After we mapped and validated the in vivo SUMOylation sites of the core HR proteins, we then addressed the functional significance of the SUMO modifications in DNA repair. Because HR protein group SUMOylation fosters physical interactions between modified repair factors promoting complex formation, we previously speculated that multiple modifications might act synergistically or have redundant roles. Indeed, SUMOylation-defective mutants of individual HR proteins did not exhibit reduced resistance to DNA-damaging treatment ((Sacher et al., 2006) and data not shown). We thus wondered whether elimination of the complete set of SUMOylation sites in the core HR proteins would affect the efficiency of recombination-mediated DNA repair.
For that purpose, we constructed a mutant strain in which we combined SUMOylation-defective variants of the core repair proteins RPA, Rad52 and Rad59 (rfa1-K133R,K170R,K427R rfa2-K199R rfa3-K46R rad52-K10R,K11R,K220R rad59- K207R,K228R,K238R) that were generated by pop-in/pop-out allele replacement method. The resulting strain (termed KR mutant) carried 11 substitutions of the SUMO-acceptor lysines to arginines in the HR proteins that form SUMO-SIM- assisted repair complexes after DNA damage required for efficient loading of the Rad51 recombinase onto resected DSB ends.
RESULTS
The generated KR strain grew at WT rates in the absence of DNA damage, however a substantial growth delay of roughly 4 hours was observed upon chronic exposure to MMS (Fig. 29A). Furthermore, both spontaneous and MMS-induced interchromosomal recombination rates measured between his1 heteroalleles in KR
mutant diploid cells were significantly reduced, similar to cells lacking SUMO ligase Siz2 (Fig. 29B). In general, the KR mutant and Siz2 deficient ( siz2) strains exhibit largely similar phenotypes and are epistatic (Fig. 29) under tested conditions, suggesting that in the recombination-mediated DNA repair Siz2 acts mainly through SUMO modification of HR factors. Thus, Siz2-dependent protein group SUMOylation facilitates HR and supports efficient repair of DNA lesions that rely on recombination.
Figure 29. Protein Group SUMOylation is Required for Efficient HR
(A) Cells deficient in HR protein SUMOylation either due to the absence of Siz2 ( siz2) or the lack of 11 SUMO-acceptor sites in the core HR proteins RPA, Rad52 and Rad59 (KR) show delayed growth upon chronic exposure to DNA damage. Growth curves for WT cells, cells deficient in Rad52 ( rad52) or Siz2 ( siz2), and cells deficient in SUMOylation sites of HR proteins (KR) and also in combination with siz2 (KR siz2), were measured in YPD medium or YPD containing 0.02% MMS. (B) Deficiency in HR protein SUMOylation results in reduction of spontaneous and MMS-induced recombination. The interchromosomal recombination rates between chromosomalhis1 heteroalleles were measured by fluctuation analysis in WT cells, cells lacking Siz2 ( siz2) and mutant cells deficient in SUMOylation sites of HR proteins (KR). The results are the average of at least three independent studies and error bars represent SD.
RESULTS
III.5.2 SUMOylation in HR Accelerates DSB Repair by Promoting Rad51 Loading
After determining that protein group SUMOylation facilitates HR, we next addressed how it might be mechanistically achieved using yeast mating-type switching mechanism as a case study (Connolly et al., 1988; Weiffenbach and Haber, 1981; White and Haber, 1990). During mating-type switching inS. cerevisiae, a single DSB is generated by the HO endonuclease specifically at the mating type (MAT) locus, which is subsequently repaired by HR using either of two homologous donor sequences (HML or HMRa, located on the same chromosome III with MAT). We first confirmed that a single DSB is sufficient to trigger protein group SUMOylation in HR using donor deficient ( hml hmr) strain defective in repair (Fig. 30). Indeed, SUMOylation wave was induced, however to a much lesser extent compared with MMS-treatment that generates many DNA lesions with long tracts of ssDNA.
