Introduction

DNA double-strand breaks (DSBs) are considered the most cytotoxic type of DNA damage, as unrepaired or improperly repaired DSBs may result in deleterious consequences, including cellular senescence, cell death, or even tumorigenesis [1, 2]. Two pathways, nonhomologous end joining (NHEJ) and homologous recombination (HR), have evolved in eukaryotes to repair DSBs. The NHEJ pathway can be further divided into two independent subpathways: canonical NHEJ (c-NHEJ) and alternative NHEJ (alt-NHEJ). In the c-NHEJ subpathway, the repair is completed without the requirement for homology. This subpathway is activated by the recruitment of the Ku70/Ku80 heterodimer to DSB sites, which constructs a platform that has high affinity for DNA-PKcs, thereby facilitating the recruitment of DNA-PKcs to assemble the DNA-PK holoenzyme at DNA damage sites [3]. The additional c-NHEJ factors include Artemis, XLF, XRCC4 and LIG4 [4, 5]. The alt-NHEJ subpathway is less well-characterized, and it frequently utilizes microhomologies of at least 2 nucleotides and usually more nucleotides at both ends of DSBs to complete the repair [6, 7]. In comparison to c-NHEJ, alt-NHEJ is highly mutagenic and may result in large deletions at the repaired junctions or even chromosomal rearrangements [6, 8, 9]. Alt-NHEJ is independent of DNA-PKcs, Ku70 or Ku80, and the major factors involved in this pathway are PARP1, CtIP, Polθ, and LIG3 [10, 11]. Among these factors, Polθ is considered the rate-limiting enzyme in alt-NHEJ. It is recruited to DSB sites in a PARP1-dependent manner [12], and its helicase activity promotes the removal of RPA from the resected single-stranded DNA, thereby facilitating the completion of strand annealing and end joining by alt-NHEJ [13].

Early studies suggested that alt-NHEJ is active mostly in c-NHEJ-deficient cells [14,15,16], but alt-NHEJ also functions when the c-NHEJ subpathway is intact in different biological contexts [7, 17,18,19], suggesting that there might exist mechanisms regulating the balance of c-NHEJ and alt-NHEJ in certain circumstances. Although great progress has been made in determining the mechanisms regulating the choice between the two major repair pathways—c-NHEJ and HR—upon the occurrence of DSBs [20,21,22,23,24,25], less attention has been given to elucidating how the choice between the two NHEJ subpathways is made. The cell cycle phase might be one of the factors determining the choice of the subpathway, as alt-NHEJ is often initiated by single-strand end resection, which is active mainly in the S and G2 phases [26,27,28], while c-NHEJ is active throughout the cell cycle [29,30,31]. In fact, the major enzyme CtIP, which participates in the end resection step, is not expressed in the G1 phase [32,33,34], and another enzyme, PARP1, which is essential in alt-NHEJ, also exhibits low expression in the G1 phase [35], indicating that the alt-NHEJ subpathway is possibly not as active as the dominant c-NHEJ subpathway in the G1 phase. Moreover, Ku80 and PARP1 play crucial roles in the process of choosing the two NHEJ subpathways [36].

cGAS (cyclic GMP-AMP synthase), the main cytosolic DNA sensor, recognizes cytoplasmic DNA derived from either viral/bacterial infection or self-DNA [37]. Upon DNA binding, cGAS catalyzes the synthesis of the second messenger cyclic guanosine monophosphate-adenosine monophosphate (cGAMP), which binds to the adaptor protein STING (stimulator of interferon genes), activating the TBK1-IRF3-mediated innate immune signaling cascade [38,39,40]. In addition to its canonical DNA sensing function in the cytosol, cGAS also plays multiple noncanonical roles in the nucleus [41,42,43,44,45,46]. With the aid of NONO, cGAS senses HIV DNA in the nucleus [47]. It may also recruit the methyltransferase PRMT5 to the promoters of type I IFN genes in the nucleus [48]. Moreover, cGAS plays dual roles in regulating genomic stability. On the one hand, it slows replication fork progression to stabilize the genome [41]; on the other hand, studies from other groups and our group also indicate that cGAS negatively regulates genomic stability by suppressing HR through distinct mechanisms [43, 44]. However, although changes in NHEJ efficiency were not observed in cells with cGAS overexpression or depletion, the reporters used to quantify NHEJ efficiency can actually reflect the efficiency of both c-NHEJ and alt-NHEJ [44]. Therefore, whether and how nuclear cGAS participates in the regulation of the balance between c-NHEJ and alt-NHEJ remain to be determined.

Here, we found that the cGAS-Ku80 complex regulated the balance between c-NHEJ and alt-NHEJ. Mechanistically, in a Ku80-dependent manner, cGAS promoted c-NHEJ by enhancing the USP7-DNA-PKcs interaction, thereby inhibiting DNA-PKcs ubiquitination and promoting DNA-PKcs stability. In contrast, with the help of Ku80, cGAS directly bound to the Polθ promoter and suppressed its transcription, thereby suppressing alt-NHEJ.

Results

cGAS promotes c-NHEJ and inhibits alt-NHEJ

Since the expression of both PARP1 and CtIP, the two critical factors in alt-NHEJ, is extremely low in G1-arrested cells, we proposed that the efficiency of alt-NHEJ is possibly low in cells in the G1 phase. First, we examined the cell cycle status of proliferating and confluent HCA2-hTERT cells through EdU assay, Western blot with an antibody against the proliferation marker Ki67 and cell cycle analysis. These assays revealed that confluent cells were indeed arrested in G1-phase (Fig. S1A–C). Then, we examined the efficiency of alt-NHEJ using the EJ2-GFP reporter (Fig. S1D), which was designed for the quantification of alt-NHEJ activity [49], in G1 cells. We found that less than 0.1% of G1-arrested cells were GFP+, while ~2% of cells in the population of proliferating cells exhibited green fluorescence (Fig. S1E). In addition, consistent with previous reports, we found that HR activity is nearly absent in cells in the G1-arrested confluent HCA2-H15c cells (Fig. S1F–G) [50, 51]. Taken together, these data indicate that the majority of DSBs are repaired by the c-NHEJ subpathway in G1-arrested cells.

