Reconstitution of Uracil DNA Glycosylase-initiated Base Excision Repair in Herpes Simplex Virus-1

Abstract

Herpes simplex virus-1 is a large double-stranded DNA virus that is self-sufficient in a number of genome transactions. Hence, the virus encodes its own DNA replication apparatus and is capable of mediating recombination reactions. We recently reported that the catalytic subunit of the HSV-1 DNA polymerase (UL30) exhibits apurinic/apyrimidinic and 5′-deoxyribose phosphate lyase activities that are integral to base excision repair. Base excision repair is required to maintain genome stability as a means to counter the accumulation of unusual bases and to protect from the loss of DNA bases. Here we have reconstituted a system with purified HSV-1 and human proteins that perform all the steps of uracil DNA glycosylase-initiated base excision repair. In this system nucleotide incorporation is dependent on the HSV-1 uracil DNA glycosylase (UL2), human AP endonuclease, and the HSV-1 DNA polymerase. Completion of base excision repair can be mediated by T4 DNA ligase as well as human DNA ligase I or ligase IIIα-XRCC1 complex. Of these, ligase IIIα-XRCC1 is the most efficient. Moreover, ligase IIIα-XRCC1 confers specificity onto the reaction in as much as it allows ligation to occur in the presence of the HSV-1 DNA polymerase processivity factor (UL42) and prevents base excision repair from occurring with heterologous DNA polymerases. Completion of base excision repair in this system is also dependent on the incorporation of the correct nucleotide. These findings demonstrate that the HSV-1 proteins in combination with cellular factors that are not encoded by the virus are capable of performing base excision repair. These results have implications on the role of base excision repair in viral genome maintenance during lytic replication and reactivation from latency.

Herpes simplex virus-1 (HSV-1)² is a large double-stranded DNA virus with a genome of ∼152 kilobase pairs (for reviews, see Refs. 1 and 2). HSV-1 switches between lytic replication in epithelial cells and a state of latency in sensory neurons during which there is no detectable DNA replication (1). Viral DNA replication is mediated by seven essential virus-encoded factors (3–5). Of these, two encode subunits of the viral replicase (for review, see Refs. 6 and 7). The catalytic subunit (UL30) exhibits DNA polymerase (Pol), 3′-5′ proofreading exonuclease, and RNase H activities (8–11). UL30 exists as a heterodimer with the UL42 protein that confers a high degree of processivity on the Pol (11–17).

Viral DNA replication is accompanied by vigorous recombination that leads to the formation of large networks of viral DNA replication intermediates (18). The HSV-1 single-strand DNA-binding protein (ICP8) has been shown to play a major role in mediating these recombination reactions (19–21). One role for the high frequency of recombination is to restart DNA replication at sites of fork collapse. Further mechanisms that contribute to genome maintenance are processes that survey and repair damage to the DNA to ensure the availability of a robust replication template. In this regard base excision repair (BER) is essential to remove unusual bases from the DNA and to repair apurinic/apyrimidinic (AP) sites resulting from spontaneous base loss (for review, see Ref. 22). With respect to HSV-1, a recent study showed that viral DNA from infected cultured fibroblasts contains a steady state of 2.8–5.9 AP sites per viral genome equivalent (23). Because AP sites are non-instructional, the failure to repair such sites would terminate viral replication. Indeed, UL30 cannot replicate beyond a model AP site (tetrahydrofuran residue) (23), indicating that the virus must enable a process to repair such lesions. In this regard HSV-1 possesses several enzymes that would safeguard from the accumulation of unusual bases, specifically uracil, and base loss. Hence, HSV-1 encodes a uracil DNA glycosylase (UDG) (UL2) as well as a dUTPase to reduce the pool of dUTP and prevent misincorporation by the viral Pol (24, 25). Moreover, we recently showed that the catalytic subunit of the viral Pol (UL30) exhibits AP and 5′-deoxyribose phosphate (dRP) lyase activities (26). The presence of a virus-encoded UDG and DNA lyase indicates that HSV-1 has the capacity to perform integral steps of BER, specifically for the removal of uracil. Indeed, the excision of uracil may be important for viral replication. Hence, it has been shown that uracil substitutions in the viral origins of replication alters their recognition by the viral initiator protein (27). Moreover, whereas UL2 may be dispensable for viral replication in fibroblast (24), UL2 mutants exhibit reduced neurovirulence and a decreased frequency of reactivation from latency (28). Thus, UDG action in HSV-1 may be important for viral reactivation after quiescence in neuronal cells during which the genome may accumulate uracil as a result of spontaneous deamination of cytosine. In another herpesvirus, cytomegalovirus, the viral UDG was shown to be required for the transition to late-phase DNA replication (29, 30). Consequently, it is possible that BER plays a significant role in various aspects of the herpesvirus life cycle.

