inhibitor. 20 units of E. coli endonuclease III (which displays AP lyase activity) was then added and incubation continued at 37 °C for an additional 60 minutes. This DNA, now containing a single nucleotide gap flanked by a
5’-phosphate and 3’-4-hydroxy-2-pentenal-5-phosphate, was incubated at a concentration of 250 nM with 14 nM ASFV APE. The reaction buffer during incubation with APE was the same as that described above, but now supplemented with 15 mM MgCl2 and containing EDTA at a reduced final concentration of 0.5 mM. At the indicated time points aliquots were withdrawn and supplemented with sodium borohydride (NaBH4) to a final concentration of 285 nM; this was done to stabilize any residual, intact abasic sites that might have escaped processing by endonuclease III. Aliquots were incubated on ice for at least 30 minutes before adding tRNA and ammonium acetate to concentrations of
0.06 µg/mL and 2.5 M, respectively. Ethanol was added to 70% and the DNA allowed to precipitate for at least 30 minutes on ice. DNA was pelleted at 4 °C in a microfuge for at least 20 minutes, the supernatant aspirated, and the pellet allowed to air dry prior to being resuspended in formamide containing xylene cyanol and bromophenol blue. Samples were resolved on a 19% denaturing polyacrylamide gel run under standard conditions.
In Figure 3.13 the 4-hydroxy-2-pentenal-5-phosphate-containing substrate migrates as a ∼25.5mer. In the presence of ASFV APE, its size is reduced to that of a
25mer – consistent with 3’-phosphodiesterase activity. Importantly, this finding establishes for the first time that a complete pathway for processing AP sites exists in
ASFV: following AP site incision by the AP lyase activity of Pol X, ASFV APE is
65
Figure 3.13 ASFV APE is also a 3’-phosphodiesterase. (A) Substrate consisted of a 45mer duplex containing a single nucleotide gap flanked by 5’ phosphate and the 3’ polymerase-blocking group 4-hydroxy-2-pentenal-5-phosphate – which is simply abbreviated as BG in this figure. See Figure 1.1 for detailed chemical structures. The red arrow indicates the site of cleavage. (B) ASFV APE removes 4-hydroxy-2-pentenal-5- phosphate to generate a single nucleotide gap flanked by a 3’-hydroxyl and a 5’- phosphate.
capable of removing 4-hydroxy-2-pentenal-5-phosphate, thereby generating a single nucleotide gap that can be filled by Pol X and subsequently ligated by ASFV DNA ligase.
3.8.3 3’→5’ Exonuclease
Human Ape1 possesses a 3’→5’ exonuclease activity that acts preferentially on mismatched nicks (75). This “editing” function appears to have evolved for the purpose of enhancing BER fidelity since Pol β is relatively error-prone and does not itself possess a proofreading function (75). We probed for such an editing activity in ASFV APE using the following assay.
200 nM nicked C:T or A:G (as shown in Figure 2.4) were incubated with 10 nM
ASFV APE at 37 °C in buffer that consisted of 50 mM Tris-borate, 100 mM KCl, 15% glycerol (v/v), pH→7.8 with KOH while at 37 °C, plus 15 mM MgCl2, 300 µM DTT, and
66
Figure 3.14 3’→5’ exonuclease activity of ASFV APE. (A) Substrate consisted of a mismatched nick, as would be generated by error prone gap filling. Base pair nomenclature is identical to that described for DNA ligation. The red arrow indicates the site of incision. (B) Excision of the nascent 3’ nucleotide from a mismatched nick as a function of time.
1.5 mg/mL BSA. At the indicated time points, aliquots were quenched in formamide containing xylene cyanol and bromophenol blue and were resolved on a 19% denaturing polyacrylamide gel.
Figure 3.14 shows very clearly that though ASFV APE does in fact possess
3’→5’ exonuclease activity, the efficiency varies considerably depending on the identity of the mismatched base pair. Since this editing activity is potentially very important to the fidelity of DNA repair in ASFV, it is examined in greater detail in Section 3.9.
3.9 Mismatch Editing: Implications for the Fidelity of AP Site Repair
Demonstration that ASFV APE can edit mismatched nicks raises the obvious question of how this activity influences the type and frequency of base substitutions that are being introduced into the viral genome by the concerted action of Pol X and ASFV DNA ligase.
The following three subsections begin to address this question.
67 3.9.1 5’-2-Deoxyribose-5-phosphate Influences Editing Efficiency
As outlined in Figure 1.1, 3’→5’ exonuclease editing of a mismatched nick by ASFV
APE could conceivably occur both before or after removal of 5’-2-deoxyribose-5- phosphate (5’-dRP). We were curious to know whether the presence of 5’-dRP influences the efficiency of mismatch editing. This is a non-trivial inquiry since this issue could conceivably have very interesting implications for the fidelity of the repair pathway as a whole. If, for example, 5’-dRP inhibited mismatch editing, then the APE- initiated route of AP site repair would be expected to generate point mutations more frequently than repair via the AP lyase-initiated route – in which 5’-dRP is never present at the site of a nick. The salient feature of such a hypothetical system is that it could enable an organism to toggle between the low and high fidelity repair pathways by controlling the manner in which repair is initiated.
In order to determine how the extra-helical 5’-dRP moiety influences the efficiency of mismatch editing by ASFV APE, we constructed a series of mismatched nicked substrates possessing either 5’-phosphate or 5’-Pi-THF (the 5’-phosphorylated tetrahydrofuran analog of 5’-dRP) (Figure 3.15). Assays were conducted at 37 °C in 50 mM Tris-borate, 100 mM KCl, 15% glycerol (v/v), pH→7.8 with KOH while at 37 °C, plus 15 mM MgCl2, 300 µM DTT, and 1.5 mg/mL BSA with 200 nM DNA and 7 nM
ASFV APE. At the indicated time points, aliquots were quenched in formamide containing xylene cyanol and bromophenol blue and were resolved on a 19% denaturing polyacrylamide gel.
68 The time courses shown in Figure 3.15 indicate that for both the efficiently edited
C:T mismatch and the inefficiently edited G:G mismatch the presence of the 5’-dRP analog inhibits 3’→5’ exonuclease activity, though only modestly; this is consistent with what has also been observed for human Ape1 (76). Accordingly, it appears that the fidelity of AP site repair in ASFV may be similar regardless of the route by which repair is effected. Note that this argument, however, does not take into consideration the influence that 5’-dRP might have on the fidelity of gap filling by Pol X.
Figure 3.15 Influence of a 5’-dRP analog (5’-Pi-THF) on the 3’→5’ mismatched nick editing activity of ASFV APE. (A) Structure of the substrate. Base pair nomenclature is the same as that described in Figure 2.4. The time courses illustrate excision of the nascent, mismatched nucleotide to generate a single nucleotide gap; specifically, the 26mer is being cut down to a 25mer. (B) Editing of nicked C:T. (C) Editing of nicked G:G. Notice the different scales for the y-axes.
69 3.9.2 Editing Efficiency as a Function of 3’ Base Pair
Hoping to elucidate which selective pressures might have guided evolution of the ASFV
DNA repair system, we next examined the base pair specificity of ASFV APE’s 3’→5’ exonuclease activity. 5’-phosphate-containing nicked substrates (as in Figure 2.4) at 200 nM were incubated with 13 nM ASFV APE at 37 °C in 50 mM Tris-borate, 100 mM
KCl, 15% glycerol (v/v), pH→7.8 with KOH while at 37 °C, plus 15 mM MgCl2, 300
µM DTT, and 1.5 mg/mL BSA (note that this is the optimized DNA ligase buffer, without ATP). Time courses for these reactions are shown in Figure 3.16.
base pairs with "A" in the templating position base pairs with "T" in the templating position A:A T:A 40 A:T 80 T:T A:G T:G A:C T:C 30 60
20 40 [25mer] (nM) [25mer] (nM)
10 20
0 0 0 5 10 15 20 25 30 0 5 10 15 20 25 30
time (minutes) time (minutes)
base pairs with "G" in the templating position base pairs with "C" in the templating position 140 G:A C:A 80 G:T 120 C:T G:G C:G G:C 100 C:C 60 80
40 60
[25mer] (nM) [25mer] (nM) 40 20 20 0 0 0 5 10 15 20 25 30 0 5 10 15 20 25 30
time (minutes) time (minutes)
Figure 3.16 Time courses for ASFV APE 3’→5’ exonuclease activity against all 16 possible base pair combinations at the 3’-OH side of a nick. Substrates were as shown in Figure 2.4.
70 To better analyze trends in the base pair specificity of ASFV APE, and in order to make comparisons between the efficiency of mismatched nick synthesis (by Pol X), mismatched nick editing (by ASFV APE), and mismatched nick ligation (by ASFV DNA ligase), the data from Figure 3.16 were converted to average velocities (product concentration at the final time point, divided by the length of this time point in minutes).
While these values aren’t as rigorous as true kinetic parameters, they do provide a
A Pol X vs. ASFV APE 5 Pol X 1000 ASFV APE 4
) -1 s
-1 100 3 (M d,app (nM/min.)
/K 2 10 pol k
Pol X catalytic efficiency 1 ASFV APE average velocity 1
0
T T T T G A G C G A G A A C C C : : : : : : : : : : : : : : : : T T A T G C T C G G C A A G A C base pair B ASFV DNA ligase vs. ASFV APE 5 1e+7 ASFV DNA ligase ASFV APE 4 1e+6 ) -1 s 3 -1 (M
M 1e+5 /K
2 (nM/min.)
cat k 1e+4 1 ASFV APE average velocity ASFV DNA ligase catalytic efficiency 0
T T T T A C G C C G A G C A A G : : : : : : : : : : : : : : : : T C T A T T G G C A C G C A G A base pair
Figure 3.17 Comparison of ASFV APE’s mismatched nick editing activity (plotted as average velocity; see text) with the catalytic efficiencies of Pol X (A) and ASFV DNA ligase (B). Data for Pol X and ASFV DNA ligase have been plotted using a logarithmic scale in order to accommodate the broad range of catalytic efficiencies.
71 reasonably accurate picture of the “editing profile” of ASFV APE. Figure 3.17 juxtaposes the average velocities of mismatched nick editing by ASFV APE with the catalytic efficiencies of Pol X and ASFV DNA ligase, as a function of 3’ base pair.
Noteworthy trends are as follows: i) though it can act as a 3’→5’ exonuclease on correctly matched nicks, ASFV APE is generally more efficient on mismatched nicks – suggesting that this enzyme is capable of enhancing the fidelity of ASFV AP site repair; ii) there is not a strong correlation between the specificities of ASFV APE and Pol X; and iii) excepting the fact that they both act preferentially on C:T (a fact that we return to momentarily), there is no correlation between the specificities of ASFV APE and DNA ligase.
Plotting the ASFV APE mismatch editing data in descending order of efficiency readily elucidates a striking trend (Figure 3.18A): the six most efficiently edited mismatches contain pyrimidines in the nascent (nicked) strand, while the six least efficiently edited mismatches contain purines in this strand. One explanation for this is that the efficiency of editing is largely dependent not on 3’ base pair shape/size or hydrogen bonding capability, but on base stacking interactions at this position. Without exception, the larger purines – which show higher stacking stabilization than their smaller pyrimidine counterparts (77) – are excised from the nascent strand less efficiently. This trend is highlighted in Figure 3.18B where a plot of average velocity for all eight base pairs (both matched and mismatched) that consist of one purine and one pyrimidine indicates that the permutation with the purine in the nascent strand is always edited less efficiently than the permutation with the pyrimidine in the nascent strand. That hydrogen bonding capability is unimportant is highlighted by the fact that G:T and T:G, both of
72 A B mismatch specificity of ASFV APE's ASFV APE's 3'->5' exonuclease activity 3'-5' exonuclease activity preferentially removes nascent pyrimidines 5 3.0
2.5 4 2.0 3 1.5 (nM/min.) 2 (nM/min.) 1.0 average velocity average velocity average 1 0.5
0 0.0
T C T T C C A A A T T G G G G C A A C G : : : : : : : : : : : : : : : : : : : : T C G T T A T C A G C A G A G T A C G C base pair base pair
Figure 3.18 Analysis of mismatched nick editing by ASFV APE. (A) Average velocity as a function of base pair. The central vertical line divides base pairs containing a purine in the nascent strand from base pairs containing a pyrimidine in the nascent strand. (B) A focus on the eight base pairs consisting of both a purine and a pyrimidine. The vertical lines have been inserted to emphasize that among the pairs of base pair permutations, the base pair containing pyrimidine in the nascent strand is always edited most efficiently.
which are capable of “wobble” hydrogen bonding, display markedly different efficiencies of editing. We summarize our interpretation of the data as follows: the efficiency of ASFV APE’s 3’→5’ exonuclease activity is inversely proportional to the stacking ability of the nucleotide being excised.
An apparent problem with the above interpretation is that the trend in editing efficiencies does not show a perfect inverse correlation with the strength of base stacking interactions. Kool and colleagues have shown the order of base “stacking ability” to be
A > G ≥ T = C (77). Despite this, ASFV APE 3’→5’ exonuclease activity removes A more efficiently than G. This discrepancy might be rationalized by the fact that the strength of base stacking is dependent not only on the nucleotide of interest, but also on the identity of the neighbors it is stacking with. Unique sequence contexts have been
73 identified which demonstrate aberrant stacking stabilization or destabilization (78).
Perhaps the sequence of the duplex used in our assays represents one such exception.
This seems unlikely though, so we propose a refinement of our explanation that includes steric constraints imposed by the protein.
Though a complex structure of E. coli endonuclease IV bound to a mismatched nick is not available, a structure of the protein bound to abasic DNA (the THF analog) has been solved (79). Centered at the position of the abasic lesion, endonuclease IV bends the DNA duplex by about 90°, and forces both the abasic ribose and the nucleotide
C
A
B
Figure 3.19 E. coli endonuclease IV complexed with abasic (THF) DNA (1QUM). This structure represents the post cleavage complex. For clarity, a loop that would otherwise obscure the THF residue has been truncated. The three green spheres are the Zn2+ ions that coordinate the scissile phosphate. (A) The 3’-OH upstream of the incised THF residue. (B) The C1’ position of the THF residue; note that it is pointing away from the helical axis. (C) The base directly opposite to the THF residue; note its extra-helical orientation.