Therefore, to test whether cGAS regulates c-NHEJ, we first induced G1 arrest in the NHEJ-I9a reporter fibroblast line by culturing them to a confluent state (Fig. S2A) and then determined whether cGAS is recruited to I-SceI-induced DSB sites using a ChIP assay [51, 52]. We found that cGAS was recruited to DNA damage sites in G1-arrested confluent HCA2-I9a cells (Fig. 1A), suggesting that cGAS might play roles in c-NHEJ. Next, using both a chromosomal assay [35, 51] and an extrachromosomal assay [53], we analyzed whether overexpressing or depleting cGAS affected NHEJ efficiency in HCA2-hTERT cells in the G1 phase. In contrast to the observation that cGAS overexpression or depletion did not change NHEJ efficiency in proliferating cells [44] (Fig. S2B, C), overexpressing cGAS significantly increased NHEJ efficiency (Figs. 1B and S2D), while depleting cGAS using siRNA or shRNA significantly decreased NHEJ efficiency in G1-arrested cells (Figs. 1C and S2E), indicating that cGAS is probably an important positive regulator of c-NHEJ. Consequently, the comet assay indicated that overexpressing cGAS significantly promoted genomic stability, while depleting cGAS resulted in a decrease in genomic stability in cells with G1 arrest (Fig. 1D–G).

Fig. 1: cGAS promotes c-NHEJ and suppresses alt-NHEJ.
figure 1

A ChIP assay showing cGAS-GFP enrichment at DSBs induced by I-SceI in G1-arrested HCA2-I9a cells. B Analysis of c-NHEJ efficiency in G1-arrested HCA2-I9a cells overexpressing cGAS. Confluent HCA2-I9a cells were transfected with vectors expressing I-SceI, DsRed2-N1 and cGAS-HA. On Day 3, cells were harvested for FACS analysis. A representative Western blotting image showing cGAS overexpression is also presented. C Analysis of c-NHEJ efficiency in G1-arrested HCA2-I9a cells with cGAS depletion. A representative Western blotting image showing cGAS depletion is also presented. D, E Evaluation of genomic stability in G1-arrested HCA2-hTERT cells overexpressing cGAS using the comet assay. Over 100 cells for each group were analyzed D. Representative pictures for each group are also shown in E. F, G Evaluation of genomic stability in G1-arrested HCA2-hTERT cells with cGAS depletion using the comet assay. Over 100 cells for each group were analyzed F. Representative pictures for each group are also shown in G. H Effects of cGAS on NHEJ in unsynchronized HCA2-I9a cells with or without PARP1 depletion. I Analysis of alt-NHEJ efficiency in HCA2-hTERT cells overexpressing cGAS. Cells were transfected with the in vitro I-SceI-digested EJ2-GFP vector for the analysis of alt-NHEJ efficiency and the internal control pDsRed2-N1 plasmid. On Day 3 post-transfection, cells were harvested for FACS analysis. J Analysis of alt-NHEJ efficiency in tail skin fibroblasts from WT, cGAS+/−, and cGAS−/− mice. Data are representative of 3 independent experiments. Data are expressed as the mean ± S.D. values, and comet assay data are presented as the mean ± SEM. Student’s t test or one-way ANOVA with Tukey’s multiple comparison test was used for statistical analysis. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001, n.s., not significant.

Full size image

To reconcile the seemingly contradictory results [44] (Figs. 1B–C and S2B–E), we hypothesized that cGAS might promote c-NHEJ at the expense of inhibiting alt-NHEJ in proliferating cells. To test this hypothesis, we examined whether overexpressing cGAS increases NHEJ efficiency in NHEJ reporter cells with the alt-NHEJ subpathway disrupted. We therefore knocked down the critical alt-NHEJ factor PARP1 using siRNA against PARP1 and then overexpressed cGAS before NHEJ efficiency was analyzed. We found that cGAS overexpression significantly improved NHEJ efficiency in PARP1-depleted proliferating cells in which only c-NHEJ is functional, while in control proliferating cells overexpressing cGAS failed to increase NHEJ efficiency, indicating that cGAS is a positive regulator in c-NHEJ (Fig. 1H). Moreover, we employed a GFP-based reporter cassette, which has 8 nucleotides of microhomology flanking the I-SceI recognition sites to enable the quantification of alt-NHEJ activity [49] (Fig. S1D), to test whether cGAS suppresses alt-NHEJ. We found that overexpressing cGAS significantly inhibited alt-NHEJ (Fig. 1I). In contrast, in a dose-dependent manner, alt-NHEJ activity was significantly increased in both cGAS+/− and cGAS−/− mouse tail fibroblasts compared to wild-type mouse tail fibroblasts (Fig. 1J).

Taken together, these data suggest that cGAS regulates the balance between the two NHEJ subpathways by promoting c-NHEJ and suppressing alt-NHEJ.

cGAS increases DNA-PKcs protein stability

We then set out to determine how cGAS promotes c-NHEJ. We first examined the changes in the protein levels of c-NHEJ factors, including DNA-PKcs, Ku70, Ku80, LIG4, XLF and XRCC4, in cGAS-overexpressing cells. We found that in HCA2-hTERT cells with a vector encoding cGAS integrated into the genome, the protein level of DNA-PKcs was increased but those of the other factors were not (Fig. 2A). Transient overexpression of cGAS in HCA2-hTERT cells also promoted the expression of DNA-PKcs in a dose-dependent manner (Fig. 2B). In contrast, knocking down cGAS drastically suppressed the expression of DNA-PKcs (Fig. 2C). Moreover, in cGAS+/− and cGAS−/− mouse tail fibroblasts, DNA-PKcs expression was reduced (Fig. 2D).

