XRCC1 prevents toxic PARP1 trapping during DNA base excision repair

1DNA base excision repair (BER) is a highly conserved pathway that is present in all organisms and required for repair of a broad range of endogenous and exogenous DNA base damage. In mammals, the canonical BER pathway involves removal of the damaged base by a DNA glycosylase, incision of the resulting abasic site by AP endonuclease (apurinic/apyrimidinic endonuclease; APE), replacement of the missing nucleotide and removal of the terminal sugar phosphate by DNA polymerase β (POLβ), and ligation of the resulting nick by DNA ligase I (LIG1) or DNA ligase III (LIG3). The importance of BER in mammals is illustrated by the embryonic lethality observed in mice in which key components such as APE1 or POLβ are deleted. Intriguingly, in addition to these core components, mammalian cells employ a number of additional proteins to accelerate BER, such as poly(ADP-ribose) polymerase-1 (PARP1), PARP2, and the molecular scaffold protein XRCC1. PARP1 and PARP2 are sensor proteins that detect and are activated by DNA strand breaks, resulting in the post-translational modification of themselves and other proteins with ADP-ribose. PARP1 and PARP2 fulfil multiple roles at DNA strand breaks. For example, PARP enzymes can modify chromatin structure directly by histone ribosylation and/or indirectly by recruitment and/or regulation of specific chromatin remodelers. In addition, PARP activity can promote recruitment of other DNA repair proteins to accelerate repair of DNA strand breaks, of which XRCC1 and its protein partners are among the most important.

2The importance of XRCC1 is illustrated by the observations that deletion of this gene in mouse is embryonic lethal and that hereditary mutations in human XRCC1 result in progressive neurodegenerative disease. At “direct” SSBs, such as those resulting from oxidative attack and disintegration of deoxyribose, XRCC1 binds, recruits, and stimulates DNA polynucleotide kinase phosphatase (PNKP). PNKP is ideally suited to repair of DNA strand breaks induced by oxidative damage to deoxyribose and also those induced by abortive topoisomerase I activity because it possesses the DNA kinase and DNA phosphatase activities that can restore normal 3′ and 5′ termini at such DNA breaks. In contrast to direct SSBs, the role of XRCC1 during BER is less clear because PNKP is required for only a subset of the BER events that are accelerated by XRCC1. Moreover, although XRCC1 also interacts with and stabilizes the BER proteins POLβ and LIG3, loss of these interactions individually only partially reduces XRCC1 functionality during BER. This is in contrast to the interaction of XRCC1 with poly(ADP-ribose), which is essential for XRCC1 functionality during BER. Consequently, the critical role of XRCC1 protein complexes during BER has remained elusive. Here, we have identified this role. We show that assembly of POLβ and LIG3 by XRCC1 into protein complexes is required to limit PARP1 engagement and activity during BER, which otherwise results in PARP1 “trapping” on BER intermediates in a manner reminiscent of that induced by clinical PARP inhibitors. XRCC1 is thus an endogenous “anti-trapper” that prevents toxic binding of PARP1 to SSB intermediates during BER, enabling their rapid repair and maintaining genome integrity.

3To address the role of XRCC1 during BER, we first examined its functional relationship with PARP1 because the ability of XRCC1 to interact directly with poly(ADP-ribose) is critical for this role. To do this, we employed human RPE-1 cells in which PARP1, PARP2, and/or XRCC1 were deleted by gene editing. As expected, RPE-1 cells lacking PARP1 alone exhibited relatively little sensitivity to the simple alkylating agent methyl methanesulfonate (MMS) compared with cells lacking XRCC1, unless PARP2 was also deleted (Figure 1A). This is consistent with the established enzymatic redundancy of PARP1 and PARP2 during BER. Unexpectedly, however, deletion of PARP1 restored almost normal levels of MMS sensitivity in XRCC1−/− cells (Figure 1A). Similar results were observed in human TK6 cells in which PARP1 and/or XRCC1 were deleted by homologous recombination-mediated gene targeting, indicating that this result was not specific to RPE-1 cells or an artifact of Cas9-mediated gene editing (Figure 1B). In contrast, PARP1 deletion did not reduce the sensitivity of XRCC1−/− RPE-1 cells to camptothecin, a genotoxin that induces SSBs independently of BER by promoting the abortive activity of topoisomerase 1 (Figure 1C). These data indicate that the essential role of XRCC1 during BER is to suppress PARP1-induced cytotoxicity.

