The RAD51 recombinase protects mitotic chromatin in human cells

The RAD51 recombinase plays critical roles in safeguarding genome integrity, which is fundamentally important for all living cells. While interphase functions of RAD51 in maintaining genome stability are well-characterised, its role in mitosis remains contentious. In this study, we show that RAD51 protects under-replicated DNA in mitotic human cells and, in this way, promotes mitotic DNA synthesis (MiDAS) and successful chromosome segregation. In cells experiencing mild replication stress, MiDAS was detected irrespective of mitotically generated DNA damage. MiDAS broadly required de novo RAD51 recruitment to single-stranded DNA, which was supported by the phosphorylation of RAD51 by the key mitotic regulator Polo-like kinase 1. Importantly, acute inhibition of MiDAS delayed anaphase onset and induced centromere fragility, suggesting a mechanism that prevents the satisfaction of the spindle assembly checkpoint while chromosomal replication remains incomplete. This study hence identifies an unexpected function of RAD51 in promoting genomic stability in mitosis.


Supplementary Tables
Supplementary Table 1: ssDNA template used to edit RAD51 gene by CRISPR/Cas9 Supplementary Table 2: Primers used to screen RAD51 knock-in mutation at S14

Supplementary Reference
Supplementary Fig. 1: Evaluation of DNA synthesis upon mild replicative stress in mitosis.
a. Schematic diagram of experimental procedure for U2OS cell synchronisation. FACS profiles showing mitotic index (pS10-H3) and DNA content (PI) of cells exposed to RO3306 for 5 hours (left) and at 30 min after mitotic release (m-R) from RO3306 block (right) are also shown. The G1, S and G2 populations were estimated by the Watson Pragmatic algorithm (FlowJo), based on the PI staining of interphase cells. b. Schematic diagram of experimental procedure, as in (a), combined with 30 min EdU-labelling prior to m-R (G2), 30 min EdU-   labelling prior to m-R followed by a 30-min 100 µM thymidine chase upon m-R (G2+Thy); 30 min EdUlabelling upon m-R (M, 0'); 25 min EdU-labelling at 5 min upon m-R (M, 5'); 20 min EdU-labelling at 10 min upon m-R (M, 10'); and 15 min EdU-labelling at 15 min upon m-R (M, 15'). Cells were collected 30 min after m-R by mitotic shake-off (m-SO) and analysed for EdU incorporation. c. Quantification of the number of EdU foci per mitotic cell following EdU-labelling protocol as in (b  outside this range are drawn as individual dots. p-values were calculated by a Kruskal-Wallis test followed by Dunn's multiple comparison test. Asterisks indicate ****p value ≤ 0.0001. c. Representative images of cells quantified in (b). Scale bar indicates 5 µm. d. Western blot assessing RAD51 and Lamin A (loading control) levels in asynchronous U2OS Flp-In T-REx cells expressing the indicated FLAG-RAD51 variants from the tet-O promoter. Based on a single experiment. e-f. Quantification of the number of EdU foci (e) and -H2AX foci (f) per mitotic cell in doxycycline treated cells. Data was obtained from four independent experiments; n=400 per condition. Data distribution is represented by Tukey box-and-whisker plots. Bounds of box are 25-75th percentile, center shows the median and '+' marks the mean. Whiskers indicate ±1.5xIQR, data outside this range are drawn as individual dots. p-values were calculated by a Kruskal-Wallis test followed by Dunn's multiple comparison test. Asterisks indicate ***p value ≤ 0.001. Source data are provided as a Source Data file.  positive MiDAS via a RAD52-mediated BIR-like mechanism. RAD51-mediated DNA protection prevents nucleolytic attack of both DNA repair intermediates and continuing replication forks in mitosis, thereby promoting MiDAS at broken and unbroken forks. Upon mitotic RAD52 inhibition (AICAR), broken forks no longer engage in BIR-like MiDAS, resulting in the reduction of H2AX-positive mitotic EdU foci; decreased mitotic RPA foci formation and persistent mitotic DNA breaks. In the absence of mitotic RAD51 recruitment (B02), RPA-bound ssDNA structures at mitotic DNA repair intermediates and ongoing replication forks are vulnerable to nucleolytic attack, resulting in the reduction of both H2AX-positive and H2AX-negative mitotic EdU foci and the increased induction mitotic DNA breaks. The mitotic inhibition of B-family DNA polymerases (APH-M) abolishes MiDAS at both broken and unbroken replication forks, leading to the accumulation of RAD52-dependent DNA intermediates at broken forks (which are undetectable by PFGE) and H2AX-associated fork stalling at unbroken forks. b. Schematic diagram of experimental procedure for U2OS cell synchronisation combined with siRNA transfection, targeting RAD51 (siR51) or RAD52 (siR52). c. Following cell synchronisation according to workflow in (b), the total cell population was collected by trypsinization 30 min after m-R and analysed by western blot for the depletion of the corresponding protein.