Fig. 2: cGAS increases DNA-PKcs protein stability.
figure 2

A Western blotting analysis of the protein levels of critical c-NHEJ factors in HCA2-hTERT cells overexpressing cGAS. B The expression of DNA-PKcs in HCA2-TERT cells transfected with different amounts of cGAS vectors. C The expression of DNA-PKcs in HCA2-hTERT cells with cGAS depletion using siRNA. D The expression of DNA-PKcs in tail skin fibroblasts from WT, cGAS+/− and cGAS−/− mice. E Analysis of the relative DNA-PKcs mRNA level in HCA2-hTERT cells overexpressing cGAS. F Representative Western blotting image showing DNA-PKcs protein levels in cGAS-overexpressing cells treated with cycloheximide (CHX, 100 μg/ml) for the indicated time window. Representative Western blotting result indicating cGAS overexpression is shown on the right. n.s., not significant.

Full size image

We then examined whether the regulation of the DNA-PKcs protein level by cGAS occurs through transcriptional mechanisms. We examined the mRNA level of DNA-PKcs in cGAS-overexpressing cells and did not observe any significant change in the mRNA level (Fig. 2E), suggesting that the regulation does not occur through transcriptional mechanisms. In contrast, treatment with cycloheximide (CHX), a protein synthesis inhibitor, greatly inhibited the degradation rate of DNA-PKcs in cGAS-overexpressing cells (Fig. 2F), suggesting that cGAS promotes DNA-PKcs protein stability to increase its protein level.

USP7 deubiquitinates DNA-PKcs to promote its protein stability

Since RNF144A is the E3 ligase that regulates the ubiquitination and protein stability of DNA-PKcs, we first examined whether cGAS diminishes the interaction between RNF144A and DNA-PKcs, thereby stabilizing the DNA-PKcs protein [54]. We found that cGAS overexpression did not disrupt the interaction of the two factors (Fig. S3A). In addition, the further co-IP experiments showed that simultaneous overexpression of cGAS and Ku80 also didn’t impede the binding of DNA-PKcs to RNF144A (Fig. S3B), suggesting that cGAS-mediated stabilization of DNA-PKcs does not occur through the RNF144A-DNA-PKcs regulatory axis. Therefore, we hypothesized that cGAS might promote the interaction between DNA-PKcs and an unknown deubiquitinase.

To test this hypothesis, we performed mass spectrometry analysis with HEK293T cell lysates that were subjected to an immunoprecipitation assay with an antibody against DNA-PKcs. We identified four deubiquitinases, USP5, USP7, USP10 and USP36, that potentially interact with DNA-PKcs (Fig. S3C). By overexpressing the vectors encoding each deubiquitinase, we found that only USP7 and not the other three deubiquitinases resulted in an increase in the DNA-PKcs protein level; in addition, overexpressing an enzymatically dead USP7 mutant resulted in loss of this promoting effect (Fig. 3A, B). In contrast, knocking down USP7 reduced the protein level of DNA-PKcs (Fig. 3C).

Fig. 3: cGAS promotes USP7-mediated DNA-PKcs deubiquitination and stabilization.
figure 3

A Western blotting analysis of the DNA-PKcs protein level in HEK293T cells transfected with the USP5, USP10 or USP36 overexpression vector. B Western blotting analysis of the DNA-PKcs protein level in cells transfected with the USP7 or enzymatically dead USP7 (USP7m) overexpression vector. C The expression of DNA-PKcs in HEK293T cells with USP7 depletion. D, E Co-IP analysis of the interaction between endogenous USP7 and DNA-PKcs in HEK293T cells using antibodies against USP7 or DNA-PKcs. F In vitro co-IP analysis of the interaction between USP7 and DNA-PKcs. The two purified recombinant proteins were incubated, and a co-IP assay with an antibody against GST was performed, followed by Western blotting analysis. CBB, Coomassie Brilliant Blue. G Analysis of the level of K48-linked ubiquitination of DNA-PKcs in USP7-depleted HEK293T cells. Cells were treated with MG132 (10 μM, 12 h) before being harvested for co-IP and Western blotting analysis. H Co-IP analysis of the interaction between USP7 and DNA-PKcs in HEK293T cells overexpressing cGAS-HA. I Western blotting analysis of the DNA-PKcs protein level in control and USP7-depleted HEK293T cells with or without cGAS overexpression.

Full size image

We next examined whether USP7 interacts with DNA-PKcs to regulate its deubiquitination. The results of the in vivo co-IP assay confirmed that USP7 interacted with endogenous DNA-PKcs (Figs. 3D, E and S3D). The results of the in vitro co-IP experiments also revealed that USP7 interacted directly with DNA-PKcs (Fig. 3F). In addition, we separated USP7 into its TRAF, CAT and UBL domains [55] and sought to determine which ___domain mediated it interaction with DNA-PKcs using a co-IP assay. We found that both the TRAF and UBL domains interacted with DNA-PKcs (Fig. S3E). In addition, we found that depleting USP7 drastically increased the ubiquitination level of DNA-PKcs (Fig. 3G). Collectively, these data suggest that USP7 deubiquitinates DNA-PKcs to promote its protein stability.

We then sought to determine whether cGAS regulates DNA-PKcs protein stability by changing the USP7-DNA-PKcs interaction. We found that overexpressing cGAS indeed enhanced the USP7-DNA-PKcs interaction (Figs. 3H and S3F). We also observed that depleting USP7 abolished the stabilization of the DNA-PKcs protein by cGAS (Fig. 3I). These data indicate that cGAS enhances the USP7-DNA-PKcs interaction to stabilize DNA-PKcs.

