Haploid Saccharomyces can change mating type through HO-endonuclease cleavage of an expressor locus, MAT, followed by gene conversion using one of two repository loci, HML or HMR, as donor. The mating type of a cell dictates which repository locus is used as donor, with a cells using HML and α cells using HMR. This preference is established in part by RE, a locus on the left arm of chromosome III that activates the surrounding region, including HML, for recombination in a cells, an activity suppressed by α2 protein in α cells. We have examined the ability of RE to stimulate different forms of interchromosomal recombination. We found that RE exerted an effect on interchromosomal mating-type switching and on intrachromosomal homologous recombination but not on interchromosomal homologous recombination. Also, even in the absence of RE, MATα still influenced donor preference in interchromosomal mating-type switching, supporting a role of α2 in donor preference independent of RE. These results suggest a model in which RE affects competition between productive and nonproductive recombination outcomes. In interchromosome gene conversion, RE enhances both productive and nonproductive pathways, whereas in intrachromosomal gene conversion and mating-type switching, RE enhances only the productive pathway.
HAPLOID Saccharomyces cells have the remarkable potential to change mating type as often as every generation (reviewed in Haber 1998; Bi and Broach 1999). The mating type of a haploid cell is dictated by the particular allele, a or α, present at the mating-type locus, MAT, located near the center of chromosome III. Mating-type switching is initiated by a double-strand break at the MAT locus, catalyzed by an endonuclease encoded by HO (Strathernet al. 1982). Switching then occurs by a gene conversion event that replaces the mating information at the MAT locus with the opposite mating information present at either of two repository mating loci, HML and HMR, located at the opposite ends of chromosome III (180 and 90 kb, respectively, away from MAT). This results in replacement of one mating-type allele at MAT with a copy of the opposite mating-type allele taken from either HML or HMR.
Mating-type switching follows a precise developmental pattern (Strathern and Herskowitz 1979). This precise choreography results from the intricate pattern of transcriptional regulation of the HO gene and from a highly regulated interaction between distant regions of chromosome III. Namely, mothers, but not daughters, transcribe the HO gene and do so only during the G1 phase of the cell cycle. Accordingly, only mothers are capable of switching cell type and the switch occurs prior to DNA replication in the mother (Nasmyth 1993; Longet al. 1997; Nasmyth and Jansen 1997). In addition, cell type dictates which donor locus is selected for participation in the gene conversion event. a cells predominantly select HML, which normally contains α mating information, whereas α cells select HMR, which normally contains a mating information (Figure 1, bottom). This pattern ensures that most of the switching events result in a change of mating type, rather than a futile replacement of the MAT allele with the same sequence (Klaret al. 1982).
The fact that cells can select between HML and HMR loci implies that the two loci possess distinguishable features that are recognized in a cell type-specific manner. Previous studies have shown that this discrimination does not derive from the different alleles resident at the donor loci, from the unique sequences flanking either locus, from any of the DNA sequences distal to either locus on chromosome III, or from any preorganization of the donor and acceptor loci within the nucleus (Weiler and Broach 1992; Simonet al. 2002). Sequences flanking MAT also do not function in the process of preferential selection. Rather, the left arm of chromosome III exhibits a cell type-dependent difference in ability to participate in recombination. A large (>40 kb) region of the left arm of chromosome III containing HML can undergo intragenic recombination between separated heteroalleles of a test gene at a rate up to 20 times higher in MATa cells than in MATα cells (Wu and Haber 1995). The MATa cell-specific enhancement of recombination of the left arm of chromosome III depends on a small (<2 kb) segment ∼30 kb from the telomere, 16 kb from HML (Wu and Haber 1996). Deletion of this recombinational enhancer (RE) causes MATa cells to choose HMR (the wrong donor) instead of HML, without altering the donor preference in MATα cells (HMR is preferred as usual; Wu and Haber 1996; Szetoet al. 1997). RE contains binding sites for the Mcm1 and forkhead (Fkh1 and Fkh2) transcription factors and FKH1 deletion or certain inactivating mutations of Mcm1 eliminate the ability of RE to promote selection of HML in MATa cells (Wuet al. 1998; Sunet al. 2002).
