Standard media were used for the culture of yeast and bacterial strains (Sherman et al., 1986). LB-AMP (Luria broth containing 100 μg/mL ampicillin) was used for the culture of the Escherichia coli DH1 strains. Media used for the culture of yeast cells included YPD (yeast extract, peptone, dextrose), SC (synthetic complete, 2% dextrose), SC-HIS (SC lacking histidine), and SC-ADE (SC-lacking adenine). YPD-Kan (G418) plates contain YPD supplemented with 50 μg/mL G418 (Sigma). YP(A)D contains YPD with 80 mg/L adenine. Gene knockouts were performed by one-step gene replacement (Rothstein, 1983), using standard lithium acetate transformation protocols (Gietz et al., 1995).

All the yeast strains, including BY4741 (Brachmann et al., 1998), and YA148 (Lesser and Guthrie, 1994) are of the S288c genetic background. The wild-type diploid strain containing the his3 recombination substrates (Fasullo and Davis, 1987) on one copy of chromosomes II and IV has been previously described. Checkpoint and recombination mutants were constructed by gene disruptions and genetic crosses. Essentially, two sets of isogenic haploid strains were constructed; one set contains the recombination substrates, and the other does not. The homozygous diploid strains, rad9 (YB134), rad51 (YB170), and mec1-21 (YB325), which also contain the recombination substrates, have been previously described (Fasullo et al., 1998; Fasullo et al., 2001; Fasullo et al., 2010). Additional homozygous diploid strains were obtained by knocking out RAD51, RAD55 RAD57, RAD50, RAD54, MRE11, XRS2 in respective MAT a and MATα haploid strains, using plasmids pRAD51Δ (Shinohara et al., 1992), pST11 (Lovett and Mortiner, 1987), pSM51 (Schild), pNKY83 (Alani et al., 1989), pSM31 (Schild), pSK-MRE11delta (Bressan et al., 1999), pET139 (Ivanov et al., 1994), respectively. The rad9 rad51, rad9 rad50, rad9 mre11, rad9 rad54 double mutants were constructed by knocking out the RAD genes in MATa rad9 and MATα rad9 haploids. The rad51 rad1 double mutants were obtained by knocking our RAD51 in rad1 haploid strains, and the diploid was obtained by mating the haploid double mutants. The rad51 rad9 double mutants were made by genetic crosses and mating the haploid double mutants. All disruptions were confirmed by radiation sensitivities and the haploid strains were then mated to generate the diploid strains used in this study. The radiation sensitivities of diploid mutants were confirmed by exposing inoculated YPD plates to 60 J/m² and 6 krads of X rays and imaging after 3 days of 30°C incubation (see Supplementary Figure S1). PCR was used to confirm rad59::KanMX gene disruptions. The presence of the his3-Δ200 allele was confirmed by PCR analysis using the primers 5′-CAC​GGC​AGA​GAC​CAA​TCA​GTA-3′ and 5′-GCA​CTC​CTG​ATT​CCG​CTA​ATA-3’.

Haploids containing the recombination substrates for measuring unequal sister chromatid recombination (uSCR) and diploid strains containing ade2-a and ade2-n for measuring homolog recombination were previously described (Fasullo et al., 2010). The mec1-21 rad51 haploid strain containing the his3 recombination substrates was constructed by knocking out RAD51 in the mec1-21 haploids. Construction of the mec1-21 mutant diploid strains has been previously described (Fasullo et al., 2010). The mec1-21 rad9 double mutant was constructed by knocking out RAD9 in the MAT a and MATα mec1-21 haploids and then mating the haploid double mutants.

The rates of spontaneous, mitotic events that generate either SCE, heteroallelic or translocations were determined by the method of the median (Lea and Coulson, 1949), as performed by Esposito et al. (1982), using 11 independent colonies for each rate calculation. This method is essentially the same as that employed by Spell and Jinks-Roberston (2004). In all rate calculations, the number of cell divisions was estimated by the number of viable cells that formed colonies. Statistical analysis difference was determined by the by the Mann-Whitney U test, where there are at least two independent calculations for each mutant strain (Zar, 1996). To measure the rates of homolog recombination between ade2 heteroalleles, colonies were obtained from cells inoculated on YP(A)D solid medium to suppress expression of the adenine pathway. To determine whether the rate of recombination increased in the diploid mutants, compared to the single mutants, the interaction factor was calculated according to David et al. (2016), see Supplementary Table 3.

