The role of Exo1p exonuclease in DNA end resection to generate gene conversion tracts in Saccharomyces cerevisiae

Abstract

The yeast Exo1p nuclease functions in multiple cellular roles: resection of DNA ends generated during recombination, telomere stability, DNA mismatch repair, and expansion of gaps formed during the repair of UV-induced DNA damage. In this study, we performed high-resolution mapping of spontaneous and UV-induced recombination events between homologs in exo1 strains, comparing the results with spontaneous and UV-induced recombination events in wild-type strains. One important comparison was the lengths of gene conversion tracts. Gene conversion events are usually interpreted as reflecting heteroduplex formation between interacting DNA molecules, followed by repair of mismatches within the heteroduplex. In most models of recombination, the length of the gene conversion tract is a function of the length of single-stranded DNA generated by end resection. Since the Exo1p has an important role in end resection, a reduction in the lengths of gene conversion tracts in exo1 strains was expected. In accordance with this expectation, gene conversion tract lengths associated with spontaneous crossovers in exo1 strains were reduced about twofold relative to wild type. For UV-induced events, conversion tract lengths associated with crossovers were also shorter for the exo1 strain than for the wild-type strain (3.2 and 7.6 kb, respectively). Unexpectedly, however, the lengths of conversion tracts that were unassociated with crossovers were longer in the exo1 strain than in the wild-type strain (6.2 and 4.8 kb, respectively). Alternative models of recombination in which the lengths of conversion tracts are determined by break-induced replication or oversynthesis during strand invasion are proposed to account for these observations.

Keywords: DNA repair; Saccharomyces cerevisiae; gene conversion tract length; mitotic recombination; ultraviolet light.

Figure 1

Patterns of LOH resulting from crossovers, BIR events, and gene conversions. The homologs derived from W303a and YJM789 are shown in red and black, respectively. Ovals and circles indicate centromeres. In A–D, the initiating DSB is on the YJM789-derived homolog. The green, red, and black lines on the right summarize patterns of heterozygosity and homozygosity in the daughter cells after the recombination event. In this depiction, green, red, and black signify heterozygosity, homozygosity for SNPs derived from the W303a homolog, and homozygosity for SNPs derived from the YJM789 homolog, respectively. (A) Crossover initiated with an SCB associated with a 3:1 gene conversion. The blue rectangle outlines the conversion event. (B) Gene conversion (3:1) unassociated with a crossover, initiated by an SCB. (C) Mitotic crossover and an associated 4:0 conversion event. Replication of a chromosome broken in G₁ results in two broken sister chromatids. The repair of this DSCB (one repair event associated with a crossover) generates the observed LOH pattern. (D) BIR event initiated by an SCB.

Figure 2

Colony-sectoring assay for detecting mitotic crossovers. The diploids in our study are homozygous for the ade2-1 ochre mutation. Diploids with this mutation form red, pink, or white colonies, depending on whether they contain zero, one, or two copies of the SUP4-o ochre suppressor tRNA gene, respectively. In our experiments, the diploids were heterozygous for the SUP4-o that was located near the end of either the left arm of chromosome V or the right arm of chromosome IV. These diploids form pink colonies. However, a crossover between SUP4-o and the centromere results in a red/white sectored colony. Note that only half of the crossovers are detected by the sectoring assay. If chromatids 1 and 4, and 2 and 3, cosegregate, a sectored colony is not formed.

Figure 3

Conversion tract lengths in wild-type and exo1 strains. Based on the location of LOH regions, we measured conversion tract length in exo1 strains, both spontaneous events and events induced by UV. These data were compared with measurements performed in wild-type strains (St. Charles and Petes 2013; Yin and Petes 2013). (A) Comparison of UV-induced crossover-associated gene conversion tracts in wild-type and exo1 strains. A total of 107 and 27 conversion tracts were examined in the wild-type and exo1 strains, respectively. (B) Comparison of UV-induced conversion tracts in the exo1 strain. Twenty-seven of the conversion tracts were crossover associated and 54 were unassociated with crossovers. (C) Comparison of spontaneous crossover-associated conversion tracts in wild-type and exo1 strains. One hundred two conversion tracts were analyzed in the exo1 strain and 139 were examined in the wild-type strain.