Of the three transcriptional regulators encoded by the two MAT alleles—a1, α1, and α2—only α2 plays a role in donor selection during mating-type interconversion. Cells expressing α2 (typically α cells) select HMR as donor while cells lacking α2 (typically a cells) select HML as donor, independent of other mating-type gene products (Szeto and Broach 1997). α2 functions in transcriptional regulation by interacting with Mcm1 and Tup1/Ssn6 to repress a-specific genes in α cells. α2 acts similarly to inactivate RE by binding to two α2/Mcm1 sites (named DPS1 and DPS2) located within RE to initiate organization of an extended region of highly ordered nucleosomes over the locus (Szetoet al. 1997; Weiss and Simpson 1997). The interactions of α2 with DPS1/2, Mcm1, and Tup1 are all essential for α2's role in preventing selection of HML during interconversion, since mutations of α2 that disrupt its interaction with its DNA target, with Mcm1, or with Tup1 abolish normal donor preference in α cells (but not in a cells; Szeto and Broach 1997).
These observations provide a model for the molecular basis of donor preference (Figure 1, bottom). In MATa cells, binding of Mcm1 to DPS1 and DPS2 within RE activates the intrachromosomal recombination potential of the left arm of chromosome III by facilitating recruitment of, or acting in conjunction with, Fkh1. Accordingly, HML becomes the preferred donor. In MATα cells, binding of α2 adjacent to Mcm1 at DPS1 and DPS2 establishes a repressive domain over RE and thereby prevents activation of the recombination potential of the left arm. As a result, HMR becomes the preferred donor, due to an as yet undefined intrinsic bias for HMR as donor (Wuet al. 1996). This model provides a heuristic framework for cataloging the roles of all the known players in donor selection. However, the fundamental details of how long-range interactions between donor and acceptor loci occur and how the recombination enhancer affects recombination potential over a large domain remain to be explained.
Here we present results from our studies on the nature of RE obtained by examining its ability to stimulate different forms of interchromosomal recombination. In one series of experiments, we examined donor preference during mating-type switching in cells in which the MAT locus resided on chromosome II while the donor loci remained on chromosome III. In a second series of experiments we examined the effect of RE on interchromosomal homologous recombination between heteroallelic pairs located at various positions on chromosome III homologs. We found that RE exerted an effect on interchromosomal mating-type switching and on intrachromosomal homologous recombination but did not do so on interchromosomal homologous recombination. We also found that even in the absence of RE, MATα still influenced donor preference in interchromosomal mating-type switching, supporting a role of α2 in donor preference independent of its affect on RE. We discuss possible mechanisms that would account for the different effects of RE on different types of recombination.
MATERIALS AND METHODS
Plasmid construction: The lys2::MATa-TRP1 plasmid, B2407, and the lys2::MATα-LEU2 plasmid, B2422, were created by inserting the genomic HindIII fragments spanning MATa and MATα into pRS404 and pRS405, respectively. For MATa, the resulting plasmid was digested with BglII and resealed by ligation in the presence of complementary oligonucleotides (5′-GATCAAAGTAAT-3′ and 5′-GATCATTACTTT-3′). This removed the BglII site without affecting the MATa1 amino acid sequence. For both plasmids the LYS2 targeting sequence from pFV8 (Volkert and Broach 1986) was then cloned into the XhoI site.
The MAT deletion plasmid B2430 was constructed by oligonucleotide mutagenesis of the genomic HindIII fragment spanning MATα cloned into vector pSELECT, creating a precise deletion of W through Z1 marked with a BglII site. The URA3 gene was then inserted at the BglII site.
Switching the URA3 marker of pKSW72 (Weileret al. 1995) to HIS3 created the HO deletion plasmid B2428. Plasmid B2574 contains a ste5Δ construct. STE5 was amplified by PCR (5′-GACGAGCTCTGTATTTGTTCAAGACCG-3′ and 5′-GACGCATGCTTGCTACATCAGCAAC-3′) and cloned into pUC19 at the SacI/SphI sites. The XbaI/SpeI fragment was then replaced with HIS3.