74 opposite to it to adopt extra-helical conformations (Figure 3.19). If endonuclease IV, and by analogy ASFV APE, interacts with a mismatched nick in analogous manner, the nascent 3’ nucleotide of the nick would need to be twisted into an extra-helical conformation to help position the scissile phosphate adjacent to the enzyme’s tri-zinc center (Figure 3.20). While extraction of the base from its intra-helical base-stacking
A B
O O O O P P O O O O O OH P O O O O O O OH P O OH + O + M O M
base flipping
O O OH P O O O O O P O + O M
Figure 3.20 A hypothetical base flipping mechanism for 3’→5’ exonuclease activity in the endonuclease IV family of APEs. (A) A highly schematic rendering of abasic DNA after cleavage by E. coli endonuclease IV as shown in the complex structure of Figure 3.19; a true AP site is shown here instead of the THF analog. The extra-helical orientation of the AP site appears to be a requirement for strand incision – perhaps by facilitating correct positioning of the scissile phosphate. (B) A schematic rendering of the hypothetical base flipping that may also be required for 3’→5’ exonuclease editing of a mismatched nick. The necessity of flipping the nascent, mismatched nucleotide may explain the empirical observation that nascent purines are edited less efficiently than are nascent pyrimidines.
75 environment would be thermodynamically more costly for the purines than for the pyrimidines (hence the trends described above), perhaps the differences between excision of “A” and “G” can be accounted for by different interactions with the nascent base while it is in the extra-helical conformation.
3.9.3 Mismatch Ligation vs. Mismatch Editing: Competition Assays
The current data suggests that Pol X, ASFV APE and ASFV DNA ligase comprise the fidelity-determining components of the ASFV AP site repair system. When Pol X catalyzes misincorporation at a single nucleotide gap, the mismatched nick so generated may then either be ligated by ASFV DNA ligase or edited by ASFV APE to regenerate the single nucleotide gap – affording another opportunity for correct synthesis by Pol X.
We therefore asked the following question: When ASFV APE and DNA ligase are present at equimolar concentrations, to what extent are mismatch editing and mismatch ligation effected?
To answer the above question “competition” assays were conducted employing mismatched nicked substrates (as shown in Figure 2.4). Reactions were conducted at 37
°C in 50 mM Tris-borate, 100 mM KCl, 15% glycerol (v/v), pH→7.8 with KOH while at
37 °C, plus 15 mM MgCl2, 1 mM ATP, 300 µM DTT, and 1.5 mg/mL BSA with 300 nM
DNA. For the C:T mismatch, ASFV APE and DNA ligase were each present at 100 pM.
For the A:G mismatch, ASFV APE and DNA ligase were each present at 30 nM. After
40 minutes, reactions were quenched and analyzed as described elsewhere. The bar height in Figure 3.21 reflects the amount (in nM) of each product type (either 45mer ligation product or 25mer exonuclease product) averaged from three independent 76 experiments. The error bars indicate ± standard deviation. Note that some DNA ligases are able to catalyze phosphodiester bond formation across single nucleotide gaps. It is therefore conceivable that in these competition assays ligation product might be formed after exonuclease editing of the mismatched nick. To examine this possibility ASFV
DNA ligase was tested for gap ligation activity under the same conditions as the competition assay. No such activity was observed.
For the C:T mismatch – the preferred substrate for ASFV DNA ligase – ligation is much more efficient than is nick editing. For the A:G mismatch – the least efficient substrate for ASFV DNA ligase – though editing is favored, a significant extent of ligation is still observed (Figure 3.21). Though this in vitro scenario is admittedly a very rough approximation of what might arise inside a cell, the data suggests that the editing activity of ASFV APE is not so potent as to preclude the generation of mismatches within
80 25mer; 3'-5' exo product 45mer; ligation product
60
40 [product] (nM) [product] 20
0 C:T A:G base pair
Figure 3.21 Mismatch editing vs. mismatch ligation: competition assays between ASFV APE and DNA ligase. Bar heights indicate the concentration of each product type. Error bars denote standard deviation.
77 the ASFV genome. It is, however, important to note that these assays were performed with Mg2+ as the sole metal ion. In the presence of Ni2+, Mn2+, or Co2+ – which activate the endonuclease activity of ASFV APE preferentially – it is conceivable that the extent of editing will increase.
Besides relative catalytic efficiencies, the fidelity of BER in ASFV will also clearly depend on the relative concentrations of each protein at sites of viral DNA repair.
Since no information is available regarding the intracellular concentration of ASFV enzymes, and since ASFV APE is required for virus viability when infecting VERO cells
(71), the influence of ASFV APE on the fidelity of viral BER and on virus variability in general remains uncertain.
78 CHAPTE R 4
5’-2-DEOXYRIBOSE-5-PHOSPHATE REMOVAL: IDENTIFICATION AND CHARACTERIZATION OF A “MISSING” ACTIVITY IN ASFV AP SITE REPAIR
4.1 An Apparent Bottleneck in ASFV APE-initiated AP Site Repair
Though our characterization of ASFV APE (in Chapter 3) established the existence of a complete ASFV-encoded AP site repair pathway (the AP lyase initiated route shown in
Figure 1.1), it also presented a conundrum. It was not clear how 5’-2-deoxyribose-5- phosphate (5’-dRP), generated upon ASFV APE incision of an AP site, would be removed. In eukaryotes 5’-dRP is excised, at least in part, by the lyase activity of Pol β
(80). However, the lyase domain of Pol β is entirely absent in its viral homologue Pol X, and in spite of their efforts the Salas group was only able to identify in Pol X a lyase activity towards AP sites, but not 5’-dRP (71). Since 5’-dRP is labile, it might be argued that a catalyst need not participate in this step of AP site processing. However, the half- life for this reaction under physiological conditions is on the order of 30 hours (81) – presenting an apparent bottleneck in the ASFV APE-initiated route of AP site repair.
Three potential remedies to this metabolic challenge are as follows: i) ASFV might recruit a 5’-dRP removal activity from the host cell (considering our discussion of
79 the ASFV-encoded PCNA analog in Chapter 5, this would appear to be feasible); ii) the microenvironment in which ASFV DNA repair occurs might be rich in small molecule effectors of 5’-dRP loss [amino acids and polyamines such as spermidine can function in this capacity by promoting β-elimination (81, 82)]; iii) one or more of the ASFV DNA repair proteins might themselves possess a previously unidentified 5’-dRP removal activity. In light of the fact that DNA repair proteins frequently display multiple – and sometimes very diverse – activities, the third option seemed most reasonable to us and we hypothesized that 5’-dRP removal activity would be present in either ASFV APE or
ASFV DNA ligase.
In this chapter we first probe for the hypothesized repair activity without specific attention to the mechanism(s) by which it is being effected. Subsequently, the question of mechanism is addressed.
4.2 Consideration of Potential Substrate Scenarios
5’-dRP removal can conceivably occur before or after gap filling by a polymerase.
Current evidence suggests that in eukaryotes Pol β removes 5’-dRP after gap filling (83).
However, this may not indicate a preference by Pol β for nicked 5’-dRP vs. gapped 5’- dRP (it shows efficient activity on gapped 5’-dRP); rather, this finding seems to be a consequence of the polymerization activity being more efficient than the lyase activity
(regardless of the context in which 5’-dRP is presented) (80). In order to maximize the likelihood of identifying a 5’-dRP removal activity amongst ASFV-encoded repair proteins, we employed substrates containing 5’-dRP in either the context of a 1-
80 nucleotide gap or in the context of a nick. Preparation of these substrates is described below.
4.3 Assay Details
Oligonucleotide sequences and lengths were essentially as described in Figure 2.4. Assay buffer consisted of 25 mM HEPES, 25 mM KCl, pH→7.5 with KOH while at 37 °C, supplemented with 1 mM EDTA, 300 µM DTT, and 1.5 mg/mL BSA. All reactions were run at 37 °C. So that chemistry occurring on both sides of the lesion could be monitored, substrate was labeled on both strands comprising the nick, as shown in Figure 4.1.
Nicked 5’-dRP was generated by incubating 120 pmoles of nicked duplex, containing 5’- phosphorylated deoxyuridine downstream of the nick (Figure 4.1), with 27 “units” of uracil DNA glycosylase (UDG; from NEB) at 37 °C for 20 minutes before the addition of a 10% molar excess of UDG inhibitor. The 5’-dRP-containing single nucleotide gap so generated was then supplemented with 2.5 nM Klenow Fragment (this was a mutant, lacking 3’→5’ exonuclease activity), 100 µM dTTP, and 10 mM MgCl2 and incubated
UDG
Figure 4.1 In situ generation of substrates containing 5’-dRP. Nicked, uracil-containing duplex was treated with UDG to yield a single nucleotide gap flanked by 3’-OH and 5’- dRP (“gapped 5’-dRP”, shown above right). This was either used directly, or was subjected to gap filling by Klenow Fragment to give the nicked variant (“nicked 5’- dRP”). Note the presence of 32P label on the oligonucleotides both upstream and downstream of the lesion.
81 for an additional 5 minutes before adding EDTA to a concentration of 25 mM. In order to keep reaction conditions as similar as possible for assays involving the different substrates, gapped 5’-dRP was prepared as described above, except that dTTP was excluded entirely, and magnesium was excluded from the Klenow incubation but was then added after the addition of EDTA.
So that sets of reactions could be conducted using identical preparations of DNA, large batches of substrate were prepared, placed on ice, and then aliquoted immediately before use. 5’-dRP removal assays were conducted in triplicate, with 150 nM DNA and the indicated concentration of protein. After preincubating both protein and DNA at 37
°C for five minutes, reactions were initiated and incubation continued for an additional
60 minutes. Reactions were quenched by placing tubes on ice and adding fresh sodium borohydride to a final concentration of 285 mM; without this reduction step, 5’-dRP is quantitatively lost during electrophoresis at 50 °C. After at least 30 minutes on ice, tRNA and ammonium acetate were added to concentrations of 50 pg/µL and 2.9 M, respectively. DNA was then precipitated by the addition of ethanol to a final concentration of 70% and placing the tubes in a dry ice/ethanol bath for 30 minutes.
After centrifugation at 4 °C for at least 20 minutes, supernatant was aspirated and the
DNA pellet dried briefly in a speed vac. Resuspension was with formamide before resolving samples on a 19% gel.
ASFV APE and DNA ligase were prepared as described in previous sections.
Expression and purification of Pol β were as described (84). Expression and purification of Pol X were as described (18) but with the following exceptions. Induction of protein
82 expression was for 9.5 hours. The pH of the purification buffer was reduced from 9.0 to
8.0. After gel filtration, Pol X-containing fractions were pooled and dialyzed into buffer consisting of 10 mM sodium phosphate, 10% glycerol, 50 µM EDTA, pH 7.3 at 4°C, 1 mM DTT, and 0.004% Tween 20 (v/v). This sample was applied to a hydroxyapatite column (1.2 cm i.d.; bed volume ∼10 mL) that had been equilibrated in the same buffer.
The first washing step was with 100 mL of the buffer described above, but containing 80 mM sodium phosphate and 1% Tween 20 (v/v). A similar second washing step was conducted in the absence of detergent. Pol X was then eluted, in the absence of detergent, using a linear sodium phosphate gradient (80-320 mM; 100 mL total volume) at a flow rate of ∼0.8 mL/min., collecting 2.5 mL fractions. Pol X-containing fractions were pooled, concentrated, and dialyzed against buffer consisting of 50 mM Tris-HCl,
150 mM KCl, pH→8.0 at 4 °C with KOH, 1 mM DTT. Samples were flash frozen in N2
(l) as 50% glycerol (v/v) stocks prior to storage at –80 °C. Subsequently, working aliquots were kept at –20 °C.
4.4 5’-dRP Removal from a Single Nucleotide Gap
Figure 4.2 shows representative lanes for reactions examining 5’-dRP removal in the context of a single nucleotide gap. As expected, the upstream oligonucleotide (25mer) is unmodified during the course of these assays. Distinct differences in the relative intensities of the 19mer and 5’-dRP-19mer bands are readily visible amongst the different reactions. These differences in extent of 5’-dRP removal are presented quantitatively in
Figure 4.3. In this bar plot “t = 0” represents the amount of 5’-dRP-19mer remaining
83
A B
Figure 4.2 5’-dRP removal from a single nucleotide gap. (A) Doubly labeled substrate. The fragment directly opposite to the templating “A” represents 5’-dRP. The red arrow indicates the cleavage site. (B) Gel demonstrating differences in the efficiency of 5’-dRP removal from a gap. The protein and its concentration are indicated above each lane. “Heat” indicates that the protein was heat denatured prior to incubation with substrate. The arrow superimposed on the surface of the gel points to a very faint band which was unique to the ASFV DNA ligase reactions, as explained in the text.
after the 5 minute preincubation period in assay buffer only. “None” represents the amount of 5’-dRP-19mer remaining after 60 minutes in assay buffer only. Spontaneous
5’-dRP loss is clearly appreciable during the course of the reaction. However, each of the
ASFV proteins increases the amount of 5’-dRP lost beyond what is observed for background (denoted by the horizontal dotted line). Each of the proteins is considered individually below.
