Replication-associated base excision repair/single-strand break repair regulates PARG inhibitor response via the PRMT1/PRMT5/ATR axis

Abstract

Poly(ADP-ribose) polymerases 1 and 2 (PARP1/PARP2), and poly(ADP-ribose) glycohydrolase (PARG), modulate the level of poly(ADP-ribose) (PAR), a post-translational protein modification, in response to DNA damage or replication stress. Here, we find that replication-dependent and PARP1/PARP2-mediated PARylation recruits the base excision repair (BER)/single-strand break repair (SSBR) scaffold protein XRCC1 and the associated factors DNA polymerase β (POLB), aprataxin (APTX), and DNA ligase isoform 3 (LIG3). Further, these BER/SSBR proteins promote resistance to inhibitors of PARP1/PARP2 and PARG, as loss of these proteins sensitizes glioblastoma and ovarian cancer cells to each. In addition, depletion of these replication-associated BER/SSBR factors leads to enhanced PAR levels and PARG inhibitor-induced activation of the ATR/CHK1 S-phase checkpoint kinases. Both PARG inhibition and ATR inhibition lead to elevated ATM- and DNA-PK-dependent KAP1 phosphorylation. In turn, inhibition of either ATR or CHK1 enhances the cellular response to PARG inhibitors. Finally, inhibition of the ATR regulators PRMT1 or PRMT5 synergizes with PARG inhibition, implicating replication-associated BER/SSBR and PARylation in the activation of the PRMT1/PRMT5/ATR axis. This study highlights the role of BER/SSBR in protecting the cell during S-phase to suppress PARylation-induced checkpoint activation, which may suggest a potential intervention strategy for PARG inhibitor-resistant tumors.

Graphical Abstract

Figure 1.

Replication-dependent PARP1/PARP2 activation. (A) PAR immunoblot of whole cell lysates prepared from LN428 cells exposed to vehicle (DMSO), NRH (100 μM), PARGi (PDD00017273, 10 µM), or NRH + PARGi for 8 h, with β-ACTIN shown as a loading control. (B) PAR immunoblot of whole cell lysates prepared from ES-2 cells exposed to vehicle (DMSO), NRH (100 μM), PARGi (PDD00017273, 10 µM), or NRH + PARGi for 8 h, with β-ACTIN shown as a loading control. (C) Immunoblot of PARP1, PARP2, and H3 (loading control) documenting PARP1-KO, PARP2-KO, and PARP1 + PARP2 double-KO via CRISPR/Cas9 in LN428 cells. (D) PAR immunoblot of whole cell lysates prepared from LN428, LN428/PARP1-KO, LN428/PARP2-KO, and LN428/PARP1-KO/PARP2-KO cells exposed to PARGi (PDD00017273) or NRH + PARGi for 8 h, with β-ACTIN shown as a loading control. (E) PAR immunoblot of whole-cell lysates prepared from LN428 cells (non-synchronized; Lanes 1–3), following serum starvation (synchronized/non-replicating cells; G1 phase, Lane 4), and after release following normal media replenishment (replicating; S-phase, Lanes 5–8). (F) PAR immunoblot of whole cell lysates prepared from ES-2 cells following double thymidine block (synchronized/non-replicating cells; G1 phase, Lane 1) and after release following normal media replenishment (replicating; S-phase, Lanes 2–5), with β-ACTIN shown as a loading control.

Figure 2.

Identification of BER/SSBR factors during replication. (A) Graphic depicting the PARP1–XRCC1 Split-TurboID system. The biotinylation protein TurboID was split into an N-terminal domain (Turbo-N) and a C-terminal domain (Turbo-C). The Turbo-N domain was fused in-frame to the C-terminus of XRCC1, while the Turbo-C domain was fused in-frame to the C-terminus of PARP1. Upon the induction of DNA damage or replication stress, the co-localization of XRCC1 and PARP1 causes the reassembly and activation of TurboID and the initiation of proximity biotinylation [96, 109]. Created in BioRender. Sobol, R. (2025) https://BioRender.com/p82xg0b. (B) Graphic depicting the experimental outline of proximity labeling (PARP1–XRCC1/Split-TurboID) to capture activated PARP1–XRCC1 complexed proteins during replication. Created in BioRender. Sobol, R. (2025) https://BioRender.com/zoadg6m. (C) Streptavidin capture of biotinylated proteins in LN428/PARP1–XRCC1/Split-TurboID cells. Cells were arrested by serum starvation and then released from the block by media change. Arrested (−) and actively replicating (+) cells were exposed to biotin (100 µM, 60 min) for 1 h. Biotinylated proteins were captured with streptavidin-coated beads and probed by immunoblot. (D) Streptavidin capture of biotinylated proteins in ES-2/PARP1–XRCC1/Split-TurboID cells. Cells were arrested in G1 for 48 h by CDK4/6 inhibition and then released from the block by media change. All cells were exposed to 10 µM PARGi and 100 µM NRH for 6 h, and biotinylation was then performed with 100 µM biotin for 1 h. Biotinylated proteins were captured with streptavidin-coated beads and probed by immunoblot.