Supplementary
Based on a single experiment. d-f. Quantification of the number of EdU foci (d) and -H2AX foci (e) per mitotic cell. Quantification of the number of EdU foci that colocalise with -H2AX ( -H2AX + ) and the number of EdU foci that do not colocalise with -H2AX ( -H2AX -) per mitotic cell (f). Data were obtained from three independent experiments (approximately 100 cells analysed per repeat); n= the total number of cells quantified per condition as indicated on graph. Data distribution is represented by Tukey box-and-whisker plots. Bounds of box are 25-75th percentile, center shows the median and '+' marks the mean. Whiskers indicate ±1.5xIQR, data outside this range are drawn as individual dots. p-values were calculated by a Kruskal-Wallis test followed by Dunn's multiple comparison test. Asterisks indicate ****p value ≤ 0.0001. Source data are provided as a Source Data file. Supplementary Fig. 4: Evaluation of endogenous RAD51 knock-in mutation at S14. a. DNA sequence and corresponding amino acid sequence of the RAD51 S14 locus targeted by gRNA for CRISPR/Cas9-mediated gene editing. b. Schematic illustration of workflow for knock-in of RAD51 S14 missense mutations by CRISPR/Cas9 gene editing. c. The total number of CRISPR/Cas9 targeted U2OS clones screened by PCR, and the number of clones validated by Sanger sequencing. Two rounds (Round 1 and 2) of the whole screening procedure, depicted in (b), were attempted. d. Representative Sanger sequencing chromatogram of the RAD51 S14 locus obtained for the U2OS RAD51 S14D/S14D homozygous knock-in mutant cell lines. e. Graphic representation of the indel size distribution for all reads of the RAD51 S14 locus in the HEK293 knockin mutant cell line as sequenced by next generation sequencing (miSeq). The corresponding DNA sequences detected are displayed below. f. Asynchronous populations of the indicated cell lines were assessed by western blot against S14 phosphorylated RAD51 (RAD51-pS14), RAD51, BRCA2, PALB2, PLK1, phosphorylated PLK1 (PLK1-pT210), Cyclin E, p21, pS10-H3 (mitotic marker) and -H2AX. Similar results were obtained Parental S14D g NT IR S14A S14A +WT P a r e n t a l S14A S14A +WT from multiple experiments. g. HEK293 cell line was assessed for RAD51 foci in untreated conditions (NT) or 3 hours after irradiation (IR) at 4 Gy. Data were obtained from three independent experiments (150 cells analysed per repeat; n=450 per condition). Data distribution is represented by Tukey box-and-whisker plots. Bounds of box are 25-75th percentile, center shows the median and '+' marks the mean. Whiskers indicate ±1.5xIQR, data outside this range are drawn as individual dots. p-values were calculated by a Kruskal-Wallis test followed by Dunn's multiple comparison test. Asterisks indicate ***p value ≤ 0.001; ****p value ≤ 0.0001. h. Cell survival of HEK293 cell line upon exposure to aphidicolin (APH) was assessed by WST-1 assay. Data are presented as mean values, error bars represent SEM; data were obtained from three independent experiments. p-values were calculated by a two-way ANOVA followed by Tukey's multiple comparisons test. Asterisks indicate significant differences compared to the parental HEK293 cell line: *p value ≤ 0.05; **p value ≤ 0.01; ***p value ≤ 0.001; ****p value ≤ 0.0001. i. Proliferation of the CRISPR gene-edited RAD51 S14A/S14A mutant cell line (S14A), the WT RAD51 complemented cell line (S14A+WT) and the corresponding HEK293 parental cell line (Parental) was assessed following 96-hour incubation with 0.5 µM aphidicolin (APH) or non-treated control (NT). The division index for each cell line was determined by Cell Trace TM cell proliferation assay based on a single experiment. j. Statistical differences of cell survival between the RAD51 S14A/S14A (S14) and the WT RAD51 complemented (S14A+WT) cell lines, as assayed in (h). p-values were calculated by a two-way ANOVA followed by Tukey's multiple