ASFV APE That ASFV APE enhances the extent of 5’-dRP loss from a nick
came as a bit of a surprise; since the reaction was performed in triplicate, we deem the
84 5'-dRP loss in the context of a single nucleotide gap
100
80
60
40
[19mer-5'-dRP] (nM) 20
0
Figure 4.3 5’-dRP removal from a single nucleotide gap. Bar heights represent the concentration of 5’-dRP-19mer remaining after a 60 minute incubation at 37 °C with the indicated protein; protein concentrations are shown in parentheses. “t = 0” and “none” denote 5 minute and 60 minute, respectively, incubations in assay buffer only. “heat” indicates that the protein was heat denatured prior to incubation with substrate. Error bars reflect the standard deviation for three independent reactions.
data to be trustworthy. Note that this reaction is relatively inefficient; with the protein concentration exceeding that of the substrate by more than 2-fold, ASFV APE was still unable to effect complete 5’-dRP removal in 60 minutes. In Figure 3.12 we demonstrated that ASFV APE incises the sugar phosphate backbone 5’ to the THF abasic site analog, generating a single nucleotide gap flanked by 3’-OH and 5’-Pi-THF (which mimics 5’- dRP). Why wasn’t the 5’-Pi-THF generated in that assay subsequently excised by ASFV
APE? The two most reasonable explanations are that: i) the enzyme concentration was
100-fold lower in the previous assay, and the time points were shorter – preventing us
85 from detecting this minor activity of ASFV APE; and ii) ASFV APE removes 5’-dRP by a lyase mechanism and is therefore only active against the aldehydic form of the substrate
(which the THF substrate cannot equilibrate with).
Since 5’-dRP removal would, to our knowledge, represent a novel activity for an
APE, it will be probed more rigorously in due course; considering the modest efficiency of this activity, it will be particularly important to prove that it is not being contributed by an E. coli contaminant.
Pol X The similar structure/chemical reactivity of 5’-dRP and AP sites made us very suspicious of Salas’ claim that Pol X is an AP lyase, but not a 5’-dRP lyase. It was for this reason that we included this protein in our analyses even though it had not been part of our initial hypothesis regarding 5’-dRP removal activity amongst ASFV DNA repair proteins. Pol X clearly shows an increase in the extent of 5’-dRP loss as a function of protein concentration (Figure 4.3). Importantly, boiling Pol X beforehand abolishes this catalytic affect. This heat lability is consistent with catalysis being dependent on native protein structure, rather than, for example, free primary amines that might be present in the protein prep.
Why are we detecting 5’-dRP lyase activity in Pol X when the Salas group did not? Almost certainly this discrepancy can be attributed to differences in assay conditions. Their assay contained 10 mM free MgCl2 while ours contained ∼150 nM; for another 5’-dRP removing enzyme, magnesium concentration appears to be an important parameter (85). We also used higher protein concentrations and allowed the reaction to run for a greater length of time – critical differences considering the relatively poor efficiency of Pol X’s 5’-dRP removal activity.
86 ASFV DNA ligase Similar to Pol X, ASFV DNA ligase increases the extent of
5’-dRP loss in a manner that is dependent on protein concentration, and this catalytic function is heat labile. Relative to the activities described above for ASFV APE and Pol
X, ASFV DNA ligase appears to be quite proficient at 5’-dRP removal. We momentarily thought this represented a novel activity for a DNA ligase, but this is merely a rediscovery of something published 6 years previous; though it is not common knowledge, AP lyase activity in an ATP-dependent DNA ligase was suggested (86), and then confirmed (85) in 1998.
It should be noted here that despite the very low magnesium concentration in this assay, ASFV DNA ligase still catalyzed the formation of a 45mer sized product indicative of ligation activity (this constituted ∼1% of the total product; not shown in
Figure 4.2). This issue was analyzed in separate experiments, enabling us to conclude that ASFV DNA ligase is capable both of adenylylating and ligating 5’-dRP in the context of a single nucleotide gap (data not shown). This essentially represents the reverse reaction of an AP endonuclease – regenerating an abasic site – and has been described previously for four different ATP-dependent DNA ligases (85). In Figure 4.2 the black arrow superimposed on the gel marks the position of a very faint band which was consistently observable only in the DNA ligase reactions; the size of this band is consistent with it being adenylylated 5’-dRP-19mer.
Pol β As expected based on previous characterization of the enzyme (80), Pol β displayed robust 5’-dRP removal activity – serving as a good positive control.
87 4.5 5’-dRP Removal from a Nick
We next examined, in analogous manner, the ability of ASFV proteins to remove 5’-dRP from a nick. Figure 4.4 shows representative lanes from each of these reactions. On the basis of the relative intensity of the 26mer band, it is clear that nicked 5’-dRP comprised the major species in this assay. However, the presence of a small percentage of 25mer suggests that some of the substrate molecules either retained their initial uracil (see
Figure 4.1) and were therefore not subject to gap filling, or did undergo uracil removal, but simply remained, for unknown reasons, refractory to gap filling by Klenow Fragment.
The presence of a minor 27mer band suggests that some molecules sustained two
A B
Figure 4.4 5’-dRP removal from a nick. (A) Doubly labeled substrate. The fragment protruding from the helix downstream of the nick represents 5’-dRP. The red arrow indicates the cleavage site. (B) Gel demonstrating differences in the efficiency of 5’-dRP removal from a nick. The protein and its concentration are indicated above each lane. “Heat” indicates that the protein was heat denatured prior to incubation with substrate.
88 incorporations of dTMP, rather than the intended single incorporation. This would obviously alter the substrate structure dramatically, giving, perhaps, a 3’ mismatched nick and a 5’one nucleotide extra-helical protrusion containing 5’-dRP. Despite these problems (which we were not able to remedy by altering the assay conditions), at least
90% of the substrate molecules exist as the intended nicked 5’-dRP.
Figure 4.4 demonstrates qualitatively that nicked 5’-dRP is processed with different efficiencies by the ASFV enzymes. Data for these reactions is presented quantitatively in Figure 4.5.
5'-dRP loss in the context of a nick 100
80
60
40
[19mer-5'-dRP] (nM) 20
0
Figure 4.5 5’-dRP removal from a nick. Bar heights represent the concentration of 5’- dRP-19mer remaining after a 60 minute incubation at 37 °C with the indicated protein; protein concentrations are shown in parentheses. “t = 0” and “none” denote 5 minute and 60 minute, respectively, incubations in assay buffer only. “heat” indicates that the protein was heat denatured prior to incubation with substrate. Error bars reflect the standard deviation for three independent reactions.
89 ASFV APE ASFV APE’s activity against 5’-dRP varies markedly depending on the context in which this moiety is presented. While ASFV APE effects 5’-dRP removal from a gap (Figures 4.2 and 4.3), in the context of a nick this enzyme actually reduces the extent of 5’-dRP loss relative to background (Figure 4.5). This “protective” affect has been reproduced on numerous occasions, but since it is modest it should not be over interpreted. Some attempt at analysis does however seem warranted.
The value in ASFV APE being able to remove 5’-dRP from a gap is obvious: the enzyme could single-handedly convert an AP site into a polymerase-usable, ligation- compatible substrate. However, the value in ASFV APE protecting 5’-dRP at the site of a nick is less clear, and in fact, would at first glance appear to be detrimental because 5’- dRP should inhibit DNA ligation [by preventing the 5’-phosphate from being adenylylated (83)].
The benefit of APE binding to nicked 5’-dRP is likely rooted in the fact that this would afford protection from solvent. Abasic ribose is subject to oxidative damage – which can render it much more difficult to process (87-89). Accordingly, if gap filling were to occur prior to 5’-dRP removal, though at this point ASFV APE cannot contribute catalytically, it might help protect 5’-dRP until its excision from the duplex by Pol X or
ASFV DNA ligase (vide infra). It is also conceivable that ASFV APE bound to nicked
5’-dRP can serve to recruit the multi-functional ASFV DNA ligase to these sites (as discussed below, ASFV DNA ligase can remove 5’-dRP from a nick before ligating it).
Such transient, highly context-dependent protein-protein interactions seem to be a common feature in DNA repair (73, 90). For example, Ape1 appears to first displace glycosylases from their tightly bound abasic product, then catalyze strand cleavage, and
90 finally “hand off” the single nucleotide gap so generated to Pol β for repair synthesis
[discussed in (73)].
The above discussion prompts further questioning. Do other APEs bind to and thereby protect nicked 5’-dRP from spontaneous β-elimination? What additional specific purposes might this interaction serve in vivo?
Our assays against both nicked and gapped 5’-dRP were conducted in the presence of very low magnesium concentration. Since we demonstrated ASFV APE to be similar to endonuclease IV in its metal cofactor requirements (Chapter 3), it will be interesting to see how its 5’-dRP removal/protective activity varies as a function of metal ion identity and concentration.
Pol X Similar to its activity against gapped 5’-dRP, Pol X enhances 5’-dRP removal from a nick as well. The activity of Pol X in these two different contexts appears to be similar.
ASFV DNA ligase This enzyme removes 5’-dRP from a nick with efficiency comparable to that with which it removes 5’-dRP from a gap. The incomplete inactivation of this activity by heat denaturation might simply be attributable to the fact that in the denatured state the protein is still expected to furnish a large number of solvent exposed basic residues that could promote 5’-dRP removal via Schiff base formation with the ring opened aldehyde (note that ASFV APE contains 40 lysine residues).
ASFV DNA ligase catalyzed 45mer (ligation product; not shown in Figure 4.4) formation from nicked 5’-dRP much more efficiently than it did from gapped 5’-dRP.
This is reasonable considering that: i) in the context of a nick, all molecules which have undergone 5’-dRP loss represent canonical DNA ligase substrates; and ii) in the context 91 of a gap only those molecules still containing 5’-dRP are ligatable (the enzyme absolutely requires adjacent 3-OH and 5’-phosphate termini), and these are apparently poor substrates owing to their conformational flexibility.
Pol β As expected, Pol β removes 5’-dRP from a nick very efficiently.
4.6 AP Site Incision by ASFV Proteins
Owing to the above findings regarding 5’-dRP removal by ASFV DNA ligase, and considering the similarities between 5’-dRP and AP sites, we asked whether this protein might also be capable of incising AP sites; though this activity was already demonstrated for ASFV APE in Chapter 3 and for Pol X by the Salas group (71), we examine them again here for purposes of comparison. AP substrate was prepared in a manner identical to gapped 5’-dRP (see Section 4.3) except that intact (rather than nicked), uracil- containing duplex was the starting material. For this series of assays it is most informative to examine the quantitative representation of the data first, and to then examine the gels – which provide some mechanistic insights.
Figure 4.6 shows that after 60 minutes at 37 °C in assay buffer only, virtually no degradation of the abasic site is observed (compared to t = 0). This is not a consequence of inefficient uracil removal. Post-UDG incubation an aliquot of substrate was boiled to effect β-elimination at AP sites; uracil removal was found to be ∼70% complete. The stability of AP sites has been noted previously (81) and stands in marked contrast to the lability of 5’-dRP (in both the nicked and gapped contexts). The best explanation for this is likely that 5’-dRP, tethered at just its 3’ terminus, is more free to adopt extra-helical,
92 AP site incision
140
120
100
80
60 [AP 45mer] (nM) 40
20
0
Figure 4.6 AP site incision. Bar heights represent the concentration of unincised abasic 45mer remaining after a 60 minute incubation at 37 °C with the indicated protein; protein concentrations are shown in parentheses. “t = 0” and “none” denote 5 minute and 60 minute, respectively, incubations in assay buffer only. “heat” indicates that the protein was heat denatured prior to incubation with substrate. Error bars reflect the standard deviation for three independent reactions.
solvent exposed conformations that make the C1’ aldehyde accessible – facilitating β- elimination (in the context of a nick, 5-dRP must be extra-helical). In contrast, the abasic ribose of an AP site experiences fewer degrees of freedom, and is therefore expected to adopt a solvent exposed, extra-helical conformation to a lesser extent than 5’-dRP; note that structural studies regarding the conformation adopted by AP sites is conflicting (91,
92).
93 In light of our characterization of ASFV APE in Chapter 3, its activity against the
AP site was expected. Consistent with their activity against 5’-dRP, Pol X and ASFV
DNA ligase are both capable of incising an AP site. These activities increase as a function of protein concentration, and are heat labile – suggestive of protein mediated, rather than small molecule mediated, catalysis. Analogous to their relative activities against nicked and gapped 5’-dRP, ASFV DNA ligase incises AP sites much more efficiently than Pol X; interestingly, the ligase is also much more efficient than Pol β in this regard.
Contrary to 5’-dRP – which is linked to DNA by just its 3’-phospohate – AP sites are secured within the duplex by attachment both 5’ and 3’ to the lesion. Accordingly,
AP site repair can be initiated on either side of the lesion, a fact that has important implications for down stream processing events (see the two routes of AP site repair in
Figure 1.1). By examining gel banding patterns we can draw some conclusions regarding the manner in which ASFV proteins process AP sites. Note that in these experiments,
45mer substrate migrates as a closely spaced doublet, the lower band corresponding to the 5’-32Pi-labeled species, the upper band corresponding to the 3’-32Pi-ddA (i.e. ddAMP)-labeled species (Figure 4.7). It is also worthwhile to note the very faint 19mer bands present in the reactions for heat denatured Pol X and ASFV DNA ligase. As mentioned previously, we attribute this apparent catalysis to the high concentration of solvent exposed basic residues that are likely to persist after protein denaturation.
ASFV APE Even at the very low free magnesium concentration (∼150 nM) used in these assays, ASFV APE shows robust incision of the AP site. This activity very clearly occurs via a hydrolytic mechanism, with strand cleavage occurring upstream of 94 the AP site to give an upstream 3’-OH-containing 25mer and a downstream 5’-dRP-
19mer – which decomposes spontaneously and enzymatically (based on our findings in
Section 4.4) to give 19mer (Figure 4.7).
Pol X In the presence of 350 nM Pol X, a 19mer-sized product is formed (Figure
4.7) – consistent with cleavage occurring 3’ to the AP site. Depending on the mechanism by which this cleavage is effected, the upstream product could consist of 25mer
A B
Figure 4.7 AP site incision. (A) Doubly labeled substrate. For simplicity the upper strand is shown here with 32P-label at both the 5’ and the 3’ termini. However, the duplex was constructed as a mixture of molecules containing label at either the 5’ terminus or the 3’ terminus. (B) Gel demonstrating differences in the efficiency of AP site incision. The protein and its concentration are listed above each lane. “Heat” indicates that the protein was heat denatured prior to incubation with substrate. The two bands labeled with “?” are discussed in the text.
