Dbf4-dependent kinase promotes cell cycle controlled resection of DNA double-strand breaks and repair by homologous recombination

Abstract

DNA double-strand breaks (DSBs) can be repaired by several pathways. In eukaryotes, DSB repair pathway choice occurs at the level of DNA end resection and is controlled by the cell cycle. Upon cell cycle-dependent activation, cyclin-dependent kinases (CDKs) phosphorylate resection proteins and thereby stimulate end resection and repair by homologous recombination (HR). However, inability of CDK phospho-mimetic mutants to bypass this cell cycle regulation, suggests that additional cell cycle regulators may be important. Here, we identify Dbf4-dependent kinase (DDK) as a second major cell cycle regulator of DNA end resection. Using inducible genetic and chemical inhibition of DDK in budding yeast and human cells, we show that end resection and HR require activation by DDK. Mechanistically, DDK phosphorylates at least two resection nucleases in budding yeast: the Mre11 activator Sae2, which promotes resection initiation, as well as the Dna2 nuclease, which promotes resection elongation. Notably, synthetic activation of DDK allows limited resection and HR in G1 cells, suggesting that DDK is a key component of DSB repair pathway selection.

Fig. 1. DDK phosphorylates HR proteins and is required for HR.

a DDK mutants are hypersensitive to camptothecin (CPT). Five-fold serial dilutions of bob1-1, bob1-1 dbf4Δ, and bob1-1 cdc7Δ strains were spotted on YPD or YPD containing CPT at the indicated concentrations. Data are representative of n = 2 biological replicates. b Scheme of the gene conversion assay used to study HR rates. A DSB generated by the HO endonuclease at 491 kb (Chr. IV) can be repaired using a homologous donor template at 795 kb (Chr. IV). The latter carries a unique 23-bp sequence, allowing for quantification of the recombination products via qPCR; arrowheads denote PCR primer locations. Gene conversion using the donor template disrupts the HO cut site, allowing cell survival to the chronic induction of the endonuclease as an alternative readout. ‘cfu’ stands for colony-forming unit. c DDK is required for HR. qPCR analysis of HR using system as in (b) in cells arrested in M-phase. Cells lacking the donor template are used as control. n = 3, box plot shows mean with values of biological replicates, error bars denote SD. Reported p-values were calculated using a two-tailed unpaired t-test. See also Supplementary Fig. 1a, b. d Class-I peptides derived from phospho-proteomic experiment were subjected to analysis of variance (ANOVA) test, with permutation-based false discovery rate (FDR) cutoff of 0.05. ANOVA significant phospho-peptides were then subjected to a hierarchical clustering in Perseus (v1.6.5.0). The calculated z-scores are shown in the heat-map and the different clusters are highlighted. The DDK cluster shows specific phospho-peptides downregulated in bob1-1 dbf4Δ cells. n = 4 biological replicates. See also Supplementary Fig. 1c–h. e Motif sequence generated for phospho-peptides enriched in the DDK cluster showing the 3 positions upstream and downstream the modified S/T. f Heat-map depicting the z-score, highlighting phospho-peptides from the DDK cluster after filtering in Perseus (v1.6.5.0) for GOBP:DNA repair (GO: 0006281). Gene name and modified residues are reported. GOBP = Gene Ontology Biological Process. Source data are provided as a Source Data file.

Fig. 2. DDK promotes DNA end resection and phosphorylates resection nucleases.