Most of the plasmids used for heteroallelic recombination were obtained from Michael Lichten. Each contains one of two arg4 mutant alleles (arg4-nsp or arg4-bgl), the URA3 marker, and a fragment of yeast chromosome III with a unique restriction site for targeting via circular integration (Wu and Lichten 1994, 1995). Plasmids B2572 and B2573 were created by cloning a 2-kb PCR fragment centromere proximal to HMR and inserting it into the EcoRI site of plasmids pMJ113 and pMJ-115, respectively (Wu and Lichten 1994). Plasmids pMJ173 and pMJ174 were designed for targeting to CHA1, pMJ121 and pMJ-123 to HIS4, pMJ483 and pMJ486 to RIM1, pMJ113 and pMJ115 to URA3, and B2572 and B2573 to HMR.
Strain construction: All strains are listed in Table 1. Strains for analysis of mating-type switching were derived from strain Y2201 (Szeto and Broach 1997). Strains Y2901 and Y2902 were created by integrating a plasmid containing HIS3 adjacent to the Z2 region of MAT in strain Y2201. The integration creates a small duplication of Z2 to the first HindIII site of the TSM1 gene. Strain Y3149 was obtained by transforming strain Y2201 sequentially with BglII-digested B2407 (lys2-MATa-TRP1) and BglII-digested B2422 (lys2-MATα-LEU2). Strain Y3149 was then transformed with the HindIII fragment of plasmid B2430 (which deletes one of the endogenous MAT loci from W-Z1) to yield strain Y2934. Strains Y2946 and Y2947 were constructed from strains Y2901 and Y2902, respectively, by a two-step gene replacement using plasmid pLS70 (YIpURA3-dps1Δ dps2Δ), scoring for the introduction of the dps1Δ dps2Δ alleles by Southern analysis (Szetoet al. 1997). The reΔ construct [a URA3 replacement of RE from BlpI (chromosome III position 28,961) to PmeI (31,213)] was then integrated on the homologous chromosome. Strain Y2948 was created from strain Y2934 in a similar manner, except for using a HIS3 allele in the reΔ construct (B2436). All integrations were confirmed by Southern analysis.
All strains for analysis of heteroallelic recombination were derived from strains Y2811 and Y2812, isogenic derivatives of S288C kindly provided by Mark Rose. ARG4 was precisely deleted using a targeted deletion cassette marked with HIS3 (Baudinet al. 1993). Integration of the arg4 recombination cassette at each locus was confirmed by Southern blotting and by segregation analysis. Cassettes were constructed so that all arg4 alleles were in the same orientation relative to chromosome III. The heteroallelic diploids of different mating types were constructed by mating the MATa and MATα versions of complementary heteroalleles at the same locus and then deleting MAT on either homolog, using plasmid B2438 (matΔ:: LEU2). RE was deleted as above as needed except that the reΔ was marked with TRP1 using plasmid B2434.
Switching assays: Switching assays were performed as described (Weileret al. 1995) with minor modifications. Briefly, diploid strains containing the mata1Δ-inc allele at HMR, the matα1α2Δ-inc allele at HML, and a genetically marked MAT locus were sporulated and then dissected on YEPD plates. After 3 days of growth at 30°, segregants were replica plated to drop-out plates to determine auxotrophies, which identified the initial mating type of the germinated spore as well as other relevant genetic features. Colony PCR to determine the allele at MAT was performed using 5′-GACTCTACCCAGATTTGTATTAGACG-3′ and 5′-GATAAGAACAAAGAATGATGCTAAGAATTGA-3′ as primers. PCR products were run on a 0.7% agarose gel in 1× TAE. Colonies predominantly exhibiting the mata1 allele were scored as selecting HMR as donor, colonies predominantly exhibiting matα1α2 were scored as selecting HML as donor, and colonies that failed to exhibit one predominant allele were scored as ambiguous.
