Histone H3 serine-57 is a CHK1 substrate whose phosphorylation affects DNA repair

Abstract

Histone post-translational modifications promote a chromatin environment that controls transcription, DNA replication and repair, but surprisingly few phosphorylations have been documented. We report the discovery of histone H3 serine-57 phosphorylation (H3S57ph) and show that it is implicated in different DNA repair pathways from fungi to vertebrates. We identified CHK1 as a major human H3S57 kinase, and disrupting or constitutively mimicking H3S57ph had opposing effects on rate of recovery from replication stress, 53BP1 chromatin binding, and dependency on RAD52. In fission yeast, mutation of all H3 alleles to S57A abrogated DNA repair by both non-homologous end-joining and homologous recombination, while cells with phospho-mimicking S57D alleles were partly compromised for both repair pathways, presented aberrant Rad52 foci and were strongly sensitised to replication stress. Mechanistically, H3S57ph loosens DNA-histone contacts, increasing nucleosome mobility, and interacts with H3K56. Our results suggest that dynamic phosphorylation of H3S57 is required for DNA repair and recovery from replication stress, opening avenues for investigating the role of this modification in other DNA-related processes.

Fig. 1. Identification of Histone H3 Ser-57 phosphorylation (H3S57ph) in vertebrates.

a MS/MS spectrum of a histone H3 peptide bearing phosphorylated Ser-57 noted by the ions y7, y8 and y9 with neutral loss (represented by *); 5 different egg extracts were used. b Western blot analysis (WB) of chromatin purified from interphase Xenopus egg extracts during DNA replication, treated with buffer or lambda phosphatase, amidoblack for loading control; n = 2 independent experiments. c WB (18% PAGE) of acid-extracted histones from a panel of human cell lines; n = 1 experiment. d WB (18% PAGE) of sucrose gradient-separated nucleosome fractions. H3.cs1 recognises the clipped form of H3; n = 2 independent experiments. e WB (18% PAGE) of purified, native mono-nucleosomes after in vitro cleavage of their N-terminal tails by trypsin; n = 3 independent experiments. Source data are provided as Source Data file.

Fig. 2. H3S57ph levels are regulated during the cell cycle.

a Representative immunofluorescence image of a U2OS cell stained with DAPI (DNA) and H3S57ph antibody; n > 10 independent experiments, over 100 cells were analysed each time. Scale bar, 10 µm. b Plotting of H3S57ph and cyclin A signal intensity quantified from immunofluorescence images. Cyclin A staining was used as a cell cycle phase marker (mean ± SD of 1102 and 1247 U2OS cells, and 362 MDA-MB-231 cells; U2OS, n = 2, MDA-MB-231, n = 1 experiment). c FACS profiles of U2OS cells, asynchronously growing (AS) or synchronised with either double thymidine block (DTB) or nocodazole (NOC), and released for the indicated time; n > 5 independent experiments; 10,000 cells an analysed. The gating strategy is presented in Supplementary Fig. 11. d Western blot analysis of cells from the synchronisation experiment in c. H3S10ph was used as mitotic marker and to confirm non-cross reactivity with anti-H3S57ph antibody; amidoblack was used for loading control; n = 2 independent experiments. e WB of proteins present at replication forks, purified by iPOND. Pulse, 10 min EdU or 30 min upon HU; Chase, 1 h thymidine upon pulse; HU, 5 mM for 3 h or 0.2 M for 20 h; No Clk, sample without click reaction for negative control; n = 2 independent experiments. Source data are provided as Source Data file.

Fig. 3. Checkpoint kinase 1 (CHK1) is the H3S57ph kinase in human cells.

a Radar plot with activity values of 190 kinases in kinase assays with H3_51-72 peptide as substrate (n = 1 experiment). Only kinases above the threshold are labelled, see Supplementary Data 1 for full results. In vitro kinase assays using γ³³P-ATP, recombinant CHK1 and H3 peptides, with the indicated modifications (b; n = 4) or full-length H3 (c; n = 4–5) as substrates; error bars, mean ± SD; results are normalised to, and represented as fold increase over, the control assays without kinase. d In vitro kinase assays as in c but with non-radioactive ATP and analysed by western blotting; n = 1 experiment. The histone extract was used as positive control for the WB. U2OS cells were treated with CHK1 inhibitors (SCH900776, 5 µM; CHIR-124, 2 µM) for indicated times, and analysed by WB of acid-extracted histones (e), FACS cell cycle profiles (f), or by immunofluorescence of fixed cells (g); n = 2 independent experiments. All lanes come from the same membrane/exposure, unrelated lanes were removed. DNA was stained with DAPI; scale bar, 20 µm. Knockdown of CHK1 by RNA interference in U2OS cells, analysed subsequently by WB (h) of total cell lysates (top) and acid-extracted histones (bottom), and FACS cell cycle profiles (i); amidoblack staining was used for loading control; siCtrl, siRNA targeting the firefly luciferase gene; n = 2 independent experiments. Source data are provided as Source Data file.