Figure 3.

BER/SSBR factors XRCC1, POLB, APTX, and LIG3 regulate PARylation in response to replication stress. Immunoblot of (A) LN428 and LN428/XRCC1-KO, (B) LN428 and LN428/POLB-KO, (C) LN428 and LN428/APTX-KO, (D) LN428 and LN428/LIG3-KO, (E) ES-2 and ES-2/XRCC1-KO, (F) ES-2 and ES-2/POLB-KO, (G) ES-2 and ES-2/APTX-KO, and (H) ES-2 and ES-2/LIG3-KO whole cell lysates. Blots on the left show the respective knockout of XRCC1, POLB, APTX, or LIG3. The blots on the right show the comparative analysis of replication-dependent PAR formation following 8 h exposure of cells to vehicle (DMSO), NRH (100 μM), PARGi (PDD00017273, 10 µM), or NRH + PARGi. β-ACTIN or H3 was used as a loading control.

Figure 4.

BER/SSBR depletion overcomes PARPi resistance in glioblastoma and ovarian cancer cells. (A) Immunoblot for XRCC1 in LN428 and LN428/XRCC1-KO cells, with H3 used as a loading control. (B) Viable LN428 and LN428/XRCC1-KO cells (%) exposed to either ABT-888 or BMN-673. (C) Immunoblot for XRCC1 in ES-2 and ES-2/XRCC1-KO cells, with H3 used as a loading control. (D) Viable ES-2 and ES-2/XRCC1-KO cells (%) exposed to either ABT-888 or BMN-673. (E) Immunoblot for XRCC1 in PARP inhibitor-resistant C4-2 and C4-2/XRCC1-KO cells, with H3 used as a loading control. (F) Viable cells (%) in response to BMN-673 exposure: PEO1, C4-2, and C4-2/XRCC1-KO cells. (G) Immunoblot for POLB in C4-2 and C4-2/POLB-KO cells, with H3 used as a loading control. (H) Viable cells (%) in response to BMN-673 exposure: PEO1, C4-2, and C4-2/POLB-KO cells. (I) Immunoblot for APTX in C4-2 and C4-2/APTX-KO cells, with H3 used as a loading control. (J) Viable cells (%) in response to BMN-673 exposure: PEO1, C4-2, and C4-2/APTX-KO cells. (K) Immunoblot for LIG3 in C4-2 and C4-2/LIG-KO cells, with H3 used as a loading control. (L) Viable cells (%) in response to BMN-673 exposure: PEO1, C4-2, and C4-2/LIG3-KO cells. For panels (H), (J), and (L), the dotted lines were taken from panel (F) to simplify comparison between the knockouts and C4-2 and PEO1 cells. For panels (B), (D), (F), (H), (J), and (L), viability was assayed after 120 h of exposure to the indicated compounds.

Figure 5.

BER/SSBR depletion overcomes PARGi resistance in glioblastoma and ovarian cancer cells. Viable cells (%) in response to PARGi (PDD00017273) in (A) LN428 and LN428/XRCC1-KO cells, (B) LN428 and LN428/POLB-KO cells, (C) LN428 and LN428/APTX-KO cells, or (D) LN428 and LN428/LIG3-KO cells. Cells (%) showing caspase-3/7⁺ activity in response to PARGi (PDD00017273) in (E) LN428 and LN428/XRCC1-KO cells, (F) LN428 and LN428/POLB-KO cells, (G) LN428 and LN428/APTX-KO cells, or (H) LN428 and LN428/LIG3-KO cells. Viable cells (%) in response to PARGi (PDD00017273) in (I) ES-2 and ES-2/XRCC1-KO cells, (J) ES-2 and ES-2/POLB-KO cells, (K) ES-2 and ES-2/APTX-KO cells, or (L) ES-2 and ES-2/LIG3-KO cells. Cells (%) showing caspase-3/7⁺ activity in response to PARGi (PDD00017273) in (M) ES-2 and ES-2/XRCC1-KO cells, (N) ES-2 and ES-2/POLB-KO cells, (O) ES-2 and ES-2/APTX-KO cells, or (P) ES-2 and ES-2/LIG3-KO cells. For the viability assays, viability was assayed after 120 h of exposure to the indicated compound, and for the caspase activation assays, the activation of caspase-3/7⁺ was assayed 48 h after drug exposure. For panels (B), (C), and (D), the dotted line was taken from panel (A) to simplify the comparison between LN428 and the corresponding knockout cells. For panels (J), (K), and (L), the dotted line was taken from panel (I) to simplify the comparison between ES-2 and the corresponding knockout cells. Where indicated, P<* .05, P<** .01, P<*** .001, P<**** .0001; two-way ANOVA.

Figure 6.