cGAS stabilizes the DNA-PKcs protein in a Ku80-dependent manner

We then examined whether cGAS directly interacts with DNA-PKcs to promote its protein stability. Surprisingly, we did not observe an interaction between cGAS and DNA-PKcs, XRCC4, LIG4 or XLF by co-IP (Fig. 4A). Instead, we found that cGAS interacted with Ku70 and Ku80 (Fig. 4A). An in vitro co-IP assay further demonstrated that Ku80 but not Ku70 directly interacted with cGAS (Fig. 4B), consistent with a previous finding [56]. Co-IP experiments also revealed that the NTase core ___domain in cGAS mediated its interaction with Ku80 (Fig. S4A, B), while both the Core and vWA domains in Ku80 were critical to its interaction with cGAS (Fig. S4C). Moreover, we demonstrated that the cGAS-Ku80 interaction was enhanced in response to DNA damage induced by X-ray irradiation (Figs. 4C and S5A), confirming that the formation of the cGAS-Ku80 complex is important in DNA repair. Subcellular fractionation followed by co-IP experiments revealed that the cGAS-Ku80 interaction occurred in both cytosol and nucleus, but DNA damage led to the enhancement of the cGAS-Ku80 interaction in the nucleus but not in the cytosol (Fig. 4D). To examine whether cGAS and Ku80 colocalize at DNA damage sites, we performed laser microirradiation experiments in U2OS cells with vectors encoding cGAS-GFP transfected, and performed immunofluorescence experiments with an anti-Ku80 antibody. The result indicated that cGAS and Ku80 colocalized at DNA damage sites (Fig. 4E).

Fig. 4: cGAS interacts with Ku80 to promote USP7-mediated stabilization of DNA-PKcs.
figure 4

A Co-IP analysis of the interaction between cGAS and c-NHEJ factors in HEK293T cells. Cells were transfected with cGAS-HA and lysed for the co-IP assay. B In vitro co-IP analysis of the interaction between cGAS and Ku80 or Ku70. The purified recombinant proteins were incubated, and a co-IP assay with an antibody against GST was performed, followed by Western blotting analysis. C Co-IP analysis of the interaction between endogenous cGAS and Ku80 in HCA2-hTERT cells exposed to X-ray irradiation (8 Gy, 0.5 h). D Co-IP analysis of the interaction between cGAS-HA and Ku80 following subcellular fractionation. E Immunofluorescence images of cGAS-GFP and endogenous Ku80 upon laser microirradiation in U2OS cells. Scale bar, 5 μm. F Co-IP analysis of the endogenous USP7-DNA-PKcs interaction in control or Ku80-depleted cells with or without cGAS overexpression. G Analysis of DNA-PKcs expression in Ku80-depleted cells with or without cGAS overexpression. H Co-IP analysis of the interaction between endogenous USP7 and Ku80. I Co-IP analysis of the interaction between endogenous USP7 and Ku80 with or without cGAS overexpression.

Full size image

Since cGAS enhanced the USP7-DNA-PKcs interaction and directly interacted with Ku80 but not DNA-PKcs, we hypothesized that cGAS-mediated stabilization of the DNA-PKcs protein might occur through strengthening of the Ku80-DNA-PKcs interaction. Indeed, co-IP experiments revealed that overexpressing cGAS enhanced the Ku80-DNA-PKcs interaction, while depleting cGAS partially reduced the interaction between Ku80 and DNA-PKcs (Fig. S5B, C). We found that knocking down Ku80 abolished the cGAS-mediated enhancement of the USP7-DNA-PKcs interaction (Figs. 4F and S5D). Importantly, we observed that depleting Ku80 abolished the stabilization of the DNA-PKcs protein by cGAS (Fig. 4G). Since Ku80 usually forms a heterodimer with Ku70 to regulate c-NHEJ, so we tested whether cGAS-mediated stabilization of DNA-PKcs is dependent on Ku70. Western blotting results showed that depleting Ku70 also eliminated the stimulatory effect of cGAS on DNA-PKcs protein stability (Fig. S5E). In addition, we found that Ku80 had a strong interaction with USP7 (Figs. 4H and S5F–G), and overexpressing cGAS weakened the Ku80-USP7 interaction (Figs. 4I and S5H). Therefore, we proposed that cGAS might compete with USP7 for binding to Ku80, thereby releasing more USP7 to interact with DNA-PKcs. We also ruled out the possibility that cGAS promotes c-NHEJ by increasing DNA-PKcs enzymatic activity, as the in vitro biochemical assay showed that cGAS did not stimulate DNA-PKcs autophosphorylation or p53 phosphorylation (Fig. S5I).

Thus, we propose that cGAS interacts with Ku80 to promote the USP7-mediated deubiquitination of DNA-PKcs, thereby stabilizing the DNA-PKcs protein and promoting c-NHEJ repair.

cGAS inhibits alt-NHEJ by suppressing the transcription of Polθ in a Ku80-dependent manner

To understand how cGAS inhibits the alt-NHEJ subpathway, we first tested the expression of key factors involved in the alt-NHEJ subpathway, PARP1, LIG3 and XRCC1, in cells overexpressing cGAS. We did not observe any significant changes in the expression levels of these factors (Fig. 5A). Polθ is the rate-limiting factor in alt-NHEJ, but no high-quality commercial antibody is available for Western blot analysis. We therefore performed qRT‒PCR to analyze whether cGAS affects Polθ transcription. Interestingly, we found that overexpressing cGAS significantly inhibited the transcription of Polθ (Fig. 5B). Knocking down cGAS by siRNA significantly promoted Polθ transcription (Fig. 5C). We then cloned the promoter of Polθ and fused it to the firefly luciferase gene (Fig. 5D). Using this vector, we found that overexpressing cGAS significantly suppressed Polθ promoter activity (Fig. 5E). A further ChIP assay demonstrated that cGAS was present in the promoter region of Polθ (Fig. 5F–G). Importantly, DNA damage induced by IR may stimulate the recruitment of cGAS to the Polθ promoter (Fig. 5H), suggesting that the regulation of alt-NHEJ by cGAS is stress dependent. We also analyzed the repair efficiency of alt-NHEJ in Polθ-depleted cells overexpressing cGAS. We found that overexpressing cGAS did not further reduce alt-NHEJ repair efficiency in Polθ-depleted cells (Figs. 5I and S6A). These data suggest that upon the occurrence of DNA damage, cGAS is recruited to the promoter of Polθ to suppress the expression of Polθ, thereby leading to a reduction in alt-NHEJ repair efficiency.