Fig. 4. Cells overexpressing H3S57 mutants use different pathways of replication stress recovery.

a U2OS cells expressing an empty vector or FLAG-tagged H3.1wt, S57A or S57D constructs under the CMV promoter, in initial populations and two isolated clones, were plated at low density and monitored for proliferation during 5 days. Experiments were performed in triplicates and results are shown as mean ± SD. b Left, overlay image of FACS profiles of cells collected 8 h after release from a 20 h HU block. Right, bar graph representing the percentage of S-phase cells at the HU + 8 h timepoint. *p = 0.01, ***p = 0.0005 (Student’s T test); error bars, mean ± SD; n = 3 independent experiments. The gating strategy is presented in Supplementary Fig. 11. c Analysis of γH2A.X by FACS in non-treated cells, or cells treated with HU for 3 h, and allowed to recover for 1 h; DAPI was used to counterstain DNA; the fluorochrome used is indicated (AF, AlexaFluor). Doublets and debris were excluded, and 10 000 cells were analysed per sample. Gates were arbitrarily set in EV control conditions and implemented on the other conditions. The gating strategy is provided in Supplementary Fig. 11. Representative images from 3 independent experiments are shown. d As in c, but ssDNA was assessed by analysis of native BrdU incorporation in native conditions, in either control cells or cells exposed to HU for 4 h (n = 2 independent experiments). e As in c, but 53BP1 was analysed (n = 2 independent experiments). f As in c, but γH2A.X was analysed in cells previously submitted to two rounds (48 h) of siRNA treatment to downregulate RAD52. Final gates were repositioned manually where necessary due to drift; n = 1 experiment. Source data are provided as Source Data file.

Fig. 5. H3S57 is phosphorylated and regulates DNA damage repair pathway choice in fission yeast.

a Whole cell extracts from the indicated strains were analysed by Western blot to detect H3S57ph and histone H3. Each wild-type (WT) and H3S57A pair represents a set of independent biological repeats (n = 3). b Lambda phosphatase treatment results in loss of detection of H3S57ph by a modification-specific antibody. Dephosphorylation does not alter overall H3 levels. Phosphorylation of the Cdc2 cyclin-dependent kinase is used as a control for the efficiency of treatment. A representative experiment is shown (n = 3). c Cells with the indicated genotypes were grown to exponential phase in YE4S and spotted on plates at 10-fold dilutions. Images were taken on the following days after spotting: day 2 for UV 200 J/s; day 3 for YES 32 °C, MMS 0.007%, MMS 0.004%, CPT 6 µM, HU 6 mM; and day 6 for HU 10 mM. d Live-cell imaging of Rad52-GFP in exponentially growing cells with the indicated genotypes. Scale bar: 5 µm. Graphs show the percentage of cells with no Rad52 focus, 1-2 foci, and aberrant/3 or more foci per nucleus. Averages (bars) and standard errors (whiskers) of three independent experiments (dots) are shown, n > 200 cells were counted for each experiment. e Analysis of the repair of double-stranded DNA breaks via non-homologous end joining (NHEJ) or homologous recombination (HR) in wild-type, H3S57A, and H3S57D strains. The re-ligation (NHEJ) or genomic integration (HR) of a linearised plasmid was assessed. A control vector was co-transformed for internal normalisation. The efficiency of DNA repair for each experiment was normalised to the wild-type sample. For comparison, strains defective for NHEJ (lig4Δ) and HR (rad52Δ) were included. Graphs show averages (bars) and standard error (whiskers). Dots indicate independent experiments; n = 4 for NHEJ and n = 5 for HR. Source data are provided as Source Data file.

Fig. 6. H3S57ph increases nucleosome dynamics and interacts with H3K56.

a Comparison of final structures of molecular dynamics (MD) simulation of unphosphorylated H3 (green), and H3S57ph (violet). b RMSD along time of residues 133–162 (DNA in the entry/exit point of the nucleosome and in contact with H3), for unphosphorylated H3 (black) and H3S57ph (red). c Representative simulation snapshot showing the intra-molecular K56-S57ph interaction. d Left, western blot analysis of in vitro acetylation assays using CPB and synthetic full-length H3, unphosphorylated or phosphorylated on S57 (H3S57ph). Right, bar graph with the quantification of the images of 4 independent WB experiments, representing the percentage of H3K56 acetylation between the two constructs; error bars, mean ± SD; p < 0.0001; n = 4 independent experiments. e Distribution of the sequenced MNase-digested fragments of around 135 bp and around 147 bp for budding yeast strains expressing WT, H3S57A and H3S57E mutants. Source data are provided as Source Data file.