H3K4 methylation by SETD1A/BOD1L facilitates RIF1-dependent NHEJ

Abstract

The 53BP1-RIF1-shieldin pathway maintains genome stability by suppressing nucleolytic degradation of DNA ends at double-strand breaks (DSBs). Although RIF1 interacts with damaged chromatin via phospho-53BP1 and facilitates recruitment of the shieldin complex to DSBs, it is unclear whether other regulatory cues contribute to this response. Here, we implicate methylation of histone H3 at lysine 4 by SETD1A-BOD1L in the recruitment of RIF1 to DSBs. Compromising SETD1A or BOD1L expression or deregulating H3K4 methylation allows uncontrolled resection of DNA ends, impairs end-joining of dysfunctional telomeres, and abrogates class switch recombination. Moreover, defects in RIF1 localization to DSBs are evident in patient cells bearing loss-of-function mutations in SETD1A. Loss of SETD1A-dependent RIF1 recruitment in BRCA1-deficient cells restores homologous recombination and leads to resistance to poly(ADP-ribose)polymerase inhibition, reinforcing the clinical relevance of these observations. Mechanistically, RIF1 binds directly to methylated H3K4, facilitating its recruitment to, or stabilization at, DSBs.

Keywords: 53BP1; BOD1L; H3K4 methylation; PARP inhibitors; RIF1; SETD1A; class switch recombination; double-strand break repair; shieldin.

Graphical abstract

Figure 1

BOD1L and SETD1A interact with RIF1 (A) Murine BOD1L-GFP complexes from HeLa cells were analyzed by mass spectrometry. Unique peptide counts of selected hits are shown. (B) HeLa nuclear cell extracts were immunoprecipitated with the denoted antibodies, and inputs and immunoprecipitates analyzed by immunoblotting. (C) Whole-cell extracts (WCEs) of HeLa cells expressing mouse BOD1L-GFP were immunoprecipitated with the denoted antibodies in the presence/absence of hydroxyurea, and inputs and immunoprecipitates analyzed as above. (D) RIF1 was immunoprecipitated from HeLa WCE in the presence/absence of benzonase, and inputs and immunoprecipitates analyzed as above. (E) BOD1L was immunoprecipitated from HeLa WCE, and inputs and immunoprecipitates analyzed as above. (F) HeLa cells were transfected with the indicated siRNAs for 72 h, and WCE analyzed by immunoblotting. Protein levels were quantified by ImageJ and expressed as a ratio compared with control cells. (G) HeLa nuclear cell extracts were incubated with GST or GST-BOD1L fragments, complexes were isolated by glutathione-sepharose and analyzed by immunoblotting. Data in all cases are representative of ≥2 independent experiments.

Figure 2

BOD1L and SETD1A are required for RIF1 recruitment to DSBs and for efficient DSB repair (A and B) HeLa cells were transfected with the indicated siRNAs for 72 h, exposed to ionizing radiation (IR), and immunostained with antibodies to CENPF and RIF1. Representative fluorescence microscopy images are shown (A); scale bars, 10 μm. Foci formation was quantified (B). (C) U-2-OS-FokI cells were transfected with the indicated siRNAs for 48 h, treated with 4-OHT and immunostained with antibodies to CENPF and RIF1. Representative images are shown above (scale bars, 10 μm), and fluorescence intensity per FokI-focus was quantified using ImageJ. Lines denote mean values from three independent experiments. (D) Chromatin isolated from cells in (C) was immunoprecipitated with the denoted antibodies and quantified by qPCR. A schematic of the relative positions of 4 qPCR amplicons is shown, and normalized amounts of RIF1 bound at FokI-induced double-strand breaks in cells across all amplicons is indicated. (E) Patient LCL cells haploinsufficient for SETD1A were exposed to IR, immunostained with an antibody to RIF1, and foci formation enumerated. (F) HeLa cells were transfected with the indicated siRNAs, irradiated, left to form colonies for 14 days, and then stained with methylene blue and colonies counted. (G) HeLa cells were transfected as in (F), exposed to IR, left for 24 h, and micronuclei formation assessed. (H and I) HeLa cells from (F) were irradiated, immunostained with antibodies to γH2AX (H) or 53BP1 (I), and foci formation enumerated. (J and K) Bod1l^F/F and Bod1l^+/+ MEFs were treated with 4-OHT, irradiated, immunostained with antibodies against γH2AX (J) or 53BP1 (K), and foci formation quantified. Plots in all cases represent data from three independent experiments; error bars = mean ± SEM, p values: unpaired two-tailed t tests except (C) (Mann-Whitney) and (F) (two-way ANOVA). ^∗p ≤ 0.05, ^∗∗p ≤ 0.01 and ^∗∗∗p ≤ 0.001. See also Figures S1–S3.

