Genetic and physical interactions between Polη and Rev1 in response to UV-induced DNA damage in mammalian cells

In response to UV irradiation, translesion DNA synthesis (TLS) utilizes specialized DNA polymerases to bypass replication-blocking lesions. In a well-established polymerase switch model, Polη is thought to be a preferred TLS polymerase to insert correct nucleotides across from the thymine dimer, and Rev1 plays a scaffold role through physical interaction with Polη and the Rev7 subunit of Polζ for continual DNA synthesis. Defective Polη causes a variant form of xeroderma pigmentosum (XPV), a disease with predisposition to sunlight-induced skin cancer. Previous studies revealed that expression of Rev1 alone is sufficient to confer enhanced UV damage tolerance in mammalian cells, which depends on its physical interaction with Polζ but is independent of Polη, a conclusion that appears to contradict current literature on the critical roles of Polη in TLS. To test a hypothesis that the Rev1 catalytic activity is required to backup Polη in TLS, we found that the Rev1 polymerase-dead mutation is synergistic with either Polη mutation or the Polη-interaction mutation in response to UV-induced DNA damage. On the other hand, functional complementation of polH cells by Polη relies on its physical interaction with Rev1. Hence, our studies reveal critical interactions between Rev1 and Polη in response to UV damage.

www.nature.com/scientificreports/ It has been well accepted that Rev1 functions as a scaffold for polymerase switch during TLS in response to UV irradiation 26,27 , in which its catalytic activity is dispensable, as the Rev1 polymerase-dead mutation does not confer increased sensitivity to UV-induced killing and mutagenesis 28 . Rev1 can be recruited to the damage site through enhanced affinity for monoubiquitinated PCNA via its PCNA-binding BRCT domain 29,30 and ubiquitinbinding UBM motifs 31 . The Rev1-CTD contains two separate domains to interact with Y-family polymerases including Polη, Polι and Polκ, and the Rev7 subunit of Polζ [32][33][34][35] . We recently attempted to address detailed scaffold roles of Rev1 in response to UV-induced DNA damage and surprisingly found that UV damage tolerance conferred by ectopic expression of Rev1 is dependent on its interaction with Rev7 but independent of Polη interaction 36 . The current study further addressed genetic and physical interactions between Rev1 and Polη, which allowed us to conclude that the Rev1 polymerase can play a backup role in the absence of Polη and that Polη requires its RIR motifs to protect cells from UV-induced DNA damage.

Results
Synergistic interaction between Rev1 polymerase and Polη-interaction mutations. We previously reported that UV damage tolerance mediated by PCNA-Ub fusion is dependent on Rev1 but independent of Polη 37 . Since Rev1 interacts with Polη and Rev7 through its Rev1-CTD that can be further divided into two subdomains 34,35 , we screened a large number of reported point mutations in this region 34,35,38,39 and identified four mutations either specifically affecting Polη but not Rev7 binding (L1170A and V1188A), or disrupting Rev7 but not Polη binding (Y1242A and L1246A). By using an RPA nuclear focus formation assay as an indication of TLS activity after UV irradiation 37,40 , it was found that Rev1-L1170A and Rev1-V1188A protected cells to a level comparable to that of Rev1, while CTD-Y1242A and CTD-L1246A lost Rev1 functions 36 ( Supplementary  Fig. S1), which indicates that DDT provided by Rev1 does not require its physical interaction with Polη.
