Non-PARylated PARP1 stimulates the strand annealing activity of RECQL4
Our previous study showed that among the RecQ helicases, RECQL4 has the most robust ssDNA annealing activity 18. As the biochemical activities of RecQ helicases are regulated by interacting proteins 18, we explored whether PARP1 affects the ssDNA annealing activities of RECQL4 and BLM. These two helicases are recruited early in response to DNA damage and participate in multiple DNA repair pathways and comparisons between these two helicases are of particular interest as reported in past literature including our own (13). These helicases were assayed for ssDNA annealing activity in the presence of increasing amounts of non-PARylated PARP1 (PARP1) (1 to 80 nM), PAR chains, or PARylated PARP1. The results show that non-PARylated PARP1, but not PAR/PARylated PARP1, stimulates RECQL4’s ssDNA annealing activity approximately 2-fold in a dose-dependent manner (Figure 1A-C, and Supplemental Figure 1A). In contrast, neither non-PARylated PARP1, nor PAR, significantly influenced the efficiency of the ssDNA annealing activity of BLM (Figure 1D-F), suggesting that it is specific to RECQL4. These data are summarized in Table 1.
RECQL4-catalyzed helicase activity is inhibited by PARP1
Previously, we showed that PARP1 and PAR inhibited the helicase activity of RECQL5 and WRN 35. Here, we studied the effect of PARP1 and PAR on RECQL4 and BLM helicase activities using a forked-heteroduplex DNA substrate (Supplemental Figure S9). It has been reported that RECQL4 exhibits weak helicase activity related to its strong intrinsic strand annealing activity in vitro 13,15,36. Our results show that RECQL4’s helicase activity was inhibited in a dose-dependent manner by PARP1 but was unaffected by the presence of PAR (Figure 1G-I), while BLM’s helicase activity is unaffected by the presence of both PAR or PARP1 (Figure 1J-L). These data are summarized in Table 1. PARP1 possesses unique DNA binding motifs 40. Surprisingly, the introduction of PARP1 did not impede BLM's helicase function; instead, it selectively hindered only RECQL4's helicase activity. This observation implies that PARP1's inhibition of RECQL4's helicase activity may not solely be attributable to PARP1's DNA binding capabilities.
PARP1-mediated PARylation is required for the early recruitment of RECQL4 to DSBs
PARP1 directly recognizes DNA DSBs or single-strand breaks and promotes the rapid recruitment of DDR proteins to DNA damage in a PAR-dependent manner 32,41. Previously, we demonstrated that RECQL4 is recruited to DSBs sites as early as two seconds after DNA damage and independent of DDR proteins ATM or DNA-PKcs 14. We used U2OS WT or PARP1 KO cells (Supplemental Figure S2) to investigate the role of PARP1 in RECQL4 recruitment. Equal amounts of GFP-tagged RECQL4/BLM were expressed in U2OS WT/PARP1 KO cells (Supplemental Figure S5) and RECQL4 recruitment to DSBs was examined using 12% laser micro-irradiation, which is known to induce DSB 14,42. As depicted in Figure 2A, GFP-RECQL4 is not recruited to double-strand breaks (DSBs) in PARP1 knockout cells,but is observed in cells expressing GFP-RECQL4 and PARP1. Olaparib, a PARP1 and PARP2 inhibitor (PARPi), also severely inhibited recruitment of GFP-RECQL4 to DSBs (Figure 2B). Similar results were obtained in PARP1 KO cells and in PARPi inhibitor-treated HeLa cells (Supplemental Figure 3A and 3B), thus PARP1-dependent early recruitment of RECQL4 to DSB sites is not cell type specific. PARP1 is responsible for 90% of the synthesized PAR chains in cells in response to DNA damage, and these chains are rapidly hydrolyzed by the exo- and endoglycosidase activity of poly(ADP-ribose) glycohydrolase 1 (PARG) 43. Thus, we tested how PARG might alter the recruitment of RECQL4. To