In mammalian single-nucleotide BER initiated by monofunctional DNA glycosylases, the resulting AP sites are incised hydrolytically at the 5′ side by AP endonuclease (APE), generating a 3′-OH. This is followed by template-directed incorporation of one nucleotide by Pol β to generate a 5′-dRP flap (22, 31, 32). The 5′-dRP residue is subsequently removed by the 5′-dRP lyase activity of Pol β to leave a nick with a 3′-OH and 5′-phosphate that is ligated by DNA ligase I or the physiologically more relevant ligase IIIα-XRCC1 complex (for review, see Refs. 33 and 34). Here we show that the HSV-1 UDG (UL2) and Pol (UL30) cooperate with human APE and human ligase IIIα-XRCC1 complex to perform BER in vitro. This finding has implications on the role of BER in viral genome maintenance during lytic replication and in the emergence of the virus from neuronal latency.

EXPERIMENTAL PROCEDURES

Enzymes and Reagents

HSV-1 UL30 and UL42 were expressed in Spodoptera frugiperda cells and purified as described (35). The experiments described here used UL30 fraction V and UL42 fraction IV. HSV-1 UL2 was purified to homogeneity from Escherichia coli as a thioredoxin fusion protein as described in the supplemental material. E. coli UDG was obtained from New England Biolabs. Human DNA Pol β and APE were from Trevigen. Calf thymus DNA Pol δ was a kind gift from Drs. Antero So and Kathleen Downey (University of Miami Miller School of Medicine). T4 DNA ligase was from either Promega or New England Biolabs as indicated. Human DNA ligase I and ligase IIIα-XRCC1 complex were kind gifts from Dr. Alan Tomkinson (University of Maryland School of Medicine). T4 DNA Pol and exonuclease-deficient Klenow Pol were obtained from Roche Applied Science and New England Biolabs, respectively. Unlabeled deoxyribonucleoside 5′-triphosphates (disodium salts) were purchased from GE Healthcare. [α-³²P]dCTP (800 Ci/mmol) was from PerkinElmer Life Sciences.

DNA Substrate

Oligonucleotides PBAZ8 (56-mer) and PBAZ9 (42-mer) were synthesized with 5′ phosphates by Operon Biotechnologies. Their sequences are shown as part of Fig. 1A. Both oligonucleotides form hairpins with complementary four-base overhangs. To construct the substrate, 2 nmol of PBAZ9 was annealed with a 1.5-fold molar excess of PBAZ8 followed by ligation with 800 units of T4 ligase (New England Biolabs) for 2 h at room temperature. The reaction was subjected to denaturing polyacrylamide gel electrophoresis, and the product corresponding to the covalently closed oligonucleotide was excised, purified, and quantified using an extinction coefficient of 605719.5 m⁻¹ cm⁻¹ at 260 nm. This DNA exhibited an electrophoretic mobility that was faster than the linear 98-mer produced by a single ligation event between PBAZ8 and PBAZ9 (see Fig. 1B). As expected of the covalently closed DNA, it was not labeled by terminal deoxynucleotidyltransferase and was converted into the slower migrating linear 98-mer upon treatment with UDG and APE (see Fig. 6A).

FIGURE 1.

BER substrate. A, schematic representation of the covalently closed duplex oligonucleotide substrate for the BER reactions. The portion of the substrate corresponding to PBAZ8 is shown in italics, whereas that corresponding to PBAZ9 is shown in regular font. The unique uracil is shown in bold and underlined. The DraI cleavage site is indicated by the arrows. B, SYBR Gold-stained image showing 0.25 pmol of the various oligonucleotides, ligation products, and purified BER substrate. Lane 1, 100-mer; lane 2, unligated PBAZ8; lane 3, unligated PBAZ9; lane 4, self-ligated PBAZ8; lane 5, self-ligated PBAZ9; lane 6, ligation of PBAZ8 and PBAZ9; lane 7, purified BER substrate. The positions of linear 100-mer marker, PBAZ8 (56-mer), PBAZ9 (42-mer), substrate, and single-ligation product (98-mer) are as indicated.