Fluctuation test: A slightly modified method of the median was used to determine recombination rates (Lea and Coulson 1949). Cells were plated for YEPD single colonies and grown for 2-3 days at 30°. Individual colonies were recovered in their entirety by coring and the majority of cells were plated on SC-Arg plates to determine the number of arginine prototrophs in the colony. A sample was diluted and plated on YEPD for viable cell count. Between 7 and 11 separate colonies were examined for each rate determination. Triplicate experiments using the same strain on different days yielded values that differed by <30%. For examination of methyl methanesulfonate (MMS)-induced recombination, cells were plated on media containing 0.1% MMS. For examination of gamma ray-induced recombination, the cultures were plated and directly irradiated with a 90 rad/min cesium source for 22 min.
Donor preference is maintained during interchromosomal mating-type switching: Mating-type interconversion normally takes place between donor and recipient loci on the same chromosome. To ask whether donor preference depends on this topology and, more broadly, what types of recombination are influenced by the recombination enhancer, we examined the efficiency of HML and HMR selection in strains in which the MAT locus and the donor loci were located on different chromosomes. This question has been addressed previously, but technical issues limited the reliability of the answer. In fact, two separate but similar experiments yielded two different outcomes (Weiler and Broach 1992; Wuet al. 1997).
To examine interchromosomal mating-type switching, we constructed an HO/HO diploid strain in which MATa and MATα were inserted at the LYS2 loci on the two chromosome II homologs and one chromosome III MAT locus was deleted and marked with URA3. The strain was further designed so that the initial switching event in spore clones became fixed due to import of the HO-refractory inc mutation from either donor locus. The donor locus from which the mating cassette derived in this initial switch could be determined by PCR analysis of the MAT locus resident at lys2 after growth of the spore clone, since the sizes of the mating cassettes derived from the donor loci were different from each other and from the initial allele residing at MAT.
To conduct the switching experiment, this strain, Y2934, was sporulated and tetrads were dissected. Marker analysis was performed and spores clones inheriting chromosome III with the matΔ::URA3 locus were assayed to determine whether the MAT locus at lys2 had undergone a mating-type switch and, if so, which donor provided the mating cassette. Results of this analysis, presented in Figure 2, demonstrated that normal donor preference was maintained in switching when MAT resided on chromosome II and the donor loci on chromosome III. Switching was efficient in that 80-90% of spore clones exhibited predominant replacement of the initial MAT allele by a cassette from one of the donor loci. In addition, preference was maintained. A total of 79% of lys2::MATa spore clones used HML as donor while only 12% used HMR (comparable to 79% HML and 6% HMR as donor following intrachromosomal switching in MATa spore clones from the parental strain Y2201; Figure 2). Similarly, 70% of lys2::MATα cells used HMR as donor while only 12% used HML. While this is somewhat less efficient than that seen in intrachromosomal switching, likely for reasons discussed below, donor preference is essentially maintained during mating-type switching from donor loci on chromosome III to the expressor locus on chromosome II.