explore this, PARG and CFP-RECQL4 were expressed in U2OS WT cells and then the cells were subjected to micro-irradiation and the CFP-RECQL4 localization was monitored. While expression of PARG completely suppressed CFP-RECQL4 stable recruitment, the treatment with ATRi or DNA PKci (supplemental Figure S3E) did not. Thus, RECQL4 recruitment to DSB requires PARP-mediated synthesis of PAR chains. Since laser-induced DNA damage can lead to various types of DNA damage, we also used the topoisomerase II poison, etoposide, to induce more specific DNA double-strand breaks. We performed proximity ligation assays (PLA) using RECQL4 and gH2AX antibodies to examine RECQL4 recruitment to DNA damage sites after treating cells with different concentrations of etoposide. As illustrated in Figure 2E, U2OS WT cells exhibited positive PLA foci in a dose-dependent manner. In contrast, PARP1 knockout and PARG-expressing WT cells (supplemental Figure S4 A) displayed a lack of PLA foci (Figure 2E). Similar outcomes were observed in chromatin fractionation assays. Etoposide treatment of U2OS WT cells increased RECQL4 binding to chromatin in a dose-dependent manner, whereas this effect was not observed in PARP1 KO or PARG-expressing cells (Supplemental Figure S4 B).
RECQL4 interacts with BLM and stimulates BLM helicase activity on DNA fork substrates in vitro 44. This functional interaction led us to investigate whether BLM’s recruitment was also affected by PARP1. However, PARP1 KO or pretreatment of cells with PARPi only delayed the recruitment of GFP-BLM to DSBs (Figure 2C and 2D). Similar results were obtained in PARP1 KO cells and in PARPi inhibitor-treated HeLa cells (Supplemental Figure 3C and 3D).
Recruitment of RECQL5 and WRN are also delayed in PARP1 KO cells and in wild type cells treated with PARPi 35. Thus, the impact of PARP1 depletion on recruitment to DSBs differs for RECQL4 compared to WRN, BLM and RECQL5. For RECQL4, recruitment is completely diminished in the absence of PARP1, while for WRN, BLM, and RECQL5, the kinetics of recruitment is slower in the absence of PARP1. These results are summarized in Table 2.
RECQL4 interacts with PARP1
PARP1 recognizes DNA breaks, and PARP1 activity is involved in recruitment of multiple DNA repair proteins 25. Our in vitro assays show that non-PARylated PARP1 enhances RECQL4 annealing function. To further investigate the interaction between RECQL4 and PARP1, Flag-tagged RECQL4 was expressed in U2OS WT/PARP1 KO cells, before and after gamma irradiation, and in the presence or absence of Olaparib. PARP1-RECQL4 interaction and RECQL4 PARylation was analyzed by immunoprecipitating Flag-tagged RECQL4 and probing for PARP1 and anti-PAR antibodies. IR treatment induced gH2AX but did not change protein levels of RECQL4 or PARP1 (Figure 3A). The immunoprecipitants were treated with benzonase to remove DNA and were then analyzed using antibodies to Flag-tag and endogenous PARP1. The results show a basal association between RECQL4 and PARP1 without DNA damage exposure. Interestingly, DNA damage significantly enhanced RECQL4 association with PARP1 and treatment with Olaparib reduced the interaction (Figure 3B). Because the co-IP was performed in the presence of nuclease or benzonase, the interaction is not bridged by DNA. These findings are in agreement with a previous report showing that RECQL4 and PARP1 interact in human cells 45.
To further investigate whether PARP1 targets RECQL4 for PARylation, the benzonase-treated immunoprecipitants were probed using anti-PAR antibody. As shown in Figure 3B, DNA damage increased PAR streaks on RECQL4, whereas PARP1KO cells/olaparib-treated cells did not stimulate RECQL4 PARylation. This indicates a crucial role of PARP1 in RECQL4 PARylation in response to DNA damage.