FIGURE 6.

Direct visualization of BER intermediates and product formation. Reactions were performed as described under “Experimental Procedures” with the exception of DNA substrate, which was at 100 nm and deoxyribonucleoside 5′-triphosphates (4 μm) as indicated. A, SYBR Gold-stained image of reaction products. B, storage phosphorimage of the same gel. Lane 1, 100-mer; lane 2, DNA only (substrate); lane 3, UL2, APE and [α-³²P]dCTP; lane 4, UL2, APE, UL30,and [α-³²P]dCTP; lane 5, UL2, APE, ligase IIIα-XRCC1, and [α-³²P]dCTP; lane 6, UL2, APE, UL30, ligase IIIα-XRCC1, and [α-³²P]dCTP; lane 7, UL2, APE, UL30, ligase IIIα-XRCC1, and dGTP; lane 8, UL2, APE, UL30, ligase IIIα-XRCC1, and [α-³²P]dCTP, dGTP, dATP, and dTTP. The positions of linear 100-mer marker, substrate, nicked product (N), ligated product (L), and exonuclease product (X) are as indicated.

BER Assay

Unless otherwise stated, reactions (10 μl) contained 10 nm (molecules) DNA substrate in 20 mm HEPES-NaOH, pH 7.5, 100 μg/ml bovine serum albumin, 10% glycerol, 5 mm MgCl₂, 4 mm ATP, 4 μm [α-³²P]dCTP (∼60 Ci/mmol), and the following proteins as indicated: UL2 (200 nm), E. coli UDG (1 unit), APE (1.25 units), UL30 (100 nm), UL42 (100 nm), T4 ligase (Promega), ligase I (50 nm), ligase IIIα-XRCC1 (50 nm), Pol β (100 nm), Pol δ (400 nm), or exonuclease-deficient Klenow Pol (1 unit). After 40 min at 37 °C, the reactions were combined with an equal volume of stop buffer (95% formamide, 20 mm EDTA, 0.05% bromphenol blue, and 0.05% xylene cyanol), heated for 3 min at 75 °C, and chilled on ice before electrophoresis through 15% polyacrylamide, 8 m urea gels in GTG buffer (89 mm Tris, pH 9.0, 28.5 mm taurine, and 0.5 mm EDTA). Radiolabeled reaction products were visualized by storage phosphor analysis with a GE Healthcare Storm 820. Product formation was quantified by determining the arbitrary counts in each band using ImageQuant Version 5.2. Alternatively, DNA was visualized by SYBR Gold (Invitrogen) staining using a Bio-Rad VersaDoc 1000 imager and the Quantity One Version 4.3.0 software.

RESULTS

Design and Construction of the BER Substrate

To study BER with the purified HSV-1 factors, we designed a DNA substrate that would not be degraded by the potent 3′-5′ exonuclease activity of UL30, which would otherwise lead to nonspecific labeling of DNA ends. To this end we constructed a covalently closed duplex oligonucleotide. This was achieved by ligating two hairpin oligonucleotides (PBAZ8 and PBAZ9) that possess complementary 4-base overhangs, generating a covalently closed 98-mer oligonucleotide with a 45-base duplex region containing a unique uracil and a DraI restriction endonuclease site (Fig. 1A). The construction of this covalently closed duplex oligonucleotide (double-hairpin) substrate is described under “Experimental Procedures.” Fig. 1B shows the purity and migration of the substrate relative to the individual hairpins and single-ligation product (linear 98-mer). In this substrate removal of the uridine would allow template directed incorporation of dCMP. Consequently, BER activity was routinely measured by the incorporation of ³²P dCMP but also by direct visualization of the DNA using SYBR Gold staining.

UL2, APE, and UL30 Promote the Initial Steps of BER

In the first instance we examined the ability of UL2, APE, and UL30 to mediate the incorporation of [³²P]dCMP into the substrate. Fig. 2 shows that the complete set of proteins was capable of labeling the substrate (lane 1), generating a single species with a mobility slightly faster than a 100-mer marker, corresponding to the full-length 98-mer from which the uridine was excised and replaced with dCMP. This product is identical to that formed with E. coli UDG, APE, and Pol β (lane 6), a combination that is known to mediate BER (31). Importantly, with either combination of proteins, the signal was dependent on UDG (UL2 or E. coli UDG; lanes 2 and 7), APE (lanes 3 and 8) and Pol (lanes 4 and 9). As intended, given the fact that the substrate is covalently closed at both ends, there was no nonspecific labeling of the ends as a result of exonuclease action followed by resynthesis. The UL2- and APE-dependent incorporation of dCMP by UL30 demonstrates that this combination of enzymes is capable of performing the initial excision and incorporation steps of BER, leading to a nicked (non-ligated) product, designated as N.