4To identify the mechanism by which XRCC1 suppresses PARP1-induced cytotoxicity, we examined the effect of deleting these proteins on BER directly, using alkaline comet assays. Strikingly, PARP1 deletion prevented the appearance of elevated MMS-induced SSBs in XRCC1−/− cells, suggesting that the accumulation of these BER intermediates was the result of PARP1 (Figure 1D). Again, this was not an artifact of gene editing because depletion of PARP1 using small interfering RNA (siRNA) gave similar results (Figures 1A, right panel, and 1E). Restoration by PARP1 deletion of normal steady-state levels of MMS-induced SSBs in XRCC1−/− cells could reflect either reduced SSB induction during BER or increased SSB repair. To address this question, we co-incubated cells with the nucleoside analog arabinosyl adenine (Ara-A), a DNA polymerase inhibitor that can block DNA repair synthesis by promoting chain termination. As expected, co-incubation with Ara-A increased the steady-state level of MMS-induced SSBs in wild-type RPE-1 cells, consistent with inhibition of BER (Figure 1F). More importantly, Ara-A increased the level of MMS-induced SSBs in XRCC1−/−/PARP1−/− cells to the same extent and to a level similar as that induced in XRCC1−/− cells (Figure 1F). These data demonstrate that PARP1 deletion prevents accumulation of MMS-induced SSBs in XRCC1−/− cells not by preventing SSB induction but by restoring normal rates of SSB repair.

5In contrast to PARP1, depletion of PARP2 failed to suppress the elevated level of MMS-induced SSBs in XRCC1−/− cells or their increased sensitivity to MMS (Figures S1A and S1B). This result did not reflect inefficient PARP2 depletion because PARP2 siRNA reduced PARP2 protein to levels that were undetectable in western blots and increased MMS-induced SSBs in PARP1−/− cells as efficiently as PARP2 deletion (Figure S1A). In addition, as expected, PARP2 depletion greatly increased the sensitivity of PARP1−/− cells to MMS (Figure S1B). Collectively, these data demonstrate that the elevated accumulation of SSBs and sensitivity of XRCC1−/− cells to MMS is due to the presence of PARP1.

6It is well established that pharmacological PARP inhibitors prolong the engagement of PARP proteins at DNA breaks and thereby slow and/or block their repair. This phenomenon is called PARP trapping and underpins the clinical utility of PARP inhibitors as anti-cancer therapeutic agents. We therefore wanted to find out whether the PARP1-dependent accumulation of SSBs in XRCC1−/− cells reflected a similar phenomenon. Indeed, consistent with this idea, although XRCC1 deletion and PARP inhibitor increased the steady-state level of SSBs in MMS-treated RPE-1 cells, the combination of both did not increase this level above that induced by the PARP inhibitor alone (Figure 1G). We therefore examined whether MMS treatment resulted in accumulation of PARP1 in chromatin in XRCC1−/− cells because this is a measure of PARP1 trapping. Indeed, although PARP inhibitor was required to trigger accumulation of high levels of PARP1 in chromatin in wild-type RPE-1 cells during 1-h incubation with MMS (Figure 2A, lanes 10 and 12), most if not all cellular PARP1 accumulated in chromatin in XRCC1−/− cells during treatment with MMS, even in the absence of PARP inhibitor (Figure 2A, lanes 6 and 14). Similar results were observed when we employed primary patient fibroblasts from XRCC1-mutated disease, suggesting that increased PARP1 trapping during BER is also a feature of this human genetic disease (Figure S1C). Notably, PARP2 also accumulated in the chromatin of XRCC1−/− cells during MMS treatment but, as indicated above, PARP2 did not measurably increased SSB accumulation, nor was it cytotoxic (Figure S1D). These data implicate endogenous PARP1 trapping as the source of SSB repair defects and cytotoxicity in XRCC1−/− cells during BER.

7PARP1 trapping induced by pharmacological inhibitors reflects the effect of altered allosteric regulation and reduced PARP1 auto-ribosylation, which results in excessive engagement of the weakly auto-modified enzyme at SSBs and blockage of their access and repair by other enzymes. To examine whether PARP1 might become trapped in XRCC1−/− cells by a similar mechanism, we measured the extent of PARP1 auto-ribosylation at various times during MMS treatment, using a poly(ADP-ribose)-specific detection reagent (Figure 2B). Although PARP1 auto-ribosylation was much higher in XRCC1−/− cells than in wild-type RPE-1 cells at early times (up to 15 min) after MMS addition, it declined progressively thereafter, resulting, after 60 min, in levels less than those in wild-type cells (Figure 2B). Importantly, the decline in PARP1 auto-ribosylation in XRCC1−/− RPE-1 cells was again accompanied by accumulation in chromatin of almost all cellular PARP1 (Figure 2C, top panel, lane 20). This accumulation was not simply a reflection of the high steady-state level of SSBs in XRCC1−/− cells because, by the end of the time course, high levels of SSBs were also present in wild-type RPE-1 cells (Figure 2D). Interestingly, there was a progressive reduction in PARP1 signal in the soluble fraction of XRCC1−/− cells even at early times (0–30 min) after addition of MMS despite the absence of a corresponding accumulation in chromatin at these times (Figure 2C, top panel, lanes 6–9 and 16–19). This decline reflected the effect of auto-ribosylation on detection of PARP1 by anti-PARP1 antibodies because the PARP1 signal in the soluble fraction was restored by pre-treatment of the protein samples with recombinant poly(ADP-ribose) glycohydrolase (PARG) immediately prior to SDS-PAGE (Figure 2C, bottom panel, lanes 7–9).

