RAD18 directs DNA double-strand break repair by homologous recombination to post-replicative chromatin

Abstract

RAD18 is an E3 ubiquitin ligase that prevents replication fork collapse by promoting DNA translesion synthesis and template switching. Besides this classical role, RAD18 has been implicated in homologous recombination; however, this function is incompletely understood. Here, we show that RAD18 is recruited to DNA lesions by monoubiquitination of histone H2A at K15 and counteracts accumulation of 53BP1. Super-resolution microscopy revealed that RAD18 localizes to the proximity of DNA double strand breaks and limits the distribution of 53BP1 to the peripheral chromatin nanodomains. Whereas auto-ubiquitination of RAD18 mediated by RAD6 inhibits its recruitment to DNA breaks, interaction with SLF1 promotes RAD18 accumulation at DNA breaks in the post-replicative chromatin by recognition of histone H4K20me0. Surprisingly, suppression of 53BP1 function by RAD18 is not involved in homologous recombination and rather leads to reduction of non-homologous end joining. Instead, we provide evidence that RAD18 promotes HR repair by recruiting the SMC5/6 complex to DNA breaks. Finally, we identified several new loss-of-function mutations in RAD18 in cancer patients suggesting that RAD18 could be involved in cancer development.

Graphical Abstract

Figure 1.

RAD18 negatively regulates 53BP1 at DSBs by recognizing H2A ubiquitinated by RNF168. (A) Relative NHEJ and HR repair in traffic light reporter U2OS cells treated with control (NC) or RAD18 siRNA. The repair efficiency was normalized to the total repair efficiency in the siNC treated cells (mean + SD, n = 4, two-tailed t-test). (B) High-content microscopy quantification of RAD51 foci in EdU positive U2OS parental or RAD18–KO cells upon IR-irradiation (mean + SD, n = 3, two-way ANOVA). (C) Quantification of ssDNA by native BrdU staining in U2OS parental or RAD18–KO cells irradiated with 10 Gy (mean + SD, n= 3, two-way ANOVA). (D) Quantification of 53BP1 foci in U2OS parental or RAD18–KO cells upon IR-irradiation (mean + SD, n = 3, two-way ANOVA). (E) Quantification of 53BP1 foci in U2OS parental cells and RAD18–KO cells transfected with wild-type EGFP–RAD18–WT, EGFP–RAD18–ΔUBZ or EGFP empty vector control. Where indicated, cells were IR-irradiated 2 h before fixation (mean + SD, n = 3, two-tailed t-test). (F) U2OS cells were treated with control, RNF8, or RNF168 siRNA for 48 h, and stained for RAD18. Where indicated, cells were IR-irradiated 2 h before fixation. The scale bar corresponds to 10 μm. (G) Quantification of (F), RAD18 foci count is shown (mean + SD, n = 3, two-tailed t-test). (H) U2OS cells were treated with control or RNF8 siRNA 48 h and transfected with FLAG-RNF168 plasmid 19 h before IR-irradiation. Cells were fixed after 2 h and stained for FLAG and RAD18. The scale bar corresponds to 10 μm. (I) Quantification of (H), RAD18 foci count in the FLAG positive and negative cells is shown (mean + SD, n = 3, two-tailed t-test). (J) Co-immunoprecipitation of Ub-H2A form HEK293 cells transfected with EGFP–RAD18–WT, EGFP–RAD18–ΔUBZ or EGFP empty vector control. Where indicated, plasmids were co-transfected with FLAG–RNF168 plasmid. Cells were lysed 24 h after transfection, treated with bensonase and incubated with GFP trap. Co-precipitated H2A was analyzed using immunoblotting. A representative experiment from two repeats is shown. (K) Schematic representation of variants in RAD18 protein sequence. Bottom, a heatmap showing phenotype of RAD18 variants in four different assays (from Supplementary Figure S3). The RAD18-ΔUBZ is included in the scheme as a negative control. The color legend is normalized to range from 0 (empty EGFP vector) to 1 (wild-type RAD18).

Figure 2.

RAD18 excludes 53BP1 to the periphery of IR-induced foci. (A) STED images of U2OS cells fixed 3 h after IR-irradiation with 2 Gy and stained with indicated antibodies (scale bars 10 μm and 200 nm). (B) Quantification of A, number of RAD18 and 53BP1 nanodomains in individual repair foci. (C) Quantification of (A), mean distance of RAD18 and 53BP1 nanodomains from the foci center. For (B) and (C), the mean ± SD is shown, n ≥ 311 foci from 26 cells from two independent experiments. (D) Quantification of (E), minimal distance of RAD18 and 53BP1 nanodomains from the foci center. The mean ± SD is shown, Mann–Whitney test, n ≥ 300 foci form 26 cells from two independent experiments. (E) STED images of DNA repair foci in parental U2OS and RAD18–KO cells fixed 3 h after IR-irradiation with 2 Gy and stained with BRCA1 and 53BP1 antibodies. Bottom, signal average is shown of 300 foci (scale bar 200 nm). (F) Quantification of E, 53BP1 nanodomains count in individual repair foci. The mean ± SD is shown, Mann–Whitney test, n ≥ 300 foci form 26 cells from two independent experiments. (G) Model showing arrangement of BRCA1 (brown), RAD18 (green) and 53BP1 (red) nanodomains at chromatin loops formed around DSB.