95 containing either 3’-2-deoxryribose-5-phosphate (hydrolase mechanism) or 3’-4- hydroxy-2- pentenal-5-phosphate (lyase mechanism) (see Figure 4.8 for detailed structures). Both of these products would migrate more slowly than 25mer, but on the basis of gel mobility alone they probably could not be differentiated unequivocally. In order to highlight this ambiguity, the band migrating slower than 25mer in the Pol X reaction has simply been labeled as “25.5mer” in Figure 4.7. This uncertainty, however, is diminished by including in this analysis data that is presented later in this chapter: since Pol X can be reductively trapped on an AP site [vide infra and (71)], which is indicative of a lyase mechanism, we can – with a high level of confidence – assign the band of interest to be 25mer containing 3’-4-hydroxy-2- pentenal-5-phosphate.
To our surprise, a third band, migrating faster than 25mer is also present in the Pol X reaction (Figure 4.7). Note that this band was not detectable in the Pol X AP lyase assays conducted by the Salas group because their substrates were only labeled downstream of the AP site (71); this highlights the value of using labels on both sides of a lesion. When the repertoire of common DNA repair mechanisms is considered, and if we assume that precedents are instructive, the band migrating below 25mer is likely to be 3’- phosphorylated 25mer (phosphate contributes minimally to size, but increases the oligo’s negative charge – hence the faster migration rate) generated upon degradation of 3’-4- hydroxy-2- pentenal-5-phosphate. As shown in Figure 4.8, this process might occur by hydrolysis, or by δ-elimination. In order to identify which mechanism is being employed, HPLC could be used to analyze the deoxyribose derivative being released
(these differ for the hydrolytic and δ-elimination mechanisms).
96
5'
O O O P O O P O O O HO O O OH hydrolysis OH H H H H2 H OH OH hydrolysis O 5 O H O P O 25mer 2 O P O O O O O P O 3' O 3' 5' O OH H H O O O P O O P O HO O O O 5' OH H H 19mer hydrolysis O H 3' O 2 O P O O P O O beta-elimination O 5' O (lyase) OH
H H 3' O O O P O O P O O delta-elimination O OH O
H O 3' O P O O O O H
3'
Figure 4.8 Different routes for AP site processing that give products consistent with those generated by Pol X and ASFV DNA ligase. We only show pathways that adhere to the following criteria: i) cleavage is first effected 3’ to the abasic site, cleavage 5’ to the lesion then follows (this scenario is consistent with the observed banding patterns – in which a 5’-substituted 19mer is never observed); and ii) the 5’ cleavage event must generate a 3’-phosphate. For purposes of comparison, the abasic residue is shown only in the aldehydic form. In the upper, hydrolytic route, incision of the AP site produces 3’-2- deoxyribose-5-phosphate. This 3’ moiety can subsequently only be removed via hydrolysis. In the lower, lyase initiated route, incision of the AP site generates 3’-4- hydroxy-2- pentenal-5-phosphate – which can subsequently be removed either by hydrolysis or by δ-elimination (which takes advantage of the fact that the C4’ hydrogen is allylic, and therefore acidic, in this α,β-unsaturated aldehyde).
97 Though a number of enzymes effecting cleavage both 5’ and 3’ to AP sites have been characterized, including E. coli Fpg glycosylase (93) and endonuclease VIII (94),
Pol X appears to be the first example of a polymerase sporting both these activities. Is this a unique feature of Pol X, or has it been overlooked in other polymerases due to the use of assay techniques that ignore chemistry occurring upstream of the AP site? A casual examination of the literature indicates that of the polymerases known to have 5’- dRP lyase activity [Pol β (80), Pol ι (95), Pol η (95), Pol κ (95), Pol γ (96), Pol λ, E. coli
Pol I (96), T7 polymerase (96), and MMLV RT (96)], only Pol β has been assayed against unincised AP sites (80, 83, 97) – but none of these studies monitored chemistry occurring upstream of the lesion. We are currently unable to say whether Pol β is capable of incising 5’ to an AP site since all our assays to date have used relatively low
Pol β concentrations.
Lack of interest in polymerase-mediated AP site incision is understandable considering that APEs process these sites so efficiently/dominantly, and that polymerase are relatively inefficient AP lyases. Even if the mechanism we have proposed for Pol X holds up under more rigorous analysis, the reality is that it is relatively inefficient and therefore may not be biologically important. Even if it represented a robust activity, this route of repair would not present obvious benefits (over the APE-initiated route) because it still requires ASFV APE or another 3’-phosphomonoesterase to remove the polymerase-blocking 3’-phosphate group (note that we have not directly assayed for
ASFV APE 3’-phosphatase activity, but assume it exists on the basis of its diesterase activity and on homology to other APEs shown to be phosphomonoesterases).
98 ASFV DNA Ligase The banding pattern seen when 50 nM ASFV DNA ligase is incubated with an AP site is qualitatively very similar to that observed for Pol X, prompting us to conclude that the ligase is also capable of effecting cleavage on both sides of an AP site. Quantitatively, the ligase catalyzes both these reactions more efficiently than Pol X; this is especially obvious in light of the fact that for the reactions shown in Figures 4.6 and 4.7 the ligase was present at a concentration 7-fold lower than the concentration of Pol X. Accordingly, it would seem that ASFV DNA ligase might contribute to AP site processing to an extent that is biologically relevant.
Bogenhagen and Pinz, in their description of AP lyase activity in T4 DNA ligase, show a gel in which the enzyme has obviously catalyzed incision both 5’ and 3’ to an AP site (85); however, since incision upstream of the AP site was not particularly efficient it was basically left undiscussed. In light of this previous finding, our results for ASFV
DNA ligase suggest that the capacity to incise both 5’ and 3’ to AP sites may be common amongst the ATP-dependent DNA ligases.
Pol β As expected based on its previous characterization, Pol β cleaves 3’ to the
AP site, generating a downstream 5’-phosphorylated 19mer. The upstream product band is very faint and not visible in this particular image.
99 4.7 Summary of ASFV Proteins’ Activity Against the Three Abasic Lesions
Findings from the 5’-dRP and AP site processing experiments, described in Sections 4.4-
4.6, are summarized below in Figure 4.9.
activity against: enzyme gapped 5’-dRP (mechanism) nicked 5’-dRP (mechanism) AP site (mechanism) ASFV APE moderate activity (possibly a no activity; actually protective strong activity (incision 5’ lyase mechanism) of 5’-dRP (no info.) to AP site via hydrolysis) Pol X moderate activity (no info.) moderate activity (no info.) moderate activity (perhaps β- and δ- elimination) ASFV DNA strong activity (no info.) strong activity (no info.) strong activity (perhaps β- ligase and δ-elimination)
Figure 4.9 Summary of ASFV proteins’ activity against three different abasic lesions. Each activity is characterized as non-existent, moderate, or strong. Information about the mechanism of action is listed in parenthesis, when available.
4.8 Reductive Trapping Assay: Discerning Between Lyase and Hydrolase Mechanisms
Many of the experiments described up to this point were incapable of providing direct mechanistic information about the activities being observed. In this section we attempt to discern, unequivocally, whether processing of nicked 5’-dRP, gapped 5’-dRP, and AP sites is occurring via hydrolase or lyase mechanisms. DNA hydrolases typically employ active site metal ions to activate a water molecule, which then acts as the nucleophile in phosphodiester bond cleavage. In the lyase mechanism an active site primary amine attacks the C1’ open chain aldehyde of the abasic ribose to form a Schiff base (imine) - which then acts as an electron sink to promote β-elimination (Figure 4.10). This Schiff base intermediate can often be trapped by reduction with a hydride donor such as sodium
100 O O lysine-E O 32 O 32 O 32 O lysine-E -O P + -O P -O P H :NH2 O O O OH OH O OH NH+ O NaBH H H 4 H O H H 32 O O O O -O P lysine-E O OH NH H H O H lysine-E O O O adduct 1 32 O :NH2 32 O 32 O P P lysine-E P lysine-E -O -O -O O OH O O OH NH+ O OH :NH
H H H
H2O O O O NaBH4
O 32 O -O P lysine-E O OH NH H H
O adduct 2
Figure 4.10 Reductive trapping to probe for a lyase mechanism. The figure shows 5’-2- deoxyribose-5-phosphate (5’-dRP) being removed via a lyase mechanism. In the presence of sodium borohydride, the two imine intermediates can be trapped, resulting in a protein-DNA or protein-dRP adduct that can be detected by SDS PAGE. The assay works similarly for AP sites.
borohydride (Figure 4.10); trapping of the protein-substrate adduct is this manner is considered strong proof for the lyase mechanism.
Assay buffer and substrates were prepared as described in previous sections of this chapter except that DNA only contained 5’-32Pi label. Ice cold DNA and protein were mixed and allowed to incubate on ice for two minutes before adding fresh NaBH4 to a final concentration of 120 mM; final DNA concentration was 150 nM; final protein concentrations were 350 nM for ASFV APE and Pol X, and 50 nM for ASFV DNA
101 ligase and Pol β. Immediately after the addition of reducing agent, tubes were transferred to a 25 °C water bath, where incubation was continued for 90 minutes. Reactions were quenched by the addition of SDS loading buffer to a concentration of 1x, and were then precipitated by the addition of –20 °C trichloroacetic acid to a final concentration of 10%.
Samples were placed on ice for 30 minutes before pelleting protein in a microfuge at 4
°C. Protein pellets were washed with –20 °C acetone and then air dried before resuspension in 1x SDS loading buffer. Samples were resolved on 10% denaturing polyacrylamide gels (containing 5% stacking portions) run inside a plate incubator in order to maintain a temperature of ∼60 °C. High temperature reduced the number of bands visualized per lane, apparently by minimizing DNA secondary structure; similar difficulties with DNA secondary structure during SDS PAGE have been described by other groups as well (98). Including 8 M urea in the SDS gel also helped to alleviate the complication.
There were a few exceptions to the above protocol. The Pol β reactions were only incubated on ice for 15 seconds (rather than two minutes) before the addition of reducing agent and transfer to the water bath. NaCl was substituted for NaBH4 in specified reactions. Prior to quenching, some reactions were supplemented to attain 50 mM KCl, 2 mM free Mg2+, and 1 unit of DNaseI (Invitrogen). These were incubated at 25 °C for 60 minutes, than at 37 °C for 15 minutes. Quenching was as described above. Prior to quenching, other sets of samples were supplemented with 2 µg of trypsin; incubation and quenching were then as described for the DNaseI samples.
102 Prior to analyzing the data for each ASFV protein it is worth mentioning the robust ∼70 kD and ∼250 kD bands that are observed in all four of the below gels. Some characteristics of these bands are as follows: i) they contain protein (trypsin destroys them), ii) they contain the 32P label – otherwise they wouldn’t be visible – but don’t appear to contain large strands of DNA (because DNaseI does not reduce their size2), iii) they are more robust in the presence of borohydride – suggesting that they are retaining
32P label via a lyase-like mechanism, and iv) they appear to be dependent on the 32P label being removed from the DNA (neither the AP site nor the THF substrate give rise to these high MW bands). Our hypothesis is that the ∼70 kD band is BSA (which has a MW of
∼65 and is present in these assays at the very high concentration of 1.5 mg/mL – which is roughly 23 µM). It would seem that this highly basic protein – which contains 59 lysine residues (30-35 of which are solvent exposed) – is simply interacting nonspecifically with the 5’-32P-labeled ribose derivative that is liberated during the course of the reaction.
This is consistent with all data except for the fact that this band is still present when the catalytic protein (ASFV APE, Pol X, ASFV DNA ligase, and Pol β) has been heat denatured; with the catalyst inactivated we would expect for there to be little release of the 5’-32P-labeled ribose derivative and therefore minimal opportunity for adduct formation with BSA. This glitch can be overcome by assuming that the BSA might itself be responsible for 5’-dRP removal from the DNA, but this is not consistent with our activity assays in Sections 4.4-4.6. These issues should be easy to resolve and haven’t been dealt with yet simply due to lack of time. The band/smearing at about 250 kD is in
2 Alternatively, DNA may in fact be present, but the large size of these proteins may preclude detection of the modest drop in MW afforded by DNaseI digestion. 103 the vicinity of the stacking/separating gel interface and may therefore represent soluble
aggregate that wasn’t able to penetrate the separating gel very far.
ASFV APE In Figure 4.11 a band of about 40 kD is observable in the presence
of both nicked and gapped 5’-dRP substrates. That the band for gapped 5-dRP is more
intense than the band for nicked 5’-dRP is consistent with our previous finding that
ASFV APE is a better catalyst with the former. Since in the activity assay of Section 4.5
ASFV APE did not remove 5’-dRP from a nick, but instead actually protected it, it is
A B
AP =
N =
G =
T =
Figure 4.11 Reductive trapping of ASFV APE on abasic site variants. (A) Substrate structures and their abbreviations. AP site, AP; nicked 5’-dRP, N; gapped 5’-dRP, G; gapped tetrahydrofuran, T. Note the position of the 32P label. (B) SDS polyacrylamide gel displaying the products of the hydride trapping reactions. In lane 6, NaCl was substituted for NaBH4. The red arrow indicates the location of free ASFV APE.
104 surprising that the protein can still be trapped on this substrate; this raises questions about the specificity of this assay when performed at such high protein concentrations.
ASFV APE should have converted the AP site to gapped 5’-dRP – which the protein can be trapped on very efficiently (lane 3 of Figure 4.11). Why, then, do we not observe a distinct adduct band with the AP substrate. The answer to this lies in the fact that the AP substrate used here was labeled at the 5’ terminus only (since the reactive 5’- dRP is attached to the downstream oligonucleotide, observation of the covalent adduct is impossible in this particular assay).