8The reduced PARP1 auto-ribosylation in XRCC1−/− cells 60 min after MMS treatment, compared to wild-type cells, appeared to involve reduced poly(ADP-ribose) chain length and/or branching complexity, as suggested by the faster electrophoretic mobility of the auto-ribosylated protein when fractionated extensively by SDS-PAGE and detected by a poly(ADP-ribose)-specific detection reagent (Figure 3A, compare lanes 3 and 4). That this signal reflected short chains of poly(ADP-ribose) rather than mono(ADP-ribose) was confirmed by its sensitivity to treatment with recombinant PARG immediately prior to electrophoresis (Figure 3B, compare lanes 5 and 6 with lanes 7 and 8). The reduced chain length/complexity of the poly(ADP-ribose) in XRCC1−/− cells was a result of reduced PARP1 activity rather than increased PARG activity because co-incubation with a PARG inhibitor during the final 5 min of MMS treatment failed to restore PARP1 auto-ribosylation in XRCC1−/− cells to the level observed in wild-type cells (Figure 3A, compare lanes 5–8).

9Our data suggest that although PARP1 is initially engaged excessively with BER intermediates and hyperactive in XRCC1−/− cells during MMS treatment, it becomes progressively less active and unable to dissociate from BER intermediates thereafter. In support of this idea, we confirmed that the PARP1 present in XRCC1−/− cells following MMS treatment for 60 min was incapable of reactivation by a second burst of SSBs (Figure 3C). Although H2O2 triggered extensive PARP1 auto-ribosylation in wild-type and XRCC1−/− cells prior to MMS treatment (Figure 3C, lanes 2 and 6), it failed to do so in XRCC1−/− cells pre-treated with MMS (Figure 3C, lanes 4 and 8). This conclusion was also supported by live-cell imaging experiments showing that GFP-tagged PARP1 was unable to efficiently relocate to sites of 405-nm laser damage in Xrcc1−/− mouse embryonic fibroblasts (MEFs) if these cells were first treated with MMS (Figure S2A). In addition, fluorescence recovery after photobleaching experiments confirmed that the mobility of GFP-tagged PARP1 following MMS treatment was greatly reduced in Xrcc1−/− MEFs compared with similarly treated wild-type MEFs (Figure S2B).

10Next we addressed the cause of the excessive PARP1 engagement and hyperactivity in XRCC1−/− cells during BER. This was not simply a result of the elevated SSBs in XRCC1−/− cells because, as indicated above (Figure 1D), the elevated SSBs were a consequence of excessive PARP1 engagement during BER rather than a cause. We reasoned that assembly of POLβ and LIG3 into protein complexes by XRCC1 might be important to compete with and/or limit PARP1 engagement and activation during BER. Such a scenario would be consistent with the idea that BER intermediates are handed from one enzyme in the pathway to the next in a coordinated relay that protects the intermediates from unnecessary engagement or attack by other enzymes. Consistent with this idea, biochemical experiments employing purified human proteins confirmed that POLβ and LIG3 suppressed PARP1 activation at SSBs created by APE1 during BER and that the presence of XRCC1 greatly promoted this suppression (Figure 3D).

11To further address this idea, we expressed truncated Myc-His-XRCC1161–406 protein lacking the N-terminal and C-terminal domains that bind POLβ and LIG3 in XRCC1−/− U2OS cells. We employed U2OS cells rather than RPE-1 cells for these experiments because of their greater transfection efficiency. Notably, Myc-His-XRCC1161–406 was unable to suppress the initial PARP1 hyperactivity in XRCC1−/− U2OS cells (Figure 4A, compare lanes 2, 7, and 12), or its subsequent inactivation (Figure 4A, compare lanes 5, 10, and 15) and accumulation in chromatin (Figure 4B, compare lanes 12, 15, and 18). Consistent with this, Myc-His-XRCC1161–406 was also unable to fully suppress accumulation of SSBs in XRCC1−/− U2OS cells during BER (Figure 4C). The inability of Myc-His-XRCC1161–406 to regulate PARP1 engagement during BER was not due to protein instability because the truncated protein was recruited into chromatin, although, as expected, it was unable to promote recruitment of POLβ or LIG3 (Figure 4B). These data indicate that XRCC1 regulates PARP1 engagement and activity during BER in part by assembling POLβ and LIG3 into protein complexes that can limit PARP1 access to BER intermediates.

12Because excessive PARP1 engagement and activity can result in NAD+ depletion, we wanted to find out whether the subsequent decline in PARP1 activity during BER in the absence of XRCC1 protein complexes reflected NAD+ exhaustion. Consistent with this idea, NAD+ has been reported to be depleted more rapidly by PARP activity in XRCC1 mutant Chinese hamster ovary (CHO) cells than in wild-type cells during MMS treatment, and we confirmed that this was the case in the XRCC1−/− RPE-1 cells employed here (Figure S3). To test directly whether NAD+ depletion was responsible for the decline in PARP1 activity in XRCC1−/− RPE-1 cells, we treated the latter with MMS for 60 min to accumulate trapped PARP1 and then incubated total cell lysates from these cells and wild-type controls with or without NAD+ supplement in vitro. As expected, lysates prepared from MMS-treated XRCC1−/− cells exhibited very little auto-ribosylated PARP1 when incubated in the absence of NAD+ supplement compared with wild-type lysates incubated in parallel (Figure 4D, compare lanes 2 and 4). However, supplementation with NAD+ not only stimulated ADP-ribosylation in XRCC1−/− cell lysates, but it increased it above that observed in wild-type cell lysates (Figure 4D, compare lanes 2 and 4 with lanes 6 and 8). This increased ADP-ribosylation reflected the activation of PARP molecules by BER intermediates because NAD+ only weakly stimulated ADP-ribosylation in cell lysates prepared from cells not pre-treated with MMS (Figure 4D, lanes 5 and 7). We conclude from these experiments that the progressive decline in PARP1 auto-ribosylation in XRCC1−/− cells during BER is a result of NAD+ exhaustion.