Fig. 5: cGAS inhibits the transcription of Polθ.
figure 5

A Western blotting analysis of the expression of critical factors in alt-NHEJ in HCA2-hTERT cells with or without cGAS overexpression. B RT‒qPCR analysis of the relative Polθ mRNA level in cGAS-overexpressing HCA2-hTERT cells. Total RNA was extracted and reverse-transcribed to cDNA for quantitative PCR. C RT‒qPCR analysis of the relative Polθ mRNA level in HCA2-hTERT cells with cGAS depletion. D Schematic diagram of the vector with the Polθ promoter driving the expression of firefly luciferase (PPolθ-Luc). E Analysis of Polθ promoter activity in cells overexpressing cGAS using the PPolθ-Luc reporter. F ChIP assay showing the recruitment of endogenous cGAS to Polθ promoter regions in HCA2-hTERT cells. G ChIP assay showing the recruitment of cGAS-HA to Polθ promoter regions in HEK293T cells. Vectors encoding cGAS-HA were transfected into HEK293T cells. Twenty-four hours post-transfection, cells were harvested for a ChIP assay with an antibody against the HA tag. H ChIP assay showing that DNA damage induced by X-ray irradiation (8 Gy, 30 min) stimulated the recruitment of cGAS-HA to the Polθ promoter. I Analysis of relative alt-NHEJ efficiency in Polθ-depleted HCA2-hTERT cells with or without cGAS overexpression. Data are representative of 3 independent experiments. Data are expressed as the mean ± SD values. Student’s t test or one-way ANOVA with Tukey’s multiple comparison test was used for statistical analysis. *P < 0.05, ***P < 0.001, ****P < 0.0001, n.s., not significant.

Full size image

Alt-NHEJ is more error-prone than c-NHEJ, so we set out to determine whether cGAS promotes DSB repair fidelity in a Polθ-dependent way. Using our well-established assay for examining the fidelity of DSB repair [53] (Fig. S6B), we found that depleting cGAS greatly diminished the repair fidelity, while depleting Polθ in cGAS knockdown cells rescued the reduction in repair fidelity (Fig. S6C), confirming that cGAS participates in regulating alt-NHEJ through targeting Polθ.

Since treating BRCA1/BRCA2 mutant cancers with Polθ inhibitors causes synthetic lethality [57], we therefore examined whether cGAS may inhibit survival of BRCA-deficient HCC1937 cells. Indeed, we found that overexpressing cGAS suppressed cell survival of HCC1937 cells (Fig. S6D–F), but had no effect on HR repair-proficient MDA-MB-231 cells (Fig. S6G–I), indicating that cGAS is a promising therapeutic target for tumor cells with HR deficiency.

Since cGAS interacts with Ku80, we then further examined whether the regulation of Polθ promoter activity by cGAS is dependent on Ku80. We first tested whether Ku80 also regulates Polθ transcription. We found that overexpressing Ku80 indeed inhibited Polθ mRNA expression, while knocking down Ku80 promoted Polθ mRNA expression (Fig. 6A, B). A luciferase assay also revealed that Ku80 overexpression suppressed Polθ promoter activity (Fig. 6C). Consistent with this finding, Ku80 overexpression significantly attenuated the alt-NHEJ repair efficiency, and knocking down Ku80 by siRNA greatly promoted alt-NHEJ (Fig. 6D, E). However, depleting Ku70 or inhibiting DNA-PKcs catalytic activity had no significant effect on Polθ mRNA level (Fig. S7A, B). A ChIP assay revealed that Ku80 was also recruited to the Polθ promoter (Fig. 6F). Then, we examined whether cGAS and Ku80 function together to suppress Polθ promoter activity. We found that simultaneous overexpression of cGAS and Ku80 did not have a further suppressive effect on Polθ mRNA expression (Fig. 6G), suggesting that the two factors act in the same pathway in downregulating Polθ promoter activity. Importantly, we found that depleting Ku80 abolished the recruitment of cGAS to the Polθ promoter (Fig. 6H). Consequently, depleting Ku80 abolished the suppressive effect of cGAS on Polθ mRNA expression (Fig. 6I). In addition, we performed laser microirradiation experiments to test whether cGAS affects the recruitment of Ku80 to DSB sites. The results showed that neither overexpressing cGAS nor depleting cGAS affected the recruitment of Ku80-GFP to DNA damage sites (Fig. S8A–F).

Fig. 6: cGAS inhibits alt-NHEJ by suppressing Polθ transcription in a Ku80-dependent manner.
figure 6

A RT‒qPCR analysis of the relative Polθ mRNA level in Ku80-overexpressing HCA2-hTERT cells. B RT‒qPCR analysis of the relative Polθ mRNA level in HCA2-hTERT cells with Ku80 depletion. C Analysis of Polθ promoter activity in cells overexpressing Ku80 using the PPolθ-Luc reporter. D Analysis of alt-NHEJ efficiency in Ku80-overexpressing HCA2-hTERT cells using EJ2-GFP reporters. E Analysis of alt-NHEJ efficiency in HCA2-hTERT cells with Ku80 depletion using EJ2-GFP reporters. F ChIP assay showing the recruitment of Ku80 to Polθ promoter regions. Vectors expressing Ku80-HA were transfected into HEK293T cells. Twenty-four hours post-transfection, cells were harvested for a ChIP assay with an antibody against HA. G RT‒qPCR analysis of the relative Polθ mRNA level in HCA2-hTERT cells overexpressing cGAS and/or Ku80. H ChIP assay showing that depleting Ku80 by siKu80 impaired the enrichment of cGAS at the Polθ promoter in HEK293T cells. I RT‒qPCR analysis of the relative Polθ mRNA level in control or cGAS-overexpressing HCA2-hTERT cells with or without Ku80 depletion. Data are representative of 3 independent experiments. Data are expressed as the mean ± S.D. values. Student’s t test or one-way ANOVA with Tukey’s multiple comparison test was used for statistical analysis. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001, n.s., not significant.

Full size image

Taken together, these results indicate that in a Ku80-dependent manner DNA damage induces the recruitment of cGAS to the Polθ promoter region to repress its transcription, thereby inhibiting alt-NHEJ repair.

