XRCC1 protects transcription from toxic PARP1 activity during DNA base excision repair

Genetic defects in the repair of DNA single-strand breaks (SSBs) can result in neurological disease triggered by toxic activity of the single-strand-break sensor protein PARP1. However, the mechanism(s) by which this toxic PARP1 activity triggers cellular dysfunction are unclear. Here we show that human cells lacking XRCC1 fail to rapidly recover transcription following DNA base damage, a phenotype also observed in patient-derived fibroblasts with XRCC1 mutations and Xrcc1−/− mouse neurons. This defect is caused by excessive/aberrant PARP1 activity during DNA base excision repair, resulting from the loss of PARP1 regulation by XRCC1. We show that aberrant PARP1 activity suppresses transcriptional recovery during base excision repair by promoting excessive recruitment and activity of the ubiquitin protease USP3, which as a result reduces the level of monoubiquitinated histones important for normal transcriptional regulation. Importantly, inhibition and/or deletion of PARP1 or USP3 restores transcriptional recovery in XRCC1−/− cells, highlighting PARP1 and USP3 as possible therapeutic targets in neurological disease.

D NA single-strand breaks (SSBs) are one of the most common types of DNA damage in cells, arising at a frequency of tens-of-thousands per cell per day 1 . The threat posed by SSBs is illustrated by hereditary genetic diseases in which the repair of SSBs is defective, resulting in neurological pathologies such as cerebellar ataxia, neurodevelopmental delay and seizures 2,3 . Single-strand breaks are rapidly detected by poly-ADP-ribose polymerase 1 (PARP1), which following binding to the DNA break is catalytically activated and modifies itself and other proteins, such as histones, with ADP-ribose [4][5][6] . ADP-ribosylation accelerates SSB repair in a number of ways-for example, by modifying the structure of chromatin in the vicinity of the break and/or recruiting specific DNA repair factors such as XRCC1 (refs. 4,6 ). The extensive auto-ribosylation of PARP1 leads to disassociation of the enzyme from the SSB, enabling other DNA repair enzymes to access and repair the break 7 .
XRCC1 is a scaffold protein that is recruited to SSBs by PARP1 and/or PARP2 activity through direct interaction between poly(ADP-ribose) and the central BRCT domain in XRCC1 (refs. [8][9][10] ). XRCC1 in turn interacts with and recruits the enzymes required for SSB repair, including DNA polymerase β 11,12 , DNA ligase III 13 , polynucleotide kinase/phosphatase 14,15 and aprataxin 16,17 . Hereditary mutations in XRCC1 and some of these protein partners result in neurodevelopmental disorders and/or progressive neurodegeneration 2,3 . Intriguingly, the deletion and/or inhibition of PARP1 greatly reduces or prevents these pathologies in an Xrcc1-defective mouse model of SSB-associated neurological disease, highlighting excessive/aberrant PARP1 activity as a source of SSB-induced neuropathology 18,19 . However, the molecular mechanism by which excessive/aberrant PARP1 activity triggers cellular dysfunction remains unclear.
Here we show that prolonged PARP1 activity at unrepaired DNA base damage suppresses the recovery of transcription by promoting the excessive recruitment and activity of the ubiquitin-specific protease USP3. Remarkably, excessive USP3 activity reduces global levels of monoubiquitinated histones important for transcriptional regulation, such as histone H2A and H2B monoubiquitination at K119 and K120 (H2AmUb and H2BmUb, respectively). Importantly, either PARP1 or USP3 inhibition/depletion rescue normal levels of histone monoubiquitination and transcription recovery following DNA base damage, highlighting these enzymes as possible therapeutic targets in the treatment of base excision repair (BER)-defective neurological disease.

