The FANCD2-FANCI heterodimer coordinates chromatin openness and cell cycle progression throughout DNA double-strand break repair

Abstract

The FANCD2-FANCI heterodimer contributes to DNA repair at interstrand crosslinks and sites of replication stress. This complex has been physically and mechanistically linked to double-strand break (DSB) repair, but its role in that process remains undefined. Here, we show that the FANCD2-FANCI heterodimer dynamically interacts with open chromatin regions, including transient DSB-induced open chromatin, where it can be stabilized through co-activation by the DNA repair kinase ATM and the Fanconi anemia core ubiquitin ligase. The loaded FANCD2-FANCI heterodimer stabilizes open chromatin and promotes resection and loading of RPA through increased association of BRCA1 and BLM. Chromatin-loaded FANCD2-FANCI has a second, distinct function promoting a G2 cell cycle arrest that is dependent on the ATR-CHK1-WEE1 axis. Our results support a two-step genome surveillance model in which FANCD2-FANCI monitors open chromatin sites and is stably loaded to coordinate DNA repair activities in response to signaling from a DNA repair kinase.

Keywords: ATM kinase; CP: Immunology; DNA repair; Fanconi anemia; cell cycle; chromatin; end resection; homologous recombination.

Figure 1.. The FANCD2-FANCI heterodimer binds double- and single-stranded DNA at DSB sites

(A) DSB induction schematic showing (1) addition of 300 nM 4OHT driving nuclear localization of the ER-AsiSI fusion protein and digestion of target sites or (2) electroporation of Cas9 and guide RNA targeting single or multiple genomic loci. (B) Stranded ChIP-seq schematic showing resected or double-stranded DNA, read profiles consistent with resected or double-stranded DNA, and skew profiles generated by subtracting reads mapping to the − strand from reads mapping to the + strand. (C) Stranded ChIP-seq data presented as reads mapping to distinct strands (top) or skew (bottom) at 122 AsiSI sites 4 h after DSB induction. Individual reads in FANCD2 ChIP-seq datasets spanning the DSB site were plotted as a frequency (right). Immunoprecipitations were performed using FANCD2 (FD2) or FANCI (FI) antibodies from DIvA-U2OS cells as indicated. Line plots are representative of n = 2 biological replicates and dot plots present all data from n = 2 replicates in each condition. (D) Stranded ChIP-seq data presented as reads mapping to distinct strands (top) or skew (bottom) at 50 AluGG sites 16 h after DSB induction. Immunoprecipitations were performed using FD2 antibody from U2OS cells. All plots are representative of n = 2 biological replicates. (E) Single-stranded DNA defined by KAS-ATAC-seq in GM12878 cells reproduced from PRJNA1103095 for the LMNB1 site (top), 122 AsiSI sites (middle), or 50 AluGG sites (bottom). All plots are the average of n = 2 biological replicates. (F) Stranded ChIP-seq data presented as reads mapping to distinct strands (top) or skew (bottom) at the LMNB1 cut site 16 h after DSB induction. Schematic shows the LMNB1 divergent transcript (DT) and LMNB1 transcript (LMNB1). Immunoprecipitations were performed using FD2 or FI antibodies from U2OS cells as indicated. Line plots are representative of n = 2 biological replicates. (G) Stranded ChIP-seq data presented as reads mapping to distinct strands (top) or skew (bottom) at the LMNB1 locus in its natural (noDSB) context. Schematic shows the LMNB1 DT and LMNB1. Immunoprecipitations were performed using FD2 antibody from U2OS cells. Line plots are representative of n = 2 biological replicates.

Figure 2.. The FANCD2-FANCI heterodimer interacts dynamically with induced and stable open chromatin sites in human cells

(A) ATAC-seq and ChIP-seq data presented as average and individual signal at 122 AsiSI sites either uncut (noDSB) or 4 h after DSB induction (+4OHT) and showing chromatin openness or repair factor recruitment. Immunoprecipitations were performed using MRE11 or FANCD2 antibodies from DIvA-U2OS cells as indicated. All plots are representative of n = 2 biological replicates. (B) Strong (≥0.7), moderate (≥0.5), and weak (<0.5) Spearman correlations between signal at all AsiSI recognition sites in the human genome (1,228 in silico predicted sites) generated by ATAC-seq without DSBs (noDSB), FANCD2 ChIP-seq without DSBs (noDSB), FANCD2 ChIP-seq with DSBs (+4OHT), and MRE11 ChIP-seq with DSBs (+4OHT). Data were generated from n = 2 biological replicates for each condition. (C) ATAC-seq signal at single (HBB) or multiple (50 AluGG) DSB sites without (noDSB) or 4 h after (+DSB) DSB induction in U2OS cells. Each panel shown is n = 1 biological replicate. (D) ATAC-seq and ChIP-seq data presented as signal at the F7 AsiSI site showing open chromatin via ATAC-seq (top), recruitment of FANCD2 via ChIP-seq without DSBs (second from top), recruitment of MRE11 via ChIP-seq with DSBs (second from bottom), and recruitment of FANCD2-FANCI via ChIP-seq with DSBs (bottom). Immunoprecipitations were performed using MRE11, FANCD2, or FANCI antibodies from DIvA-U2OS cells. Data shown are representative of n = 2 biological replicates