Several cancer-associated cGAS mutants lose the ability to promote c-NHEJ

Since defects in key c-NHEJ factors are often associated with high incidence of tumorigenesis, we analyzed whether cancer-associated mutations in cGAS cause the loss of the stimulatory effect of cGAS on c-NHEJ. Using vectors encoding cancer-associated cGAS mutants [58], we performed NHEJ analysis in HCA2-I9a cell. We found that 14 out of 37 mutants failed to stimulate c-NHEJ (Fig. S9A), and these mutants retained their canonical function in innate immunity (Fig. S9B). These data suggest that these 14 mutants might promote tumorigenesis through disrupting its positive function in c-NHEJ.

Discussion

As the key DNA sensor, cGAS is mainly localized in the cytosol to perform its canonical innate immune function, but its presence in the nucleus is also critical to a number of biological processes, including the stabilization of replication forks and noncanonical immune functions [41,42,43,44, 46]. Our previous study revealed for the first time a negative function of cGAS, namely, that DNA damage-induced nuclear translocation of cGAS suppresses DNA repair by HR, thereby promoting tumorigenesis [44]. However, this protumorigenic effect of cGAS occurs only in proliferating cells in which HR is active [59]. Our recent study showed that nuclear cGAS promotes TRIM41-mediated ubiquitination and degradation of LINE-1 ORF2 protein to repress LINE-1 (L1) retrotransposition to stabilize genomes in human cells [58]. Together with our finding of nuclear cGAS regulating the two NHEJ subpathways, we propose that cGAS plays multifaceted roles in regulating genomic stability (Fig. S10). In most quiescent somatic cells, the c-NHEJ subpathway is the dominant DSB repair pathway, and our present finding that nuclear cGAS plays a critical role in promoting c-NHEJ suggests that cGAS might safeguard genome integrity to prevent the onset of aging or tumorigenesis. In addition, since cGAS is abundant in immune cells such as T and B lymphocytes, its enhancing effect on the DNA-PKcs protein level to promote c-NHEJ might also be critical to the adaptive immune functions of T cells and B cells, in which DNA-PKcs is required for efficient V(D)J recombination to generate T-cell receptors or antibodies upon bacteria or virus infections [17, 60, 61]. However, the hypothesis that cGAS might be essential in these physiological processes warrants further investigation.

Although a previous study reported that DNA-PKcs protein stability can be regulated by the E3 ubiquitin ligase RNF144A [54], our findings, for the first time, reveal the USP7-mediated deubiquitination mechanism underlying DNA-PKcs protein stabilization. Since elevated NHEJ is often associated with radio- or chemoresistance during cancer treatment [62,63,64], targeting the USP7-DNA-PKcs axis might provide another approach to overcome radio- or chemoresistance. In this study, using mass spectrometry, we also identified three other deubiquitinases, USP5, USP10 and USP36, that could potentially interact with DNA-PKcs upon the occurrence of DNA damage. Since overexpressing any of these three deubiquitinases did not change the protein level of DNA-PKcs, whether and how they regulate DNA-PKcs enzymatic activity remains to be determined.

The cGAS-Ku80 interaction has been reported in two previous studies [56, 65], and this complex has been suggested to potentiate the antiviral function of cGAS [56], which apparently occurs in the cytosol. However, Ku80 is a major factor involved in NHEJ and is mainly localized in the nucleus [66, 67]. Our study provides another explanation for why the two factors form a complex from the perspective of nuclear functions. In fact, a previous report indicated that Ku80 competes with PARP1 to control the choice between c-NHEJ and alt-NHEJ. Whether cGAS participates in the regulation of the competition between Ku80 and PARP1 at DNA damage sites remains to be further studied.

Our findings also indicate that cGAS functions as a transcriptional repressor to inhibit the transcription of Polθ in a Ku80-dependent manner. Since cGAS has been shown to bind to histones and ChIP assays have revealed that cGAS is a chromatin-associated factor in different types of cells, whether the cGAS-Ku80 complex regulates the expression of other genes in a DNA damage-dependent manner remains to be determined.

Materials and methods

Cell culture and cell treatment

HCA2-hTERT cells (immortalized foreskin fibroblasts) and HCA2-I9a cells (derived from HCA2-hTERT cells harboring a single copy of the NHEJ reporter cassette in the genome) were cultured in MEM (HyClone, Cat. # SH30234), HEK293T cells (human embryonic kidney epithelial cells) and MDA-MB-231 cells were cultured in DMEM (KeyGen Biotech, Cat. # KGM12800), and HCC1937 cells were cultured in RPMI-1640 medium (KeyGen Biotech, Cat. # KGL1501-500) supplemented with 10% FBS (Life Technologies, Cat. # 16000), 1% nonessential amino acids (Gibco, Cat. # 11140-050) and 1% penicillin/streptomycin (Gibco, Cat. # 15140-122). All cells were maintained in a humidified incubator at 37 °C in 5% CO2. Cells stably expressing HA-cGAS or control HCA2-hTERT cells were treated with cycloheximide (CHX) (APExBIO, Cat. # A8244), harvested at different time points and then lysed. Protein levels were measured by Western blot analysis. To induce DSBs, cells were treated with ionizing radiation (IR) at a dose of 8 Gy.