The above observations are highly unexpected, as it has been well established in both yeast and mammalian cells that Polη plays a critical role in TLS in response to UV irradiation, and that the Rev1-Polη interaction is critical during this process 3,41 . Based on these observations, we wished to test a hypothesis that in the absence of Rev1-Polη interaction, Rev1 uses its own catalytic activity to initiate TLS in response to UV-induced DNA damage. To this end, we cloned Rev1-DE (Rev1-D568A, E569A, polymerase dead), Rev1-L1170A and the corresponding double mutant Rev1-DE-1170 into pEGFP-C1 as GFP fusions. These plasmids were transfected into 293T cells, with vector pEGFP-C1 as a negative control and Rev1-GFP as a positive control, followed by two functional assays as previously described 37 . For the RPA nuclear focus formation assay, typical RPA2-positiveand negative-cells are illustrated in Fig. 1A. Compared with vector-transfected control, ectopic expression of Rev1 can significantly reduce the percentage of cells with RPA2-positive foci after UV irradiation (Fig. 1B,C), and confer UV damage tolerance (Figs. 1D and S2A). Under the above experimental conditions, expression of either Rev1-DE or Rev1-1170 is sufficient to bring the percentage of RPA2-positive cells to the wild-type level (Fig. 1B,C) and confer near wild-type level UV tolerance (Fig. 1D). In sharp contrast, ectopic expression of the Rev1-DE-1170 double mutant did not confer UV tolerance in either assay in comparison to the control transfected cells (Fig. 1). We ruled out the possibility that the lack of DNA-damage tolerance function of Rev1-DE-1170 was due to altered gene expression or protein stability ( Fig. 2A). The above observations collectively allow us to conclude that the Rev1 polymerase and Polη-interaction mutations are synergistic in response to UV-induced DNA damage.
Rev1 and Polη play alternative roles at the insertion step of TLS in bypassing UV-induced lesions. The Rev1-L1170A mutation likely affects interaction with all three Y-family polymerases, namely Polη, Polι and Polκ 32,33 . Since Polη plays a critical role in cellular response to UV irradiation, we hypothesized that the Rev1 catalytic activity is to back up Polη in the Rev1-L1170A background. To test this hypothesis, we asked whether compromised Polη could replace the Rev1-L1170A mutation. To this end, we depleted the endogenous Polη by siRNA to approximately 16% of the wild-type level (Fig. S3, also see Fig. 4B) while expressing GFP-REV1 and its various mutations. GFP-REV1 and its mutants expressed equally well in siPolη cells in comparison to non-specific siNC cells (Figs. 2A and S3), while siPolη specifically reduced the transcript level of the POLH gene, but not other relevant genes encoding Rev1 and Polζ subunits (Fig. 2B). After depletion of Polη from 293T cells, the percentage of RPA2-positive cells almost doubled that of the control group, while ectopic expression of Rev1 or Rev1-1170 caused a decrease in RPA2-positive cells to the same extent. Interestingly, expression of Rev1-DE alone provided UV resistance to siNC cells, but failed to protect Polη-depleted cells, in which the percentage of RPA2-positive cells was similar to that of Rev1-DE-1170-transfected cells (Fig. 2C). The above observations indicate that the Polη depletion is epistatic to Rev1-1170A and additive to Rev1-DE. The additive effect between Polη depletion and the Rev1-DE mutation was also seen in a 20 J/m 2 UV-induced cell survival assay, in which Rev1-DE-transfected cells behave like Rev1 after siNC treatment, but like empty vector after siPolη treatment (Fig. 2D,E). Based on the above observations, we infer that Rev1 and Polη play alternative roles at the insertion step of TLS upon UV irradiation. UV damage tolerance conferred by Polη is partially dependent on its interaction with Rev1. Our observations that the Rev1 interaction with Polη is dispensable appear to contradict a notion of functional importance of physical interaction between Rev1 and other Y-family polymerases during TLS 3,41 . One possibility is that our study was under the Rev1 ectopic expression condition, in which excessive Rev1 is sufficient to provide backup catalytic activity during TLS. It has been previously reported that Polη interacts with Rev1 through residues 369-491 20 and 509-557 33 , designated as RIR1 and RIR2, respectively (Fig. 3A), and that an FF motif is critical for this interaction 22  www.nature.com/scientificreports/ tal conditions, ectopic expression of GFP-POLH and its mutant forms resulted in approximately sixfold more GFP-Polη over endogenous Polη, as judged by western blot analysis (Fig. 3B). GFP-Polη transfection reduced UV-induced RPA2-positive cells (Fig. 3C,D) and protected cells from killing by UV (Figs. 3E and S2B). We then assessed whether the UV damage tolerance conferred by Polη was dependent on its interaction with Rev1. The Polη-RIR1 mutation seemed to have a moderate effect on the resistance provided by Polη. In contrast, the Polη-RIR2 mutation had a dramatic effect on the Polη function in both RPA2 foci (Fig. 3C,D) and cell survival (Fig. 3E) assays. When T-REx-293 cells were transfected with Polη in which both RIR motifs were mutated, the UV-induced RPA2-positive cells further increased over