Figure 30. A Single DSB is Sufficient to Trigger SUMOylation of HR Proteins
A single DSB is generated by HO endonuclease at the MAT locus of the yeast mating-type switching system (top; map of chromosome III) to follow the induction of SUMOylation of HR proteins. HO is expressed from a galactose-inducibleGAL-promoter, and repair is prevented by deletion of HML and HMR homologous donor sequences (donor deficient strain). Two hours after HO induction, mono- and di-SUMOylated species of RPA (Rfa1) and Rad52 appear. Detection by Western blotting following Ni-NTA pull-down of HisSUMO conjugates. MMS-induced SUMOylation is shown for comparison.
RESULTS
Because Siz2-mediated protein group SUMOylation of the core HR factors fosters physical interactions within the RPA-Rad52-Rad59 complexes that are required for efficient loading of the Rad51 recombinase onto resected DSB ends, we next asked whether Rad51 nucleofilament formation in mating-type switching system is affected in the absence of Siz2. The DSB formation was rapidly induced at the
MAT locus of donor deficient WT and siz2 cells by constant expression of HO endonuclease under the control of GAL-promoter (Fig. 31A). When we then monitored the recruitment of the HR factors at sites 0.2 kb and 5.7 kb distal to the persistent DSB by chromatin immunoprecipitation (ChIP), we observed that RPA (Rfa19Myc) was efficiently loaded to the resected ssDNA with similar kinetics in both WT and siz2 cells (Fig. 31B, top). However, recruitment of the Rad51 recombinase was substantially delayed when HR protein group SUMOylation was abolished (Fig. 31B, bottom).
Inefficient Rad51 nucleofilament formation in the absence of HR protein SUMOylation should result in a delay of DSB repair, similar to the delay observed when DSB-end resection is compromised in cells lacking MRX complex ( rad50) (Ivanov et al., 1994). We therefore transiently induced the DSB and directly Figure 31. Rad51 Loading at the DSB is Affected in the Absence of SUMOylation in HR
(A) A single unrepairable DSB was induced by HO-endonuclease at MAT of cells lacking HML and HMR (donor-deficient strain). DSB-induction was monitored in WT and Siz2- deficient cells ( siz2) by real-time quantitative PCR with primers (PMAT) spanning the HO-cut
site. (B) ChIP directed against C-terminally Myc-tagged Rfa1 and Rad51 at 0.2 and 5.7 kb distal from DSB was performed 1, 2, and 3 hrs after HO-induction to compare loading of RPA (top) and Rad51 (bottom) in the absence of HR protein SUMOylation. The results are the average of at least three independent studies and error bars represent SEM.
RESULTS
monitored the speed of its repair (mating type switching) in donor proficient cells that harbor homologous sequence (HML ) as well as efficiently induce SUMOylation wave in HR (Fig. 32A). Indeed, we found that the DSB repair was significantly delayed in theKRmutant strain (Fig. 32B) compared to WT. Thus, HR protein group SUMOylation accelerates DSB repair by fostering SUMO-SIM-assisted repair protein complex formation that is required for efficient Rad51 nucleofilament assembly at resected DSB ends.
Figure 32. The Speed of DSB Repair is Reduced in the Absence of SUMOylation in HR
(A) A single DSB can trigger SUMOylation of HR proteins independent of the absence (top panel) or presence (bottom panel) of homology required for repair. Experiment was conducted similar to (Fig. 30). (B) A single repairable DSB (arrow indicates repair reaction) was transiently induced by HO at MATa in WT cells, cells lacking the MRX subunit Rad50 ( rad50) defective in DSB-resection initiation, or the Rad52 protein ( rad52), and the KR mutant. Repair kinetics was measured by real-time quantitative PCR (with primers PAand PB)
following the appearance of repaired productY at MAT, 1, 2, and 3 hrs after inactivation of HO (shift from galactose to glucose-containing media). The results are the average of at least three independent studies and error bars represent SEM.