Figure 3.. FANCD2-FANCI loading at open chromatin substrates requires the FA core complex and is cell cycle dependent

(A) ChIP-seq profiles showing FANCD2 binding (U2OS cells) and FANCI binding (U2OS cells); DNase hypersensitivity profiles (MG63 cells) at 200 non-TSS DNase hypersensitive sites defined by the ENCODE consortium. Data shown are representative of n = 2 biological replicates. (B) ChIP-seq profiles showing FANCD2 binding 16 h after electroporation with Cas9 RNP targeting the RAB11A gene in patient-derived fibroblast cell lines expressing FANCD2 (PD20 RV:D2) or lacking the FA core components FANCA (PD220) or FANCC (PD331). Data shown are representative of n = 2 biological replicates. (C) ChIP-seq profiles showing FANCD2 binding to DNase hypersensitive sites in patient-derived fibroblast cell lines expressing FANCD2 (PD20 RV:D2) or lacking the FA core component FANCC (PD331). Data shown are representative of n = 2 biological replicates. (D) ChIP-seq profiles showing FANCD2 binding to DNase hypersensitive sites in U2OS cells with and without inhibition of the deubiquitinase USP1 by 30 μM ML323 for 4 h. Data shown are representative of n = 2 biological replicates. (E) QIBC immunofluorescence tracking EdU incorporation, DAPI incorporation, and FANCD2 chromatin binding in non-extracted (top row) or pre-extracted (bottom row) DIvA-U2OS cells. Data are presented as pseudo cell cycle plots (left), as box and whisker plots showing FANCD2 mean intensity as a function of cell cycle (middle), and as example IF images (right). Data shown are generated from at least m = 12 images and n = 2,000 cells, and the p values were derived from Shapiro-Wilk test, followed by non-parametric Kruskal-Wallis test, followed by Dunn test. ****p ≤ 0.0001. Scale bar represents 20 μm.

Figure 4.. ATM activation stabilizes FANCD2-FANCI at DSBs and DSB-adjacent DNase hypersensitive sites

(A) ChIP-seq summary data showing MRE11 and FANCD2 binding as a function of time after electroporation with Cas9 RNP targeting the RAB11A locus (left) and the fraction of reads ending at the DSB site in each condition (right). Data shown are generated from n = 2 biological replicates in U2OS cells. (B) ChIP-seq profiles showing FANCD2 binding to DSBs (top) or open chromatin sites (bottom) in untreated (DMSO), 100 μM mirin-treated, or 100 μM PFM01-treated conditions (4 h). Data shown are representative of n = 2 biological replicates in DIvA-U2OS cells simultaneously treated with 300 nM 4OHT for 4 h. (C) ChIP-seq profiles showing FANCD2 binding to DSBs (top) or open chromatin sites (bottom) in untreated (DMSO) or 2 μM NSC-105808-treated (DNA2i) conditions (4 h). Data shown are representative of n = 2 biological replicates in DIvA-U2OS cells simultaneously treated with 300 nM 4OHT for 4 h. (D) ChIP-seq profiles showing FANCD2 binding to DSBs (top) or open chromatin sites (bottom) in untreated (DMSO) or ATM inhibitor-treated (10 μM KU-55933, 1 μM M4076, or 0.1 μM AZD1390) conditions (4 h). Data shown are representative of n = 2 biological replicates in DIvA-U2OS cells simultaneously treated with 300 nM 4OHT for 4 h. (E) DNase hypersensitivity profiles (MG63 cells) at non-TSS DNase hypersensitive sites defined by the ENCODE consortium. ChIP-seq profiles showing FANCD2 binding (DIvA-U2OS cells) and MRE11 binding (DIvA-U2OS cells) at DSB-adjacent DNase-hypersensitive sites. These sites arise more than 1.5 kb but less than 50 kb upstream and downstream of 122 AsiSI cut sites (top) and average signal at these sites is presented without (middle; noDSB) or with (bottom; +DSB) DSB induction, and with ATM inhibitor treatment (10 μM KU-55933). Data shown are representative of n = 2 biological replicates in DIvA-U2OS cells. For DSB induction, cells were treated with 300 nM 4OHT (with or without ATMi) for 4 h.