the level in Polη-RIR2 transfected cells (Fig. 3C,D), indicating that Polη-RIR1 contributes moderately to Rev1 binding. The above observations collectively allow us to conclude that the Polη-Rev1 interaction is mainly through the Polη-RIR2 motif, and that UV damage tolerance conferred by Polη overexpression appears to be partially dependent on its interaction with Rev1. www.nature.com/scientificreports/ POLH defective cells were sensitive to UV-induced DNA damage. Loss of Polη activity is responsible for the XPV cells 13,14 . To further investigate the role of Rev1 and Polη in TLS, we established POLH-inactivated cell lines by knocking out the XPV/POLH gene from 293T cells using a CRISPR/Cas9 method [42][43][44] . One of the cell lines, POLH-1, contains a homozygous 2-bp deletion at the second exon, causing a frameshift mutation (Fig. 4A). A western blot analysis compared endogenous Polη levels in 293T cells, siPolη-treated cells and the isogenic POLH-1 cells. While siPolη treatment reduced cellular Polη by 84%, Polη is undetectable in the POLH-1 cells (Fig. 4B). Compared with the parental 293T cells, POLH-1 cells displayed a relatively normal proliferation rate in the absence of UV irradiation; however, upon 5 J/m 2 UV irradiation, the POLH-1 cells stopped proliferation over 3 days, whereas the proliferation of 293T cells was only moderately affected (Fig. 4C). 293T cells displayed a characteristic increase in RPA2-positive cells with increasing doses of UV irradiation, reaching approximately 20% at 8 J/m 2 . In contrast, POLH-1 cells dramatically increased RPA2-positive cells to over 40% upon 2 J/

Effects of Rev1 and its mutant derivatives on tolerance to UV irradiation in POLH-1 cells. Pre-
viously, we hypothesized that the catalytic activity of Rev1 plays a role in nucleotide insertion during TLS either when Rev1 cannot interact with Polη or when the endogenous Polη is reduced. However, the above results are subject to a different interpretation, as they were not obtained from a strict genetic system. With the creation of POLH-1 cells, we were able to critically test our original hypothesis in a clean genetic background. Indeed, in comparison to vector-transfected cells, ectopic expression of REV1 in POLH-1 (Fig. 5A) could reduce UVinduced RPA2-positive cells (Fig. 5B,C). Under the above experimental conditions, expression of REV1-L1170A could rescue POLH-1 cells to the wild-type REV1 level, whereas expression of REV1-DE or the double mutation was no longer able to protect POLH-1 cells (Fig. 5B,C). Similarly, expression of REV1 or REV1-L1170A protected POLH-1 cells from killing by UV irradiation to the same level, while expression of REV1-DE or REV1- DE-1170 had no protective effect (Fig. 5D). These results, together with previous observations 36 , clearly show that when REV1 is overexpressed, its Rev7 interaction is absolutely required for cellular tolerance against UV damage, while either its catalytic activity or Polη, but not both, is dispensable.

Effects of Polη and its mutant derivatives on tolerance to UV irradiation in POLH-1 cells. Phys-
ical interaction between Polη and Rev1 may facilitate TLS; however, it is unclear whether Polη-mediated DDT can bypass the requirement for Rev1. To test whether RIR mutations affect the affinity of Polη for Rev1, we co-transfected POLH-1 cells with GFP-Rev1-CTD and Polη-RIR mutant derivatives followed by co-IP against www.nature.com/scientificreports/ GFP and western blot analysis against Polη. Figure 6A shows that, compared to wild-type Polη, the Polη-RIR2 mutation had a stronger effect on binding to GFP-Rev1-CTD than the Polη-RIR1 mutation, and the Polη-RIRD double mutation further reduced its affinity for GFP-Rev1-CTD. Hence, RIR2 appears to play a major role in the Polη-Rev1 interaction. The remaining coimmunoprecipitated Polη may come from indirect interactions, as both Rev1 and Polη interact with PCNA and Ub 3,16,29,31 , or from other putative RIR motifs found in Polη 22 . We transfected plasmids producing GFP-Polη and its RIR mutant forms in POLH-1 cells and monitored their response to UV irradiation. Mutant forms of GFP-Polη did not affect their expression and protein stability in POLH-1 cells under our experimental conditions (Fig. 6B). Expression of POLH in POLH-1 reduced UVinduced RPA2-positive cells to a great extent, and expression of POLH-RIR1 and POLH-RIR2 mutations partially restored the Polη function in POLH-1 cells, while expression of the POLH-RIRD double mutant form of Polη further reduced its rescue ability (Fig. 6C,D). Hence, Polη-RIR1 and Polη-RIR2 mutations appear to be additive in affecting Rev1 interaction, which plays a critical role in TLS. In a cell survival assay, expression of POLH restored www.nature.com/scientificreports/ POLH-1 cell tolerance to 2 J/m 2 UV and expression of POLH-RIR1 had a similar effect. In contrast, expression of either POLH-RIR2 or POLH-RIRD failed to rescue POLH-1 cells from killing by UV irradiation (Fig. 6E), indicating that the RIR2 motif plays a critical role in Rev1 interaction and is absolutely required during TLS in response to UV under our experimental conditions. These observations collectively allow us to conclude that interaction with Rev1 is critical for Polη to function in response to UV irradiation.