Results
XRCC1 promotes the recovery of transcription following oxidative damage. Given that SSBs can slow or block the physical progression of RNA polymerases [20][21][22] , we wondered whether SSB repair-defective cells exhibit defects in transcription following DNA damage. We therefore examined the impact of XRCC1 deletion on global transcription in human RPE-1 cells following treatment with H 2 O 2 (a physiologically relevant source of oxidative SSBs) by pulse labelling with 5-ethynyluridine (EU). We observed a robust inhibition of global transcription immediately after treatment with H 2 O 2 in both wild-type (WT) and XRCC1 −/− RPE-1 cells (Fig. 1a,b). More importantly, whereas the levels of transcription recovered significantly within 2 h following H 2 O 2 treatment in the WT RPE-1 cells, they declined further in the XRCC1 −/− cells (Fig. 1a,b). The lack of transcription recovery in XRCC1 −/− cells was accompanied by a progressive loss of RNA polymerase I (RNAPI) foci (RPA194; Fig. 1a,c). Given that we did not detect a reduction in the global levels of RPA194 in the XRCC1 −/− cells, this most probably reflected the dissociation of RNAPI from nucleolar sites of ribosomal RNA transcription rather than degradation of the RNAP (Extended Data Fig. 1a).
Although most of the EU signal detected in our pulse-labelling experiments was nucleolar, the recovery of EU incorporation in the XRCC1 −/− RPE-1 cells was also reduced in the nucleoplasm (Fig. 1a), suggesting that RNA polymerase II (RNAPII)-dependent XRCC1 protects transcription from toxic PARP1 activity during DNA base excision repair Articles NATURE CEll BIoloGy transcription was also affected. To confirm this, we compared the levels of hyperphosphorylated RNAPII, which is an established measure of the RNAPII transcriptional activity 23,24 , in WT and XRCC1 −/− RPE-1 cells before and after H 2 O 2 treatment. The level of RNAPII hyperphosphorylation was markedly reduced in the XRCC1 −/− RPE-1 cells following H 2 O 2 treatment, an observation we verified with antibodies specific to two phosphorylated sites (S2 and S5) within the carboxy (C)-terminal domain of RNAPII that are associated with transcriptional activity (Fig. 1d) 25 . The loss of hyperphosphorylated RNAPII seemed to involve, at least partly, proteolysis of RNAPII, given that incubation with proteasome inhibitors restored normal levels of the hyperphosphorylated RNAP (Extended Data Fig. 1b). However, proteasome inhibitors did not restore normal levels of transcription recovery, indicating that degradation of hyperphosphorylated RNAPII was a result of failed transcription recovery rather than a cause (Extended Data Fig. 1b-e). Importantly, the defect in transcription recovery was neither limited to RPE-1 cells nor an off-target effect of gene editing because we observed similar results in XRCC1 −/− U2OS cells and the defect in transcriptional recovery in these cells was corrected by stable expression of recombinant human XRCC1 (Extended Data Fig. 2a-d). Collectively, these data indicate that XRCC1 is required for the rapid recovery of transcription by both RNAPI and RNAPII following oxidative stress.

Reduced transcriptional recovery in XRCC1-deficient cells and neurons is a result of PARP1.
Hereditary mutations in human XRCC1 result in cerebellar ataxia, a phenotype that is also observed in Xrcc Nes-cre mice in which Xrcc1 is conditionally deleted in the brain 18,19,26 . This phenotype is largely rescued by the deletion of Parp1, the primary sensor of unrepaired SSBs in mammalian cells 18,19 . We therefore examined the impact of PARP1 deletion on the recovery of transcription in XRCC1 −/− cells. Strikingly, PARP1 deletion rescued the transcriptional recovery in XRCC1 −/− RPE-1 cells, as measured by EU pulse labelling, RNAPI foci and RNAPII phosphorylation ( Fig. 2a-d). In addition, transcription recovery was rescued in both XRCC1 −/− RPE-1 and U2OS cells by incubation with PARP inhibitor (PARPi), confirming that the transcription defect was the result of toxic/aberrant ADP-ribosylation ( Fig. 2e and Extended Data Fig. 3). Consistent with this idea, the defect observed in the XRCC1 −/− cells was recapitulated in WT RPE-1 cells by incubation with an inhibitor of PARG (the enzyme primarily responsible for poly(ADP-ribose) degradation), which suppressed the recovery of global transcription for up to 8 h following H 2 O 2 treatment (Fig. 2f-h). Importantly, reduced transcriptional recovery following oxidative stress was also observed in patient-derived human fibroblasts with XRCC1 mutations as well as Xrcc1 Nes-cre mouse cerebellar neurones, and these defects were again prevented by PARPi, implicating PARP1-induced transcriptional repression in neurological disease ( Fig. 3 and Extended Data Fig. 4).