a A DSB generated by the HO endonuclease in the LEU2 open reading frame can be repaired via single-strand annealing (SSA) upon extensive resection that reveals a region of homology 25 kb distant from the DSB, followed by annealing of the homologous sequences and ligation. RAD51 is deleted to suppress HR. b Five-fold serial dilutions of reported strains were spotted on YPD (as control) or YPGal (for chronic pGAL-HO-induction) plates. Data are representative of n = 3 biological replicates. c Strand-specific loss of DNA and enrichment of RPA-bound ssDNA is used to measure resection at an unrepairable DSB, generated by the HO endonuclease at the MAT locus. d, e DDK is required for efficient DNA end resection. d Total DNA shows preferential loss of 5′ DNA strands 2 and 4 h after DSB induction. Loss of DNA is less pronounced after IAA-induced degradation of Dbf4-3AID. Strand-specific read coverage is normalized to read coverage before DSB induction. e Strand-specific accumulation of 3′-ssDNA enriched by RPA-ChIP is diminished after IAA-induced degradation of Dbf4-3AID compared to mock-treated cells. RPA-ChIP signals at the DSB are normalized to DSB-independent RPA signals occurring throughout the genome. Data are representative of n = 4 biological replicates. See also Supplementary Fig. 2e, f. f DDK inhibition leads to defective resection in human cells. M-phase-arrested U2OS cells were treated with the DDK inhibitor XL413 (or mock-treated) before DNA damage induction with zeocin (or mock-treatment). RPA foci (RPA70 subunit) were counted as proxy for resection. Left, representative images of cells (DAPI, RPA70). Scale bars: 10 μm. Right, quantification of RPA foci number per cell. Blue bars represent the mean. Reported p-values were calculated using a two-tailed Mann-Whitney test. Mock/mock n = 171 cells, mock/zeocin n = 230 cells, XL413/mock n = 169 cells, XL413/zeocin n = 265 cells; pooled from n = 2 biological replicates. See also Supplementary Fig. 2g–j. g Sae2 and Dna2 display a cell cycle and DDK-dependent shift in electrophoretic mobility. Cells expressing 9Myc-tagged resection proteins as indicated were arrested either in G1 or M-phase and samples run on gels. Data are representative of n = 2–4 biological replicates. See also Supplementary Fig. 2k. Source data are provided as a Source Data file.

Fig. 3. DDK regulates short-range resection via phosphorylation of Sae2.

a A model CDK (mammalian CDK2-CycA) and budding yeast DDK can phosphorylate Sae2 in vitro. Top, autoradiography monitoring incorporation of radioactive phosphate; Bottom, same gel stained with Coomassie Blue. Note that DDK can auto-phosphorylate, as previously shown^,. Data are representative of n = 2 independent experiments. b, c Sae2 (b) or Sae2-S267E (c) were phosphorylated by DDK or mock-treated, and added to the MRX complex to monitor endonucleolytic clipping of DNA. Left, quantification of cleavage products resolved on gels such as shown on the right. n = 3 independent experiments, shown is the mean with values of replicates, error bars denote SD. Reported p-values were calculated using a two-tailed unpaired t-test. d Inverted Alu repeats (Alu-IR) at the LYS2 locus induce Sae2-MRX endonucleolytic cleavage and recombination with a locus carrying a truncated version of LYS2 (lys2::Δ5’). e Recombination rates were calculated using a fluctuation analysis. n = 3, 7–8 fluctuations used per replicate per strain. Box plot shows mean with values of biological replicates, error bars denote SD. f ssDNA accumulation is monitored by resistance to restriction enzyme cleavage after DSB induction via the HO nuclease at the MAT locus. The exo1Δ sgs1-AID dna2-AID background make the assay specific for Sae2-MRX-dependent short-range resection. RS = restriction site. g, h Depletion of Dbf4 induces defect in Sae2-MRX mediated resection. ssDNA accumulation upon DSB induction measured 98 bp downstream (g) and 120 bp upstream (h) the DSB via qPCR after digestion with restriction nucleases RsaI and MseI, respectively. n = 6 biological replicates, shown is mean with values of replicates, error bars denote SD. Reported p-values were calculated using a two-tailed unpaired t-test. See also Supplementary Fig. 3c, d. i Scheme of Sae2 highlighting S/T-D/E sites and S267. j Cells of indicated strains were arrested in G1 or M-phase to monitor the Sae2 phospho-shift. Data are representative of n = 3 biological replicates. See also Supplementary Fig. 3i. k, l Five-fold serial dilutions of indicated strains were grown on YPD plates or YPD plates supplemented with CPT at the indicated concentrations. Data are representative of n = 3 biological replicates. See also Supplementary Fig. 3i, l. Source data are provided as a Source Data file.