Discussion
In mammalian cells, PCNA plays crucial roles in DNA replication and repair 45 . PCNA interacts and travels with all three replicative polymerases during chromosomal DNA replication. When DNA damage stalls the replication fork, PCNA can be ubiquitinated at its K164 residue by Rad6-Rad18, switching to a DDT mode 18 . Monoubiquitinated PCNA enhances affinity for Y-family polymerases 16 including Polη, Polι, Polκ and Rev1, all of which contain PCNA-and Ub-binding domains 46 . In response to UV irradiation, both Polη and Rev1 are colocalized to the damage sites in the form of nuclear foci 20 . Although subject to debate, our own observations 21,36 favor a previous report 20 that they are recruited to the damage site independently from each other, which raises a critical issue: what is the role of Rev1-Polη interaction during TLS? We previously reported that UV damage tolerance conferred by ectopic expression of PCNA-Ub fusion 37 and Rev1 36 depends on Rev1 and its physical interaction with Polζ, respectively, but is independent of Polη. Here we show that ectopic expression of Polη can confer additional UV damage tolerance, which requires its RIR domains. Furthermore, expression of Polη can rescue the increased UV sensitivity in Polη-defective cells that mimic the XPV syndrome. This rescue relies on Polη's physical interaction with Rev1 through RIR motifs. Our observations differ from a previous report 17 that the Polη-RIR mutations do not affect Polη rescue of XPV cells from killing by UV irradiation, but are consistent with a report 47 that expression of a polymerase-dead Polη moderately rescues UV sensitivity of Polη-null mouse cells, which depends on its interaction with Rev1. Although both RIR motifs have been reported to mediate interaction with Rev1 20,33 , we found that RIR2 plays a critical role while RIR1 may play a backup role, although RIR1 is critical for the interaction with PolD2 48 . In the absence of RIR motifs, Polη still retains certain physical interaction with Rev1, probably through cryptical RIR motifs found in Polη 22 , although it is insufficient to support the Polη TLS activity. We propose a matchmaker mechanism in which only when cells sense the presence of both Polη and Rev1 at the same damage site through their physical interaction will insertion by Polη and extension by Rev1-mediated Polζ take place to complete the two-step TLS (Fig. 7). It has been well accepted that in response to UV irradiation, Polη plays a critical catalytic role while Rev1 only plays a scaffold role; the catalytic activity of Rev1 is involved in bypassing abasic sites 49 and lesions induced by 4-nitroquinoline-1-oxide 50-52 , but is dispensable for lesions induced by UV 28 . However, some studies have indicated direct or indirect roles of Rev1 in bypassing CPD and (6-4)PP [53][54][55] . Since UV damage tolerance conferred by Rev1 is independent of its physical interaction with Polη 36 , we critically tested a hypothesis that the Rev1's catalytic activity is responsible for the observed DDT. Firstly, we found a strong synthetic effect between the catalytic and Polη-binding mutations in Rev1. While either the Rev1-DE or Rev1-L1170A mutation can provide DDT to near the wild-type Rev1 level, the double mutation behaves like a null mutation. Secondly, we ruled out a possibility that other TLS polymerases are responsible for the effect by experimental depletion of Polη, and found that once Polη is depleted, the Rev1-L1170 mutation still behaves like wild-type Rev1, while the Rev1-DE mutation is no longer able to provide DDT. Finally, using a newly created POLH-1 cell line, we showed that the polh is additive with the Rev1-DE mutation