13Finally, we examined how the excessive engagement and progressive inactivation of PARP1 that occurs in XRCC1−/− cells might impede BER. Treatment of wild-type RPE-1 cells with PARP inhibitor reduced XRCC1 and POLβ accumulation in chromatin, consistent with the idea that PARP1 trapping impedes BER by preventing BER enzymes from accessing SSB intermediates (Figure 5A, compare lanes 12 and 14). We therefore reasoned that the excessive engagement of PARP1 observed in XRCC1−/− cells may block BER by a similar mechanism. Consistent with this, POLβ was almost undetectable in the chromatin of XRCC1−/− RPE-1 cells before and after treatment with MMS (Figure 5A, lanes 15 and 16). Given the extent of this defect, we considered it unlikely that it simply reflected the effect of XRCC1 interaction on POLβ stability and/or recruitment. Indeed, PARP1 deletion fully rescued the accumulation of POLβ in chromatin in XRCC1−/− cells, during BER (Figure 5A, compare lanes 11 and 12 with lanes 15 and 16 and lanes 19 and 20). Notably, PARP1 deletion also increased the accumulation of POLβ and XRCC1, even in wild-type RPE-1 cells, suggesting that PARP1 and XRCC1 protein complexes compete continuously for SSB intermediates during BER (Figure 5A, compare lanes 12 and 18).

14In summary, we show here that the importance of XRCC1 during BER is to assemble protein complexes that can limit the engagement and activity of PARP1 at SSB intermediates (Figure 5B). XRCC1 is thus an endogenous PARP1 anti-trapper that safeguards genome integrity during BER by preventing PARP1 from impeding this essential DNA repair process in a manner reminiscent of anti-cancer PARP inhibitors.

15DNA base excision repair (BER) is a highly conserved pathway where damaged DNA bases are excised and replaced with undamaged nucleotides using a core set of enzymes comprised of DNA glycosylase, AP endonuclease, DNA polymerase, and DNA ligase activities. In addition, human cells employ several additional proteins to accelerate BER including the SSB sensors PARP1 and/or PARP2 and the molecular scaffold protein XRCC1. The importance of XRCC1 during BER is illustrated by the elevated accumulation of SSBs and cellular hypersensitivity to simple DNA base damage in cells in which XRCC1 is mutated or absent. Although it is well established that XRCC1 interacts with multiple proteins involved in BER, including the core enzymes POLβ and LIG3, the essential role fulfilled by these interactions during BER has been unclear.

16In this study, unexpectedly, our data reveal that the accumulation of SSBs and the cellular hypersensitivity to DNA base damage that are characteristic features of XRCC1−/− cells are ablated by deletion of PARP1. It is known that increased PARP1 activity resulting from elevated levels of unrepaired DNA breaks can result in cell death via NAD+ depletion, defective glycolysis, and necrosis or parthanatos. Thus, a simple interpretation of the PARP1-induced toxicity detected here is that it is a consequence of the increased level of unrepaired SSBs in XRCC1−/− cells. However, this is not the case because it is increased PARP1 engagement that causes the accumulation of SSBs in XRCC1−/− cells during BER, rather than the other way around. This work reveals that the essential role of XRCC1 during BER is to prevent excessive PARP1 engagement and activity at SSB intermediates, which otherwise blocks their access and repair by other BER enzymes. This work overturns the paradigm that there is an intrinsic requirement for XRCC1 for rapid rates of BER because, in the absence of PARP1, XRCC1 is dispensable.

17The excessive engagement of PARP1 during BER in the absence of XRCC1 is reminiscent of the effect of PARP inhibitors in wild-type cells. However, although PARP trapping by PARP inhibitors reflects reduced PARP1 auto-ribosylation and allosterically enhanced DNA binding resulting from chemical inhibition, PARP1 trapping in the absence of XRCC1 reflects the absence of BER complexes that can limit and/or regulate cycles of PARP1 association and activation. If this excessive PARP1 association and activation is prolonged sufficiently, NAD+ becomes progressively depleted, reducing the ability of PARP1 to auto-ribosylate and dissociate from BER intermediates and promoting its accumulation in chromatin. We define both the excessive PARP1 association and subsequently reduced dissociation in this scenario as forms of PARP1 trapping, both of which directly impede BER.