FIGURE 2.

Reconstitution of the initial stages of BER. Storage phosphorimage showing the products of the reconstitution reaction. Reactions were performed as described under “Experimental Procedures” with the indicated proteins. Lane 1, UL2, APE, and UL30; lane 2, APE and UL30; lane 3, UL2 and UL30; lane 4, UL2 and APE; lane 5, DNA only; lane 6, E. coli UDG, APE and Pol β; lane 7, APE and Pol β; lane 8, E. coli UDG and Pol β; lane 9, E. coli UDG and APE. The position of a linear 100-mer marker is as indicated.

Completion of BER by Ligase

Completion of BER requires ligation between the 3′-OH of the newly incorporated nucleotide (dCMP in this case) and a functional 5′-end that is generated by the removal of the 5′-dRP residue, in this case via the 5′-dRP lyase of UL30. Fig. 3 shows that a ligase-dependent product, designated as L, was formed when reactions containing UL2, APE, and UL30 (lane 1) were supplemented with either T4 ligase, ligase I, or ligase IIIα-XRCC1 (lanes 2–4, respectively). The relative efficiency of the ligation reaction was determined by the ratio of ligated to nicked product (Fig. 3). Based on this ratio, the reaction with ligase IIIα-XRCC1 appears to be the most efficient in terms of converting the nicked intermediate into the ligated product. In contrast, the reaction with ligase I was ∼6-fold less efficient. Although the reaction with T4 ligase appears to be robust, the L/N ratio was only half that with ligase IIIα-XRCC1 despite the fact that the specific activity of T4 ligase in this reaction was ∼150-fold higher than that of ligase I and ligase IIIα-XRCC1.

FIGURE 3.

Completion of BER; formation of ligation product. Reactions were performed as described under “Experimental Procedures” with the indicated proteins. Storage phosphorimage of reaction products obtained with UL2, APE, and UL30 (lane 1) and identical reactions supplemented with T4 ligase (1.5 units) (lane 2), ligase I (lane 3), or ligase IIIα-XRCC1 (lane 4) are shown. Lane 5, DNA only; lane 6, linear 100-mer. The positions of nicked product (N), ligated product (L), and of the 100-mer are as indicated. The values in italics below the lane numbers indicate the L/N ratios.

Time courses of the reactions with each of the three ligases shows that the formation of the ligation product was linear over a 60-min period (Fig. 4). In these experiments, the amount of T4 ligase had the same specific activity (as determined by the ligation of HindIII digested λ DNA) as 50 nm ligase I or ligase IIIα-XRCC1. As documented in Fig. 3, based on the ratio of nicked to ligated products, the reaction with ligase IIIα-XRCC1 was the most efficient. Interestingly, in contrast to the reactions with T4 ligase and ligase I, where the levels of nicked intermediate were maximal at the earliest time points, there was a gradual accumulation of the nicked form in the reaction with ligase IIIα-XRCC1, reaching a maximum after 20 min with a concomitant buildup of the ligated product. This may be indicative of a more concerted reaction.

FIGURE 4.

Time course of BER. 35-μl reactions were performed as described under “Experimental Procedures” with UL2, APE, and UL30 in the presence of either T4 ligase (0.001 units/μl) (A), ligase I (B), or ligase IIIα-XRCC1 (C). 5-μl aliquots were removed at the times indicated. Activity is expressed as the fraction of maximum for the nicked (N) (●) and ligation (L) (○) products. The insets in each panel show the relevant gel images.

The ligation product formed with UL2, APE, and UL30 in the presence of each of these ligases is identical to that formed with an established BER system, namely the one involving Pol β and ligase IIIα-XRCC1 (see Fig. 8, lane 4). The identification of the ligation product as a species with a faster electrophoretic mobility than the nicked intermediate, formed by excision and incorporation, is consistent with the faster electrophoretic mobility of the covalently closed duplex oligonucleotide substrate compared with the linear 98-mer (Fig. 1B), as the substrate and product have essentially the same structure. The faster electrophoretic mobility of the ligation product is also consistent with that of dumbbell oligonucleotides as well as covalently closed quadruplex DNA (36–38). In these cases it is likely that the covalently closed double-stranded nature of the DNA contributes to their faster electrophoretic mobility compared with their linear counterparts, even under denaturing conditions.