but epistatic to the Rev1-L1170A mutation. These observations collectively support a notion that Polη plays a frontline role in TLS insertion across from UV-induced lesions and that Rev1 is required for its non-catalytic role, probably through recruiting Polζ 2 . However, in the absence of Polη, Rev1 can insert nucleotide(s) across from UV-induced lesions as well as recruit Polζ 2 to the damage site (Fig. 7). A major critique for the above model is that it is based on ectopic expression of Rev1, which may not occur in untransfected cells. Interestingly, a synergistic interaction between Rev1-DE and Polη in response to UV irradiation is recently reported in budding yeast 56 , which lends strong support to the working model. Rev1 may also facilitate the assembly of Polζ 4 by physical interaction with PolD3 57 , which is likely an active form for TLS extension 58 (Fig. 7). Rev1 is a template-dependent dCMP transferase 51,59 . Since UV-induced lesions are almost exclusively on pyrimidines 60 , the dCMP insertion by Rev1 is expected to cause transversion mutations. Fortunately Rev1 has very limited catalytic activity toward major UV-induced lesions 52 , and it must be kept at bay until needed. For the co-IP assay, DNA sequences corresponding to Rev1-CTD (residues 1150-1249) were PCR-amplified and cloned into pEGFP-C1 to produce EGFP-fused proteins. POLH and its mutant coding sequences were cloned into vector pcDNA4/TO (Invitrogen). www.nature.com/scientificreports/ RNA interference. The depletion of endogenous Polη in 293T cells was performed as previously described 37 .
RPA nuclear focus formation assay. Cultured cells were seeded on poly-lysine-coated cover slips, rinsed once with ice-cold PBS (2.25 g Na 2 HPO 4 , 8 g NaCl, 0.2 g KH 2 PO 4 , 2 g KCl dissolved in 1 L ddH 2 O), treated with 0.4% NP-40 in PBS for 20 min on ice, and then fixed with 4% paraformaldehyde for 15 min. The fixed cells were rinsed 3 times with PBS, treated with methanol for 5 min, and then rinsed 4 times with PBST for 5 min each time. After incubation with 5% FBS in PBST for 45 min, cells were incubated with mouse anti-Replication protein A2 (RPA2) antibody (Abeam, Ab2175, 1:1000) overnight. The cells were washed four times with PBST, incubated with Alexa Flour 546 goat anti-mouse secondary antibody (Invitrogen, A11030, 1:1000) and 1.5 µg/ mL 4′,6-diamidino-2-phenylindole (DAPI) at room temperature for 1 h, and finally washed 4 times with PBST again. For quantitative analysis of UV-induced RPA2 focus formation, the 293T or POLH-1 cells transfected with corresponding plasmids were treated with UV. Images were taken with the same exposure time. Microscopy was performed with an inverted Olympus 10*22 microscope equipped with a 40 × immersion lens, and images were acquired using the CCD RoHs (Q26053) as previously described 37 . At least 1000 cells were counted for each treatment.

Co-immunoprecipitation (Co-IP) and western blotting.
To measure the expression levels of GFP-Rev1 and Polη or their mutant derivatives, 293T cells were transfected with the plasmids. 2 days later, cell lysates were harvested, boiled before SDS-PAGE, and detected by indicated antibodies against Polη (Abcam, ab17725) and the GFP Tag (Abmart, 7G9). The following reference protein antibodies were from Lifetech: GAPDH (GA331), β-Actin (GA321) and β-tubulin (GA311). For the co-IP assay, the POLH-1 cells transfected with GFP-Rev1-CTD and Polη RIR mutants were harvested and immunoprecipitated with GFP-Trap A (ChromoTek, gta-20) overnight. The input and the immunoprecipitated proteins were separated by SDS-PAGE and proteins of interest were detected by indicated antibodies against Polη, β-tubulin and GFP.