Figure 3

BOD1L and SETD1A suppress DSB resection (A and B) HeLa cells were transfected with the indicated siRNAs, exposed to IR, harvested at the indicated times, and WCE were analyzed by immunoblotting. (C) HeLa cells from (A) and (B) were immunostained with antibodies to CENPF and RPA2, and foci formation enumerated. (D) Bod1l^F/F and Bod1l^+/+ MEFs were treated with 4-OHT, irradiated, immunostained with antibodies against RPA2, and foci formation enumerated. (E) U-2-OS-FokI cells were transfected with the indicated siRNAs, treated with 4-OHT and immunostained with antibodies to CENPF and RPA2. Fluorescence intensity per FokI-focus was quantified using ImageJ. Lines denote mean values from three independent experiments. (F and G) HeLa cells were transfected with the indicated siRNAs, pulsed with IdU for 24 h, exposed to IR for 1 h, and labeled with anti-IdU antibody. Native tract length (F) or native IdU foci (G) were calculated or enumerated. Lines denote mean values from three independent experiments, and representative images are shown (scale bars, 10 μm). (H and I) HeLa cells from (F) were exposed to IR, immunostained with antibodies to either CENPF and RPA2 (H) or CENPF and BRCA1 (I), and foci formation assessed. (J) U-2-OS-FokI cells were transfected with the indicated siRNAs, treated with 4-OHT and immunostained with antibodies to CENPF and BRCA1. Fluorescence intensity per FokI-focus was quantified using ImageJ. Lines denote mean values from three independent experiments. (K) SETD1A patient LCL cells were exposed to IR, immunostained with an antibody to BRCA1, and foci formation enumerated. (L) HeLa cells from (H) were immunostained with antibodies to CENPF and RAD51. Plots in all cases represent data from three independent experiments; error bars = mean ± SEM, p values: unpaired two-tailed t tests except (E, F, G, and J) (Mann-Whitney). ^∗p ≤ 0.05, ^∗∗p ≤ 0.01, and ^∗∗∗p ≤ 0.001. See also Figure S4.

Figure 4

BOD1L, SETD1A, and RIF1 act together to promote NHEJ (A–C) HeLa Kyoto iCas9-TRF2 gRNA cells were transfected with the indicated siRNAs, treated with doxycycline, and the percentage of telomere end-to-end fusions enumerated (A). Representative images of telomere fusions are shown in (B) (scale bars, 10 μm), and WCE were immunoblotted with the indicated antibodies (C). (D) HeLa cells were transfected with the indicated siRNAs and a plasmid expressing dominant-negative TRF2^ΔBΔM, and the percentage of telomere end-to-end fusions enumerated. (E) HeLa cells were transfected with the indicated siRNAs and with CRISPR-Cas9 HR plasmids. Cells undergoing HR expressing fluorescent nuclear lamin A/C were quantified. (F and G) HeLa cells were transfected with the indicated siRNAs, exposed to the indicated doses of Talazoparib or olaparib, left to form colonies for 14 days, stained with methylene blue and colonies counted. (H and I) Cells from (G) were treated as above, incubated with olaparib for 24 h, and radial chromosome formation analyzed by Giemsa staining and light microscopy (H). Alternatively, cells were immunostained with antibodies to CENPF and RAD51, and foci formation enumerated (I). (J) HeLa cells were transfected with the indicated siRNAs, and with CRISPR-Cas9 HR plasmids as in (E). Relative levels of HR were enumerated. (K) HeLa cells were transfected with the indicated siRNAs, exposed to IR, immunostained with antibodies to CENPF and RIF1, and foci formation quantified. (L) U-2-OS cells were transfected with the indicated siRNAs for 72 h, pulsed for 20 min each with CldU and IdU, and exposed to 4 mM HU for 5 h (as in the schematic). DNA was visualized with antibodies to CldU and IdU, and tract length was calculated. Graph denotes average ratios of IdU:CldU label length. Plots in all cases represent data from three independent experiments; error bars = mean ± SEM, p values: one-way (A) and two-way (F and G) ANOVA; unpaired two-tailed t tests (D, E, I, J, and K); Mann-Whitney (H and L). ^∗p ≤ 0.05, ^∗∗p ≤ 0.01 and ^∗∗∗p ≤ 0.001.