comparisons test and summarised. k. Cell survival upon exposure to neocarzinostatin (NCS) was assessed by WST-1 assay. Data are presented as mean values, error bars represent SEM. Data were obtained from three independent experiments. l. Cell survival upon exposure to 0.1 µM APH was assessed by clonogenic assay. Data are presented as mean values, error bars represent SEM. Data were obtained from three independent experiments. p-value was calculated by a two-sided unpaired t test. m. Proliferation of the CRISPR gene-edited RAD51 S14D/S14D mutant cell line (S14D) and the corresponding U2OS parental cell line was assessed following 96 hour incubation with 0.5 µM aphidicolin (APH) or non-treated control (NT). The division index for each cell line was determined by Cell Trace TM cell proliferation assay based on a single experiment. Source data are provided as a Source Data file. Supplementary Fig. 5: Synchronization of RAD51 S14 knock-in mutant cell lines. a. Schematic diagram of experimental procedure used to obtain data in (b). The total cell population was collected and analysed by FACS at m-R. b. FACS analysis of mitotic index (pS10-H3) and DNA content (PI) in the CRISPR gene-edited HEK293 cell lines following cell synchronization according to the experimental procedure depicted in (a). The G1, S and G2 populations were estimated by the Watson Pragmatic algorithm (FlowJo), based on the PI staining of interphase cells. c-e. EdU incorporation and -H2AX IF signal were analysed in the parental HEK293 cell line; the CRISPR gene-edited RAD51 S14A/mutant cell line (S14A) and the wild-type RAD51 complemented RAD51 S14A/cell line (S14A+WT) following cell synchronization as depicted in Fig. 4b. Representative images of the data analysed (c). Scale bar indicates 5 m. Quantification of the number of EdU foci (d) and the number of -H2AX foci per mitotic cell (e). Data were obtained from three independent experiments (100 cells analysed per repeat); n=300 per condition. Data distribution are represented by Tukey box-and-whisker plots. Bounds of box are 25-75th percentile, center shows the median and '+' marks the mean. Whiskers indicate ±1.5xIQR, data outside this range are drawn as individual dots. p-values were calculated by a Kruskal-Wallis test followed by Dunn's multiple comparison test. Asterisks indicate ***p value ≤ 0.001; ****p value ≤ 0.0001. f. Schematic diagram of experimental procedure used to obtain data in (d). The total cell population was collected and analysed by FACS at the indicated timepoints. g. FACS analysis of mitotic index (pS10-H3) and DNA content (PI) in the U2OS parental cell line and CRISPR gene-edited RAD51 S14D/S14D mutant cell line (S14D) following cell synchronization according to the experimental procedure depicted in (f). Source data are provided as a Source Data file. wild-type and U2OS RAD51 S14D/S14D knock-in mutant cell line (S14D) upon exposure to 0.5 M AZ3146. Data are presented as mean values, error bars represent SEM. Data were obtained from three independent experiments. The significant difference between the samples was assessed by a two-sided unpaired Welch's t test. Asterisks indicate *p value ≤ 0.05. c. Survival of HEK293 cell lines upon exposure to MPS1 inhibitor (AZ3146), as determined by WST-1 assay. Data are presented as mean values, error bars represent SEM. Data were obtained from three independent experiments. Significant differences compared to the parental HEK293 cell line were determined by a two-way ANOVA followed by Tukey's multiple comparisons test. Asterisks indicate *p value ≤ 0.05; **p value ≤ 0.01; ***p value ≤ 0.001; ****p value ≤ 0.0001. d. Representative image of mitotic progression analysed in U2OS cells following cell synchronisation as in Fig. 5d. Frames displaying nuclear envelope breakdown (NEBD) and anaphase onset (AO) are indicated. e. Schematic diagram of cell synchronisation for HEK293 cell lines used