53BP1 cooperation with the REV7-shieldin complex underpins DNA structure-specific NHEJ

Abstract

53BP1 governs a specialized, context-specific branch of the classical non-homologous end joining DNA double-strand break repair pathway. Mice lacking 53bp1 (also known as Trp53bp1) are immunodeficient owing to a complete loss of immunoglobulin class-switch recombination^1,2, and reduced fidelity of long-range V(D)J recombination³. The 53BP1-dependent pathway is also responsible for pathological joining events at dysfunctional telomeres⁴, and its unrestricted activity in Brca1-deficient cellular and tumour models causes genomic instability and oncogenesis^5-7. Cells that lack core non-homologous end joining proteins are profoundly radiosensitive⁸, unlike 53BP1-deficient cells^9,10, which suggests that 53BP1 and its co-factors act on specific DNA substrates. Here we show that 53BP1 cooperates with its downstream effector protein REV7 to promote non-homologous end joining during class-switch recombination, but REV7 is not required for 53BP1-dependent V(D)J recombination. We identify shieldin-a four-subunit putative single-stranded DNA-binding complex comprising REV7, c20orf196 (SHLD1), FAM35A (SHLD2) and FLJ26957 (SHLD3)-as the factor that explains this specificity. Shieldin is essential for REV7-dependent DNA end-protection and non-homologous end joining during class-switch recombination, and supports toxic non-homologous end joining in Brca1-deficient cells, yet is dispensable for REV7-dependent interstrand cross-link repair. The 53BP1 pathway therefore comprises distinct double-strand break repair activities within chromatin and single-stranded DNA compartments, which explains both the immunological differences between 53bp1- and Rev7- deficient mice and the context specificity of the pathway.

Extended Data Figure 1. CSR characterisation in Rev7 conditional-knockout mice

(A) PCR amplicons from genomic DNA obtained by ear biopsies or purified splenic B cells (left panel), or flow cytometry sorted cells from Hardy Fractions A, B, and C (right panel) from mice with the indicated genotype. Bands of different size correspond to the Rev7^tm1c allele (Cond=475bp), Rev7⁺ allele (WT=314bp) and Rev7^tm1d allele (Flox= 255bp). Representative data; n>3 experiments. (B) Western blot analysis of REV7 protein expression in splenic B cell isolated from mice with the indicated genotype. Representative data; n=2 experiments. For gel source data, see Supplementary Figure 1. (C) CTV-labelled purified B cells were stimulated as indicated and stained for surface IgG1 (left panel) or IgG2b (centre panel) or IgG3 (right panel) on day 4. Representative of n>6experiments.

Extended Data Figure 2. B cell lineage developmental differences in Rev7- and 53bp1- deficient mice.

(A) Flow cytometry analysis of B cell development in the bone marrow (BM) of Rev7^+/+Mb1^+/Cre, Rev7^F/FMb1^+/Cre, and 53bp1^-/- Mb1^+/Cre mice; gating on B220⁺CD43⁺ (left panels, Hardy Fractions A-B-C) and on B220⁺CD43^- (right panels, Hardy Fractions D-E-F). Representative data; n>8 experiments. (B) Apoptotic indices of total, pro/pre-, immature, and mature B cell fractions in the BM (left panel) and spleen (right panel) of Rev7^+/+ Mb1^+/Cre (n=4), Rev7^F/F Mb1^+/Cre (n=4), 53bp1^-/- Mb1^+/Cre (n=2) mice. (C) Flow cytometric sub-classification of mature splenic B cell fractions in Rev7^+/+ Mb1^+/Cre (n=8), Rev7^F/F Mb1^+/Cre (n=8), 53bp1^-/- Mb1^+/Cre(n=9) mice. p-values, unpaired Students two-tailed t-test. Bars represent mean ± 95% CI. (D) Top panel: Indicated populations of pooled Pro- to Pre- B cell stage BM lymphocytes (B220⁺ CD43⁺; Hardy Fr. A-B-C) from n=2 mice per genotype were FACS sorted and used to generate whole cell extracts. Bottom panel: Immunoblot shows an absence of REV7 protein in extracts prepared from Rev7^F/F Mb1^+/Cre experimental BM, when compared to extracts prepared from Rev7^F/F Mb1^+/+ (no Cre) controls, yet equivalent levels of 53BP1, Histone H3 (loading control), and total protein (Ponceau S stain). For gel source data, see Supplementary Figure 1. (E) Left panel: Diagram of the mixed bone marrow chimera transplantation experiment. Bone marrow cells from a wild-type CD45.1⁺ donor mouse were combined with an equal number of bone marrow cells from an experimental CD45.2⁺ donor mouse, and injected into lethally irradiated recipient CD45.1⁺ mice (n=8 per genotype). After 8 weeks, the recipient bone marrow was analysed for the relative contribution of CD45.1⁺ or CD45.2⁺ cells to reconstitute the recipient mice. Right top panels: Enumeration of B cell precursors (as per Fig. 1g) of CD45.1⁺ (white circles) or CD45.2⁺ (from Rev7^F/FMb1^+/Cre, or 53bp1^-/- Mb1^+/Cre mice; black circles) cells in the bone marrow following reconstitution. Right bottom panel: stage-specific ratios of CD45.1⁺ to CD45.2⁺ grafted B cells for indicated mixed chimeras. In parallel, an additional control experiment involving wild type CD45.1 and Mb1^+/Cre mixed chimera was performed, resulting in equal CD45.1:CD45.2 reconstitution (data not shown). p­values, Multiple t-test with Holm-Sidak correction; Mean ± 95% CI.