NATURE CEll BIoloGy
The poly(ADP-ribose)-binding activity of XRCC1 promotes transcriptional recovery by directly regulating PARP1 activity.
To understand how XRCC1 prevents toxic PARP1 activity, we identified the protein domain/s important for this process. A truncated Myc-histidine (Myc-His)-tagged fragment of XRCC1 encoding only the central one-third of the protein (Myc-His-XRCC1 161-406 ) rescued transcription recovery in XRCC1 −/− U2OS cells (Fig. 4a,b and Extended Data Fig. 5a-e). This was surprising because this fragment lacks the protein-interaction domains that underpin the role of XRCC1 as a molecular scaffold 27 and so cannot support normal rates of SSB repair (Extended Data Fig. 5f,g). The ability of XRCC1 to bind directly to poly(ADP-ribose) was required to prevent PARP1-mediated transcriptional suppression, as indicated by the failure of a mutant derivative of Myc-His-XRCC1 161-406 harbouring mutations that prevent binding to this polymer (R335A/ K369A; denoted Myc-His-XRCC1 161-406,RK ) 9 to rescue transcriptional recovery, thereby suggesting that XRCC1 might regulate poly(ADP-ribosylation) directly ( Fig. 4b and Extended Data Fig. 5a-e). The most probable target of this regulation is PARP1 itself, given that this protein was the most abundant, if not the only, ribosylated protein detected in Myc-His-XRCC1 161-406 immunoprecipitates following H 2 O 2 treatment and its recovery was prevented by the mutations that prevent poly(ADP-ribose) binding ( Fig. 4c). Consistent with direct regulation of PARP1 activity by the poly(ADP-ribose)-binding domain of XRCC1, both recombinant full-length XRCC1 and XRCC1 161-406 reduced PARP1 auto-ribosylation in vitro, whereas the mutant derivatives of these proteins that cannot bind poly(ADP-ribose) were unable to do so (Fig. 4d). These data suggest that XRCC1 can suppress aberrant PARP1 activity not only by accelerating SSB repair but also by direct binding to, and suppression of, poly(ADP-ribose) chain extension. To confirm this idea, we examined whether transcriptional recovery could be rescued by the overexpression of APLF, a structurally distinct poly(ADP-ribose)-binding protein that can also negatively regulate poly(ADP-ribose) synthesis 28 . Recombinant APLF suppressed poly(ADP-ribose) chain extension in vitro, albeit less effectively than XRCC1, and overexpression of high levels of APLF restored transcriptional recovery in XRCC1 −/− U2OS cells in a manner dependent on the poly(ADP-ribose)-binding zinc finger domain 29 (Fig. 4e,f and Extended Data Fig. 5h-k). These data indicate that the negative regulation of PARP1 activity by direct binding of XRCC1 to poly(ADP-ribose) is both required and sufficient for normal transcription recovery following oxidative stress.
Prolonged transcriptional suppression is triggered by low but persistent PARP1 activity at unrepaired DNA BER intermediates. Collectively, the experiments described so far indicate that although the initial suppression of transcription by H 2 O 2 is independent of PARP1 activity and probably reflects the stalling/inefficient bypass of DNA lesions by RNAPs [20][21][22] , the prolonged suppression observed in XRCC1 −/− cells is imposed by excessive/aberrant PARP1 activity at persistent SSBs. To address this further we conducted time-course experiments to determine when and for how long transcriptional suppression occurs. Whereas WT RPE-1 cells continued to recover transcription from 2 h following H 2 O 2 treatment, with the transcription levels restored to normal within 24 h, XRCC1 −/− cells exhibited little or no transcriptional recovery for at least 9 h following H 2 O 2 treatment ( Fig. 5a and Extended Data Fig.  6a,b). Moreover, incubation with PARPi for 1 h rescued transcription even if applied 8 h after the H 2 O 2 treatment, indicating that the failure to recover transcription in the XRCC1 −/− cells was a result of prolonged and ongoing ADP-ribosylation ( Fig. 5a and Extended Data Fig. 6a,b). Consistent with this idea, although H 2 O 2 -induced protein ADP-ribosylation declined below detectable levels in both WT and XRCC1 −/− cell extracts 2 h after H 2 O 2 treatment, as measured by western blotting, we detected low but persistent levels of PARP1 auto-ribosylation in the XRCC1 −/− cell extracts following enrichment with anti-ADP-ribose antibodies (Extended Data Fig. 6c,d).
We reasoned that the low level of persistent PARP1 activity in the XRCC1 −/− cells most probably reflected SSBs arising as intermediates of BER, given that while SSBs induced directly by disintegration of oxidized deoxyribose arise immediately, SSBs arising during BER require the enzymatic excision of oxidized DNA bases and thus arise more slowly. Moreover, we recently showed that PARP1 can become 'trapped' on SSB intermediates in XRCC1 −/− cells during BER and that low but persistent levels of PARP1 auto-ribosylation is a feature of this phenomenon 30 . Consistent with this idea, short interfering RNA (siRNA)-mediated depletion of NTH1-a DNA glycosylase that excises a broad spectrum of oxidized DNA bases 31-33 -rescued the recovery of transcription in H 2 O 2 -treated XRCC1 −/− cells (Fig. 5b,c and Extended Data Fig. 6e). Furthermore, global transcription was substantially suppressed in XRCC1 −/− cells following prolonged treatment with methyl methanesulfonate (MMS), an alkylating agent that induces SSBs largely or entirely via BER ( Fig. 5d-f). This repression was also accompanied by a loss of RNAPII hyperphosphorylation, which once again was rescued by PARPi (Fig. 5f). Together, these data indicate that transcriptional recovery is suppressed in XRCC1 −/− cells by low levels of persistent PARP1 activity at unrepaired BER intermediates.
Aberrant PARP1 activity disrupts global histone monoubiquitination. Next, we examined the mechanism/s by which aberrant PARP1 activity inhibits transcription recovery during BER. Given that PARP1 can affect transcription by ADP-ribosylating histones 34-36 , we examined the impact of the deletion of histone parylation factor-1 (HPF1), a protein partner of PARP1 that is required for histone ADP-ribosylation [37][38][39][40] . However, although HPF1 deletion prevented detectable histone ADP-ribosylation in XRCC1 −/− U2OS cells as expected, it did not rescue transcription recovery, suggesting