Figure 5

BOD1L is required for CSR (A) Serum from Bod1l^+/+Cd19^+/Cre and Bod1l^F/FCd19^+/Cre mice was isolated and immunoglobulins quantified by ELISA. (B and C) CD19⁺ B cells were isolated from Bod1l^F/FR26^+/+ or Bod1l^F/FR26^CreERT2/+ mice and stimulated in vitro with the indicated factors for 96 h. Relative levels of CSR were quantified (B), and the quantity of immunoglobulins produced measured by ELISA (C). (D) CD19⁺ B cells were isolated from Bod1l^+/+Cd19^+/Cre and Bod1l^F/FCd19^+/Cre mice and stimulated in vitro with the indicated factors for 96 h. Relative levels of CSR were quantified. (E and F) Mice were immunized with NP-CGG, and NP-specific IgM or IgG were quantified in serum at the indicated time points after immunization. Plots in all cases represent data from n = 3 mice; error bars = mean ± SEM, p values: unpaired two-tailed t tests. ^∗p ≤ 0.05, ^∗∗p ≤ 0.01, and ^∗∗∗p ≤ 0.001. See also Figure S5.

Figure 6

H3K4 methylation by SETD1A is required for RIF1-dependent DNA repair (A) H3-GFP WT and K4A cells were exposed to ionizing radiation (IR), left to form colonies for 10 days, stained with methylene blue and colonies counted. (B and C) Cells from (A) were immunostained with antibodies to CENPF and RIF1 (B) or CENPF and 53BP1 (C), and foci formation enumerated. (D) Quantification of proximity ligation assay (PLA) signals between γH2AX and RIF1 in H3-GFP WT and K4A cells. (E–G) Cells from (A) were immunostained with antibodies to CENPF and RPA2 (E), CENPF and BRCA1 (F), or CENPF and REV7 (G) and foci formation assessed. (H) H3-GFP WT and K4A cells were transfected with the indicated siRNAs, exposed to olaparib, left to form colonies for 10 days, stained with methylene blue and colonies counted. (I and J) H3-GFP WT and K4A cells were transfected with the indicated siRNA, exposed to IR, and immunostained with antibodies to CENPF and RIF1 (I), or CENPF and RPA (J), and foci formation enumerated. (K) HeLa cells were transfected with constructs expressing WT or H483A KDM5A, and immunostained with antibodies to either CENPF and RIF1 or CENPF and BRCA1. Representative images are shown in Figure S6G. (L and M) U-2-OS cell lines bearing inducible full-length (FL) SETD1A or a variant lacking the SET (ΔSET) domain were transfected with the indicated siRNAs, exposed to doxycycline where denoted, and exposed to IR. RIF1 or BRCA1 foci formation was then quantified as above. (N and O) U-2-OS-FokI cells were treated with 4-OHT and/or transfected with the indicated siRNAs, chromatin isolated and ChIP was performed with the indicated antibodies. Data represent the average signal across the 4 amplicons represented in the schematic normalized to input. Plots in all cases represent data from at least three independent experiments; error bars = mean ± SEM, p values: unpaired two-tailed t tests except (A) (two-way ANOVA) and (D) (Mann-Whitney). ^∗p ≤ 0.05, ^∗∗p ≤ 0.01, and ^∗∗∗p ≤ 0.001. See also Figure S6.

Figure 7

RIF1 associates with methylated H3K4 in vitro and in vivo (A) Chromatin immunoprecipitation profiles of murine RIF11 at H3K4me1 and H3K4me3 peak sites in mESCs from ENCODE. Data are from Foti et al. (2016). (B and C) Chromatin immunoprecipitation profiles of murine RIF1 and BLISS signals over H3K4me3-positive areas in mESCs that lie outside areas defined as TSS by ENCODE. Data are from Yan et al. (2017). (D) HeLa nuclear cell extracts were incubated with biotinylated histones and analyzed by immunoblotting. (E and F) In vitro transcribed/translated RIF1 was incubated with biotinylated histones and analyzed by immunoblotting. (G) HEK-293 cells were transfected with plasmids expressing the indicated RIF1 proteins, left for 48 h, and cell lysates incubated with purified biotinylated histones and analyzed by immunoblotting. Data in (D)–(G) represent ≥2 independent experiments. (H) (Upper) Upon DSB formation in G1, pre-existing H4K20me2 and H2AK15Ub recruit 53BP1 to sites of DSBs. H3K4me catalyzed by SETD1A and its co-factor BOD1L (1 and 2) stabilize the recruitment of RIF1 (3), allowing downstream cascades (not shown). This suppresses BRCA1-dependent resection and promotes NHEJ (4). (Lower) In the absence of H3K4me (5), RIF1 recruitment is destabilized (6), leading to inappropriate BRCA1-dependent resection of DSBs (7) and mis-repair. See also Figure S7.