18The ability of XRCC1 to suppress PARP1 toxicity during BER reflects its ability to assemble BER enzymes into protein complexes because this suppression required the XRCC1 domains that interact with POLβ and LIG3. Truncated XRCC1 lacking these domains was also unable to prevent the initial hyperactivation of PARP1 during BER and its subsequent sequestration into chromatin. Truncated XRCC1 did support a small but statistically significant decrease in SSB acumulation, however, suggesting that the central DNA binding and/or poly(ADP-ribose)-binding domains that are retained in the truncated protein contribute to PARP1 regulation. This is consistent with the established requirement for these domains for proper XRCC1 function during BER. The efficacy of XRCC1 protein complexes in regulating PARP1 most likely reflects the close proximity of POLβ and LIG3 within these complexes and that these enzymes are required consecutively during BER from the point where the SSB is created by APE1 to the final step of DNA ligation. The assembly of these enzymes into XRCC1 protein complexes may thus facilitate coordinated hand-off of SSB intermediates to POLβ and LIG3 during BER in a manner similar to that proposed for APE1 and POLβ, reducing the opportunity for interference by PARP1. Interestingly, LIG3 possesses a “nick-sensing” zinc finger that is similar to those in PARP1 and is able to compete with and suppress PARP1 binding and activation by SSBs. It will be of interest to examine whether this zinc finger contributes to suppression of PARP1 trapping by XRCC1 complexes.

19It is noteworthy that the role of XRCC1 in suppressing PARP1 trapping is not observed at all types of SSBs. For example, XRCC1 and PARP1 exhibit a simple epistatic relationship during the repair of 'direct' SSBs induced by oxidative attack of deoxyribose, in which loss of these proteins alone or together slows SSB repair to a similar extent. This is also true for topoisomerase I-induced SSBs, as shown here with respect to the camptothecin sensitivity of RPE-1 cells in which PARP1 and/or XRCC1 are deleted. The BER intermediates that trap PARP1 are currently unknown, but likely candidates are the incised abasic sites that are created by APE1 because these are unique to this pathway and are the primary substrate of POLβ. Consistent with this idea, PARP1 binds tightly to these SSBs and can even become covalently crosslinked to DNA at such sites. Interestingly, PARP2 also accumulated in chromatin in XRCC1−/− cells during MMS treatment, which is expected because NAD+ is also required for PARP2 auto-ribosylation and dissociation. However, importantly, PARP2 depletion reduced neither the accumulation of SSBs in XRCC1−/− cells during BER nor their sensitivity to MMS. This may reflect a lower level of PARP2 protein/activity and/or the different mode of binding by PARP2 to BER intermediates, which lacks the high-affinity zinc-finger motifs characteristic of PARP1.

20To explain our data, we propose the following model (Figure 5B). Although PARP1 and/or PARP2 activity is required to promote BER, most likely to regulate chromatin compaction, PARP1 has a propensity to associate excessively with BER intermediates, blocking their access and repair by other BER enzymes. We define this excessive association as a form of PARP1 trapping. When NAD+ becomes sufficiently depleted by excessive cycles of PARP1 association, the ability of PARP1 to auto-ribosylate and dissociate from BER intermediates is progressively reduced, leading to even tighter trapping and accumulation of PARP1 in chromatin. By assembling POLβ and LIG3 into protein complexes, XRCC1 can promote the molecular “hand-off” of SSB intermediates from one enzyme to the next during BER, limiting the opportunity for excessive PARP1 engagement and thereby suppressing PARP1 trapping and promoting repair. Consequently, in the absence of PARP1, XRCC1 is dispensable for rapid rates of BER because this pathway can be conducted by POLβ and LIG3 and/or other DNA repair enzymes without hindrance. Intriguingly, we also detected increased PARP1 trapping in primary human fibroblasts from XRCC1-mutated disease, and we reported recently that loss of cerebellar interneurons and cerebellar ataxia in Xrcc1-deleted mice are suppressed greatly by Parp1 deletion. It will be of interest to determine whether endogenous PARP1 trapping during BER accounts for this Parp1-dependent neuropathology in Xrcc1-defective brain and disease.

21Here we show that, in the absence of XRCC1, PARP1 can become excessively engaged on BER intermediates in a manner similar to that induced by anti-cancer PARP1 inhibitors, demonstrating that PARP1 trapping is a threat to normal genome integrity. We show that the essential role of XRCC1 during BER is as a PARP1 anti-trapper that regulates the engagement and activity of this enzyme, ensuring that this essential DNA repair pathway occurs rapidly and without obstruction.

22We show here that PARP1 trapping is an endogenous threat to genome integrity and that it is the role of XRCC1 to prevent this from happening. Although this role requires XRCC1 to bind poly(ADP-ribose) and assemble DNA BER enzymes into protein complexes, the precise details of how such complexes coordinate BER remain to be defined.

23Further information and requests for resources and reagents should be directed to and will be fulfilled by the Lead Contact, K.W.Caldecott (k.w.caldecott@sussex.ac.uk).

24All unique reagents (plasmids/cell lines etc) will be provided on request to academic laboratories without restrictions.

25All primary data are available on request to the lead contact. No code or large genomic/proteomic datasets are associated with this work.