Figure 3.

Auto-ubiquitination of RAD18 interferes with its localization to DSBs. (A) Quantification of EGFP–RAD18 foci in U2OS RAD18–KO cells transfected with EGFP–RAD18–WT or EGFP–RAD18–ΔRAD6BD. Cells were IR-irradiated 2 h before fixation (mean + SD, n = 3, two-tailed t-test). (B) Quantification of RAD18 foci in U2OS cells treated with control or RAD6 siRNA for 48 h. Where indicated, cells were IR-irradiated 2 h before fixation (mean + SD, n = 3, two-tailed t-test). (C) Quantification of (D), RAD18 foci in U2OS cells transfected or not with RAD6–WT–FLAG or catalytically inactive RAD6–C88A–FLAG mutant. Where indicated, cells were IR-irradiated 2 h prior fixation (mean + SD, n = 3, two-tailed t-test). (D) U2OS cells were transfected with RAD6–WT–FLAG or RAD6–C88A–FLAG. Where indicated, cells were IR-irradiated 2 h prior fixation. The scale bar corresponds to 10 μm. (E) U2OS cells were transfected with EGFP–RAD18–WT or EGFP–RAD18–4KR mutant. Where indicated, cells were co-transfected with RAD6–FLAG plasmid. RAD18 and UbRAD18 levels were analyzed using immunoblotting. A representative experiment form two independent repeats is shown. (F) Quantification of EGFP–RAD18 foci in U2OS RAD18–KO cells transfected with EGFP–RAD18–WT or EGFP–RAD18–4KR. Where indicated, cells were IR-irradiated 2 h before fixation (mean + SD, n = 3, two-tailed t-test). The scale bar in the representative image corresponds to 10 μm. (G) HEK293 cells were transfected with a plasmid coding for RAD6–FLAG (top) or co-transfected with FLAG–H2A and RNF168 plasmids (bottom). Cells were lysed 24 h after transfection, treated with bensonase and immunoprecipitated with FLAG antibody. Co-precipitated RAD18 and UbRAD18 were analyzed using immunoblotting. Asterisk indicate a non-specific signal of antibody light chains. Right, immunoblot quantification of UbRAD18/RAD18 ratio (mean ± SD, n = 3, two-tailed t-test).

Figure 4.