Targeting the S-phase checkpoint overcomes PARGi resistance in glioblastoma and ovarian cancer cells. (A) Immunoblot of ES-2/XRCC1-KO whole cell extracts for pCHK1(S345), pCHK1(S317), and CHK1, with H3 used as a loading control. The induction factor (I.F.) for CHK1 phosphorylation was determined by densitometry analysis and is listed under each lane. Cells were exposed to vehicle control (DMSO), ATRi (AZD6738, 10 µM), ATMi (KU-55933, 10 µM), DNA-PKi (KU-57788, 10 µM), and/or PARGi (PDD00017273, 10 µM) as indicated, for 24 h. (B) Immunoblot of ES-2 and ES-2/XRCC1-KO whole cell extracts for pRPA2 (S4/S8), RPA2, pCHK1 (S345), CHK1, pKAP1 (S824), and KAP1, with H3 used as a loading control. Cells were exposed to PARGi (PDD00017273, 5 µM) for the indicated time points, with DMSO as the vehicle control. (C) Quantification of PAR (green), pRPA(S4/S8) (red), and PAR/pRPA(S4/S8) co-localized (black) foci. Where indicated, P<* .05, P<** .01, P<***.001, P< ****.0001; Student’s t-test (representative images in Supplementary Fig. S4C). (D) Immunoblot of ES-2/XRCC1-KO whole cell extracts for pKAP1 (S824) and KAP1, with H3 used as a loading control. Cells were exposed to vehicle control (DMSO), ATRi (AZD6738, 10 µM), ATMi (KU-55933, 10 µM), DNA-PKi (KU-57788, 10 µM), and/or PARGi (PDD00017273, 10 µM) as indicated for 24 h. (E) Immunoblot of ES-2/XRCC1-KO whole cell extracts for pKAP1 (S824) and KAP1, with H3 used as a loading control. Top: Cells were exposed to vehicle control (DMSO), ATMi (KU-55933, 10 µM), DNA-PKi (KU-57788, 10 µM), or ATMi + DNA-PKi, with or without PARGi (PDD00017273, 10 µM) as indicated, for 24 h. Middle: Cells were exposed to vehicle control (DMSO), ATMi (KU-55933, 10 µM), DNA-PKi (KU-57788, 10 µM), or ATMi + DNA-PKi, with or without ATRi (AZD6738, 10 µM) as indicated, for 24 h. Botttom: Cells were exposed to vehicle control (DMSO), ATMi (KU-55933, 10 µM), DNA-PKi (KU-57788, 10 µM), or ATMi + DNA-PKi, with or without ATRi (AZD6738, 10 µM) + PARGi (PDD00017273, 10 µM) as indicated, for 24 h.

Figure 7.

PARGi response enhanced by co-inhibition with CHK1, ATR, and PRMT1i/PRMT5i. (A) Viable LN428 cells (%) in response to ATRi (AZD6738), PARGi (PDD00017273), and ATRi + PARGi (10 µM) treatment. (B) Viable LN428 cells (%) in response to CHK1i (MK8776), PARGi (PDD00017273), and CHK1i + PARGi (10 µM) treatment. The dotted line was taken from panel (A) for comparison. (C) Viable C4-2 cells (%) in response to ATRi (AZD6738), PARGi (PDD00017273), and ATRi + PARGi (10 µM) treatment. (D) Viable C4-2 cells (%) in response to CHK1i (MK8776), PARGi (PDD00017273), and CHK1i + PARGi (10 µM) treatment. The dotted line was taken from panel (C) for comparison. For panels (A–D), viable cells were assayed after 120 h of exposure to the indicated compounds. (E) Viable LN428 cells (%) in response to PARGi (PDD00017273) and PRMT1i (GSK336871) exposure at the indicated concentrations. (F) Viable LN428 cells (%) in response to PARGi (PDD00017273) and PRMT5i (PRT543) exposure at the indicated concentrations. The dotted line was taken from panel (E) for comparison. (G) Viable ES-2 cells (%) in response to PARGi (PDD00017273) and PRMT1i (GSK336871) exposure at the indicated concentrations. (H) Viable ES-2 cells (%) in response to PARGi (PDD00017273) and PRMT5i (PRT543) exposure at the indicated concentrations. The dotted line was taken from panel (G) for comparison. (I) Bliss synergy score for LN428 cells exposed to PARGi and PRMT1i, or PARGi and PRMT5i. (J) Bliss synergy score for ES-2 cells exposed to PARGi and PRMT1i, or PARGi and PRMT5i. (K) Model for the role of BER/SSBR in suppressing replication stress. Canonical and replication-associated BER/SSBR proteins process replication-stalling DNA lesions. Upon inhibition of BER by PARGi, the inter-S-phase checkpoint is activated via ATR/CHK1. Simultaneous inhibition of PARG and ATR/CHK1 causes cell death, and PRMT1/PRMT5 inhibition-induced cell death is synergistic with PARG inhibition [created in BioRender. Sobol, R. (2025) https://BioRender.com/xke5uat].