26Human hTERT RPE-1 cells (ATCC) and U2OS cells were incubated in a low oxygen (3%) incubator (37°C, 5% CO2). RPE-1 cells were maintained in DMEM-F12 Glutamax 10% FBS supplemented with penicillin/streptomycin. U2OS were maintained in DMEM 10% FBS supplemented with penicillin/streptomycin and L-glutamine. Human TK6 cells were cultured in 20% oxygen (37°C, 5% CO2) in RPMI 1640 medium (Nacalai Tesque, Kyoto, Japan) supplemented with heat-inactivated horse serum (10%) (GIBCO, lot no. 2017-06), Sodium pyruvate (0.1 mM), L-glutamine, and penicillin/streptomycin. The XRCC1−/−, PARP1−/− and XRCC1−/−/ PARP1−/− RPE-1 cells, and U2OS XRCC1−/− cells, used in this study were generated and characterized previously. U2OS XRCC1−/− cells complemented with full-length N-terminal Myc-His-tagged XRCC1 or truncated N-terminal Myc-His-tagged XRCC1161-406 were generated as described below (“siRNA and cell transfection”).

27The PARP inhibitor KU 0058948 hydrochloride and PARG inhibitor were purchased from Axon (Cat#2001; CAS: 763111-49-5) and Tocris (PDD0017273; Cat#5952; CAS: 1945950-21-9), respectively. Both inhibitors were dissolved in dimethyl sulfoxide (DMSO) to a stock concentration of 10 mM and used at a final working concentration of 10 μM. Methyl methanesulfonate (MMS; Sigma-Aldrich, Cat#129925) was dissolved directly into culture medium and concentration used is as indicated in Figures. Recombinant PARG enzyme was from Trevigen.

28Mouse anti-PARP1 monoclonal (Santa Cruz Cat#sc-8007; RRID:AB_628105) was employed at 1:500 rabbit polyclonal anti-PARP2 (Active Motif Cat#39743; RRID:AB_2793328) at 1:5000, rabbit anti-XRCC1 polyclonal (Novus; Cat#NBP1-87154; RRID:AB_11029388) at 1:5000, rabbit anti-DNA polymerase beta polyclonal (Millipore; Cat#6C0087) at 1:1000, rabbit anti-DNA ligase III polyclonal sera (TL25) at 1:5000, anti-poly-ADP-ribose binding reagent (Millipore; Cat# MABE1031; RRID:AB_2665467) at 1:10,000, mouse anti-alpha-Tubulin monoclonal (Sigma-Aldrich; Cat#T6074; RRID:AB_477582) at 1:10000, and rabbit anti-histone H3 polyclonal (Abcam; Cat#ab1791; RRID:AB_302613) at 1:10000.

29Non-targeting siRNA (ON-TARGETplus) and SMARTpool siRNA (25 nM) against PARP1 or PARP2 were reverse-transfected into cells using Lipofectamine® RNAiMAX (Invitrogen) according to the manufacturer’s instructions. All experiments were carried out 72 hr post-transfection. U2OS XRCC1−/− cells complemented with full-length N-terminal Myc-His-tagged XRCC1 or truncated N-terminal Myc-His-tagged XRCC1161-406 were generated by co-transfection with an empty vector (pCI-Puro) encoding resistance to puromycin and either pCD2E (empty vector), pcD2E-Myc-His-XRCC1, or pcD2E-Myc-His-XRCC1161-406. Following transfection, cells were selected against puromycin (2 μg/ml) and after one week single colonies were isolated, amplified, and validated for expression of recombinant XRCC1.

30The recruitment and exchange of GFP-tagged PARP1 was measured in wild-type and Xrcc1−/− MEFs as described previously with minor modification. Briefly, ∼1x104 MEFs were seeded on 35 mm diameter glass-bottom plate on day 1 and transiently transfected with GFP-PARP1 plasmid (1 μg) via Lipofectamine 2000 on day 2 (Invitrogen). All images were collected on day 4 (2 days after transfection) with a Nikon Ti Eclipse inverted microscope equipped with the Lu-N3 Laser Units and the A1 RMP confocal system (all from Nikon Inc, Tokyo, Japan). The micro-irradiation was generated by a 405 nm laser (energy level ∼500 μW) in ∼0.8 μm diameter nucleoli-free region of the nuclei. The GFP-PARP1 recruitment were measured in control untreated cells and cells treated with 0.3 mg/ml MMS (Sigma, Cat. 64294) for ≥ 60 min. Images were taken immediately following micro-irradiation (defined as 0 s) up to 1 minute with 2 s interval. The relative intensity of GFP-PARP1 foci is defined as the mean green fluorescent intensity at the micro-radiated area (0.8 μm diameter)/mean green fluorescent intensity in the entire nucleus. For FRAP, the cells were either untreated or pre-treated as indicated with 0.3 mg/ml MMS for 0-30 minutes or ≥ 60 minutes before micro-irradiation. A confined nucleoli-free area of ∼0.8 μm diameter in the nucleus was bleached with a 488 nm laser targeting GFP (energy level = 217 μW) and images were taken every other second to 1 minute. The recovery curve was plotted as the percentage of GFP intensity (damage region)/GFP intensity (whole nucleus) at the given time points versus before bleaching of each given cell. At least 6 cells were collected for recruitment analysis of each condition and at least 10 cells were collected for FRAP assay of each condition.