DNaseI reduces the size of the trapped complex (lane 4) to a size very similar to free ASFV APE (34.5 kD). In the presence of trypsin (lane 5), the complex band is reduced to smaller fragments as expected. No bands whatsoever are seen for the reaction with THF (lane 8). Surprisingly, a moderately intense band the size of ASFV APE is visible in the absence of hydride (lane 6); we have observed this phenomenon consistently. The MW of this species suggests that it contains protein bound to 32P label that has already been released from DNA. This may reflect the precursor (i.e. prior to reduction) to adduct # 2 shown in Figure 4.10; note that inefficient release of this lysine- bound fragmented sugar moiety has been reported for a number of DNA polymerases
(95). However, that the imine intermediate would persist at 60 °C in water (while the gel is being run) seems doubtful…so further characterization of this species would be worthwhile.
Collectively, these data suggest that the activity ASFV APE displays against gapped 5’-dRP proceeds by a lyase mechanism. Regarding substrate preference, the results of the trapping assays reported here support the previous conclusions from the
105 activity assays. Accordingly, it seems that after incising an AP site very efficiently via a hydrolytic mechanism, ASFV APE can then remove the resulting 5’-blocking group, 5’- dRP, via a less efficient lyase mechanism. In contrast, if gap filling is effected prior to
5’-dRP removal, ASFV APE apparently protects the abasic residue until it can be removed by another enzyme. This sort of scenario might play a regulatory role akin to what has been described for Ape1 in the human BER system (73).
Pol X That Pol X can be trapped on an AP site, and on nicked and gapped 5’- dRP (Figure 4.12) immediately suggests that this enzyme employs a lyase mechanism
A B
AP =
N =
G =
T =
Figure 4.12 Reductive trapping of Pol X on abasic site variants. (A) Substrate structures and their abbreviations. AP site, AP; nicked 5’-dRP, N; gapped 5’-dRP, G; gapped tetrahydrofuran, T. Note the position of the 32P label. (B) SDS polyacrylamide gel displaying the products of the hydride trapping reactions. In lane 6, NaCl was substituted for NaBH4. The red arrow indicates the location of free Pol X.
106 against each of these substrates. The size of the complex bands for nicked and gapped
5’-dRP are reasonable (20 kD protein + ∼5.7 kD downstream oligonucleotide); the size of the complex band generated with the AP site migrates more slowly than expected (the upstream oligo generated after strand incision is only ∼ 7.5 kD, so the complex should be smaller than the observed 37 kD). At present we have no explanation for this phenomenon; the stock of Pol X used here is quite pure so the prospect that this adduct results from a protein contaminant is very unlikely.
The inability of DNaseI to modify the size of the Pol X complex with gapped 5’- dRP (lane 4 of Figure 4.12) is suspicious and will require further inquiry; most likely the
DNaseI used in this assay was faulty. As expected, treatment with trypsin (lane 5), heat denaturing Pol X (lane 7), and use of the THF analog (lane 8) all abolish complex formation. Similar to what was seen for ASFV APE, a faint Pol X-sized band appears even in the absence of NaBH4. Once again, this may represent the precursor to adduct #2 in Figure 4.10.
ASFV DNA ligase ASFV DNA ligase is trapped very efficiently on the AP site and on both the 5’-dRP variants (Figure 4.13). Results for this enzyme are qualitatively similar to those for Pol X, and are therefore not discussed extensively here.
Pol β As expected, Pol β is trapped on all three abasic substrates (Figure 4.14).
Unlike the three ASFV proteins, Pol β does not show a band the size of the free protein when hydride is excluded from the assay. This may suggest that breakdown of the lysine-sugar adduct is much more efficient for this protein than it is for the ASFV proteins.
107
A B
AP =
N =
G =
T =
Figure 4.13 Reductive trapping of ASFV DNA ligase on abasic site variants. (A) Substrate structures and their abbreviations. AP site, AP; nicked 5’-dRP, N; gapped 5’- dRP, G; gapped tetrahydrofuran, T. Note the position of the 32P label. (B) SDS polyacrylamide gel displaying the products of the hydride trapping reactions. In lane 6, NaCl was substituted for NaBH4. The red arrow indicates the location of free ASFV DNA ligase.
108
A B
AP =
N =
G =
T =
Figure 4.14 Reductive trapping of Pol β on abasic site variants. (A) Substrate structures and their abbreviations. AP site, AP; nicked 5’-dRP, N; gapped 5’-dRP, G; gapped tetrahydrofuran, T. Note the position of the 32P label. (B) SDS polyacrylamide gel displaying the products of the hydride trapping reactions. In lane 6, NaCl was substituted for NaBH4. The red arrow indicates the location of free Pol β. In the reaction involving AP substrate only, the concentration of Pol β was increased from 50 nM to 1.25 µM.
109 4.9 Adenylylation of ASFV DNA Ligase Enhances its Interaction with Nicked 5’-dRP
Since a lysine ε-NH2 serves as the site of adenylylation in ATP-dependent DNA ligases and is also the nucleophile in the standard 5’-dRP/AP lyase reaction (80), we asked the following questions about ASFV DNA ligase: i) Is the presumed lysine nucleophile that effects 5’-dRP removal the same lysine that gets charged with AMP?; and ii) If different lysine residues are employed for the ligase and lyase functions, how does the presence of the adenylylated lysine influence 5’-dRP processing?
Trapping assays were conducted with nicked 5’-dRP in a manner similar to those described in Section 4.8, but with one major modification. Whereas the assays conducted in Section 4.8 were performed with a DNA ligase stock containing both adenylylated and unadenylylated proteins, here we homogenized the protein – forcing it to be all one form or the other. Before being mixed with nicked 5’-dRP, ASFV DNA ligase was preincubated in one of the following ways. One aliquot of protein was preincubated with
ATP in the presence of MgCl2 to quantitatively adenylylate the enzyme (when purified from E. coli approximately 25% of the active enzyme is unadenylylated). The second aliquot of protein was preincubated with unlabeled, nicked DNA in the presence of
MgCl2 to quantitatively deadenylylate the enzyme (cold DNA was present in 4-fold excess over enzyme during this preincubation). Both periods of preincubation were at room temperature for roughly 20 minutes. At the end of this time, trapping assays were conducted in a manner similar to those described above.
Figure 4.15 demonstrates that the protein-DNA adduct is efficiently trapped using
ASFV DNA ligase that is quantitatively charged with AMP. This suggests that different residues are acting as the nucleophile in the lyase and ligase reactions. Further proof of
110
Figure 4.15 Adenylylation enhances the efficiency with which ASFV DNA ligase is trapped on nicked 5’-dRP. The “+” and “-“ lanes denote enzyme that was quantitatively adenylylated or deadenylylated, respectively, prior to running the trapping assay.
this will be sought by repeating the assay with unlabeled DNA, and protein that has been charged with [α-32P]ATP; after incubation with nicked 5’-dRP and hydride, an increase in the MW of the ligase-AMP band will be proof that different residues are employed in the lyase and ligase reactions.
Adenylylation of DNA ligases is known to effect conformational changes which apparently enhance their affinity for nicked substrate (28). Considering that deadenylylation reduces the efficiency with which ASFV DNA ligase is trapped on nicked 5’-dRP (Figure 4.15), it seems likely that the conformational changes presumed to be important for nick sensing in this enzyme are also important for recognition of 5’- dRP.3 Interestingly, this does not appear to be universally true as we have found, in cursory assays, that adenylylation mitigates the extent to which T4 DNA ligase is trapped on nicked 5’-dRP.
3 An alternative explanation for the reduced amount of trapped complex when using deadenylated ligase is that the presence of the residual unlabeled, nicked DNA (used to decharge the enzyme) interferes with trapping (by competing with the labeled 5’-dRP-containing substrate for the ligase). We will examine this issue in further detail by digesting the residual DNA after the decharging step with DNase, inactivating DNase with EDTA, and then repeating the trapping assay.
111 CHAPTER 5
DNA REPAIR IN ASFV
5.1 Chapter Overview
In this chapter we attempt to address issues that could not be conveniently dealt with in previous chapters; the disparate nature of this chapter’s sections does not, however, indicate a lack of importance (i.e. this is not simply a compilation of minutia). Herein we: i) assess the completeness of the ASFV DNA repair system, ii) analyze the substrate preference and translesion bypass capability of Pol X, and iii) reassess the mutagenic
DNA repair hypothesis.
5.2 Completeness of the ASFV DNA Repair System
5.2.1 Reconstitution of AP Site Repair
ASFV-encoded activities necessary for processing an abasic site by both the APE- and the AP lyase-initiated routes were demonstrated, individually, in Chapters 2-4. To
112 examine more rigorously the integrity of the ASFV AP site repair system, we attempted to reconstitute the APE-initiated route (the APE route was chosen instead of the AP lyase route since, on the basis of relative efficiencies, the former is expected to be employed predominantly). Assay buffer consisted of 50 mM Tris-borate, 100 mM KCl, 15% glycerol (v/v), pH → 7.8 at 37 °C with KOH, supplemented with 300 µM DTT, 15 mM
MgCl2, and 1.5 mg/mL BSA. As indicated in Figure 5.1, proteins and cosolvents were added to the DNA substrate in distinct phases. Final concentrations were as follows:
0.031 units/µL of uracil DNA glycosylase (UDG; from New England Biolabs), 28 nM
ASFV APE, 46 nM Pol X, 800 µM dTTP, 1 mM ATP, 47 nM ASFV DNA ligase, 155 nM A:U DNA substrate.
In Figure 5.1 the “percentage” of each DNA species is plotted as a function of time, where percentage = (the intensity of the band corresponding to the species of interest)/(summation of the intensities of all the bands observable at the given time point)*100. The data would be more appropriately represented in units of molar concentration, with the 5’ and 3’ labeled species being quantitated separately. However, since proteins and cosolvents were added sequentially – causing concentrations to fluctuate slightly with time – we have simply used the more facile unit of percentage here.
Simultaneous treatment of the uracil-containing substrate with UDG and ASFV
APE results in conversion of 45mer to 25mer and 5’-dRP-19mer, as expected (Figure
5.1). Subsequent addition of Pol X and dTTP gives rise to gap-filling (loss of 25mer with concomitant formation of 26mer) but no appreciable acceleration of 5’-dRP loss. Upon addition of ASFV DNA ligase and ATP, there is a burst in 45mer formation (with 113 concomitant consumption of 26mer and 19mer). This most likely corresponds to ligation of the nicked substrate molecules for which 5’-dRP had already been removed, whereas the second slower phase of ligation would appear to be limited by the rate of 5’-dRP removal.
Though it was expected (on the basis of findings presented in previous chapters), this reconstitution assay demonstrates unequivocally that ASFV would be capable of repairing abasic sites by the APE-initiated route without having to recruit host-derived repair factors.
UDG + ASFV APE ASFV DNA ligase + ATP incubation Pol X + dTTP incubation incubation
100 % 19mer % 5'-dRP-19mer % 25mer 80 % 26mer % 45mer
60
% species % 40
ligase+ATP incubation 20
0
0 50 100 150 200 time (min.)
Figure 5.1 Reconstitution of the APE-initiated route of AP site repair using ASFV proteins. Substrate, intermediate, and product structures are shown around the periphery. For simplicity, the time course plots the percentage (rather than the concentration) of each DNA species. The dashed vertical lines denote distinct incubation periods, for the indicated protein(s) and cosolvent, prior to addition of the next protein and cosolvent. See text for details.
114 5.2.2 A Missing DNA Glycosylase
Though its AP site repair capabilities have now been demonstrated, an ASFV-encoded activity for generating AP sites has not been identified; genes with homology to a known
DNA glycosylase are not present in the viral genome (16). Surprisingly, we have found that each of our preparations of ASFV APE is capable of incising a uracil-containing substrate similar to the one shown in Figure 5.1. Subsequent investigation indicated that this is due to contamination by E. coli UDG. After purification of non-tagged ASFV
APE by DEAE, P-11, hydroxyapatite, and S-100 chromatographies, the UDG contaminant was still present. Moreover, when His-tagged ASFV APE was bound to a cobalt column and washed with buffer containing 1 M KCl or 1% NP-40 detergent the
UDG contaminant was not entirely removed (data not shown), suggesting that interaction between these two proteins is likely to be specific.
UDG shares 57% sequence homology with its human counterpart UNG2 (when the targeting sequence of UNG2 is excluded). Because UNG2 exists, albeit at low concentrations, in the cytoplasm (99) we asked whether ASFV APE might recruit this glycosylase to sites of viral DNA repair; note that specific interaction between an APE and a DNA glycosylase has been reported for a number of systems (100, 101). To begin probing this possibility, the human UNG2 gene was subcloned and the His-tagged protein expressed and purified in a manner analogous to that described for ASFV APE in Chapter
3 (Figure 5.2). In pull-down assays, UNG2 did appear to bind non-tagged ASFV APE; however, in the control experiment non-tagged ASFV APE showed high affinity for the cobalt resin – calling into question the apparent interaction between UNG2 and ASFV
115
Figure 5.2 Silver-stained SDS polyacrylamide gel of human uracil DNA glycosylase (UNG2) after purification on cobalt and S-100 columns.
APE. In light of the persistent co-purification of ASFV APE and E. coli UDG, and the precedence for interaction between these two classes of proteins, we have not accepted the negative result from the pull-down assay as conclusive and the potential interaction between these proteins will be further probed by native PAGE in the near future.
If specific interaction between ASFV APE and UNG2 can in fact be demonstrated, it raises the following questions: i) Which, if any, other glycosylases can bind to ASFV APE?, and ii) Do these interactions occur in vivo to an extent sufficient to actually contribute to ASFV DNA repair? It is interesting to note that there is precedence for viral recruitment of a host DNA glycosylase: human UNG2 is specifically incorporated into the HIV capsid in order to minimize the level of uracil that persists in the genome after replication (102-104).