SLF1 promotes RAD18 binding to the post-replicative chromatin. (A) Quantification of RAD18 foci in U2OS fixed 2 h upon IR-irradiation. Cell cycle phases were distinguished based on the total DAPI and EdU nuclear intensities. The foci count was normalized to the total DAPI nuclear intensity of 2n cells (mean ± SD, n = 514). (B) Quantification of mean RAD18/53BP1/γH2AX intensity in γH2AX foci plotted against the total nuclear DAPI intensity.) RPE-iCut cells were treated overnight with Shield-1 and doxycycline to induce expression of Cas9 endonuclease that was targeted to DNA upon transfection of mixture of 12 single cutter sgRNAs. Cells were fixed 6 h post transfection. Color legend shows cell cycle phases distinguished by gating on the total EdU and DAPI nuclear intensities (mean ± SD, n ≥ 4700). A representative experiment form two independent repeats is shown. (C) U2OS cells transfected with wild-type EGFP-SLF1 or ΔBRCT and ΔARD variants were fixed 2 h after IR-irradiation. The scale bar in the representative image corresponds to 10 μm. Right, quantification shows EGFP-SLF1 foci count in G1 and S cell cycle phases distinguished by gating on the total EdU and DAPI nuclear intensities (mean + SD, n = 3, two-tailed t-test). (D) HEK293 cells were transfected with the wild-type EGFP-SLF1 or ΔBRCT and ΔARD variants, or EGFP empty vector control. Cells were lysed 24 h after transfection, treated with bensonase, incubated with GFP trap and extensively washed. Immunoprecipitated EGFP-SLF1 was then incubated with acid histone extracts. Pulled-down H4K20me0 was analyzed using immunoblotting. A representative experiment form two independent repeats is shown. (E) Quantification of RAD18 foci count in EdU positive U2OS cells treated with control or two different SLF1 siRNAs for 48 h. Where indicated, cells were IR-irradiated 2 h before fixation (mean + SD, n = 3, two-tailed t-test). (F) Quantification of RAD18 and 53BP1 foci count in U2OS cells treated with control or SETD8 siRNA for 24 h. Where indicated, cells were IR-irradiated 2 h before fixation. G1 cells were distinguished by gating on the total EdU and DAPI nuclear intensities (mean + SD, n = 3, two-tailed t-test). (G) Quantification of EGFP-SLF1 foci in G1 cells transfected with EGFP-SLF1 and treated as in F and G (mean + SD, n = 3, two-tailed t-test). (H) Quantification of EGFP–RAD18 foci count in U2OS–RAD18–KO cells transfected with EGFP–RAD18–WT or EGFP–RAD18–S442/444A and fixed 2 h after IR-irradiation. G1 and S cells were distinguished by gating on the total EdU and DAPI nuclear intensities (mean + SD, n = 3, two-tailed t-test). (I) PLA quantification of RAD18–GFP–SLF1 interaction. U2OS cells were transfected with SLF1 variants and IR-irradiated as indicated 2 h prior fixation. Cells were then probed for PLA with RAD18 and GFP antibodies. Relative total nuclear PLA intensity is shown that is normalized to intensity of SLF1–WT in S cells (mean + SD, n = 3, single-sample two-tailed t-test). (J) Quantification of RAD18 foci in EdU positive U2OS cells transfected with wild-type EGFP-SLF1 or ΔBRCT and ΔARD variants and IR-irradiated 2 h before fixation (mean + SD, n = 3, two-tailed t-test). (K) Colocalization of RAD18 and EGFP with EdU in U2OS cells transfected with EGFP-SLF1 variants. Colocalization in early S cells was evaluated by Pearson's correlation coefficient. At least 26 cells per condition were analyzed in three independent experiments (mean + SD, two-tailed t-test).

Figure 5.

RAD18 promotes recruitment of SMC5 to DNA lesions. (A) Relative HR (right) and NHEJ (left) repair in traffic light reporter U2OS cells treated with indicated siRNAs. The repair efficiency was normalized to the total repair efficiency in control siRNA treated cells (mean + SD, n ≥ 3, two-tailed t-test). (B) Relative proliferation of parental U2OS and RAD18–KO cells treated with control or 53BP1 siRNA was evaluated using resazurin viability assay 7 days after IR-irradiation with indicated doses (mean ± SD, n ≥ 3, two-way ANOVA). (C) Colocalization of SMC5 with γH2AX foci in parental U2OS and RAD18–KO. Cells were incubated with EdU prior IR-irradiation, pre-extracted and fixed at indicated time points. Incorporated EdU was Click-iT-labelled and cells were stained for SMC5 and γH2AX. EdU positive cells are shown (arrows indicate SMC5 foci colocalizing with γH2AX foci, scale bar 10 μm). Right, quantification of cell fraction with >1 SMC5 foci colocalizing with γH2AX foci. EdU positive cells were analyzed (mean + SD, n = 3, two-tailed t-test). (D) Quantification of the mean nuclear SMC5 intensity in parental U2OS and RAD18–KO cells. Where indicated, cells were pre-extracted prior fixation (mean ± SD, n = 300, Mann–Whitney test). (E) Relative HR and NHEJ repair in traffic light reporter U2OS cells treated with indicated siRNAs. The repair efficiency was normalized to the total repair efficiency in control siRNA treated cells (mean + SD, n ≥ 3, two-tailed t-test). (F) Quantification of RAD51 foci in U2OS cells treated with indicated siRNAs for 48 h. Cells were incubated with EdU prior IR-irradiation, pre-extracted and fixed after 6 h. Incorporated EdU was Click-iT-labelled and cells were stained for RAD51. EdU positive cells were analyzed (mean + SD, n = 3, two-tailed t-test). (G) Relative proliferation of parental U2OS and RAD18–KO cells treated with control or SMC5 siRNA was evaluated using resazurin viability assay 7 days after IR-irradiation with indicated doses (mean ± SD, n ≥ 3, two-way ANOVA). (H) Model describing RAD18 function at DSBs. RAD18 is recruited to the post-replicative chromatin by bimodal recognition of H2AK15Ub and H4K20me0 with its UBZ domain and with ARD domain of its interacting partner SLF1, respectively. RAD18 then inhibits NHEJ by limiting 53BP1 activity and promotes HR by recruiting the SLF2/SMC5/6 complex. RAD18 accumulation at DSBs is inhibited by RAD6-mediated auto-ubiquitination.