31Recombinant histidine-tagged human XRCC1 (XRCC1-His) and DNA ligase IIIα (His-LIG3), and untagged DNA polymerase β (POLβ), were expressed in and purified from 1-2 l of E.coli culture essentially as described previously, using an AKTA Pure FPLC system. In brief, recombinant XRCC1-His and His-LIG3 were purified by sequential metal-chelate affinity chromatography and gel filtration using 1ml HisTrap (GE Healthcare) HiLoad 16/600 Superdex 200 (GE Healthcare), respectively. Recombinant human POLβ was purified by sequential anion exchange, cation exchange, and gel filtration using HiTrap Q HP (2x1ml; GE healthcare), HiTrap SP (1ml; GE Healthcare) and HiLoad 16/600 Superdex 200 columns.

32PARP1 reactions (20 μL) were incubated at room temp for 5 min and contained 50 mM Tris-HCl pH7.9, 100 mM NaCl, 10 mM MgCl2, 100 μg/ml BSA, 0.3 μM each of the indicated recombinant proteins, 10 μM NAD+ (where indicated), and 0.15 μM of fluorescein labeled double-hairpin duplex substrate containing a single uracil residue (IDT; Integrated DNA Technologies). Reactions were terminated by the addition of SDS-PAGE sample buffer and heating at 95°C. for 5 minutes. The duplex hairpin substrate was generated by heating/cooling the single strand oligonucleotide 5′- GCACGGCGCATCAGCTGCAGAACAACTGCAGCTGATGCGC/deoxyU/GTGCGGATCCGGTGCAAC/iFluorT/AAGCACCGGATCC-3′ to form the duplex followed by ligation of the residual nick with T4 ligase and gel purification. The substrate possessed a single internal fluorescein dT residue for visualization/purification. The substrate was pre-incubated or not as indicated with uracil DNA glycosylase and AP endonuclease (UDG/APE1; 1.2 U each; NEB) to create a SSB intermediate of BER (cleaved abasic site) prior to the addition, where indicated, of recombinant PARP1, XRCC1, POLB, LIG3.

33XRCC1 was disrupted with knockout constructs prepared using primers 5′-GTAGTAAAAGACAGATGCCCACAGTCCACA-3′ and 5′-CTGGCTGCTGCAGGACACGACATGGCGGAG-3′ for the left arm and 5′-ACTCACTGTGCAGAAAATCTTCTCAAGGCA-3′ and 5′-AACCACCATACCTGGCTATTATTCTTTAAA-3′ for the right arm. The TALEN vector was designed to recognize the following sequences: 5′-TGACATGCCGGAGATCCG-3′ and 5′-GTCCTGCAGCAGCCAGGA-3′. PARP1 was disrupted with knockout constructs prepared using primers 5′-TGGGGAGTAGTGCTTTGTTTGGATATATCC-3′ and 5′-CTGGAGAATCAAACAGACAGCAATGCTCAT-3′ for the left arm and 5′-GTAAGATCTTGGGGGCCCAGATCCCTGAAC-3′ and 5′-CTTAAATTCCAAATGGCTGGCAACCTACCT-3′ for the right arm. We used CRISPR/Cas9 and the guide sequence 5′-GAAGTACGTGCAAGGGGTGTATGG-3′ to facilitate gene targeting. For gene targeting, we used the maker genes DT-ApA/NEOR (provided by the Laboratory for Animal Resources and Genetic Engineering, Center for Developmental Biology, RIKEN Kobe; http://www.clst.riken.jp/arg/cassette.html) and DT-ApA/PUROR digested with ApaI and AflII. Wild-type TK6 cells were transfected with the above-mentioned targeting vectors (2 μg), and the expression vector (10 μg) for TALEN or CRISPR (pX330; Addgene, US), employing the NEON transfection system (Invitrogen, CA) at 1500 V 20 msec. β-actin transcripts were used as a positive control for the RT-PCR analysis using primers 5′-GATGGTGGGCATGGGTCAGAAGGATTCC-3′ and 5′-GTCCAGGGCGAGGTAGCACAGCTTCTC-3′.

34Cells were treated with the indicated concentration of MMS in the presence/absence of 10 μM PARP inhibitor (KU0058948) for 15 min at 37°C in medium, and alkaline comet assays conducted essentially as described previously. In brief, following the indicated treatment of 5x104 cells in suspension (1 ml) with MMS, cells were washed once in ice-cold PBS and resuspended in 0.45 mL PBS. Aliquots (0.15 ml) of the cell suspension mixed with an equal volume of low gelling agarose (at 42°C), spread on agarose pre-coated frosted slides and set on ice, and then lysed in alkaline lysis buffer (2.5M NaCl, 100 mM EDTA, 10 mM Tris-Cl, 1% v/v DMSO, 1% v/v Triton X-100, pH 10) at 4°C for 60 min. Samples were then pre-incubated for 45 min and subject to electrophoresis at 12V for 25 min in electrophoresis buffer (50mM NaOH, 1mM EDTA, 1% DMSO, pH13) at 4°C. Slides were neutralized in 0.4M Tris-Cl pH 7.4 prior to staining in PBS containing SYBR Green (Sigma, 1:10000 dilution) and 0.04 mg/ml p-Phenyledenediamine dihydrochloride anti-fade (Fisher 417481000). Comet tail moments (an arbitrary-unit measure of DNA strand breaks) from 50-100 cells per sample were scored blinded and using automated Comet Assay IV software (Perceptive Instruments, UK).