116 5.2.3 Searching for NIR Activity
A glycosylase-independent alternative to BER has recently been elucidated. In nucleotide incision repair (NIR), an APE hydrolyzes the sugar phosphate backbone 5’ to a nucleotide containing a damaged base to generate a 3’-OH and a 5’-“dangling nucleotide” (Figure 5.3). APE’s from both the exonuclease III [human Ape1 (105)] and the endonuclease IV [yeast Apn1 (106), E. coli endonuclease IV(107)] families have been demonstrated to posses NIR activity.
In the presence of Mg2+, ASFV APE did not display NIR activity against 45mer duplex substrates containing either 5,6-dihydrothymine (opposite to adenine) or 8-oxo- guanine (opposite to cytosine) (data not shown); the former lesion is processed efficiently by other APEs (106, 107). It is conceivable, on the basis of ASFV APE’s homology to both endonuclease IV and Apn1, that in the presence of its preferred metal cofactors
(Ni2+, Co2+, or Mn2+; see Figure 3.11) the ASFV enzyme will indeed show NIR activity.
NIR analyses with these alternative metal ions are under way.
OH O O
O P O O APE O O O O O P O O O P O P O O O O
Figure 5.3 Nucleotide incision repair. The pink wavy line represents a damaged base.
117 If NIR activity can be detected for ASFV APE, it will raise a new question: How is the 5’-dangling nucleotide product then processed? In the human system, it appears that flap endonuclease I (FEN1) removes the dangling nucleotide to generate a 5’- phosphate (107). Since a FEN1 homologue isn’t present in ASFV, NIR could be a potential dead end for the virus unless: i) a FEN1-like activity is present in one of the
ASFV-encoded proteins, or ii) host derived repair factors can be recruited (vide infra).
At our present state of knowledge, the most reasonable/least speculative way to account for the AP site processing activities of ASFV appears to be that they have simply been retained to process lesions generated via spontaneous or chemically-induced (59), rather than enzyme mediated, base loss.
5.2.4 Potential Recruitment of Host-derived Repair Factors by the ASFV-encoded PCNA Homologue
It is worthwhile to note that ASFV codes for a homologue of proliferating cell nuclear antigen (PCNA) (16), a protein known to interact with and/or promote the activity of at least 20 different DNA replication and repair proteins – including glycosylases, APEs, polymerases, ligases, etc. (108). The PCNA binding motif (the PIP box), which is found in myriad eukaryotic repair proteins, is also found in ASFV APE and DNA ligase, and a
PIP box variant is found in Pol X (Figure 5.4), suggesting that this binding interface has been conserved. This raises questions about whether host proteins are sequestered by
ASFV PCNA, and the extent to which they might compete with or supplement ASFV- derived proteins in repair and/or mutagenesis of the viral genome.
118 A
B
Figure 5.4 ASFV APE, DNA ligase, and Pol X each contain motifs known/suspected to interact with the “sliding clamps” PCNA and the β subunit of E. coli polymerase III. (A) Alignment of the canonical PCNA interacting protein box (PIP box) – found in eukaryotic DNA repair proteins – with sequences from ASFV APE and DNA ligase. While “x” represents any amino acid, “h” denotes the moderately hydrophobic residues L, I, or M, and “a” represents the hydrophobic, aromatic residues F, H, or Y. (B) Alignment of the β subunit interacting motif with the sequence of Pol X. In parts A and B, “N” represents the N-terminus of the protein, and the subsequent number is the number of amino acids that precede the sequence shown. The blue color serves to highlight residues conserved between the ASFV proteins and the canonical motifs.
5.3 Further Characterization of Pol X
5.3.1 Influence of 5’-2-Deoxyribose-5-phosphate on Gap Filling
As shown in Figure 1.1, Pol X might encounter single nucleotide gaps containing either
5’-phosphate or 5’-2-deoxyribose-5-phosphate (5’-dRP). Since the presence of 5’- phosphate enhances the gap-filling activity of Pol X (relative to a gap with 5’-OH) (17), we were interested to see what the influence of 5’-dRP might be; owing to the instability of 5’-dRP, as described previously, we used the 5’-Pi-THF analog instead. Burst assays, rather than single turnovers, were performed in order to obtain information about both the pre-steady-state and the steady-state phases of nucleotide incorporation. Incorporation of dCMP opposite to templating guanine was monitored by rapid quench. The assay consisted of 50 mM Tris-borate, 50 mM KCl, 3% glycerol (v/v), pH → 9.0 at 37 °C with
119 KOH, supplemented with 100 µg/mL BSA, 1 mM DTT, 9 mM dCTP, 10 mM free Mg2+,
200 nM gapped DNA substrate, and roughly 75 nM Pol X (all concentrations are post
mixing); note that the very high dCTP concentration of 9 mM was employed since
multiple saturation curves for G:C formation, performed by Sandeep Kumar, consistently
yielded a Kd,app of 1.0-1.5 mM.
Relative to 5’-phosphate, 5’-Pi-THF causes two very modest changes in the
kinetic properties of Pol X (Figure 5.5). First, 5’-Pi-THF decreases the single turnover
rate constant by about 8%. Even if this difference is real (i.e. is not due to error), it is so
small that it would appear to be biologically irrelevant and uninteresting. Second, 5’-Pi-
AB Pol X bursts
120
100
80 60
(nM) [26mer] 40 20 5'-Pi 5'-Pi-THF 0 0 10203040
time (seconds) C
-1 -1 substrate amplitude (nM) ksingle turnover (s ) ksteady-state (s ) 5’-Pi 71.5 0.25 0.0026 5’-Pi-THF 59.7 0.23 0.0160
Figure 5.5 Burst assays for Pol X-catalyzed gap filling (formation of G:C) in the context of either 5’-Pi or 5’-Pi-THF. (A) Gapped substrates containing either 5’-phosphate or the 5’-dRP analog 5’-Pi-THF. (B) Time course for G:C formation in both substrate contexts. (C) Kinetic parameters.
120 THF increases the steady state rate constant (obtained by dividing the steady state rate by the burst amplitude) by about 6-fold – suggesting that dissociation of the enzyme•product complex (which is the rate limiting step in the steady state) is enhanced.4 This difference, however, may be artifactual since the labile 5’-dRP (vs. the lyase-insensitive 5’-Pi-THF used in our assay) might be removed by Pol X immediately before or after gap filling at a rate similar to, or faster than, the rate of dissociation of the enzyme from a 5’-phosphate- containing product. This interpretation is consistent with the fact that for Pol β the steady state rate of gap filling is the same regardless of whether the 5’ moiety is Pi or dRP (83) – indicating that enzyme•product dissociation remains the rate limiting step.
Though the above analyses yielded results of no particular merit, it may still be worthwhile to perform a Pol X fidelity analysis for a limited set of base pairs using substrates containing either 5’-Pi or 5’-dRP (or 5’-Pi-THF). A change in fidelity as a function of the 5’ moiety of the gap would be interesting for two reasons. First, this would suggest differences in the inherent fidelity of the two alternative routes (APE vs.
AP lyase) of AP site processing. It is extremely speculative, but this sort of scenario – as discussed for mismatch editing in Chapter 4 – could enable the virus to toggle between low and high fidelity routes of repair, depending on the method by which repair is initiated. Second, a change in fidelity on going from 5’-Pi to 5’-dRP would open up the possibility that Pol X-catalyzed correct and incorrect nucleotide incorporation proceed by
4 A third, obvious difference between the two substrates is that of burst amplitude. While it might be taken to suggest that Pol X has a higher affinity for 5’-Pi-containing substrate (i.e. that the 5’-Pi-THF-containing substrate is not saturated in this assay), the lower burst amplitude for 5’-Pi-THF substrate is more simply a consequence of the pre-steady-state and the steady-state rates being too similar – which causes the burst amplitude to be underestimated. 121 different mechanisms5, which could then be probed kinetically and structurally in further detail.
5.3.2 Translesion Synthesis Capabilities
The paradigm of high-fidelity DNA replication and repair has been challenged in recent years by the discovery of error-prone DNA polymerases within each of the three kingdoms of life (109). These enzymes are generally characterized by: i) the ability to catalyze nucleotide incorporation opposite to, or down stream of, templating nucleotides that are chemically damaged – hence the title “translesion polymerase”, and ii) low- fidelity DNA synthesis (2, 110). These unique capacities, depending on the context in which they are employed, can serve two purposes. First, by synthesizing past lesions that would otherwise block the replisome, translesion polymerases help ensure survival.
Second, by replicating DNA with low-fidelity, translesion polymerases can promote genetic variability (1).
In light of the error-proneness of Pol X on undamaged templates (both in the gapped and template/primer contexts) (18), we asked the following questions: i) Is Pol X capable of performing translesion synthesis?, ii) If Pol X can bypass lesions, which
5 In the most simplistic scenario the detailed topology of enzyme and substrate atoms during the course of the reaction would look very similar for both correct and incorrect dNTP incorporation – though the free energy barriers for the microscopic steps might vary for correct vs. incorrect. Alternatively, different relative positioning of protein, DNA, and dNTP atoms might be necessary for correct vs. incorrect incorporation. Note that the latter scenario must certainly be the case; an incoming purine must “look” different than an incoming pyrimidine. The point of interest, therefore, is that of qualitative similarity. For example, it is conceivable that in order to accommodate the bulky G:G mismatch, Pol X flips the incoming base about the glycosidic bond to the syn conformation in order to allow for a Hoogstein-like base pairing scheme. In the absence of direct structural evidence, this type of mechanistic difference between correct and incorrect dNTP incorporation might be hinted at by kinetic differences such as the fidelity value changing as a function of the 5’-moiety of the gap. Important to this sort of analysis is the fact that correct incorporation is relatively unperterbed by the presence of 5’-Pi-THF. 122 lesions does it bypass most efficiently?, iii) If Pol X can bypass lesions, which nucleotides does it insert preferentially opposite to the damaged, templating base (i.e. what is its fidelity during translesion synthesis)?, and iv) Having obtained answers to the previous three questions, does the data suggest that Pol X has evolved primarily to function as A) a mutase, optimized for incorporating the wrong nucleotide opposite to undamaged, gapped templates, or B) a translesion polymerase, optimized for bypassing lesions in a template-primer context?
To begin addressing these questions, working with me, Sandeep Kumar performed running-start assays with templates containing either 8-oxo-deoxyguanosine
[8-oxo-G; one of the most abundant, and stable, oxidative lesions known (59, 111)], 5,6- dihydro-deoxythymidine (DHT; a lesion commonly formed by ionizing radiation under anoxic conditions), or the abasic site analogue THF. Reactions consisted of 50 mM Tris- borate, 100 mM KCl, 15% glycerol (v/v), pH → 7.8 at 37 °C with KOH, supplemented with 200 µg/mL BSA, 2 mM DTT, 1 mM of each dNTP, 20 mM total Mg2+, 100 nM template/primer substrate, and 400 nM Pol X.
As shown in Figure 5.6, Pol X is capable of bypassing 8-oxo-G with relatively high efficiency. Though very little “pausing” is observed immediately upstream of, opposite to, or immediately downstream of this lesion (i.e. the 19mer, 20mer, and 21mer products do not accumulate extensively), there is a minor accumulation of 20mer – suggesting that extension of the base pair containing 8-oxo-G occurs with reduced efficiency.
DHT is also bypassed by Pol X, albeit with reduced efficiency relative to 8-oxo-G
(Figure 5.7). For DHT, accumulation of both the 20mer and the 21mer products 123
C
H A H N 2 N O O H O N N O N O NH O NH O NH O N O N O N N N NH2 NH2 O
O deoxyguanosine O anti-8-oxo-deoxyguanosine O syn-8-oxo-deoxyguanosine
B
Figure 5.6 Pol X-catalyzed running-start bypass of 8-oxo-deoxyguanosine. (A) Comparison of unmodified deoxyguanosine with 8-oxo-G in both the anti and the syn conformations. NMR studies have demonstrated that in duplex DNA 8-oxo-G preferentially exists in the atypical syn conformation (112); this apparently relieves steric clash which otherwise exists between the 8-oxo group and the ribose ring. (B) Running- start substrate; “X” denotes 8-oxo-G – the position of which is highlighted by the red arrow. (C) Primer extension as a function of time. The red arrow highlights the band corresponding to the nucleotide which has been incorporated opposite to 8-oxo-G.
124
C
A O O NH NH O O N N O O O O
O deoxythymidine O 5,6-dihydro-deoxythymidine
B
Figure 5.7 Pol X-catalyzed running-start bypass of 5,6-dihydro-deoxythymidine. (A) Comparison of unmodified deoxythymidine with DHT. (B) Running-start substrate; “X” denotes DHT – the position of which is highlighted by the red arrow. (C) Primer extension as a function of time. The red arrow highlights the band corresponding to the nucleotide which has been incorporated opposite to DHT.
125
C A
O O H H
O tetrahydrofuran
B
Figure 5.8 Pol X-catalyzed running-start bypass of the abasic site analog tetrahydrofuran. (A) Structure of the THF analog. (B) Running-start substrate; “X” denotes THF – the position of which is highlighted by the red arrow. (C) Primer extension as a function of time. The red arrow highlights the band corresponding to the nucleotide which has been incorporated opposite to THF.
126 indicates reduced efficiency of incorporation opposite to and immediately downstream of the lesion.
Pol X experiences no inhibition upstream of or opposite to the abasic site analog,
THF, but it is completely incapable of extending beyond the lesion (Figure 5.8).
Though they are just cursory, the above assays indicate that Pol X does indeed possess translesion synthesis capabilities and might therefore contribute to the fitness of the virus by “rescuing” the viral replisome were it to get stuck opposite to, or adjacent to,
8-oxo-G or DHT. The lesion preference of Pol X, as well as the fidelity of lesion bypass are currently under investigation. Even if our future analyses suggest that Pol X has evolved primarily to bypass DNA lesions, its error-proneness on undamaged templates will not be discounted. As long as this enzyme has access to undamaged templates, there will be a propensity to generate mismatches.