NATURE CEll BIoloGy
that histone ribosylation is not the cause of transcriptional suppression (Extended Data Fig. 7).
As PARP1 can also affect transcription indirectly, via the recruitment of chromatin remodelling proteins 41,42 , we examined whether aberrant ADP-ribosylation affected the levels of histone marks known to promote transcriptional activity. Although the global levels of methylation and acetylation on histone H3 (H3K4me3 and H3K9Ac, respectively) were slightly, albeit not statistically   Mean relative number of RPA194 foci per nucleus  Fig. 8a,b), the global levels of H2AmUb and H2BmUb were dramatically reduced ( Fig. 6a,     The cells were pulse labelled with eU as in Fig. 1. b,

NATURE CEll BIoloGy
XRCC1 −/− cells by incubation with PARPi, demonstrating that the loss of these important histones modifications was the result of aberrant ADP-ribosylation ( Fig. 6a-d). This impact of aberrant PARP activity on histone ubiquitination was particularly pronounced given that we did not detect a reduction in total chromatin ubiquitination, as detected by the anti-ubiquitin antibody FK2 (Extended Data Fig. 8c).

PARP1 suppresses histone monoubiquination and transcription recovery by triggering aberrant recruitment of the ubiquitin protease USP3.
Finally, to identify the mechanism by which aberrant PARP1 activity reduces histone monoubiquitination, we depleted enzymes that are known to deubiquitinate histone H2A and/or H2B [46][47][48][49][50][51][52][53] . Whereas siRNA directed against either of the deubiquitylases USP22 and USP36 failed to rescue transcription recovery in XRCC1 −/− cells following H 2 O 2 treatment, siRNA to USP11 resulted in a small but significant rescue of transcription recovery (Extended Data Fig. 9a). More importantly, however, siRNA directed against USP3 resulted in almost complete rescue of transcription recovery in XRCC1 −/− RPE-1 cells following H 2 O 2 treatment, whether measured by EU pulse labelling ( Fig. 7a and Extended Data Fig.  9a) or RNAPII hyperphosphorylation (Fig. 7b). Consistent with these data, USP3 siRNA prevented the PARP1-dependent reduction in histone monoubiquitination following H 2 O 2 treatment ( Fig. 7c and Extended Data Fig. 9b). The USP3 siRNA also reduced the loss of histone monoubiquitination in XRCC1 −/− cells during continuous treatment with MMS ( Fig. 7d and Extended Data Fig. 9c) throughout the 3 h time course. We also detected the accumulation of green fluorescent protein (GFP)-tagged USP3 in the chromatin of the XRCC1 −/− cells during MMS treatment, which was prevented by treatment with PARPi, providing support for PARP1 activity as a source of aberrant USP3 recruitment during BER (Fig. 8a). Interestingly, the amino (N)-terminal zinc finger domain of USP3 was sufficient for PARP-dependent recruitment at BER intermediates. However, we were unable to detect any evidence of direct binding of this domain to poly(ADP-ribose) (Extended Data Fig. 10a-f), which suggests that PARP1 mediates USP3 recruitment indirectly, perhaps via its well-established role in regulating chromatin compaction 6,54,55 .