Figure 5.. FANCD2-FANCI loading changes the recruitment of DNA repair proteins and chromatin factors in the vicinity of DSBs

(A) ATAC-seq signal in U2OS cells at single (HBB) or multiple (50 AluGG) DSB sites in non-targeting control cells without DSBs (gray line; noDSB), in non-targeting control cells 4 h after DSB induction (purple line; +DSB), or FANCD2-depleted cells 4 h after DSB induction (blue line; +DSB; FD2−). Each panel shown is n = 1 biological replicate. (B) qPCR-based DNA resection assay showing the percentage of single-stranded DNA (%ssDNA) at three different distances from the DSB in non-targeting control (siNTC) and FANCD2-depleted (siFANCD2) DIvA-U2OS cells 4 and 24 h after DSB induction using 300 nM 4OHT. Data were generated from n = 3 technical replicates. (C) ChIP-seq profiles showing FANCD2-FANCI, BRCA1, BLM, and RPA signal at AsiSI DSBs in CRISPRi non-targeting control (WT) and FANCD2-depleted (FD2−) DIvA-AID-U2OS-ZIM3 cells 4 h after DSB induction using 300 nM 4OHT. Data shown are representative of n = 2 biological replicates. (D) Stranded ChIP-seq profiles showing FANCD2-FANCI, BRCA1, BLM, and RPA signal at AsiSI DSBs in CRISPRi non-targeting control (WT) and FANCD2-depleted (FD2−) DIvA-AID-U2OS-ZIM3 cells 4 h after DSB induction using 300 nM 4OHT. Data shown are representative of n = 2 biological replicates. (E) QIBC immunofluorescence tracking EdU incorporation, DAPI incorporation, and RPA or BRCA1 colocalization with chromatin-bound FANCD2 in pre-extracted DIvA-U2OS cells 24 h after DSB induction using 300 nM 4OHT. Example IF images are presented at left and Pearson correlations as a function of cell cycle are plotted at right as box and whisker plots. Data shown are generated from at least m = 12 images and n = 2,000 cells. Scale bar represents 2 μm.

Figure 6.. Chromatin-loaded FANCD2 promotes an ATR-dependent G2 arrest

(A) Flow cytometry measurement of G2 cell cycle abundance in non-targeting control (siNTC) or FANCD2-depleted (siFANCD2) DIvA-U2OS cells at the indicated times after DSB induction using 300 nM 4OHT. Error bars indicate SEM. Data shown were generated from n = 2 biological replicates. (B) Flow cytometry cell cycle distributions of non-targeting control (siNTC) or FANCD2-depleted (siFANCD2) DIvA-U2OS cells either asynchronous (noDSB) or 24 h after DSB induction (+DSB). Left plot represents G1 phase, middle plot represents S phase, and right plot represents G2 phase. Error bars indicate SEM. Data shown were generated from n = 2 biological replicates, and the p values were derived from ordinary two-way ANOVA with full model, followed by Sidak’s multiple comparisons test, with a single pooled variance. ns, not significant; *p ≤ 0.05. (C) Co-culture experiments comparing growth rates of CRISPRi non-targeting control (WT) and FANCD2 knockdown (FD2−) DIvA-AID-U2OS-ZIM3 cells continuously grown in 300 nM 4OHT (to induce DSBs) relative to undamaged controls. Error bars indicate SEM. Data shown are generated from n = 2 biological replicates. (D) Cell cycle distributions measured by QIBC of untreated (DMSO) or ATR inhibitor-treated (1 μM AZ20) DIvA-U2OS cells 24 h after DSB induction. Error bars indicate SEM. Data shown are generated from at least m = 12 images and n = 2,000 cells, and the p values were derived from repeated-measure two-way ANOVA with the Geisser-Greenhouse correction, followed by Sidak’s multiple comparisons test, with individual variances computed for each comparison. **p ≤ 0.01. (E) Western blots showing phospho-CHK1 (p-CHK1) and phospho-WEE1 (p-WEE1) signal at indicated times after release from a double thymidine block into media containing 300 nM 4OHT in non-targeting control (siNTC) or FANCD2-depleted (siFANCD2) DIvA-U2OS cells. Data shown are representative of n = 3 biological replicates.