FIGURE 8.

Ligase IIIα-XRCC1 confers specificity onto the BER reaction. Storage phosphor image of reactions performed with either ligase IIIα-XRCC1 (lanes 1–6) or ligase I (lanes 7–12). Standard reactions (lanes 1 and 7) contained UL2, APE, and UL30. Lanes 2 and 8, standard reactions with UL42; lanes 3 and 9, substitution of UL2 with E. coli UDG; lanes 4 and 10, substitution of UL30 with Pol β; lanes 5 and 11, substitution of UL30 with exonuclease-deficient Klenow Pol; lanes 6 and 12, substitution of UL30 with Pol δ. The positions of nicked (N) and ligated (L) products are as indicated.

To demonstrate that the ligase-dependent product is indeed the result of bona fide ligation, we subjected the purified BER reaction products to a number of treatments. Thus, as indicated in Fig. 1A, digestion of the ligated product with DraI would generate an internally labeled 50-mer, whereas digestion of the nicked product would generate an internally labeled 4-mer. Fig. 5 shows that DraI treatment generated a 50-mer, indicating that ligation had occurred. In addition, a covalently closed ligation product would be resistant to the action of a 3′-5′ exonuclease such as that associated with T4 Pol. Indeed, Fig. 5 shows that the ligation product was resistant to the exonucleolytic action of T4 Pol, whereas the nicked product was sensitive. In contrast, neither product was affected when treated with the exonuclease-deficient Klenow Pol. These findings demonstrate that the product formed by the addition of ligase is a bona fide ligation product.

FIGURE 5.

Characterization of BER products. Reactions were performed as described under “Experimental Procedures” with UL2, APE, UL30, and 100 nm ligase I. After the reaction the DNA was purified by ethanol precipitation and treated for 60 min at 37°C as indicated. Storage phosphorimage of reaction products. Lane 1, no treatment; lane 2, DraI (5 units); lane 3, exonuclease-deficient Klenow Pol (1 unit); lane 4, T4 Pol (1 unit). The positions of nicked product (N), ligated product (L) and of 100- and 50-mer markers are as indicated.

Characterization of BER

Thus far, incorporation of ³²P dCMP into unlabeled substrate was used to detect BER. To directly visualize the conversion of substrate into the nicked intermediate and product formation, SYBR Gold staining was used with concomitant incorporation of [³²P]dCMP. Figs. 6, A and B, are the SYBR Gold and storage phosphorimages of the same gel, respectively. Incubation of the substrate with UL2 and APE led to essentially complete conversion of the substrate to the slower migrating intermediate (lane 3), indicating that neither UDG or APE activities were limiting. When this reaction was supplemented with UL30, the same unlabeled DNA species was observed with concomitant appearance of the radiolabeled nicked product (N) (lane 4), indicating that template-directed nucleotide incorporation had occurred. In addition, a minor species with a slightly faster electrophoretic mobility was observed by direct staining (designated as X) but was not radiolabeled. Presumably, this species is because of UL30 exonuclease digestion beyond a point to permit template-directed incorporation of dCMP. In fact, the presumed exonuclease reaction product (X) was not detected when UL30 was omitted from the reaction (lanes 3 and 5). Thus, reactions with UL2, APE, and ligase IIIα-XRCC1 converted the substrate to the incised intermediate without nucleotide incorporation (lane 5). The products of the complete reaction (UL2, APE, UL30, and ligase IIIα-XRCC1) are shown in lane 6. The SYBR Gold image depicts the formation of a product that has the same electrophoretic mobility as the substrate, i.e. the covalently closed product of the BER reaction. This was accompanied by the appearance of the radiolabeled ligation product (L) in the storage phosphorimage. Although it is not possible to ascertain the efficiency of the BER reaction using radiolabeling, analysis of the SYBR Gold-stained image indicates that ∼50% of the substrate was converted into the final ligation product. A small amount of the exonuclease digestion product (X) was also observed via direct staining, indicating that a portion of the intermediate had been digested beyond a point to permit template-directed incorporation of dCMP and subsequent ligation. Fig. 6, lane 7, shows the results of the complete reaction in which [α-³²P]dCTP as the correct templated or template-directed nucleotide was substituted with unlabeled dGTP as an incorrect non-templated or non-template-directed nucleotide. As expected, the absence of the correct nucleotide failed to produce the final ligation product (L) and instead led to increased accumulation of the exonuclease digestion product (X). Therefore, BER in this system is dependent on the incorporation of the correct nucleotide. Finally, formation of the exonuclease digestion product (X) was prevented when the complete reaction containing [α-³²P]dCTP was supplemented with the remaining three nucleotides (lane 8).