The number of His⁺ recombinants stimulated by DNA-damaging agents was determined by subtracting the spontaneous frequency from the stimulated frequency and multiplying by 10⁷, the approximate number of cells plated, as done previously (Fasullo et al., 1994). The significance of the differences between mutants and Rad⁺ (diploid) strains was determined by using the two-tailed paired sample t-test (Zar, 1996). Protocols used to assess the recombinogenicity of UV and X rays have been described elsewhere (Fasullo et al., 1994). The X-ray radiation source was purchased from Rad Source, Inc. (Wheeling, Ill.), and the dose rate was 440 rads/min. For measuring radiation-associated stimulation of translocations, cells were washed twice in sterile H₂O, resuspended in H₂O, irradiated, and then plated on selective medium (SC-His) for selecting for recombinants and inoculating an aliquot of the appropriate dilution on YPD medium to measure viability. Typically, 20–200 colonies were counted on SC-HIS plates, and 100–300 colonies were counted on YPD plates. The UV source emitted 260 nm UV light at a dose rate of 2 J/m². One-way ANOVA followed by Dunnet’s test was used to determine the statistical significance of differences between radiation-associated frequencies (Zar, 1996).

Rates of spontaneous, mitotic translocations were measured in diploid and haploid strains using his3 constructs that are shown in Figure 1. We measured rates of spontaneous translocations in both diploid single and double mutants that were defective in RAD51, RAD55, RAD57 and the checkpoint gene RAD9. The rate of spontaneous translocations in the Rad⁺ diploid was 3 × 10⁻⁸ and was similar to previous measurements (4 × 10⁻⁸, average) obtained from three independent Rad⁺ diploids (Fasullo et al., 1998; Fasullo et al., 2004; Fasullo et al., 2010). Knocking out either RAD51, RAD55, or RAD57 conferred a ten-fold or greater increase in translocation frequency (Table 1). Diploid mutants defective in MRE11, RAD50,or XRS2, which encode the MRX complex, exhibited a 22–27-fold increase in the rate of homology-directed translocations. However, knocking out RAD54, only conferred a two-fold increase in recombination, compared to wild type, which was not significant (P > 0.05, Mann-Whitney), while knocking out RAD59, conferred a two-fold decrease; RAD54 functions at multiple stages in the initiation and processing of recombination intermediates (Heyer et al., 2006). These data indicate that knocking out the RAD51 paralogs, RAD55 and RAD57, confers a similar phenotype as that of knocking out RAD51, and that knocking out genes encoding the MRX complex conferred the highest increase in translocation frequencies among the RAD50 group. Thus, specific genes that function in recombinational repair of DSBs suppress homologous recombination between short repeated sequences on non-homologous chromosomes.

Checkpoint and DNA repair mutants exhibit higher rates of spontaneous translocation events in diploid cells; however, it is unclear whether these pathways function independently to suppress recombination. We therefore determined whether knocking out both checkpoint and recombination genes would lead to a synergistic increase in recombination. Failure to arrest the cell cycle in G₂ can also increase the frequency of DNA damage-associated chromosomal rearrangements. Compared to wild type, the rad9 diploid mutants exhibit a seven-fold increase in the rate of spontaneous recombination, similar to previous reports (Fasullo et al., 1998). We measured rates of translocations in double mutants defective in both rad9 and in recombinational repair genes (Table 1). The rate of translocations in rad9 mutants defective in either RAD51, RAD55, or RAD57 were similar and 57–77-fold higher than the wild-type rate. We also observed a 55-fold increase in rate of translocations in rad9 rad54 double mutant, compared to the rate in wild type. The fold increase in rates of the double mutant in comparison to the single mutants was thus synergistic, i.e., the increase was greater than additive (Perez-Perez et al., 2009), as indicated by the positive IF (see Supplementary Table 3). However, the synergistic increase in recombination is dependent on ploidy; the rate of spontaneous translocations was (2.0 ± 0.2) × 10⁻⁸ in the rad9 rad51 haploid, a rate below 5.4 × 10⁻⁸, observed in rad9 haploids. The higher rate of spontaneous recombination observed in rad51 diploids is also dependent on RAD59, indicating that RAD59 has a role in RAD51-independent ectopic recombination between these short-repeated sequences. This indicates that recombinational repair and RAD9 checkpoint and RAD51 participates in independent pathways for suppressing homology-directed translocation events in diploid strains.