Discussion
XRCC1 is a molecular scaffold protein that interacts with many of the enzymatic components of SSB repair 27

NATURE CEll BIoloGy
activity triggers this pathology have been unclear. One recent clue emerged from the discovery that the assembly of DNA repair protein complexes by XRCC1 is necessary to prevent excessive engagement and activity of PARP1 during BER, which otherwise leads to PARP1 'trapping' on BER intermediates and consequently their reduced access and repair by other BER enzymes 30 . The data presented in this study demonstrate that the low but persistent levels of PARP1 activity associated with this 'trapped' PARP1 in XRCC1-defective cells lead to prolonged transcriptional suppression. This delay in transcription recovery is not simply a result of slower BER because PARP inhibition by a chemical inhibitor or overexpression of either of two distinct poly(ADP-ribose)-binding domains is sufficient to rescue transcription recovery in XRCC1-defective cells, even in the continued presence of unrepaired BER intermediates. The recovery of transcription at BER intermediates can thus be separated mechanistically from the completion of BER. This result may indicate that although SSBs may initially impede RNAP progression, these lesions can eventually be bypassed, as long as PARP1 signalling at the unrepaired SSBs is suppressed. Alternatively, the prolonged suppression of transcription by PARP1 detected in our experiments may reflect an impact on RNAP initiation and/or elongation in trans at sites distal to unrepaired SSBs.
PARP1 is a dynamic regulator of gene activity and is able to either repress or activate gene transcription in response to different physiological stimuli [57][58][59] . PARP activity can regulate transcription in a number of ways, such as by relaxing higher-order chromatin compaction and/or regulating RNAP activity [60][61][62] . PARP1 can also recruit polycomb and NuRD transcriptional repressor complexes, thereby facilitating the initial suppression of transcription following DNA damage 41,42 . However, we did not detect a requirement for PARP1 for the initial suppression of transcription following DNA damage-the initial reduction in EU pulse labelling observed following H 2 O 2 treatment was not prevented by PARP1 deletion or inhibition. Thus, in our experiments the initial suppression of transcription probably reflects a PARP-independent mechanism, such as the pausing of RNAPII by BER intermediates 21