In the prior experiment (Figure 3B) we used PAR antibody to detect RECQL4 PARylation. To assay PAR binding ability of RECQL4, slot blots 37 were used to directly demonstrate that RECQL4 binds to PAR in vitro (Figure 3C). In addition, in vitro pull-down assays using purified recombinant His-RECQL4 or Ni-NTA alone further corroborated the RECQL4-PARP1 interaction (Figure 3D). To analyze the domain in RECQL4 that undergoes PARylation, protein extracts of HEK293T cells expressing flag-tagged RECQL4 (full length or three truncated flag-tagged variants) were employed (Figure 3E, left panel). Full-length flag-tagged RECQL4, N-terminal and the C-terminal region of RECQL4 variant were pulled down by anti-PAR antibodies (Figures 3E, right panel). The PARylated domains on RECQL4 are summarized in Figure 3F. This suggests that there are PARylation sites within the both N- and C- terminal regions of RECQL4.
Identification of interaction domains between PARP1 and RECQL4
PARP1 is a multidomain protein containing an N-terminal DNA binding domain (DBD), an auto-modification domain (AD) with a breast cancer 1 protein (BRCA1) C-terminus motif, and a C-terminal catalytic domain (catalytic) 21. To identify the protein regions that mediate interactions between PARP1 and RECQL4, we obtained V5-tagged full-length PARP1 and several deletion mutants (Figure 4A) and overexpressed them in HEK293T cells (Figure 4B). These constructs were co-transfected with or without Flag-tagged RECQL4, followed by co-IP with anti-Flag beads and western blot with anti-V5 antibody. Only domains C (amino acids 1-461), and G (amino acids 662-1014) of PARP1 (Figure 4B) co-IPed with RECQL4 with the same efficiency as full-length PARP1 (Figure 4B). These results suggest the presence of multiple points of contact between PARP1 and RECQL4.
Similarly, RECQL4 is composed of several domains including a Sld-2 like (SLD), a basic region, helicase domain (HD), and RECQ conserved (RQC) domain (Figure 4C). Consistent with a previous report, which reported that RECQL4’s C-terminus interacts with PARP1 45, we found that both N- and C-terminal regions of RECQL4 bind to PARP1 (Figure 4C and D).We also confirmed these results using proximity ligation assay (PLA) between Flag-tagged RECQL4 WT and its domains with PARP1 in U2OS WT cells (Supplementary Figure S6). Notably, a majority of RECQL4 interacting protein partners interact with the SLD domain of RECQL4 46-48. The combined regulation of these protein interactions remains to be determined, and some of them are regulated by phosphorylation and perhaps other post-translational modifications of RECQL4 12,46.
Role of RECQL4 in alt-NHEJ
Building upon our earlier findings, it is evident that PARP1 selectively enhances the annealing activity of RECQL4. Our next objective was to investigate the specific DNA repair pathway in which RECQL4 employs this annealing function. The alternative non-homologous end joining (Alt-NHEJ) pathway necessitates the microhomology annealing of DNA end products, presenting a potential context for RECQL4's annealing activity. Notably, replication protein A (RPA) has been shown previously to be a negative regulator of alt-NHEJ 49 and other RecQ helicases are known to physically and functionally interact with RPA. Recently, RECQL4 was shown to participate in c-NHEJ 50. However, what role of RECQL4 plays in alt-NHEJ is unknown. Thus, we sought to investigate the molecular mechanisms by which RECQL4 may promote alt-NHEJ. We hypothesized that RECQL4 may help in displacing RPA from ssDNA to facilitate annealing of ssDNA microhomology overhangs which is a crucial step in alt-NHEJ pathway. To test this hypothesis, we carried out an RPA displacement assay (Figure 5A) using ssDNA, which was pre-incubated with RPA, then incubated with increasing amounts of RECQL4 and radiolabeled complementary ssDNA. Adding RPA prevented spontaneous annealing of ssDNA while, interestingly, addition of RECQL4 stimulated modest ssDNA annealing to form dsDNA indicating weak displacing activity of RPA (Figure 5B and C). To explore if RECQL4's limited annealing function stemmed from its helicase activity, we utilized a helicase-inactive RECQL4 variant, K508A, in our experiments. The results, illustrated in Supplemental Figure S7, revealed that both the wild-type RECQL4 and the helicase-inactive variant exhibited comparable activities. This aligns with existing studies indicating that RECQL4's annealing capability is distinct from its helicase function13. In contrast, BLM did not show any increased ssDNA annealing in the presence of RPA (Figure 5D and 5E). The percentage of dsDNA was calculated by measuring dsDNA duplex formed before and after adding RECQL4 or BLM. The protein ratio was assessed from initial standardization assays.
