Fig 2: Rad9 favours Rad51 at donor site and D-loop extension in BIR.a Scheme of the PCR-based assay to test D-loop extension and ChIP in JRL092 background. Rad51 enrichment at the cut site on chromosome V (b) at 3 (n = 3, biologically independent experiments) and 4.5 h (n = 2, biologically independent experiments), after DSB induction in the indicated JRL092 strains. Rad51 enrichment at the donor site on chromosome XI (c) at 3 and 4.5 h after DSB formation in the indicate JRL092 strains (n = 2, biologically independent experiments). In both (b) and (c) the cells were blocked in G2/M with nocodazole. d PCR-based assay to test D-loop extension in the indicated JRL092 strains, (*) indicates the 1000 bp band of the 1 kb DNA ladder (NEB). e Quantification of the D-loop extension in four independent experiments as in (d). All the data in the figure are presented as mean ± SEM wherever n = 3 or more. Statistical analysis was done using unpaired two-tailed Student’s t-test. See accompanying Source data file.

Fig 3: A model for repeat instability during resection and gap filling at a CTG repeat.a In wildtype cells, as resection occurs, Rad9 controls the recruitment of RPA and Rad51 to ssDNA. Delayed binding of RPA or Rad51 results in hairpin formation. Contractions are a result of hairpin bypass during polymerase mediated gap filling (shown) or by SSA/end-joining events within the broken CTG repeat region (Supplementary Fig. 3g). In addition, exposed ssDNA results in increased breakage and alternative repair. b In rad9? mutants, there no delayed recruitment of RPA and Rad51 to the freshly resected ssDNA. This prevents hairpin formation and breakage at the repeat tract resulting in increased viability and decreased contractions. c In rad51?, initiation of resection is delayed. Absence of Rad51 and slowed gap filling both allow for increased DNA structure formation and contractions. Breaks that occur at the CTG repeat tract cannot repair via BIR leading to increased cell death.

Fig 4: Deletion of RAD51 impairs repair and results in increased contractions during gap filling.a For the scrm(CTG)70 template strain: viability (%) of the rad51? mutant (n = 8) is significantly decreased compared to wildtype (n = 12) (p = 0.002) and the rad9? rad51? mutant (n = 6) is significantly decreased compared to the rad51? mutant (p = 0.0002). For the (CTG)70 template strain: viability (%) of the rad51? mutant (n = 12) is significantly decreased compared to wildtype (n = 9) (p = 0.0002) and the rad9? rad51? mutant (n = 8) is significantly decreased compared to the rad51? mutant (p = 0.04). Each n value represents assays from biologically independent experiments. Statistical significance was determined using an unpaired, two-tailed Student’s t test. b No break condition: rad51? (n = 119) and rad9? rad51? (n = 96) mutants had decreased contractions compared to wildtype (n = 120; p = 0.15; p = 0.38 respectively). DSB condition: rad51? (n = 119) and rad9? rad51? (n = 120) mutants had increased contractions compared to wildtype (n = 119; p = 0.26 for both). Each n value represents a PCR of an independent colony, statistical analysis by Fisher’s exact test. c U2 repair measurement (%) on Southern blots after DSB induction. Number of replicates: scrm(CTG)70 (n = 3), rad51? scrm (CTG)70 (n = 2), (CTG)70 (n = 4), and rad51? (CTG)70 (n = 2) where each n represents biologically independent time courses. Graph shows mean ± SD. d Promoter homology repair measurement (%) on Southern blots after DSB induction. Number of replicates: scrm(CTG)70 (n = 3), rad51? scrm(CTG)70 (n = 2), (CTG)70 (n = 4), and rad51? (CTG)70 (n = 2) where each n represents biologically independent time courses. Graph shows mean ± SD. e Resection and fill-in kinetics for the (CTG)70 template in wildtype (n = 4) and rad51? (n = 2) strains 600 bp after the repeat locus after DSB induction where each n represents biologically independent time courses. f Enrichment of RPA 60 bp after the (CTG)70 repeat tract in wildtype and rad51? mutants following DSB induction. Independent biological replicates for wildtype (n = 2) and rad51? (n = 2) where each n represents biologically independent time courses. Enrichment was determined using P3 & P4 and calculated using absolute quantity and normalized to ACT1. Bars on graph depict the mean; • and ? each indicate one experimental replicate. For (a–f) source data are provided as a Source Data file.

Fig 5: Deletion of RAD9 rescues the decreased viability and gap fill-in mediated (CTG)70 contractions.a For the scrm(CTG)70 template strain, percent viability of the rad9?mutant (n = 6) is unchanged compared to wildtype (n = 12). For the (CTG)70 template strain, percent viability of the rad9? mutant (n = 5) is significantly increased compared to wildtype (n = 9) (p = 0.0002). Each n value represents assays from biologically independent experiments. Statistical significance determined using an unpaired, two-tailed Student’s t test. b In the (CTG)70 template strain, the contraction frequency in the no-break condition (n = 120) decreased in the rad9? mutant (n = 163), p = 0.04. In the DSB condition, the contraction frequency decreased more significantly in the rad9? mutant (n = 163), p = 0.001 compared to wildtype (n = 119). Each n value represents a PCR of an independent colony, statistical analysis by Fisher’s exact test. c Percent ssDNA 600 bp after the repeat locus was determined after DSB induction as in Fig. 2b; wildtype n = 4, rad9? n = 3. Graph shows mean ± SD where n represents biologically independent time courses. Statistical significance determined unpaired Student’s t test using a two-stage step-up with a false-discovery rate of 1% (Benjamini, Krieger, and Yekutieli). d U2 repair measurement (%) on Southern blots after DSB induction. Statistical significance determined unpaired Student’s t test using a two-stage step-up with a false-discovery rate of 1% (Benjamini, Krieger, and Yekutieli) where rad9? scrm (CTG)70 (n = 3) was compared to scrm(CTG)70 (n = 3) and rad9? (CTG)70 (n = 3) was compared to (CTG)70 (n = 4). Each n represents biologically independent time courses. Enrichment of (e) RPA and (f) Rad51 60 bp after the (CTG)70 repeat tract occurs earlier in the rad9? mutant following DSB induction compared to wildtype. Independent biological replicates for wildtype (n = 2) and rad9? (n = 2). Enrichment adjacent to the (CTG)70 repeat was determined using primers P3 & P4 and calculated using absolute quantity and normalized to ACT1. Bars on graph depict the mean; • and ? each indicate one experimental replicate. For (a–f) source data are provided as a Source Data file.