NATURE CEll BIoloGy
A striking feature of our work is the discovery that excessive/ aberrant PARP1 activity prolongs transcriptional suppression via recruitment of the ubiquitin protease USP3. We do not yet know how poly(ADP-ribose) synthesis recruits USP3 into chromatin at unrepaired BER intermediates. Although the N-terminal zinc finger of USP3 is sufficient for this recruitment, we have so far not detected binding of this domain to poly(ADP-ribose). PARP1 might therefore mediate USP3 recruitment indirectly, such as via its ability to regulate chromatin compaction. For example, the ability of poly(ADP-ribose) to relax chromatin structure directly 64,65 or indirectly via the recruitment of one or more chromatin remodelling enzymes 41,42,66-68 may allow the USP3 zinc finger to detect its ubiquitylated substrate/s 51 . USP3 deubiquitinates multiple substrates, including the histones H2A, γH2AX and H2B, and is implicated in the cellular response to DNA damage both in vitro and in vivo [51][52][53]69 . In particular, reduced USP3 levels are associated with increased levels of DNA breakage and DNA replication stress as well as slower S phase progression 51,52 . USP3 reduces the level of γH2AX and H2A ubiquitination at K13 and K15, which are key sites of RNF8-and RNF168-mediated ubiquitination that regulate the repair of DNA double-strand breaks 53,69,70 . However, these modifications are unlikely to explain the impact of USP3 on transcription repression in our experiments, which result from unrepaired SSBs. Moreover, these modifications are not known to regulate transcriptional activity.
USP3 also deubiquitinates H2A and H2B monoubiquitination at K119 and K120, respectively 51,52 , both of which are critical for normal transcriptional control and are regulated in response to DNA damage 42,[71][72][73][74] . Of these two modifications, it is the loss of H2BmUb that is a more likely explanation for the prolonged transcriptional suppression observed in our experiments. This is because this modification promotes transcriptional activity, whereas loss of H2AmUb ought to increase transcription. We do not yet know why loss of H2BmUb might dominate the transcriptional response in our experiments but it may relate to the different mechanisms and/ or kinetics by which H2AmUb and H2BmUb regulate transcription. For example, loss of H2BmUb can reduce RNAP progression directly, and thus rapidly, by affecting nucleosome unwinding mediated by the histone chaperone FACT ahead of RNAP 44 . In contrast, loss of H2AmUb may increase gene transcription more slowly via the progressive loss of polycomb-repressive complexes from chromatin 75 . Consistent with this idea, we did not detect changes in the levels of H3K4me3, which is a downstream target of H2AmUb 76 , in our experiments (Extended Data Fig. 8).
H2BmUb is tightly coupled with RNAP elongation 44,77,78 and promotes transcription in several ways, including facilitating histone In WT cells, RNAP pausing at BeR intermediates is accompanied by PARP1 activation, completion of BeR by XRCC1 protein complexes and transcription resumption. In XRCC1-defective cells, persistent PARP1 activity at unrepaired BeR intermediates leads to aberrant USP3 recruitment, extensive/excessive protein deubiquitination (including histones) at damaged and proximal/nearby undamaged sites, and prolonged transcriptional suppression. Note that PARP1 inhibition prevents USP3 recruitment and rescues global transcription recovery, most probably reflecting the eventual bypass of unrepaired SSBs by RNAP and/or transcription resumption at proximal/nearby undamaged sites.

NATURE CEll BIoloGy
H3K4 and H3K79 methylation 79,80 and cooperating with FACT 44 . H2BmUb is also regulated in response to DNA damage. For example, H2BmUb is introduced at DNA double-strand breaks as a result of ATM-mediated phosphorylation and recruitment of the RNF20-RNF40 heterodimer, the ubiquitin E3 ligase that promotes this modification 81 . More intriguingly, with respect to the current work, H2BmUb is deubiquitinated at sites of DNA damage that have stalled RNAP 82 . In yeast, H2BmUb is deubiquitinated by Ubp8 and Ubp10, the former of which is the ubiquitin protease most closely related to USP3. It is thus tempting to speculate that a similar process occurs at sites of stalled RNAP in human cells and that the prolonged transcriptional suppression by USP3 in BER-defective cells is a pathological extension of this process (Fig. 8b). Finally, hereditary XRCC1 mutations result in cerebellar ataxia and deletion of XRCC1 in the mouse brain results in cerebellar ataxia, seizures and juvenile mortality 18,19,26 . The impact of excessive/aberrant PARP1 activity on these phenotypes is illustrated by PARP1 inhibition and/or deletion, which alleviate the neuropathology in Xrcc1-defective neurons and mice 19 . Xrcc1-defective neurons also exhibit a defect in pre-synaptic Ca 2+ signalling, which is rescued by PARP inhibition/deletion, which in light of our recent data might reflect an impact of aberrant PARP1 activity on the expression of genes critical for normal neuronal activity 19 . Consistent with this idea, we recently demonstrated that neuronal enhancers are hotspots of SSB repair 83 , most probably as a result of cytosine demethylation during the epigenetic (re)programming of neuronal gene expression; a process in which XRCC1-mediated BER is strongly implicated [84][85][86] . It is also possible that PARP1 activity might impact on transcription in other pathophysiological conditions that induce DNA base damage, such as chronic inflammation 87 .

online content
Any methods, additional references, Nature Research reporting summaries, source data, extended data, supplementary information, acknowledgements, peer review information; details of author contributions and competing interests; and statements of data and code availability are available at https://doi.org/10.1038/ s41556-021-00792-w.