To further demonstrate that completion of BER, i.e. formation of the ligation product, is dependent on UDG (UL2), APE, Pol (UL30), 5′-dRP lyase (UL30), and ligase activities, a systematic examination of the activities of each of the individual proteins and their various combinations was performed. As expected, Fig. 7 shows that the formation of the nicked product required UL2, APE, and UL30 and that the formation of the ligation product required the addition of either T4 ligase, ligase I, or ligase IIIα-XRCC1. No product formation was observed with any of the individual enzymes or by omitting one of the required functions. The background observed in lanes 3 and 8 was because of the activity of UL30 alone and is presumably because of its exonuclease activity followed by DNA synthesis.

FIGURE 7.

Product formation is dependent on UL2, APE, UL30, and ligase. Storage phosphorimages of reactions performed as described under “Experimental Procedures” with the indicated components. Lane 1, UL2; lane 2, APE; lane 3; UL30; lane 4, T4 ligase (1.5 units); lane 5, ligase I; lane 6, ligase IIIα-XRCC1; lane 7, UL2 and APE; lane 8, APE and UL30; lane 9, UL2 and UL30; lane 10, UL2, APE and UL30; lane 11, UL2, APE, UL30 and T4 ligase (1.5 units); lane 12, UL2, APE, UL30, and ligase I; lane 13, UL2, APE, UL30, and ligase IIIα-XRCC1; lane 14, DNA only. The positions of nicked (N) and ligated (L) products are as indicated.

Ligase IIIα-XRCC1 Dictates the Specificity of BER

In HSV-1-infected cells, UL30 forms a heterodimer with its processivity factor, UL42 (11–13). Therefore, to investigate its contribution to BER, reactions containing UL2, APE, and UL30 were supplemented with equimolar UL42. As shown in Fig. 8, the addition of UL42 did not affect the formation of nicked product (lanes 2 and 8). However, the ligated product only formed in the presence of ligase IIIα-XRCC1 (lane 2). The addition of UL42 prevented ligation by ligase I (lane 8). To further examine the specificity of the reaction, UL2 and UL30 were substituted with heterologous enzymes in the presence of either ligase I or ligase IIIα-XRCC1. In the presence of ligase IIIα-XRCC1, substituting UL2 with E. coli UDG had no significant effect (lane 3), presumably because it is not limiting. Substituting UL30 with Pol β led to complete conversion of nicked product into the ligation product (lane 4), indicating, as previously documented (33), that the reaction with Pol β and ligase IIIα-XRCC1 is very efficient. Interestingly, neither nicked or ligation products were observed when UL30 was substituted with the exonuclease-deficient Klenow Pol or Pol δ (lanes 5 and 6, respectively). In contrast, when BER reactions were performed with ligase I, the reaction with Pol β was less efficient in terms of ligation (lane 10), and importantly, both the exonuclease-deficient Klenow Pol and Pol δ were capable of generating nicked and ligation products (lanes 11 and 12, respectively). These findings indicate that the factor that governs the specificity of the reaction is ligase IIIα-XRCC1; not only does it allow the physiologically relevant UL30-UL42 heterodimer or Pol β to function efficiently, but it also prevents polymerases that do not function in BER to participate.

DISCUSSION

Our findings demonstrate that HSV-1 UL2 and UL30 cooperate with human APE and specifically ligase IIIα-XRCC1 complex to mediate BER in vitro. This implies that HSV-1 has the capacity to excise uracil from DNA and repair the resulting AP site to provide a robust template for replication. This is important given the fact that AP sites occur at a frequency of 2.8–5.9 per viral genome and the fact that the viral replicase is not capable of replicating beyond these sites (23). Therefore, the ability of HSV-1 to repair these sites via BER provides a mechanism to sustain genome replication.