Extended Data Figure 3. REV7 protein expression and CSR in complemented Rev7^-/- CH12-F3.

(A) Stable Flag-HA-REV7 protein expression in indicated complemented CH12-F3 cell-lines as determined by western blotting with HA-specific antibody. Tubulin serves as loading control. Representative data; n>3 independent experiments. For gel source data, see Supplementary Figure 1. (B) Flow-cytometric plots of CSR data in Fig. b. Representative data; n>3 independent experiments. (C) Chromatin prepared from indicated CH12-F3 lines 30 h following stimulation, was subjected to ChIP with RPA34 or histone H3 (control) antibodies. Schematic, mouse Igh locus with positions of qPCR amplicons. ChIP recoveries were quantified against input DNA. Representative data, n=2 independent experiments. Mean ± SD.

Extended Data Figure 4. Distinct REV7 residues mediate its functions in ICL repair.

(A) Indicated REV7 mutants were tested for their interaction with the REV3L RBM₁ by yeast-2-hybrid. A fragment of REVL3 (aa^1775-2200) with RBM₂ mutated (P1996A, P2001A). 5-fold dilutions, n=4 biological repeats. (B) Representative histograms of indicated CH12-F3 lines, mock or MMC treated for 8 and 24 h. Numbers depict sub-G1 and G2/M populations as a proportion (%) of total events. Representative data, n=3 independent experiments. (C) Representative PI/DAPI stained metaphases from indicated CH12-F3 cell-lines following 24 h incubation with MMC; n=2 independent experiments. (D) Blind quantification of DSBs (chromatid and chromosome-type) and radials in indicated cell lines, with each metaphase scored as a single point. n=2 independent experiments, each with 50 metaphases scored per genotype and condition. Bars indicate mean. (E) Cell cycle of indicated CH12-F3 lines following MMC treatment. Proportions (%) of sub-G1, G1, S, and G2/M events. n=3 independent experiments ± SD.

Extended Data Figure 5. Identification of the REV7-Shieldin complex.

(A) N-terminal SHLD3/FLJ26957 sequences contain conserved REV7-binding motifs (RBM₁ and RBM₂). Pairwise alignment of REV3L RBM1 with SHLD3 RBM₁ and RBM₂ across multiple species. Black and grey shading indicate identical and similar residues, respectively. (B) Surface representation of REV7 crystal structure (RCSB pdb: 4FJO), pseudo-coloured according to amino-acid conservation. V, variable, and C, conserved residues. (C) Protein threading model of FAM35A identifies an RPA70-like triple OB-fold architecture. The FAM35A structural model generated using Phyre2 predicts amino acid residues 429-826 (coloured in green) adopt the same folds as Ustilago maydis RPA70 (pdb: 4GOP, chain C; 99.5% confidence in the model). Blue structural chains from RPA70 are inserted in regions of FAM35A that could not be modelled with high confidence (internal and FAM35A residues 826+). In the Shieldin complex, RPA34 (opaque red) and RPA14 (opaque yellow) subunits of the heterotrimeric RPA complex might be substituted with SHLD1, SHL3 and REV7 proteins. (D) Pairwise alignment shows 16% sequence identity across human FAM35A isoform 2 (OB-folds 1-3) and Ustilago maydis RPA70 (OB-folds 2-4; generated with Phyre2). Black and grey shading indicate identical and similar residues, respectively.