Methods
Cell lines, treatments and culture conditions. All cell lines were cultured in a low-oxygen (3%) incubator (37 °C and 5% CO 2 ). RPE-1 cells (American Type Culture Collection) were cultured in DMEM-F12 GlutaMAX medium containing 10% FBS supplemented with penicillin-streptomycin. U2OS cells (American Type Culture Collection) were cultured in DMEM containing 10% FBS supplemented with penicillin-streptomycin and l-glutamine. Primary fibroblasts from patients with XRCC1 mutations (XD1) 18 Table 4). Transfected cells were sorted (BD FACSMelody) for GFP + cells two days after transfection and seeded into 96-wells plates for the amplification of single-cell clones, which were subsequently validated by immunoblotting. Gene editing was confirmed by Sanger sequencing.
Primary cerebellar neurons were established from seven-day-old (P7) male and female WT and Xrcc1 Nes-Cre (ref. 18 ) mouse pups using the Papain Dissociation System (Worthington). The mice from which these neurons were obtained were maintained in accordance with the UK Animal (Scientific Procedures) Act 1986 and satisfied local institutional regulations at the University of Sussex and under the auspices of UK Home Office project licence number P3CDBCBA8. The cells were seeded onto poly-l-lysine-coated coverslips at 6 × 10 5 cells per well and maintained in Neurobasal A medium (Gibco) supplemented with B27 (Invitrogen), 5.25 mM KCl, 0.5 mM glutamine and penicillin-streptomycin in a humidified incubator (3% O 2 and 5% CO 2 at 37 °C). The cultures were employed for transcription experiments between 4 and 11 d in vitro.
Western blotting. RPE-1 cells were harvested and lysed in TEB150 lysis buffer (50 mM HEPES pH 7.4, 150 mM NaCl, 2 mM MgCl 2 , 5 mM EGTA pH 8, 1 mM dithiothreitol (DTT), 0.5% Triton X-100 and 10% glycerol) containing protease inhibitor cocktail (Roche), phosphatase inhibitor cocktail (PhosSTOP, Sigma) and 1:1,000 benzonase (Sigma) for 45 min on ice. Lysis was stopped by the addition of SDS-PAGE sample buffer and boiling for 5 min at 90 °C. The other cell lines were harvested directly in sample buffer and boiled as described for the RPE-1 cells. The protein extracts were fractionated by SDS-PAGE, transferred to nitrocellulose, blocked in 0.1% TBS-Tween containing 5% non-fat dried milk and incubated with the indicated appropriate antibodies (Supplementary Table 1).
Immunofluorescence. For immunofluorescence, cells were fixed in 4% PFA in PBS for 10 min, permeabilized in 0.2% Triton X-100 solution in PBS for 10 min, rinsed in PBS and blocked in 5% BSA solution. For the purpose of H2BmUb staining, the cells were fixed in 100% methanol at room temperature for 10 min, rinsed in PBS and blocked in 5% BSA solution. The cells were incubated with the appropriate primary antibody (Supplementary Table 1) for 1 h, washed in PBS, incubated with the appropriate fluorophore-conjugated secondary antibody (Invitrogen) for 1 h and then counterstained with DAPI. For global chromatin ubiquitination (staining with antibody to FK2), the cells were pre-extracted on ice for 2 min with 0.2% Triton X-100 in PBS, rinsed quickly in cold PBS and fixed in 4% PFA in PBS. To measure the global sites of nascent RNA synthesis, the cells were pulse labelled in medium containing 1 mM EU (Abcam) for 20 min before fixation. The EU pulse-labelled cells were subjected to click chemistry using a Click-iT kit (Invitrogen) according to the manufacturer's protocol. Click chemistry was performed after the blocking step of the immunofluorescence protocol described above. All images for quantitation were acquired using the automated wide-field microscope scanR image acquisition and analysis software, with >500 cells scored per sample (Olympus). Representative images were acquired with the use a wide-field microscope (Apotome; Zeiss). Where displayed, scanR cell galleries were generated by the Olympus Analysis Software as a representative example of the cell populations that were quantified.