Alt-NHEJ employs short stretches of microhomology to facilitate double-strand break repair 51. To assess RECQL4's involvement, we measured its microhomology annealing efficiency using model substrates with 10-base microhomology, as depicted in Figure 5F. RECQL4 effectively annealed these substrates, forming double-sized products. Interestingly, only non-PARylated PARP1 stimulated RECQL4’s microhomology annealing whereas PARylated PARP1 did not show any difference from RECQL4 alone (Figure 5G).
To understand how PARP1 or PARP1 activity affect RECQL4’s interaction with alt-NHEJ proteins, we carried out GFP-RECQL4 immunoprecipitation after PARP1 knockdown or PARPi treatment before and after IR irradiation. As shown in Figure 5I, after DNA damage RECQL4 associated with alt-NHEJ proteins XRCC1, PARP1 and LIG3 in WT cells, however PARP1-depleted or PARPi-treated cells abrogated RECQL4 interaction with the alt-NHEJ proteins. In conclusion, RECQL4 removes RPA from ssDNA and promotes annealing of microhomology-mediated end joining. PARP1 and PARP1 activity were essential for RECQL4’s interaction with alt-NHEJ repair proteins.
RECQL4 is required for efficient double-strand break repair
Next, our aim was to investigate the impact of PARP1 on RECQL4's chromatin binding affinity following DNA damage. Wild-type (WT) or PARP knockout (KO) cells were treated with olaparib and/or exposed to 10Gy of irradiation. Chromatin fractions were isolated as described in the methods section. As depicted in Figure 6A, in cells without irradiation (IR) treatment, a basal level of RECQL4 was bound to chromatin (Lane 1). While RECQL4 participates in various DNA repair pathways, it also plays an important role in DNA replication initiation and interacts with replication origins. Notably, the SLD2-like domain of human RECQL4 forms a chromatin-bound protein complex during the cell cycle, involving core replication factors such as the MCM complex, CDC45, and GINS52-54. Therefore, the presence of chromatin-bound RECQL4 in the absence of DNA damage is expected. Further, RECQL4's association with chromatin appeared to be largely independent of PARP1 or its activity, as demonstrated by minimal effects on RECQL4 retention upon PARP1 depletion or olaparib treatment (Lane 2 and Lane 3). However, after IR, the purified chromatin fraction showed increased RECQL4 (Lane 4). In contrast, lanes containing PARP1 knockdown or PARP inhibitor (PARPi) treatment exhibited significantly less chromatin-bound RECQL4 (Lane 5 and Lane 6) (Figure 6A). These results suggest that RECQL4 utilizes at least two distinct modes for chromatin binding: one that is independent of PARP1 and its activity (basal binding) and another that is DNA damage- and PARP1-dependent.
PARP1 recognizes DNA lesions and is involved in the recruitment of multiple DDR proteins, which facilitates efficient DNA repair 25,55. To better dissect how PARP1 affects RECQL4 distribution at DSB sites, we took advantage of the U2OS-AsiSi DiVa system, where DSB can be induced at specific AsiSi target sequence across the human genome 56,57 . After knockdown of PARP1 or PARPi treatment, U2OS AsiSi-ER cells were treated with 4-OHT for 4 hours, then RECQL4 chromatin immune-precipitation (ChIP) using RECQL4 antibody was done. Induction of the DSBs with 4-OHT consistently increased RECQL4 occupation at DSB site (DSB1) in WT cells. Neither PARP1 knockdown or PARPi treated cells exhibited increased RECQL4 occupancy at DSB1(Figure 6B). Collectively these data demonstrate that PARP1 regulates RECQL4 binding on chromatin after DNA damage and occupancy at double-strand break sites.