Extended Data Figure 6. SHLD3 (FLJ26957) mediates 53BP1-dependent NHEJ.

(A) Immunoblot showing levels of REV7, 53BP1, and REV7 protein in whole cell lysates prepared from indicated CH12-F3 cell lines. Representative of n=2 individual experiments. For gel source data, see Supplementary Figure 1. (B) Normal proliferation of stimulated wild type and Shld3^-/-, Shld1^-/- and Shld2^-/- CH12-F3. CFSE dye dilution assay. n=2 independent experiments. (C) Summary of IgM to IgA CSR frequencies in indicated control GST or HA-SHLD2(mouse) -complemented Shld2^-/- CH12-F3. Data normalized to CSR in WT cells. n=4 independent experiments; mean ± SD. (D) IgM to IgA CSR in indicated CH12-F3 Shld3^-/- ± complementation with wild-type SHLD3 or SHLD3^P53A-P58A (RBM2 mutated). Data normalized to CSR in WT cells. Mean ± SD (n=4). (E) Summary of IgM to IgA CSR frequencies in Rev7^-/- and Rev7^-/- Shld3^-/- double knockout CH12-F3 clones. Data normalized to CSR in WT cells. n=4 independent experiments; mean ± SD. (F) Histone H3 (control) ChIP efficiencies at indicated Igh and control loci. Related to RPA-ChIP data in Fig. 4d. Representative data, n=2 independent experiments. Mean ± SD, 2-3 qPCR replicates. (G) Indicated CH12-F3 lines stimulated with CIT (30 h) were subjected to ChIP experiments with RPA34 and histone H3 (control) antibodies. Representative data, n=2 independent experiments. Mean ± SD, 2-3 qPCR replicates. (H) Cell cycle of indicated CH12-F3 lines following MMC treatment. Proportions (%) of sub-G1, G1, S and G2/M events. n=3 independent experiments; mean ± SD. (I) Change (%) of Cas9-dependent indels at the indicated sgRNA locus in KB1P-G3 cells after outgrowth in DMSO or Olaparib (300 nM) for 7 days. Representative data, 2 independent experiments. (J) As in Extended Data Fig. 6i but with 53bp1^-/- KB1P-G3. Representative data, n=2 independent experiments. (K) As in Fig. 4e but with 53bp1^-/- KB1P-G3. Representative data, n=3 independent experiments.

Figure 1. REV7 and 53BP1 cooperate during CSR, yet are functionally uncoupled in V(D)J recombination.

(A) Serum immunoglobulin in Rev7^+/+Mb1^+/Cre and Rev7^F/FMb1^+/Cre mouse cohorts; n=11 mice per genotype; p values, unpaired two-tailed t-test. Mean ± 95% CI. (B) NP-specific serum IgM (left) and IgG (right) at indicated times after NP-CGG immunisation; Representative data, n=2 independent experiments, each with 4 mice. Mean ± 95% CI. (C) Cell trace violet (CTV)-labelled splenic B cells were stimulated as indicated and stained for surface IgG1 or IgE on day 4. Representative data, n>6 mice. (D) CTV dilution in purified B cells cultured in the presence of LPS and IL-4 for 96 hours. Representative data, n>6 mice. (E) Splenic B cells cultured with the indicated stimuli (96 h) and stained for surface IgG1, IgE, IgG2, IgG3. n=4 mice per genotype. 100% CSR, mean Ig isotype switch frequency of 2 control animals in each experiment. p values, two-way ANOVA with Tukey’s correction. Mean ± 95% CI. (F) Absolute numbers of B220+ B cells in the BM (1 femur plus 1 tibia) and spleen. n=9 mice per genotype, except Rev7^+/+Mb1^+/Cre (n=8). p values, unpaired two-tailed t-test, Mean ± 95% CI. (G) Absolute numbers of B cell precursors (Hardy fractions A, B220+CD43+BP-1-CD24-; B, B220+CD43+BP-1-CD24+; C, B220+CD43+BP-1+CD24+; D, B220+CD43-IgM-IgD-; E, B220+CD43-IgM+IgD- ; and F, B220+CD43-IgM+IgD+) in the BM (one femur and one tibia) from Rev7^+/+Mb1^+/Cre (n=8), Rev7^F/FMb1^+/Cre (n=9), 53bp1^-/- Mb1^+/Cre (n=9) mice. p values, unpaired two-tailed t-test. Mean ± 95% CI. (H) Top, schematic of the Igh and Igl loci and FISH probes. Bottom, representative metaphase images showing normal and abnormal Igh and Igl loci. C, centromere-proximal; T, telomere-proximal. (I) Igh and Igl locus breakage in splenic B cells of indicated mice upon stimulation (anti-CD40 + IL-4) for 96 h. n=4 mice per genotype, between 98 and 151 metaphases were analysed from each mouse (n=4), except for one WT sample with only 45 metaphases; Multiple comparisons, one-way ANOVA. Mean ± SD.