to obtain data in (f). f. Quantification of the number of -H2AX foci per mitotic cell, detected in mitotic cells collected by m-SO. Data were obtained from three independent experiments (100 cells analysed per repeat); n=300 per cell line. p value was calculated by a Kruskal-Wallis test followed by Dunn's multiple comparison test. Asterisk indicates *p value ≤ 0.05. g. Cells were synchronized according following the protocol depicted in (e). The total cell population was collected at 30 min after m-R and analysed by FACS for mitotic index (pS10-H3) and DNA content (PI). The G1, S and G2 populations in respective cell lines were estimated by the Watson Pragmatic algorithm (FlowJo), based on the PI staining of interphase cells. h. Percentage of mitotic aberrations (lagging chromosomes, anaphase bridges) was quantified for data analysed in Fig. 5d. Data are presented as mean values, error bars represent SEM. Data were obtained from at least three independent experiments. p values were calculated by a Brown-Forsythe and Welch ANOVA test followed by Dunnett's T3 multiple comparisons test. Asterisks indicate *p value ≤ 0.05; **p value ≤ 0.01; ***p value ≤ 0.001. i. Representative images of normal (A) and aberrant mitotic chromatin (B-D) as quantified in (h). Arrows indicate lagging chromosomes, chromatin bridges or micronuclei. j. The percentage of cells that die before completing mitosis was quantified for data analysed in Fig. 5d. Data are presented as mean values, error bars represent SEM. Data were obtained from at least three independent experiments. p values were calculated by a Brown-Forsythe and Welch ANOVA test followed by Dunnett's T3 multiple comparisons test. Asterisk indicates *p value ≤ 0.05. Source data are provided as a Source Data file. Supplementary Fig. 7: Gating strategy.

Supplementary
Forward versus side scatter (FSC vs. SSC) gating was used to remove debris from the main cell population where necessary; single cells were discriminated from doublets by FL2A vs. FL2W gating. Histone H3 S10 phosphorylation staining intensity (mitotic cells) is measured on a logarithmic scale on the Y-axis while PI staining on the X-axis measures the doubling of DNA content (i.e. 2C to 4C) to identify interphase and mitotic cell populations. Cell counts in G1, S, G2 were estimated using the Watson Pragmatic algorithm in FlowJo on gated interphase cells and combined with quantification of gated mitotic cells to calculate cell cycle distributions.   Table 2: Primers used to screen RAD51 knock-in mutation at S14 name sequence WT reverse 5'-CTTGAAGCAAATGCAGATACTTCA-3' S14A reverse 5'-CTTGAAGCAAATGCAGATACTGCC-3' S14D reverse 5'-CTTGAAGCAAATGCAGATACTGAT-3' forward 5'-TGCAGCCTTTCCACAGGATT-3'

CRISPR/Cas9-mediated editing of RAD51 gene
Cells were co-transfected with (i) a plasmid carrying Cas9, GFP and a guide RNA targeting RAD51 S14 locus (AGCAAATGCAGATACTTCAGTGG) and (ii) the relevant ssDNA template (Supplementary Table 1), using Lipofectamin LTX with Plus Reagent (ThermoFisher Scientific) according to manufacturer's instructions. At 24 hours after transfection, GFP-expressing cells were sorted by FACS (MoFlo XDP, Beckman Coulter). Single clones were expanded and genomic DNA was extracted for PCR amplification of the targeted RAD51 S14 locus. Clones were screened for the desired mutation using forward primers that are perfectly complementary to either the wildtype S14 sequence or the mutated S14D or S14A sequence (Supplementary Table 2). Clones that generated a convincing PCR product with the mutation-specific primer, but not the wild-type primer, were further verified for the knock-in mutation by Sanger sequencing. For heterozygous clones, next-generation sequencing (Illumina MiSeq Next Generation Sequencer) was used to determine the allele-specific sequences at the RAD51 S14 locus. Verified clones were further tested for the absence of RAD51 phosphorylation by western blot.