The MRX complex is required in checkpoint signaling (Grenon et al., 2001). We also measured the rates of spontaneous recombination in mre11, xrs2, and rad50 diploid mutants that were defective in the RAD9 checkpoint function (Table 1). While rad9 mre11 and rad9 xrs2 double mutants exhibited over a 40-fold increase in rates of translocations., rates of translocations in rad50 and rad9 rad50 were still in the range of 20–25 ‐fold increase, compared to wild type. However, the radiation sensitivities of the rad9 mre11, rad9 xrs2, and rad9 rad50 mutants were uniformly higher than that of the mre11, xrs2, and rad50 single mutants (Supplementary Figure S1), and all the double mutants generate microcolonies after irradiation (Supplementary Table S2). Similarly, while there is a noticeable delay in the doubling time for the mre11, xrs2, and rad50 single mutants (100–110 min), the doubling time for the rad9 mre11, rad9 xrs2, and rad9 rad50 mutants was approximately the same as wild type (80–90 min). The data indicate that knocking out RAD9 in the mre11, xrs2 and rad50 single mutants increases radiation sensitivity while increasing the rate of translocations in either mre11 or xrs2 but not in rad50 mutants.

Previous studies had shown that defects in S phase checkpoint function also increase translocations that result from homologous recombination between short repeats. The mec1-21 mutant (Craven and Petes, 2000) is a mec1 hypomorph that is defective in S phase checkpoint function but retains some G₂ checkpoint function (Sun and Fasullo, 2007). We had previously measured recombination in the diploid mec1-21 mutant and observed a 23-fold increase in rates of spontaneous translocations, compared to the rate observed in the wild-type diploid (Fasullo et al., 2010). After RAD51 was knocked out in the mec1-21 diploid, the rate increased to 2.9 × 10⁻⁶, which is approximately 100-fold higher than what is observed in the wild-type diploid (Table 2). The rate of spontaneous translocations in the mec1-21 rad51 diploid is similar to the rate of spontaneous unequal siter chromatid exchange between juxtaposed his3 recombination substrates observed in Rad⁺ strains (Dong and Fasullo, 2003). However, knocking out RAD9 in the mec1-21 diploid decreased the rate of translocations, compared the rate observed in mec1-21, although it was still higher than wild type (Fasullo and Sun, 2008a). These data indicate that MEC1 and RAD51 participate in independent pathways for suppressing homology-directed chromosomal rearrangements. These data are consistent with the notion that genetic instability phenotypes are highest in S phase checkpoint mutants (Myung et al., 2001; Myung et al., 2004; Pennaneach and Kolodner, 2004).

We previously observed that the rates of spontaneous unequal sister chromatid recombination (uSCR) and homolog recombination between ade2 heteroallelic increased in mec1-21 mutants, compared to wild type (Table 2; Fasullo et al., 2010). While there is no significant difference in the rates of spontaneous uSCR in rad51 mutants and wild type, the rate of spontaneous, homolog recombination between ade2 heteroalleles cannot be detected (<1 × 10⁻⁸), in agreement with results obtained by Bai and Symington (1996). However, in diploid mec1-21 rad51 mutants recombination between ade2 heteroalleles was detected, and in the mec1-21 rad51 haploid the spontaneous rate of uSCR was three-fold higher than wild type. These data indicate that multiple types of homologous recombination events are enhanced in the mec1-21 strain, which are variably affected by knocking out RAD51.

Knocking-out RAD9 had no significant effect on the rates of unequal uSCR and homolog recombination between heteroalleles, consistent with observations that rad9 mutants do not generally exhibit mitotic phenotypes (Weinert and Hartwell, 1990). Similar to the RAD9 requirement for the high translocation frequencies observed in the mec1-21 diploid, knocking out RAD9 in diploid and haploid mec1-21 mutants also decreased homolog recombination and uSCR, respectively (Table 2). These data indicate diverse types of mitotic recombination events are increased in mec1-21 mutants which are variably affected by knocking out RAD51 and RAD9.

The DNA lesions that initiate spontaneous translocation events are unknown. One explanation for the higher relative rates of spontaneous translocations in rad51 and rad55 diploid mutants is that spontaneously generated DSBs stimulate more homology-directed translocations in these rad mutants compared to those stimulated in wild type. In addition, both rad51 and rad55 mutants are defective in DNA damage-associated uSCR (Dong and Fasullo, 2003). We therefore exposed rad51 and rad55 diploid mutants to 2, 4, 6, and 8 krads of X rays and measured frequencies of X-ray associated translocations (Figure 2). For all exposures, the X-ray-associated frequencies of translocations in the rad mutants were significantly higher than wild-type frequencies (P < 0.05). We observed a dose-dependent increase in the frequencies of recombination wild type, rad51, and rad55 mutants (P < 0.05); however, the dose-dependent differences were most significant for the rad55 mutant (P < 0.01, Dunnett’s). Compared to the frequencies of spontaneous recombination, translocation frequencies after 8 krad exposure were 13-fold and 51-fold higher for rad51 and rad55, respectively, compared to 7-fold for wild type (Figure 2). The data indicate that frequencies of X-ray-associated translocations are similar in both the rad55 mutants and rad51 mutants and suggest that knocking-out other RAD51 paralogs may also confer increased X-ray associated genetic instability.