Figure 2. Mutational screening reveals REV7 interaction surfaces essential for CSR and resection inhibition.

(A) Schematic representation of the screen. Rev7 single and combinatorial point mutant alleles were stably expressed in Rev7^-/- CH12-F3. IgM to IgA CSR was measured 40h post stimulation with anti-CD40 antibody, IL-4 and TGFβ-1 (CIT). (B) Quantification of IgM to IgA CSR in Rev7^-/- CH12-F3 cell-lines complemented as indicated. IgA switching efficiency normalized against WT REV7 complemented cells. Mutants indicated in red were excluded from further analysis due to unstable REV7 protein expression. n=3 independent experiments. Mean ± SD. (C) CSR in stimulated B cells harvested from Rev7^F/FMb1^+/Cre mice and infected with GST (control) or indicated Rev7 mutant-expressing retrovirus. n=6 mice per genotype. IgG1-positive events in infected cells as a proportion (%) of the mean IgG1-positive events in 2 WT REV7-complemented controls. Experiments n=3 (each 2 mice per genotype). Mean ± SD. (D) Representative flow cytometry plots of data in (C).

Figure 3. REV7 interacts with SHLD3 and SHLD1-SHLD2 via distinct surfaces.

(A) B cell REV7 interactome as defined by LC-MS/MS and label-free quantification (LFQ). Scatter plot depict log₂ fold enrichment of indicated Flag-immunocomplexes across 2 independent experiments. (B) Interactions between indicated control (AD/BD empty vector), SHLD3, and indicated REV7 proteins by Y2H. 5-fold dilutions, representative data (n=4). (C) Top, schematic of SHLD3, indicating sequence features and positions of alanine substitutions in each mutant protein. Bottom, interactions between REV7 and indicated SHLD3 proteins by Y2H. Representative data (n=4). (D) LC-MS/MS and LFQ analysis of Flag-HA-REV7 and Flag-HA-REV7^K129A interactomes. Scatter plot depict log₂ fold enrichment of interacting proteins across 2 independent experiments.

Figure 4. The REV7-Shieldin complex mediates 53BP1-dependent NHEJ.

(A) Representative flow cytometry plots for IgM to IgA CSR in indicated mutant CH12-F3 cell-lines. Representative data, n>3 independent experiments. (B) IgM to IgA CSR frequencies in knockout CH12-F3 cell-lines. n=5 (Shld3^-/-), n=4 (Shld1^-/-) and n=3 (Shld2^-/-) independent experiments. Mean ±SD. (C) IgM to IgA CSR in indicated CH12-F3 lines stably transduced with either control (GST), WT, or mutant SHLD3-expressing transgenes. Data normalized to CSR frequencies of WT CH12-F3. n=5 independent experiments. Mean ± SD. (D) RPA-ssDNA ChIP with indicated CH12-F3 lines stimulated with CIT (30 h). Representative data, n=2 independent experiments. Bars indicate mean. (E) Clonogenic assay using KB1P-G3 cells following deletion of indicated genes. sgRNA, gene targeted by CRISPR-Cas9 guide RNA. (F) Quantification of data in (E) and Extended Data Fig. 6k. n=3 independent experiments. Mean ±SD. (G) Metaphases prepared from indicated cell-lines following 24 h control or olaparib treatment were analysed for the presence of radial chromosomes. n=3 independent experiments (each 50 metaphases per condition). Mean ± SD. (H) Proposed model of Shieldin-53BP1 cooperation during NHEJ. The stabilization of ssDNA overhangs or ssDNA-dsDNA junctions at DSB sites by REV7-SHLD1-3 complexes promotes DSB resolution activities during NHEJ. The absence of lymphocyte development defects in Rif1^-/- mice suggests RIF1 could link 53BP1 anti-resection activities in chromatin to Shieldin activities in ssDNA compartments (arrow).