Antibodies, reagents and siRNAs

The following antibodies and reagents were used in this study: anti-cGAS (Cell Signaling, Cat. # 83623), anti-Ku80 (ABclonal, Cat. # A5862), anti-Ku70 (ABclonal, Cat. # A0883), anti-LIG4 (ABclonal, Cat. # A1743), anti-XLF (ABclonal, Cat. # A4985), anti-DNA-PKcs (Abcam, Cat. # ab70250), anti-PARP1 (ABclonal, Cat. # A3121), anti-XRCC1 (Abcam, Cat. # ab134056), anti-LIG3 (Abcam, Cat. # ab587), anti-USP7 (Cell Signaling, Cat. # 4833 S), anti-K48 (Cell Signaling, # 8081), anti-DNA-PKcs-pS2056 (Abcam, Cat. # ab18192), anti-P53-pS15 (ABclonal, Cat. # AP0083), anti-Flag (ABclonal, Cat. # AE005), anti-HA (Cell Signaling, Cat. # 3724), anti-GFP (Abcam, Cat. # ab290), anti-Tubulin (Bioworld, Cat. # AP0064), anti-β-actin (Cell Signaling, Cat. # 4970), anti-GAPDH (Proteintech, Cat. # HRP-60004), anti-cGAS (ABclonal, Cat. # A8335), anti-Lamin A/C (ABclonal, Cat. # A19524), GFP-Trap (ChromoTek, Cat. # gta-20), and HA-beads (Selleck, Cat. # B26201). The reagents used for qRT‒PCR analysis were SYBR Green qPCR Mix (Roche, Cat. 760 # 04913914001) and Transgen EasyScript® One-Step gDNA Removal and cDNA Synthesis SuperMix (Transgen, Cat. # AE311-03). The sequences of the siRNAs used in this study are as follows: sicGAS-1: GGCCUCUGCUUUGAUAACU, sicGAS-2: GCAAAAGUUAGGAAGCAAC, siKu80: AGAGGAAGCCUCUGGAAGUUC, siPARP1: CGACCUGAUCUGGAACAUCAA, siPolθ-1: CCUUAAGACUGUAGGUACU, siPolθ-2: ACACAGUAGGCGAGAGUAU, siPolθ-3: CGACUAAGAUAGAUCAUUU, siPolθ-4: ACAACAACCCUUAUCGUAAA, siKu70-1: GGAAGAGAUAGUUUGAUUU, siKu70-2: GAUGCCCUUUACUGAAAAA. The sequences of the shRNAs used in this study were as follows: shcGAS: CGTGAAGATTTCTGCACCTAA, shKu80: GAGCAGCGCTTTAACAACT, shUSP7-1: AAGCGTCCCTTTAGCATTACA, shUSP7-2: ACCCTTGGACAATATTCCTGG.

Isolation of mouse skin fibroblasts

Mouse skin fibroblasts were isolated from tail skin tissues of age-matched female mice. Seven-month-old WT, cGAS+/− and cGAS−/− mice were used for the isolation of skin fibroblasts. The isolation procedure was performed as previously reported [68]. The isolated primary mouse skin fibroblasts were cultured in DMEM supplemented with 10% FBS, 1% nonessential amino acids and 1% penicillin/streptomycin and were maintained in incubators with 5% CO2 and 3% O2 at 37 °C.

Comet assay

Cells in the G1 phase were harvested and resuspended in PBS at a concentration of 2 × 105 cells/ml and then subjected to the alkaline comet assay according to the manufacturer’s instructions (Trevigen, Cat. # 4250-050-K). The results were analyzed by CometScore software (casplab_1.2.3b2), and the tail moments were calculated and used as the measure of the amount of DNA damage.

Generation of cell lines stably expressing exogenously introduced genes

HEK293T cells were transfected with a set of plasmids, including pMD2.G, pSPAX2, and shRNA vectors targeting human Ku80, USP7, or cGAS. After 72 h, the viral supernatants were collected and strained through a 0.45 µm filter. HEK293T cells were infected with viral supernatants supplemented with polybrene (6 μg/ml). After 48 h, the infected HEK293T cells were treated with puromycin (2 μg/ml). HCA2-hTERT cells with genomic integration of the HA-cGAS expression vector were generated by electroporation with the DT-130 program on a Lonza 4D machine and selected with G418 (400 μg/ml).

Immunoprecipitation

HEK293T cells transfected with the indicated plasmids were harvested and lysed in lysis buffer (20 mM HEPES (pH 8.0), 0.2 mM EDTA, 5% glycerol, 150 mM NaCl, 1% NP40, and protease inhibitor cocktail). The cell lysate was placed on ice for 30 min, sonicated at 35% power for 10 s, and then centrifuged for 10 min at 13,000 rpm at 4 °C. The lysate supernatant was collected after being precleared with IgG and protein A/G agarose beads for 1 h at 4 °C. The collected supernatant was then incubated with antibodies, GFP-Trap or HA-beads overnight at 4 °C. On Day 2, the supernatant incubated with antibodies was added to protein A/G agarose beads for 2 h at 4 °C, and then the supernatant was washed 4 times with cold lysis buffer. The beads with bound precipitated proteins were added to 2× sample buffer and boiled for 10 min prior to Western blot analysis.

Chromatin immunoprecipitation

The ChIP assay was performed with HCA2-I9a cells and HEK293T cells in different experiments as previously reported [44, 52]. Briefly, to quantify the recruitment of cGAS to DSB sites, HCA2-I9a cells in G1 phase were transfected with the GFP-tagged cGAS vector along with the I-SceI vector. Cells were harvested at different time points (0, 4, 8 and 12 h) after transfection, and the ChIP assay was carried out using GFP-Trap. Precipitated DNA and input were quantified by quantitative PCR with the following primers (5′-CCTGAAGATTTGGGGGATTGTGCTTC-3′, 5′-CTTGGAAACACCCATGTTGAAATATC-3′). To test the enrichment of cGAS or Ku80 on the Polθ promoter, HEK293T cells were transfected with the HA-tagged cGAS vector or HA-tagged Ku80 vector. After 20 h, cells were collected and subjected to a ChIP assay using an antibody against the HA tag or IgG. To test the enrichment of endogenous cGAS on the Polθ promoter, HCA2-hTERT cells were collected for ChIP assay using the antibody against cGAS or IgG. The relative amounts of DNA from IgG, anti-HA, or anti-cGAS antibody precipitated samples and input were then quantified by real-time PCR, followed by the 2-△△CT calculation method. The relative enrichment is then calculated by comparing the percentage of each ChIP sample in total input chromatin. 1% of starting chromatin was taken as input in this study. The sequences of the primers for the Polθ promoter were as follows: 5′-GCCACGGAGAACTCTATGGTT-3′ and 5′- TCTGAGCCTGATTCTGAACGC-3′.