Fig. 4. DDK regulates long-range resection via phosphorylation of Dna2.

a Serine 236 of Dna2 is phosphorylated in a DDK-dependent manner. Data from phospho-proteomic experiment of Fig. 1d. Heat-map depicts the z-score. See also Supplementary Fig. 4a, b. b Scheme of Dna2 phosphorylation sites. Green: CDK target sites previously identified; Orange: DDK target site S236, within S/T-S/T motif. c Serine 236 phosphorylation contributes to DDK-dependent Dna2 phosphorylation shift in vivo. Cells of the indicated strains were arrested in G1 or M-phase and samples collected and loaded on a gel to monitor the Dna2 phospho-shift. Data are representative of n = 3 biological replicates. See also Supplementary Fig. 4c, d. d dna2-S236A strain is sensitive to CPT, when EXO1 is deleted. Five-fold serial dilutions of WT, exo1Δ, dna2-S236A, and dna2-S236A exo1Δ strains were grown on YPD plates or YPD plates supplemented with CPT at the indicated concentrations. Data are representative of n = 3 biological replicates. e Depletion of Dbf4 induces defects in STR-Dna2-mediated long-range resection. ssDNA accumulation upon DSB induction was measured via qPCR after digestion with the RsaI restriction nuclease at sites with indicated distances from the DSB. n = 3 biological replicates, shown is mean with values of replicates, error bars denote SD. Reported p-values were calculated using a two-tailed unpaired t-test. See also Supplementary Fig. 4f, g. f DDK-dependent resection defect accumulates in the absence of EXO1. Resection was measured as in (e), but in a background where resection was carried out by Sae2-MRX and STR-Dna2. n = 3 biological replicates, shown is mean with values of replicates, error bars denote SD. Reported p-values were calculated using a two-tailed unpaired t-test. See also Supplementary Fig. 4f, h. Source data are provided as a Source Data file.

Fig. 5. Synthetic activation of DDK activates limited DNA end resection and HR in G1 cells.

a Synthetic activation of DDK in G1-arrested cells. Codon-optimized versions of DBF4 and CDC7 are expressed from bidirectional pGAL1-10 promoter, with Dbf4 carrying D-box mutations that stabilize Dbf4^,. b Expression of DDK induces Sae2 phosphorylation in G1. Cells of the indicated strains were arrested in G1, DDK expression was induced by addition of galactose, and samples collected and loaded on a gel to monitor the Sae2 phospho-shift. Data are representative of n = 2 biological replicates. See also Supplementary Fig. 5a. c Synthetic activation of DDK in G1 allows for limited activation of DNA end resection. Strand-specific accumulation of 3′-ssDNA enriched by RPA-ChIP in G1-arrested cells indicates resection throughout time course (2, 4 h after DSB induction) in WT, sae2-S267E, GAL-DDK, and GAL-DDK sae2-S267E strains. RPA-ChIP signals at the DSB are normalized to DSB-independent RPA signals occurring throughout the genome. Data are representative of n = 2 biological replicates. See also Supplementary Fig. 5b-d. d, e Synthetic activation of DDK allows for limited recombination-mediated repair in G1. qPCR analysis of HR upon DSB induction at 491 kb (Chr. IV) using a donor template at 795 kb (Chr. IV). WT cells lacking the donor template are used as negative control. d Comparison of WT, sae2-S267E, GAL-DDK and GAL-DDK sae2-S267E strains. e Comparison of WT, GAL-DDK, yku80Δ, and GAL-DDK yku80Δ strains. n = 3, box plot shows mean with values of biological replicates, error bars denote SD. Reported p-values were calculated using a two-tailed unpaired t-test. See also Supplementary Fig. 5e-i. f Double kinase mechanism for cell cycle-regulated DNA end resection. CDK and DDK target at least two proteins, Sae2 to control Sae2-MRX-dependent short-range resection and Dna2 to control STR-Dna2-dependent long-range resection. Source data are provided as a Source Data file.