5.4 Reassessing the Mutagenic DNA Repair Hypothesis
The hypervariability of RNA viruses and the comparative stability of DNA viruses are largely a consequence of their respective replication fidelities. The mutagenic DNA pathway considered in this report is a mechanism by which a DNA virus might achieve moderate levels of sequence diversity without the need for highly error-prone genome replication. Note that high frequencies of mutation are not expected to be well tolerated in a large virus such as ASFV – whose genome is ∼180 kb (113) and consists of approximately 150 genes (16); the greater the number of genes required for viability, the more difficult it becomes to obtain a full complement of functional genes in the face of high mutation frequencies.
127 Though infidelity during replication may in fact contribute to ASFV variability
(the ASFV replicative polymerase is unstudied in this regard), we suggest that point mutations introduced during DNA repair are likely to contribute to genome diversification in ASFV. That the genes for ASFV APE, Pol X, and DNA ligase were initially retained in the ASFV genome suggests the usefulness/necessity of these proteins in viral DNA repair (the replacement of a chemically damaged nucleotide with an undamaged nucleotide). That both the Pol X and ASFV DNA ligase proteins are among the most error-prone/error-tolerant members of their respective enzyme families is not likely to be coincidental and supports their potential usefulness in viral mutagenesis (by replacing a chemically damaged nucleotide with an undamaged, though incorrect, nucleotide).
In contrast to the diversification processes associated with somatic hypermutation
– where error-prone DNA synthesis is targeted to discrete regions of immunoglobulin loci (114) – the proposed ASFV mutagenic BER system would be capable, in theory, of inducing mutations throughout the entire viral genome. This does not appear to be the case, however, since some regions of the ASFV genome are more prone to variation due to base substitution or small insertions and deletions than are other regions (15, 113).
This phenomenon is likely a consequence of the mechanism by which the viral genome is replicated and assembled. Short ASFV DNA fragments are present in the host cell nucleus at early stages of virus replication, and subsequently larger DNA intermediates are detected in the cytoplasm (12, 115). With the ASFV genome being synthesized and assembled via discreet nuclear and cytoplasmic phases of DNA processing, perhaps the extent of DNA damage and the extent of DNA repair by either host nuclear repair factors
128 or ASFV repair factors dictates the location, extent, and type of mutations sustained by the viral genome. This would explain some of the apparent inconsistencies in the literature regarding the variability of ASFV. Many questions remain, and ultimately the contribution of the mutagenic enzyme system reported here will need to be borne out by in vivo analyses.
129 BIBLIOGRAPHY
1. Rattray, A. J., and Strathern, J. N. (2003) Error-Prone DNA Polymerases: When Making a Mistake is the Only Way to Get Ahead, Annu. Rev. Genet. 37, 31-66. 2. Tippin, B., Pham, P., and Goodman, M. F. (2004) Error-prone replication for better or worse, TRENDS in Mircrobiology 12, 288-295. 3. Bjedov, I., Tenaillon, O., Gerard, B., Souza, V., Denamur, E., Radman, M., Taddei, F., and Matic, I. (2003) Stress-Induced Mutagenesis in Bacteria, Science 300, 1404-1409. 4. Schaaper, R. M., and Dunn, R. L. (1987) Spectra of spontaneous mutations in Escherichia coli strains defective in mismatch correction: the nature of in vivo DNA replication errors, Proceedings of the National Academy of Sciences 84, 6220-6224. 5. LeClerc, J. E., Li, B., Payne, W. L., and Cebula, T. A. (1996) High Mutation Frequencies Among Escherichia coli and Salmonella Pathogens, Science 274, 1208-1211. 6. Perrino, F. W., Preston, B. D., Sandell, L. L., and Loeb, L. A. (1989) Extension of mismatched 3' termini of DNA is a major determinant of the infidelity of human immunodeficiency virus type 1 reverse transcriptase, Proceedings of the National Academy of Sciences 86, 8343-8347. 7. Wolfs, T. F., Zwart, G., Bakker, M., Valk, M., Kuiken, C. L., and Goudsmit, J. (1991) Naturally occurring mutations within HIV-1 V3 genomic RNA lead to antigenic variation dependent on a single amino acid substitution, Virology 185, 195-205. 8. Huebner, A., Kruhoffer, M., Grosse, F., and Krauss, G. (1992) Fidelity of human immunodeficiency virus type I reverse transcriptase in copying natural RNA, Journal of Molecular Biology 223, 595-600. 9. Vinuela, E. (1985) African Swine Fever Virus, Current Topics in Microbiology and Immunology 116, 151-171. 10. Thomson, G. R. (1985) The epidemilogy of African swine fever: the role of free- living hosts in Africa, Onderstepoort journal of veterinary research 52, 201-209. 11. Enjuanes, L., Cubero, I., and Vinuela, E. (1977) Sensitivity of Macrophages from Different Species to African Swine Fever (ASF) Virus, Journal of General Virology 34, 455-463. 12. Rojo, G., Garcia-Beato, R., Vinuela, E., Salas, M. L., and Salas, J. (1999) Replication of African Swine Fever Virus DNA in Infected Cells, Virology 257, 524-536. 13. Brookes, S. M., Dixon, L. K., and Parkhouse, R. M. E. (1996) Assembly of African Swine Fever Virus: Quantitative Ultrastructural Analysis in Vitro and in Vivo, Virology 224, 84-92. 14. Garcia-Barreno, B., Sanz, A., Nogal, M. L., Vinuela, E., and Enjuanes, L. (1986) Monoclonal antibodies to African swine fever virus: antigenic differences among
130 field virus isolates and viruses passaged in cell culture, Journal of Virology 58, 385-392. 15. Dixon, L. K., and Wilkinson, P. J. (1988) Genetic Diversity of African Swine Fever Virus Isolates from Soft Ticks (Ornithodoros moubata) Inhabiting Warthog Burrows in Zambia, Journal of General Virology 69, 2981-2993. 16. Yanez, R. J., Rodriguez, J. M., Nogal, M. L., Yuste, L., Enriquez, C., Rodriguez, J. F., and Vinuela, E. (1995) Analysis of the Complete Nucleotide Sequence of African Swine Fever Virus, Virology 208, 249-278. 17. Oliveros, M., Yanez, R. J., Salas, M. L., Salas, J., Vinuela, E., and Blanco, L. (1997) Characterization of an African Swine Fever Virus 20-kDa DNA Polymerase Involved in DNA Repair, Journal of Biological Chemistry 272, 30899-30910. 18. Showalter, A. K., and Tsai, M.-D. (2001) A DNA Polymerase with Specificity for Five Base Pairs, Journal of the American Chemical Society 123, 1776-1777. 19. Showalter, A. K., Byeon, I.-J. L., Su, M.-I., and Tsai, M.-D. (2001) Solution structure of a viral DNA polymerase X and evidence for a mutagenic function, Nature Structural Biology 8, 942-946. 20. Lehman, I. R. (1974) DNA Ligase: Structure, Mechanism, and Function, Science 186, 790-797. 21. Meselson, M., and Weigle, J. J. (1961) Proceedings of the National Academy of Sciences 47, 857. 22. Young, E. T., and Sinsheimer, R. L. (1964) Journal of Molecular Biology 10, 562. 23. Gellert, M. (1967) Proceedings of the National Academy of Sciences 57, 148-. 24. Weiss, B., and Richardson, C. C. (1967) Proceedings of the National Academy of Sciences 57, 1021-. 25. Olivera, B. M., and Lehman, I. R. (1967) Proceedings of the National Academy of Sciences 57, 1426-. 26. Gefter, M. L., Becker, A., and Hurwitz, J. (1967) Proceedings of the National Academy of Sciences 58, 240-. 27. Cozarelli, N. R., Melechen, N. E., Jovin, T. M., and Kornberg, A. (1967) Biochemical and Biophysical Research Communications 28, 578-. 28. Odell, M., Sriskanda, V., Shuman, S., and Nikolov, D. B. (2000) Crystal Structure of Eukaryotic DNA Ligase-Adenylate Illuminates the Mechanism of Nick Sensing and Strand Joining, Molecular Cell 6, 1183-1193. 29. Smith, G. L., Chan, Y. S., and Kerr, S. M. (1989) Transcriptional mapping and nucleotide sequence of a vaccinia virus gene encoding a polypeptide with extensive homology to DNA ligases, Nucleic Acids Research 17, 9051-9062. 30. Hammond, J. M., Kerr, S. M., Smith, G. L., and Dixon, L. K. (1992) An African swine fever virus gene with homology to DNA ligases, Nucleic Acids Research 20, 2667-2671. 31. Yanez, R. J., and Vinuela, E. (1993) African Swine Fever Virus Encodes a DNA Ligase, Virology 193, 531-536. 32. Modrich, P., and Lehman, I. R. (1973) Deoxribonucleic Acid Ligase, The Journal of Biological Chemistry 248, 7502-7511.
131 33. Sriskanda, V., and Shuman, S. (2002) Role of Nucleotidyl Transferase Motif V in Strand Joining by Chlorella Virus DNA Ligase, The Journal of Biological Chemistry 277, 9661-9667. 34. Barany, F. (1991) Genetic disease detection and DNA amplification using cloned thermostable ligase, Proceedings of the National Academy of Sciences 88, 189- 193. 35. Landegren, U., Kaiser, R., Sanders, J., and Hood, L. (1988) A ligase-mediated gene detection technique, Science 241, 1077-80. 36. Bhagwat, A. S., Sanderson, R. J., and Lindahl, T. (1999) Delayed DNA joining at 3' mismatches by human DNA ligases, Nucleic Acids Research 27, 4028-4033. 37. Husain, I., Tomkinson, A. E., Burkhart, W. A., Moyer, M. B., Ramos, W., Mackey, Z. B., Besterman, J. M., and Chen, J. (1995) Purification and Characterization of DNA Ligase III from Bovine Testes, The Journal of Biological Chemistry 270, 9683-9690. 38. Tong, J., Cao, W., and Barany, F. (1999) Biochemical properties of a high fidelity DNA ligase from Thermus Species AK16D, Nucleic Acids Research 27, 788-794. 39. Tomkinson, A. E., Tappe, N. J., and Friedberg, E. C. (1992) DNA Ligase I from Saccharomyces cerevisiae: Physical and Biochemical Characterization of the CDC9 Gene Product, Biochemistry 31, 11762-11771. 40. Shuman, S. (1995) Vaccinia Virus DNA Ligase: Specificity, Fidelity, and Inhibition, Biochemistry 34, 16138-16147. 41. Luo, J., Bergstrom, D. E., and Barany, F. (1996) Improving the fidelity of Thermus thermophilus DNA ligase, Nucleic Acids Research 24, 3071-3078. 42. Liu, P., Burdzy, A., and Sowers, L. C. (2004) DNA ligases ensure fidelity by interrogating minor groove contacts, Nucleic Acids Research 32, 4503-4511. 43. Mizuuchi, K., Nobbs, T. J., Halford, S. E., Adzuma, K., and Qin, J. (1999) A new method for determining the stereochemistry of DNA cleavage reactions, Biochemistry 38, 4640-4648. 44. Sriskanda, V., and Shuman, S. (1998) Specificity and fidelity of strand joining by Chlorella virus DNA ligase, Nucleic Acids Research 26, 3536-3541. 45. Lu, J., J. Tong, Feng, H., Huang, J., Alfonso, C. L., Rock, D. L., Barany, F., and Cao, W. (2004) Unique ligation properties of eukaryotic NAD+-dependent DNA ligase from Melanoplus sanguinipes entomopoxvirus, Biochemical and Biophysical Research Communications 1701, 37-48. 46. Shuman, S., and Lima, C. D. (2004) The polynucleotide ligase and RNA capping enzyme superfamily of covalent nucleotidyltransferases, Current Opinion in Structural Biology 14, 757-764. 47. Lamarche, B. J., Showalter, A. K., and Tsai, M.-D. (2005) An Error-Prone Viral DNA Ligase, Biochemistry ASAP Article; DOI: 10.1021/bi047706g. 48. Pascal, J. M., O'Brien, P. J., Tomkinson, A. E., and Ellenberger, T. (2004) Human DNA ligase I completely encircles and partially unwinds nicked DNA, Nature 432, 473-478. 49. Subramanya, H. S., Doherty, A. J., Ashford, S. R., and Wigley, D. B. (1996) Crystal Structure of an ATP-Dependent DNA Ligase from Bacteriophage T7, Cell 85, 607-615.