FACS analysis

For analysis of NHEJ efficiency, HCA2-I9a cells were transfected with 5 μg I-SceI vector, 15 ng DsRed vector and the indicated plasmids or siRNAs. For the extrachromosomal assay to analyze NHEJ efficiency, HCA2-hTERT cells and mouse skin fibroblasts were transfected with the NHEJ reporter construct digested in vitro with the I-SceI restriction enzyme and the DsRed plasmid. At 72 h post-transfection, at least 20,000 cells were harvested and subjected to FACS analysis on a BD FACSVerse (BD Biosciences, USA). The results were further analyzed with FlowJo software. The proportions of GFP+ cells and DsRed+ cells were considered to indicate the relative efficiency of NHEJ.

Reverse transcription quantitative PCR

Total RNA was isolated with TRIzol reagent (Invitrogen), followed by reverse transcription with a kit (Transgen, Cat. # AT301). Real-time quantitative PCR was carried out on the QuantStudio PCR System. The sequences of the primers used to amplify human DNA-PKcs, Polθ and GAPDH were as follows: 5′-GCGCCATATCTGTCATCTGCTG-3′, 5′-TTATAGCGGCGCTTCAGGTCGA-3′; 5′-CTACAAGTGAAGGGAGATGAGG-3′, 5′-TCAGAGGGTTTCACCAATCC-3′; 5′-TCCTGTTCGACAGTCAGCCGCA, 5′-ACCAGGCGCCCAATACGACCA-3′.

Luciferase assay

The Polθ promoter was amplified from genomic DNA isolated from HEK293T cells using the primers 5′-GGTACCGAGCTCTTACGCGTCTATACAATGTGTGGAATCTATGAATAACATC-3′ and 5′-CTTAGATCGCAGATCTCGAGGGCAAACTCTTCTCGGCC-3′, and then the 3 kb DNA segment upstream of the Polθ transcription start site was subcloned into the pGL3-Basic plasmid. HEK293T cells were transfected with vectors encoding the Polθ promoter-firefly luciferase fusion (5 μg); Renilla luciferase (6 ng), which was used to normalize the transfection efficiency; and the indicated plasmids. At 24 h post-transfection, cells were collected, and luciferase activity was measured with a Dual-Luciferase Reporter System (Promega, Cat. # E1910) on a GloMax Luminometer (Promega, Cat. # E5311).

In vitro DNA-PKcs kinase activity assay

The DNA-PKcs kinase activity assay was performed as reported previously [69]. In brief, 500 ng GST or GST-cGAS protein was incubated with 200 ng DNA-PKcs and 185 ng Ku70/80/DNA mix containing 500 ng P53 protein in 40 μL reaction buffer (50 mM HEPES-NaOH (pH 7.4), 100 mM KCl, 10 mM MgCl2, 2 mM EGTA, 0.1 mM EDTA and 1 mM ATP) at 30 °C for 10 mins. The reaction product was analyzed by Western blotting.

GST pull-down assay

GST-tagged proteins or the GST protein were expressed in Escherichia coli by isopropyl-β-D-thiogalactopyranoside (0.1 mM) induction and purified using glutathione Sepharose 4B beads (GE Healthcare, USA). GST-tagged proteins were incubated with the purified protein in buffer (10 mM Tris-HCl (pH 8.0), 1 mM EDTA, 100 mM NaCl) overnight at 4 °C. After the reaction, the samples were centrifuged and washed three times with cold buffer. The beads were boiled with 2 × sample buffer for 10 mins, and the precipitated proteins were analyzed by Western blotting.

Laser microirradiation

The laser microirradiation assay was performed as previously reported [44]. In brief, U2OS cells cultured in glass-bottomed culture dishes (NEST, Cat # 801002) were transfected with Ku80-GFP or cGAS-GFP plasmids. At 72 h post transfection, a 405 nm UV laser was used to irradiate U2OS cells to induce DNA damage with the Leica DM6500 confocal microscopy 405-nm laser diode system. The images were acquired and analyzed using Leica LAS AF software (LAS AF3.1.0).

NHEJ fidelity analysis

The method is as previously described [53, 70]. I-SceI linearized NHEJ construct at an amount of 0.8 μg together with siControl or siPolθ were transfected into 1 × 106 control or cGAS depleted HCA2-hTERT cells. At 72 h post transfection, cells were harvested for genomic DNA extraction. Then the genomic DNA was transformed into E.coli HB101 Electro-cells (TaKaRa, Cat # 9021) for plasmid rescue. The rescued NHEJ plasmids were collected for sequencing analysis using the primer at the upstream of I-SceI recognition sites. The primer sequence is 5’GACCACTGGATTCAGAAGCGATC3’.

EdU incorporation assay

The HCA2-hTERT cells, cultured to either a proliferating or confluent state, were incubated with EdU (5-ethynyl-2′-deoxyuridine) for 2 h. EdU incorporation was detected using EdU Cell Proliferation Kit (Sangon Biotech, Cat. # E607204). Subsequently, images were captured using a fluorescence microscope, and the proportion of EdU-positive cells was determined by manually counting.

Clonogenic assay

HCC1937 cells or MDA-MB-231 cells with or without stably integrated vectors encoding cGAS were seeded in a six-well plates. Cells were incubated for 10 days, and then stained with Coomassie reagent (0.25% Coomassie, 50% methanol, and 10% acetic acid) for 6 h. Colonies containing at least 50 cells were counted.

Statistical analysis

GraphPad Prism 8.0 software was used for data analysis. Data from the NHEJ repair efficiency analysis, ChIP analysis, and luciferase assay, as well as transcript levels, are presented as the mean ± S.D. values, and data from the comet assay are presented as the mean ± SEM values. Comparisons between two groups were performed by Student’s t test, and comparisons among more than two groups were performed by one-way ANOVA with Tukey’s multiple comparison test. P < 0.05 was considered to indicate a significant difference. n.s., not significant.