In this study we set out to demonstrate that particular HSV-1 factors aided by cellular proteins are capable of promoting BER. Specifically, given their known activities, we used the HSV-1 UDG (UL2) and Pol (UL30 and UL42). Collectively, these proteins exhibit UDG, lyase, and Pol activities. In other systems BER also involves the action of APE to generate a functional 3′-OH. Therefore, we examined the role of human APE in our system. To restore the continuity of the phosphodiester backbone, completion of BER requires the action of DNA ligase. Hence, we examined the activity of different ligases in our system. In the first instance we designed a suitable DNA substrate. To circumvent nonspecific labeling of DNA ends by UL30 (i.e. exonuclease digestion followed by resynthesis), we designed a novel type of DNA structure consisting of a covalently closed DNA duplex, generated by ligating two hairpin oligonucleotides. The duplex portion of this substrate contains a unique uracil opposite a guanine, reminiscent of the outcome of a spontaneous deamination event at a G:C base pair. For the initial stages of BER, this DNA provides a substrate to examine the UDG-dependent and template-directed incorporation of ³²P dCMP, leading to the formation of a nicked (N) intermediate. Completion of BER would join available 5′-phosphate and 3′-OH ends. This phase would be dependent on both lyase, to remove the 5′-dRP and provide a free 5′-phosphate, and ligase activities to generate a covalently closed DNA structure designated as the ligated form (L).

Our data show that UL2, human APE, and UL30 are necessary and sufficient to generate the N form, which is at least representative of uracil removal, incision at the 5′ side of the AP site, and nucleotide incorporation. Furthermore, formation of the L form was shown to be dependent on the addition of ligase but also requires the 5′-dRP lyase activity of UL30, albeit not measured in the reactions, as this activity is necessary for generating a ligatable 5′-end. Using SYBR Gold staining, we directly showed that in our BER system, essentially all of the substrate is converted into the nicked intermediate, and 50% is further processed into the final ligation product in a UDG-, APE-, Pol-, and ligase-dependent manner. Moreover, formation of the ligation product also required incorporation of the correct nucleotide (dCMP), as specified by the template, presumably to fill the gap and provide a ligatable substrate. Based on the fact that APE has been shown to be dispensable for BER in systems that rely on the 3′-phosphatase activity of polynucleotide kinase after incision by NEIL1 or NEIL2 DNA glycosylase/AP lyase (39, 40), it would not have been unprecedented that after UDG and AP lyase action by UL2 and UL30, respectively, that the 3′-5′ exonuclease activity of UL30 could have removed the 5′-dRP residue and generated the N form with only these two proteins. However, our data indicate that UL2 and UL30 alone were unable to do so, indicating a requirement for APE in this system.

Interestingly, our data show that whereas both T4 ligase and ligase I were capable of mediating the ligation phase of the BER reaction, ligase IIIα-XRCC1 appeared to be more efficient. Importantly, unlike ligase I, ligase IIIα-XRCC1 also allowed the BER reaction to proceed in the presence of the Pol processivity factor (UL42) and prevented heterologous Pols (both exonuclease-deficient Klenow and Pol δ) to be active. Therefore, ligase IIIα-XRCC1 appears to govern the specificity of the reaction. This finding is consistent with the role of ligase IIIα-XRCC1 complex in mammalian short patch BER in which XRCC1 stabilizes ligase III and acts as a scaffold to mediate interactions with other BER factors including APE and Pol β (for review, see Ref. 34). The finding that ligase IIIα-XRCC1 is involved in viral BER adds to the recent report that ligase IV-XRCC4 is required for viral genome replication, presumably in the formation of endless DNA (41).

In summary, our data indicate that HSV-1 in combination with cellular APE and ligase IIIα-XRCC1 has the capacity to perform BER as a mechanism to prevent mutagenesis and replisome stalling at abasic sites during viral replication. Undoubtedly, given the frequency of AP sites in the HSV-1 genome, this is an important mechanism to safeguard the integrity of the genome as a replication template both during lytic replication and as the virus emerges from neuronal latency. In the later scenario it is easy to envision that in the absence of genome surveillance extended periods of quiescence in post-mitotic neurons may lead to accumulation of AP sites and uracil from spontaneous deamination. Consequently, BER would be a perquisite to providing a robust replication template. This notion is supported by the previous finding that uracil substitutions in the viral origins impacts recognition by UL9 protein (27) and is consistent with the reduced neurovirulence and reduced reactivation frequency of UL2 mutants (28), indicating that the HSV-1 UDG and, by extension, virus-mediated BER may play an important role during viral replication in neurons.