Human cells possess multiple different DSB repair (DSBR) pathways, including HR, c-NHEJ, alt-NHEJ, and single-strand annealing 17,38. PARP1 is involved in HR 30,58 and alt-NHEJ pathways 59, and RECQL4 is required for c-NHEJ 11 and HR 12 mediated DSBR. In contrast, BLM plays a role primarily in HR-mediated DSBR 60. DSBR pathway choice is different in different phases of the cell cycle, and we previously reported that RECQL4 was phosphorylated, and this facilitated modulated the DSB pathway choice 46. PARP1 is known to function in alt-NHEJ 61 and above we have shown that RECQL4 interacts with PARP1, and that PARP1 protein and activity are required for RECQL4’s recruitment to sites of damage. Therefore, to investigate RECQL4 in DSBR via c-NHEJ and alt-NHEJ and whether loss of PARP1 protein or activity modulates this function, we used the cell-based GFP repair assays 62. For this purpose, site-specific I-SceI endonuclease was overexpressed in cells carrying a I-SceI target cleavage site in the marker gene GFP (Figure 6C-D) 38,39. After I-SceI introduces a DSB in GFP, the extent of DSBR was monitored as change in fluorescence output by flow cytometry (FACS). Two cell lines were used, EJ2-GFP and EJ5-GFP with different combinations of RECQL4, BLM and PARP1 depletion with or without olaparib (Supplemental Figure S8 A), where EJ2 reports DSBR by alt-NHEJ and EJ5 monitors DSBR by c-NHEJ. Consistent with previous results 39, olaparib inhibited alt-NHEJ mediated DSBR but not c-NHEJ mediated DSBR (Figure 6C and 6D). The number of GFP-positive cells decreased significantly in both cell lines depleted for RECQL4 (Figure 6C and 6D). RECQL4’s role has been implicated in c-NHEJ mediated DSBR through its direct interaction with the MRN complex and through initiation of DNA end resection with CtIP 46. However, the function of RECQL4 in alt-NHEJ mediated DSBR is unknown. Interestingly, in both alt-NHEJ and c-NHEJ reporter cell lines, cells treated with RECQL4 siRNA, PARP1 siRNA, and olaparib showed significantly lower DNA repair capacity than cells treated with control siRNA. This result suggests that both PARP1 and RECQL4 play important role in alt-NHEJ and c-NHEJ pathways (Figure 6C and 6D). In contrast, siRNA targeting BLM did not substantially impair c-NHEJ mediated DSBR, but it decreased alt-NHEJ DSBR (Figure 6C and 6D). Knockdown of BLM plus or minus olaparib did not significantly change the results. This is consistent with our in vitro and DNA recruitment results showing depletion of PARP1 or PARPi treatment did not alter BLM function or recruitment. Therefore, loss of PARP1 does not appreciably alter BLM’s function in these repair assays.
Polymerase Theta (PolQ) is a critical alt-NHEJ factor in mammalian cells. PolQ inhibition suppresses alt-NHEJ at dysfunctional telomeres and hinders chromosomal translocations at non-telomeric regions 51. Genetic inactivation or inhibition of (PolQ) leads to synthetic lethality in HR-deficient BRCA2 knockout (KO) cells 17,63. To understand epistatic role of RECQL4 and PolQi, we conducted clonogenic assay in DLD1 wild-type (WT) cells transfected with either siControl or siBRCA2 (Supplemental Figure S8B). Depleting RECQL4 or treating cells with PolQi in BRCA2-reduced cells lowered cell survival to 50%, and combining these approaches further reduced survival to 20%. Additionally, applying Olaparib resulted in complete cell death, as shown in Figure 6E. This suggests a cooperative function between RECQL4 and PolQ within the same DNA repair pathway.