DNA strand-break measurements.
To measure DNA strand breaks under the conditions at which transcription is inhibited in XRCC1 −/− cells, the cells were treated with 100 μM H 2 O 2 for 20 min (the H 2 O 2 was refreshed after 10 min) in serum-free medium at 37 °C and then in drug-free medium for 2 h (Extended Data Fig. 4g). To measure the kinetics of SSB repair, the cells were treated with 50 μM H 2 O 2 for 10 min on ice and then in drug-free medium at 37 °C for the indicated time periods (Extended Data Fig. 4f). The DNA strand breaks were measured using alkaline comet assays, as described previously 91 , using the Comet Assay IV software (Perceptive Instruments).
In vitro analysis of PARP1 auto-ribosylation. Recombinant high-specific-activity PARP1 (100 nM; Trevigen) was either mock ribosylated in the absence of NAD + or auto-ribosylated in the presence of 2.5 μM NAD + (NEB) and the indicated recombinant proteins. The reactions were carried out in 50 mM Tris-HCl pH 8, 0.8 mM MgCl 2 , 1% glycerol, 1.5 mM DTT and 100 nM of single-stranded oligodeoxyribonucleotide (5′-CATATGCCGGAGATCCGCCTCC-3′; Eurogentec), in a final volume of 20 μl at room temperature for 30 min. The reaction products were heated to 90 °C for 3 min in sample buffer and subjected to SDS-PAGE. The proteins were transferred to nitrocellulose and immunoblotted with ADP-ribose binding reagent (Millipore; Supplementary Table 1).

USP3 expression construct.
To create the USP3-GFP and USP3 ZnF(1-110) -GFP expression constructs, USP3 complimentary DNA (Source Bioscience) was amplified with the use of the indicated primers (Supplementary Table 4) and cloned in-frame into the BamHI and XhoI restriction sites of the pEGFP-N1 vector (Addgene).
USP3 chromatin retention assay. U2OS cells were transiently transfected with USP3-GFP or USP3 ZnF(1-110) -GFP and H2B-mCherry constructs 24 h before the experiment. The cells were then treated with 0.01 mg ml −1 MMS for 4 h in the presence or absence of 10 μM PARPi. The cells were washed once with PBS before incubation in CSK buffer (10 mM PIPES, pH 7.0, 100 mM NaCl, 300 mM sucrose, 3 mM MgCl 2 and 0.7% Triton X-100) for 3 min at room temperature. The cells were subsequently washed in PBS and fixed in 4% PFA for 10 min. Next, the cells were counterstained with DAPI, mounted in Moviol and subjected to immunofluorescence analysis. Between 50 and 100 H2B-mCherry-expressing cells were counted per condition.
In vitro ADP-ribose binding assay. First, 96-well plates were washed with H 2 O and then incubated overnight with 0.1 mg ml −1 of calf thymus histones (Sigma) at 1 nature portfolio | reporting summary

March 2021
Corresponding author(s): keith caldecott Last updated by author(s): May 21, 2021 Reporting Summary Nature Portfolio wishes to improve the reproducibility of the work that we publish. This form provides structure for consistency and transparency in reporting. For further information on Nature Portfolio policies, see our Editorial Policies and the Editorial Policy Checklist.

Statistics
For all statistical analyses, confirm that the following items are present in the figure legend, table legend, main text, or Methods section.

n/a Confirmed
The exact sample size (n) for each experimental group/condition, given as a discrete number and unit of measurement A statement on whether measurements were taken from distinct samples or whether the same sample was measured repeatedly The statistical test(s) used AND whether they are one-or two-sided Only common tests should be described solely by name; describe more complex techniques in the Methods section.
A description of all covariates tested A description of any assumptions or corrections, such as tests of normality and adjustment for multiple comparisons A full description of the statistical parameters including central tendency (e.g. means) or other basic estimates (e.g. regression coefficient) AND variation (e.g. standard deviation) or associated estimates of uncertainty (e.g. confidence intervals) For null hypothesis testing, the test statistic (e.g. F, t, r) with confidence intervals, effect sizes, degrees of freedom and P value noted

Software and code
Policy information about availability of computer code Data collection scanR software, Comet Assay IV software,Graphpad Prism 9 Data analysis scanR software, Comet Assay IV software,Graphpad Prism 9 For manuscripts utilizing custom algorithms or software that are central to the research but not yet described in published literature, software must be made available to editors and reviewers. We strongly encourage code deposition in a community repository (e.g. GitHub). See the Nature Portfolio guidelines for submitting code & software for further information.

Data
Policy information about availability of data All manuscripts must include a data availability statement. This statement should provide the following information, where applicable: -Accession codes, unique identifiers, or web links for publicly available datasets -A description of any restrictions on data availability -For clinical datasets or third party data, please ensure that the statement adheres to our policy Data availability: Source data are provided with this study. Because of their large number,all raw microscope quantification data sets involved in this work such as Excel and scanR/comet fluorescence quantification tables (which generated the provided source data) are available from the corresponding author on reasonable request, citing the experiments of interest.  Note that full information on the approval of the study protocol must also be provided in the manuscript.