Figure 4

Distribution of spontaneous crossovers on the right arm of chromosome IV in wild-type and exo1 strains. We show the frequency with which SNPs on the right arm of chromosome IV were involved in a conversion/crossover in the strains JSC25 [wild type (St. Charles and Petes 2013)] and YYy34 (exo1). More than 100 crossovers were mapped in each strain. (A) Crossovers and associated conversions in the wild-type strain. HS3 and HS4 are hotspots for recombination, each containing inverted pairs of Ty elements. (B) Crossovers and associated conversions in the exo1 strain. (C) Location of SNPs used in mapping. Each yellow line shows a SNP position. Note that there is a region of ∼65 kb that contains only one SNP.

Figure 5

Role of Exo1p in regulating HS4 hotspot activity. The HS4 G1-specific hotspot for spontaneous mitotic recombination contains two closely linked Ty elements. Since deletion of one of the elements or expansion of the distance between the elements results in loss of hotspot activity, we previously suggested that the recombinogenic effects of HS4 likely involve the formation of a hairpin or cruciform structure (St. Charles and Petes 2013). The two Ty elements of HS4 (shown in blue) are present on the W303a-derived homolog (red) but not the YJM789-derived homolog (black). One scenario for production of a DSB in G₁ is that a nick on one strand of the inverted repeat (shown by the arrow) is expanded into a gap by Exo1p, allowing formation of a hairpin. A subsequent nick of the sequences at the tip of the hairpin would produce a DSB. The resulting broken ends would require extensive resection (another function of Exo1p) to allow pairing with the YJM789-derived homolog.

Figure 6

Patterns of crossover-associated and crossover-unassociated conversion tracts based on current models of recombination. Black and red lines show DNA strands of the two homologs with arrows marking the 3′ ends. Dotted lines show repair-associated DNA synthesis. For all of the models, we assume that conversion events unassociated with crossovers (NCO) occur as a consequence of SDSA, and events associated with crossovers (CO) reflect resolution of Holliday junctions. We assume that the two broken ends are resected to the same extent (A^WT = B^WT), although our conclusions do not require this assumption (details in text). C^WT shows the length of DNA synthesized by the invading strand. In heteroduplexes with one black strand and one red strand, we show correction of the mismatches to generate two red strands. The resulting conversion tracts are outlined in blue. (A) The length of DNA synthesized by the invading strand (C^WT) equals the amount of resection, and A^WT = B^WT. The expected conversion tract length for events associated with crossovers is about twice that for events unassociated with crossovers. (B) If C^WT < A^WT and B^WT, then the relative length of the crossover-associated tract compared to the crossover-unassociated tract is greater than in A. (C) If C^WT > A^WT and B^WT, one possibility is that, during the SDSA event, the 3′ end of the reinvading strand is displaced. Removal of this end by a branch-processing enzyme such as Rad1p/Rad10p (Mazón et al. 2012) would result in a conversion tract in the NCO pathway that is the same length as in A.

Figure 7

Two models for the formation of long gene conversion tracts that are unassociated with crossovers. The data that require explanation are (1) crossover-associated conversion events are shorter in the exo1 strain than in wild type and (2) crossover-unassociated conversions are longer than crossover-associated conversions in the exo1 strain. The labels are the same as in Figure 6. The observations can be explained by an oversynthesis model (A and B) or by a BIR model (C and D). (A) UV-induced conversion in the wild-type strain (oversynthesis model). The synthesis from the invading strand is more extensive than the amount of DNA resected in the NCO pathway. The reinvasion during SDSA displaces the 5′ end of invaded homolog, and the resulting single-stranded branch is removed (shown by a triangle). In the CO pathway, the heteroduplex region is limited by resection. (B) UV-induced conversion in the exo1 strain (oversynthesis model). The conversion tract in the NCO pathway is similar to that observed in wild type, whereas the conversion tract for the CO pathway is shorter due to less resection. (C) UV-induced conversion in the wild-type strain (BIR model). In the NCO pathway, the conversion is a consequence of BIR. Following copying of the red chromosome, the invasion is reversed, and the end generated by BIR reassociates with the broken black chromosome. In this model, the conversion tract is not a consequence of repair of mismatches in a heteroduplex. The conversion events in the CO pathway occur by the same mechanism as in A and B. (D) UV-induced conversion in the exo1 strain (BIR model). As in C, the BIR event generating the conversion tract in the NCO pathway is not limited by resection, unlike the conversion tract in the CO pathway.