132 50. Singleton, M. R., Hakansson, K., Tinson, D. J., and Wigley, D. B. (1999) Structure of the adenylation domain of an NAD+-dependent DNA ligase, Structure 7, 35-42. 51. Lee, J. Y., Chang, C., Song, H. K., Moon, J., Yang, J. K., Kim, H.-K., Kwon, S.- T., and Suh, S. W. (2000) Crystal structure of NAD+-dependent DNA ligase: modular architecture and functional implications, The EMBO Journal 19, 1119- 1129. 52. Sibanda, B. L., Critchlow, S. E., Begun, J., Pei, X. Y., Jackson, S. P., Blundell, T. L., and Pellegrini, L. (2001) Crystal structure of an Xrcc4-DNA ligase IV complex, Nature Structural Biology 8, 1015-1019. 53. Krishnan, V. V., Thornton, K. H., Thelen, M. P., and Cosman, M. (2001) Solution Structure and Backbone Dynamics of the Human DNA Ligase IIIalpha BRCT Domain, Biochemistry 40, 13158-13166. 54. Odell, M., Malinina, L., Sriskanda, V., Teplova, M., and Shuman, S. (2003) Analysis of the DNA joining repertoire of Chlorella virus DNA ligase and a new crystal structure of the ligase-adenylate intermediate, Nucleic Acids Research 31, 5090-5100. 55. Gajiwala, K. S., and Pinko, C. (2004) Structural Rearrangement Accompanying NAD+ Synthesis within a Bacterial DNA Ligase Crystal, Structure 12, 1449- 1459. 56. Kulczyk, A. W., Yang, J.-C., and Neuhaus, D. (2004) Solution Structure and DNA Binding of the Zinc-finger Comain from DNA Ligase IIIalpha, Journal of Molecular Biology 341, 723-738. 57. Li, Z., Woo, C. J., Iglesias-Ussel, M. D., Ronai, D., and Scharff, M. D. (2004) The generation of antibody diveristy through somatic hypermutation and class switch recombination, Genes and Development 18, 1-11. 58. Riley, P. A. (1994) Free radicals in biology: oxidative stress and the effects of ionizing radiation, International Journal of Radiation Biology 65, 27-33. 59. Lindahl, T. (1993) Instability and decay of the primary structure of DNA, Nature 362, 709-715. 60. Ljungman, M., and Hanawalt, P. C. (1992) Efficient protection against oxidative DNA damage in chromatin, Molecular Carcinogenesis 5, 264-269. 61. Moyer, R., Marien, K., Holde, K. v., and Bailey, G. (1989) Site-specific aflatoxin B1 adduction of sequence-positioned nucleosome core particles, Journal of Biological Chemistry 264, 12226-12231. 62. Lawley, P. D. (1989) Mutagens as carcinogens: development of current concepts, Mutation Research 213, 3-25. 63. Setlow, R. B., Swenson, P. A., and Carrier, W. L. (1963) Thymine dimers and inhibition of DNA synthesis by ultraviolet irradiation of cells, Science 142, 1464- 1466. 64. Hutchinson, F. (1985) Chemical changes induced in DNA by ionizing radiation, Progress in Nucleic Acid Research and Molecular Biology 32, 115-154. 65. Lindahl, T. (2000) Suppression of spontaneous mutagenesis in human cells by DNA base excision repair, Mutation Research 462, 129-135.
133 66. Lindahl, T., and Nyberg, B. (1972) Rate of depurination of native deoxyribonucleic acid, Biochemistry 11, 3610-3618. 67. Cappelli, E., Hazra, T., Hill, J. W., Slupphaug, G., Bogliolo, M., and Frosina, G. (2001) Rates of base excision repair are not solely dependent on levels of initiating enzymes, Carcinogenesis 22, 387-393. 68. Chen, D. S., Herman, V., and Demple, B. (1991) Two distinct human DNA diesterases that hydrolyze 3'-blocking deoxyribose fragments from oxidized DNA, Nucleic Acids Research 19, 5907-5914. 69. Friedberg, E. C., Walker, G. C., and Siede, W. (1995) DNA Repair and Mutagenesis, American Society of Microbiology, Washington, D.C. 70. Dixon, L. K., Twigg, S. R. F., Baylis, S. A., Vydelingum, S., Bristow, C., Hammond, J. M., and Smith, G. L. (1994) Nucleotide sequence of a 55 kpp region from the right end of the genome of a pathogenic African swine fever virus isolate (Malawi LIL20/1), Journal of General Virology 75, 1655-1684. 71. Garcia-Escudero, R., Garcia-Diaz, M., Salas, M. L., Blanco, L., and Salas, J. (2003) DNA Polymerase X of African Swine Fever Virus: Insertion Fidelity on Gapped DNA Substrates and AP Lyase Activity Support a Role in Base Excision Repair of Viral DNA, Journal of Molecular Biology 326, 1403-1412. 72. Corpet, F. (1988) Multiple sequence alignment with hierarchical clustering, Nucleic Acids Research 16, 10881-10890. 73. Mol, C. D., Hosfield, D. J., and Tainer, J. A. (2000) Abasic site recognition by two apurinic/apyrimidinic endonuclease families in DNA base excision repair: the 3' ends justify the means, Mutation Research 460, 211-229. 74. Levin, J. D., Shapiro, R., and Demple, B. (1991) Metalloenzymes in DNA Repair, Journal of Biological Chemistry 266, 22893-22898. 75. Chou, K.-M., and Cheng, Y.-C. (2002) An exonucleolytic activity of human apurinic/apyrimidinic endonuclease on 3' mispaired DNA, Nature 415, 655-659. 76. Wong, D., DeMott, M. S., and Demple, B. (2003) Modulation of the 3'-5'- Exonuclease Activity of Human Apurinic Endonuclease (Ape1) by Its 5'-incised Abasic DNA Product, The Journal of Biological Chemistry 278, 36242-36249. 77. Guckian, K. M., Schweitzer, B. A., Ren, R. X.-F., Sheils, C. J., Tahmassebi, D. C., and Kool, E. T. (2000) Factors Contributing to Aromatic Stacking in Water: Evaluation in the Context of DNA, Journal of the American Chemical Society 122, 2213-2222. 78. Bommarito, S., Peyret, N., and Jr., J. S. (2000) Thermodynamic parameters for DNA sequences with dangling ends, Nucleic Acids Research 28, 1929-1934. 79. Hosfield, D. J., Guan, Y., Haas, B. J., Cunningham, R. P., and Tainer, J. A. (1999) Structure of the DNA repair enzyme endonuclease IV and its DNA complex: double-nucleotide flipping at abasic sites and three-metal-ion catalysis, Cell 98, 397-408. 80. Prasad, R., Beard, W. A., Strauss, P. R., and Wilson, S. H. (1998) Human DNA Polymerase B Deoxyribose Phosphate Lyase, The Journal of Biological Chemistry 273, 15263-15270.
134 81. Price, A., and Lindahl, T. (1991) Enzymatic Release of 5'-Terminal Deoxyribose Phosphate Residues from Damaged DNA in Human Cells, Biochemistry 30, 8631-8637. 82. Lindahl, T., and Andersson, A. (1972) Rate of Chain Breakage at Apurinic Sites in Double-Stranded Deoxyribonucleic Acid, Biochemistry 11, 3618-3622. 83. Srivastava, D. K., Berg, B. J. V., Prasad, R., Molina, J. T., Beard, W. A., Tomkinson, A. E., and Wilson, S. H. (1998) Mammalian Abasic Site Base Excision Repair, Journal of Biological Chemistry 273, 21203-21209. 84. Dunlap, C. A., and Tsai, M.-D. (2002) Use of 2-aminopurine and Tryptophan Fluorescence as Probes in Kinetic Analyses of DNA Polymerase Beta, Biochemistry 41, 11226-11235. 85. Bogenhagen, D. F., and Pinz, K. G. (1998) The Action of DNA Ligase at Abasic Sites in DNA, The Journal of Biological Chemistry 273, 7888-7893. 86. Pinz, K. G., and Bogenhagen, D. F. (1998) Efficient Repair of Abasic Sites in DNA by Mitochondrial Enzymes, Molecular and Cellular Biology 18, 1257-1265. 87. Kroeger, K. M., Hashimoto, M., Kow, Y. W., and Greenberg, M. M. (2003) Cross-Linking of 2-Deoxyribonolactone and Its B-Elimination Product by Base Excision Repair Enzymes, Biochemistry 42, 2449-2455. 88. DeMott, M. S., Beyret, E., Wong, D., Bales, B. C., Hwang, J.-T., Greenberg, M. M., and Demple, B. (2002) Covalent Trapping of Human DNA Polymerase B by the Oxidative DNA Lesion 2-Deoxyribonolactone, Journal of Biological Chemistry 277, 7637-7640. 89. Greenberg, M. M., Weledji, Y. N., Kim, J., and Bales, B. C. (2004) Repair of Oxidized Abasic Sites by Exonuclease III, Endonuclease IV, and Endonuclease III, Biochemistry 43, 8178-8183. 90. Vidal, A. E., Boiteux, S., Hickson, I. D., and Radicella, J. P. (2001) XRCC1 coordinates the initial and late stages of DNA abasic site repair through protein- protein interactions, EMBO J 20, 6530-6539. 91. Hoehn, S. T., Turner, C. J., and Stubbe, J. (2001) Solution structure of an oligonucleotide containing an abasic site: evidence for an unusual deoxyribose conformation, Nucleic Acids Research 29, 3413-3423. 92. Ciarrocchi, G., Lestingi, M., Fontana, M., Spadari, S., and Montecucco, A. (1991) Correlation between anthracycline structure and human DNA ligase inhibition, Biochemical Journal 279, 141-146. 93. Bhagwat, M., and Gerlt, J. A. (1996) 3'- and 5'-Strand Cleavage Reactions Catalyzed by yhe Fpg Protein from Esherichia coli Occur via Successive B- and d-Elimination Mechanisms, Resepectively, Biochemistry 35, 659-665. 94. Jiang, D., Hatahet, Z., Melamede, R. J., Kow, Y. W., and Wallace, S. S. (1997) Characterization of Escherichia coli Endonuclease VIII, Journal of Biological Chemistry 272, 32230-32239. 95. Haracska, L., Prakash, L., and Prakash, S. (2003) A mechanism for the exclusion of low-fidelity human Y-family DNA polymerases from base excision repair, Genes and Development 17, 2777-2785.
135 96. Pinz, K. G., and Bogenhagen, D. F. (2000) Characterization of a catalytically slow AP lyase activity in DNA polymerase gamma and other family A DNA polymerases, Journal of Biological Chemistry 275, 12509-12514. 97. Piersen, C. E., McCullough, A. K., and Lloyd, R. S. (2000) AP lyases and dRPases: commonality of mechanism, Mutation Research 459, 43-53. 98. Piersen, C. E., Prasad, R., Wilson, S. H., and Lloyd, R. S. (1996) Evidence for an Imino Intermediate in the DNA Polymerase Beta Deoxyribose Phosphate Excision Reaction, The Journal of Biological Chemistry 271, 17811-17815. 99. Nagelhus, T. A., Slupphaug, G., Lindmo, T., and Krokan, H. E. (1995) Cell cycle regulation and subcellular localization of the major human uracil-DNA glycosylase, Experimental Cell Research 220, 292-297. 100. Hegde, V., Wang, M., and Deutsch, W. A. (2004) Human ribosomal protein S3 interacts with DNA base excision repair proteins hAPE/Ref-1 and hoGG1, Biochemistry 43, 14211-14217. 101. Back, J. H., Chung, J. H., Park, Y. I., Kim, K.-S., and Han, Y. S. (2003) Endonuclease IV enhances base excision repair of endonuclease III from Methanobacterium thermoautotrophicum, DNA Repair 2, 455-470. 102. Priet, S., Gros, N., Navarro, J.-M., Boretto, J., Canard, B., Querat, G., and Sire, J. (2005) HIV-1-associated uracil DNA glycosylase activity controls dUTP misincorporation in viral DNA and is essential to the HIV-1 life cycle, Molecular Cell 17, 479-490. 103. Priet, S., Navarro, J.-M., Gros, N., Querat, G., and Sire, J. (2003) Role of the human uracil DNA glycosylase UNG2 enzyme in the HIV-1 life cycle, Recent Research Developments in Molecular Biology 1, 45-50. 104. Chen, R., Rouzic, E. L., Kearney, J. A., Mansky, L. M., and Benichou, S. (2004) Vpr-mediated Incorporation of UNG2 into HIV-1 Particles Is Required to Modulate the Virus Mutation Rate and for Replication in Macrophages, Journal of Biological Chemistry 279, 28419-28425. 105. Gros, L., Ishchenko, A. A., Ide, H., Elder, R. H., and Saparbaev, M. K. (2004) The major human AP endonuclease (Ape1) is involved in the nucleotide incision repair pathway, Nucleic Acids Research 32, 73-81. 106. Ishchenko, A. A., Sanz, G., Privezentzev, C. V., Maksimenko, A. V., and Saparbaev, M. (2003) Characterization of new substrate specificities of Escherichia coli and Saccharomyces cerevisiae AP endonucleases, Nucleic Acids Research 31, 6344-6353. 107. Ischenko, A. A., and Saparbaev, M. K. (2002) Alternative nucleotide incision repair pathway for oxidative DNA damage, Nature 415, 183-187. 108. Vivona, J. B., and Kelman, Z. (2003) The diverse spectrum of sliding clamp interacting proteins, FEBS Letters 546, 167-172. 109. Ohmori, H., Friedberg, E. C., Fuchs, R. P. P., Goodman, M. F., Hanaoka, F., Hinkle, D., Kunkel, T. A., Lawrence, C. W., Livneh, Z., Nohmi, T., Prakash, L., Prakash, S., Todo, T., Walker, G. C., Wang, Z., and Woodgate, R. (2001) The Y- Family of DNA Polymerases, Molecular Cell 8, 7-8.
136 110. Johnson, R. E., Washington, M. T., Haracska, L., Prakash, S., and Prakash, L. (2000) Eukaryotic polymerases iota and zeta act sequentially to bypass DNA lesions, Nature 406, 1015-1019. 111. Dizdaroglu, M. (1985) Formation of an 8-hydroxyguanine moiety in deoxynucleic acid on gamma-irradiation in aqueous solution, Biochemistry 24, 4476-4481. 112. Kouchakdjian, M., Bodepudi, V., Shibutani, S., Eisenberg, M., Johnson, F., Grollman, A. P., and Patel, D. J. (1991) NMR Structural Studies of the Ionizing Radiation Adduct 7-Hydro-8-oxodeoxyguanosine (8-oxo-7H-dG) Opposite Deoxyguanosine in a DNA Duplex. 8-oxo-7H-dG(syn)-dA(anti) Alignment at Lesion Site, Biochemistry 30, 1403-1412. 113. Blasco, R., Aguero, M., Almendral, J. M., and Vinuela, E. (1989) Variable and Constant Regions in African Swine Fever Virus DNA, Virology 168, 330-338. 114. Wu, X., Feng, J., Komori, A., Kim, E. C., Zan, H., and Casali, P. (2003) Immunoglobulin Somatic Hypermutation: Double-Strand DNA Breaks, AID and Error-Prone DNA Repair, Journal of Clinical Immunology 23, 235-246. 115. Garcia-Beato, R., Salas, M. L., Vinuela, E., and Salas, J. (1992) Role of the Host Cell Nucleus in the Replication of African Swine Fever Virus DNA, Virology 188, 637-649.
137