35For RPE-1 and U2OS cells we conducted clonogenic survival assays. Cells were plated in 10 mm plates and 4 h later treated with indicated concentrations of MMS for 30 min or with CPT continuously. Cells were rinsed twice with PBS, incubated in drug-free medium for 10 days, and then fixed in 100% ethanol and stained with 0.05% crystal violet. The surviving fraction at each dose was calculated by dividing the average number of colonies (defined as > 50 cells) in treated dishes by the average number in untreated dishes. For TK6 cells, we employed a liquid-culture cell survival assay, as described previously. 1x106 cells were suspended in PBS (1 ml) containing horse serum (1%) and treated with MMS for 1 h. Treated cells (10 μl) were transferred to culture medium (1 ml) and cultured for 72 h. The incubated cells (100 μl) were transferred to 96-well plates, and the amount of ATP was measured using CellTiter-Glo (Promega; Cat# G7570), according to the manufacturer’s instructions. Luminescence was measured using the Fluoroskan Ascent FL (Thermo Fisher Scientific Inc., Waltham, MA).

36RPE-1, U2OS or human fibroblast cells were harvested and lysed in lysis buffer containing 100 mM, 150 mM or 150 mM KCl, respectively, and 50 mM HEPES pH 7.4, 2.5 mM MgCl2, 5 mM EDTA pH 8, 3 mM dithiothreitol (DTT), 0.5% Triton X-100, 10% glycerol, and protease inhibitor cocktail (Roche) for 45 min on ice. Soluble and chromatin-bound proteins were separated by centrifugation (15 min, 16,000 g). The pellet containing the detergent insoluble material (including chromatin) was washed twice in lysis buffer and then subjected to sonication to shear the DNA. The soluble and chromatin extracts were mixed with SDS-PAGE sample buffer and heated for 10 min at 95°C. For protein levels in whole cell extracts (WCE), cells were lysed directly in SDS-PAGE sample buffer and heated as above. All protein extracts were subjected to SDS-PAGE followed by protein transfer onto nitrocellulose membrane and western blotting with the indicated antibodies.

37NAD+ levels in RPE-1 cells were determined by a chromogenic assay as described before. Briefly, cells were treated with 0.3 mg/ml MMS in culture media at 37°C for the indicated time, washed with PBS and scraped in PBS supplemented with 100 μM PARG inhibitor (PDD00017273, Sigma), 40 μM PARP inhibitor (KU 0058948, Axon), and cOmplete protease inhibitors (04693132001, Roche). Cell pellets were resuspended in lysis buffer (20 mM sodium bicarbonate, 100 mM sodium carbonate, 0.5% Triton X-100, 10 mM nicotinamide, 100 μM PARG inhibitor, 40 μM PARP inhibitor, and cOmplete protease inhibitors, pH 10.3) and lysed by two freeze/thaw cycles. Protein concentration was normalized, lysates were transferred to a 10,000 MWCO centrifugal filter (UFC501024, Merck) and centrifuged at 14,000 × g at 4 °C for 30 min. Half of each lysate was incubated at 60 °C for 30 min to decompose NAD+. Samples were incubated in cycling buffer [100 mM tricine-NaOH (pH 8), 4 mM EDTA, 40 mM NaCl, 1.66 mM phenazine ethosulfate (PES), 0.42 mM thiazolyl blue tetrazolium bromide (MTT), 10% ethanol] at 37 °C for 5 min, and 10 U/ml alcohol dehydrogenase, reconstituted in 100 mM tricine-NaOH (pH 8), was added to drive a cycling reaction at 37 °C for 40 min. The reaction was terminated by addition of NaCl (2 M final concentration) and samples were centrifuged at 14,000 × g at 4 °C for 5 min. Reduced MTT was resuspended in 100% ethanol and the absorbance was measured at 560 nm. NAD+ concentrations were calculated by subtracting the absorbance of samples with thermal decomposition prior to the cycling reaction from those samples without thermal decomposition prior to the cycling reaction.

38Data were examined for statistical significance using GraphPad Prism 9 using one- or two-way ANOVA with appropriate post hoc tests, as indicated in the figure legends. For survival assays, plotted data are mean (+/−SEM) surviving fraction (%) calculated from the indicated number (N) of independent biological replicate experiments. For comet assays, plotted data are the comet tail moments of every individual cell in each of the indicated number (N) of independent experiments. Note that for the comet assays, individual cells within each experiment are technical replicates, and independent experiments are biological replicates.