Considering that mre11, rad50, and xrs2 diploid mutants also exhibit higher rates of spontaneous translocations, compared to wild-type diploid, we measured the radiation-associated translocations frequencies were higher in mre11, rad50, and xrs2 mutants (Figure 3). We exposed wild type, and mre11, rad50, and xrs2 mutants to both 2, 4, and 8 krads, as well as 60, 90, 120, and 150 J/m². While the radiation-associated frequencies of translocations in wild type are significantly lower than all those in the rad mutants (P < 0.05), there is more than a 100-fold increase in UV-associated frequencies (6.8 × 10⁻⁶, avg.) and X-ray associated frequencies (1.3 × 10⁻⁵, avg.) in wild type, compared to the non-irradiated wild-type control (6 × 10⁻⁸, avg.). The frequencies of X-ray and UV-associated translocations were highest in the mre11 mutants and sixfold higher than the frequencies obtained from the non-irradiated control (Figure 3); the dose dependence is significant for both X-ray and UV exposures (P < 0.05, Dunnett’s). On the other hand, the frequencies of X-ray and UV-associated translocations are similar in the xrs2 and rad50 mutant, and while frequencies of UV and X-ray associated frequencies was dose-dependent for the xrs2 mutant (P < 0.05, Dunnett’s), only the X-ray associated frequencies were significant for the rad50 mutants (P < 0.01, Dunnett’s). These data illustrate that ectopic recombination between homologous sequences can be stimulated by radiation in mre11, rad50, and xrs2 mutants, which are defective in the processing of DSBs.

We previously observed that the frequencies of radiation-associated rearrangements increase in rad51 (Fasullo et al., 2004) and rad9 mutants (Fasullo et al., 1998), defective in homologous recombination and G₂ checkpoint control, respectively. However, it was unclear whether radiation-associated translocations could be detected at radiation exposures less than 1 krad (10 Gy); on average, 0.58 DSBs are introduced per krad in a diploid yeast cell (Frankenberg-Schwager and Frankenberg, 1990; Westmoreland and Resnick, 2016). We therefore determined whether rad9, rad51, and rad9 rad51 diploid mutants exhibit higher levels of radiation-associated translocations frequencies after exposures of less than 1 krad. We measured translocations frequencies after exposure to 1 krad, 500 rads, 200 rads and 100 rads of X-radiation (Figure 4). While after 1 krad X-ray exposure we observed a seven-fold increase in the frequencies of translocations in wild type diploids, we observed a 36-fold and 47-fold increase in radiation-associated frequencies in rad9 and rad51 mutants, respectively, but only a three-fold increase in the frequencies of translocations in the double rad9 rad51 double mutant. However, in the Rad⁺ diploid we did not observe higher levels radiation-associated frequences after cells were exposed to lower levels of radiation (Figure 4). Collectively, frequencies of radiation-associated translocations were significantly different in rad9, rad51, rad9 rad51 after radiation exposures at all doses, compared to frequencies of spontaneous translocations (ANOVA, P < 0.05); radiation-associated frequencies in rad9, rad51, and rad9 rad51 mutants after exposure to 500 rad were all significantly different than frequencies of spontaneous translocation. rad9 rad51 mutant showed the highest dose-associated significance after 200 rad (Dunnett’s, P < 0.05), 500 rad (Dunnett’s, P < 0.01, and 1 krad (Dunnett’s, P < 0.001), compared to the other strains. Compared to net frequencies of translocations after 100 rad exposures in wild-type cells, those obtained in rad51 (Dunnett’s, P < 0.04) and in rad51 rad9 cells (Dunnett’s, P < 0.003) were significantly different. Interestingly, the rad9 rad51 diploid mutant was more resistant to ionizing radiation at low radiation doses than the rad51 mutant. We speculate that the rad9 rad51 diploid mutant may be able to tolerate low levels of radiation than rad51 mutant due to lack of checkpoint function. On the other hand, we speculate that His⁺ recombinants generated in rad9 rad51 mutants after exposure to higher levels of radiation may contain more chromosomal rearrangements or chromosomal loss events, which may limit their viability. These studies thus indicate that frequencies of radiation-associated rearrangements increase in yeast mutants defective in DSB repair at exposures to low doses (100 rads) of ionizing radiation.