Interchromosomal switching is delayed relative to intrachromosomal switching: We observed that lys2::MATα matΔ spores from strain Y2934 yielded significantly smaller colonies, with a higher percentage of inviable cells, than did lys2::MATa matΔ spores from the same strain or MATα spores from the parental Y2201 strain (Figure 3). This reduced growth was a consequence of initiation of mating-type interconversion since isogenic spores in which the HO gene was inactivated did not yield small colonies or poor viability (Figure 3). While this delay is likely due in part to the persistence of a double-strand break at MAT that is not healed in a timely fashion (Weiler and Broach 1992), this explanation cannot be the complete story since HO lys2::MATa matΔ spore clones do not grow slowly. One explanation for this discrepancy is that HMR is less accessible than HML in interchromosomal switching, so that the double-strand break is healed by mating-type switching less efficiently in a MATα than in a MATa background. Another explanation, though, is that the delay in healing the double-strand break at MATα causes a loss of the MATα gene by exonuclease processing and subsequent phenotypic conversion of the cell from α-mating type to a-mating type (Laney and Hochstrasser 2003). This would permit mating between these phenotypic “a” cells carrying a break at MAT and adjacent MATα cells, which, being in the daughter lineage, would not have initiated mating-type switching. This diploid cell would also suffer from double-strand-break-induced arrest, either from the inherited unrepaired break or from newly initiated HO cutting of the MATα locus. Such phenotypic conversion and mating cannot occur in a lys2::MATa cell. To address this possible explanation, we deleted STE5 from the strain. STE5 encodes a scaffold protein involved in the pheromone response pathway and is required for mating. As evident in Figure 3, lys2::MATα spore clones deleted for STE5 showed normal viability and no decrease in colony size. Further, ste5 lys2::MATα cells exhibit normal donor preference, with 82% of such spore clones selecting HMR as donor (data not shown). Thus, the decrease in colony size and cell viability seen in spores inheriting a lys2::MATα allele depends both on persistence of the double-strand break at MAT and on the ability of cells to mate. Similar results were observed for a cells attempting to switch in a strain deleted for HML (Wuet al. 1996).
The recombination enhancer influences donor preference during interchromosomal recombination: Previous studies identified a region on chromosome III required for activation of HML as donor in MATa cells and responsible for suppressing this activation in MATα cells. Deletion of this locus, RE, does not affect donor selection in MATα cells but does result in inappropriate selection of HMR in MATa cells. To enhance our understanding of the function of RE, we asked what effect RE mutations would have on donor locus selection when MAT was located on another chromosome. Control strains with MAT at the wild-type location were also constructed for comparison. As seen previously and in Figure 2, in a strain in which MAT resides at its normal location, deletion of RE results in a reversal of donor preference in MATa cells, such that HMR rather than HML becomes the preferred donor. Consistent with previous reports (Wu and Haber 1996; Szetoet al. 1997), MATα cell preference for HMR remained unchanged by deletion of RE. RE contains two Mcm1/α2p binding sites. In strains in which these two sites are precisely deleted, then selection of donor in MATa cells became random and selection of donor in MATα cells was unchanged, consistent with previous observations.
In contrast with the above results, deletion of RE in a strain with MAT resident on chromosome II resulted in essentially random selection of donor loci in a MATa background, although as with intrachromosomal recombination deletion of RE did not affect donor preference in a MATα background. Similar but less dramatic results are observed with deletion of only the Mcm1/α2p binding sites in RE, consistent with the fact that these elements play critical but not absolutely essential roles in RE function (Sunet al. 2002). The fact that MATa and MATα cells exhibit a difference in donor preference during the interchromosomal mating-type switch in the reΔ background suggests that α2p exerts an influence on donor preference through some mechanism independent of RE. This observation is discussed below.
The recombination enhancer affects intrachromosomal but not interchromosomal gene conversion: We used a reporter system developed by Michael Lichten to examine the effects of the recombination enhancer on intra- and interchromosomal gene conversion (Wu and Lichten 1995; Bordeet al. 1999). The system consists of two arg4 genes, each carrying a distinct restriction site fill-in mutation, inserted at two separate sites on chromosome III (for intrachromosomal conversion) or at equivalent sites on the two chromosome III homologs in diploids (for interchromosomal conversion). Since the reversion rate of either mutation is quite low (<10–6/cell/generation) and since the strain is deleted for ARG4 at its normal location, appearance of Arg+ prototrophs signals a recombination event between the two alleles. For interchromosomal recombination these recombination events could be gene conversion of one or the other allele or reciprocal recombination with a crossover site between the two alleles. For intrachromosomal recombination only gene conversion events are recorded, since the two arg4 alleles are in direct orientation on the chromosome and reciprocal recombination would lead to excision of intervening sequences, generally a lethal event in the haploid background. The rates of recombination between the two alleles in a strain were determined by the method of the median (Lea and Coulson 1949).
Wu and Haber (1995) previously reported a mating-type-dependent bias in intrachromosomal gene conversion between leu2 heteroalleles inserted at various sites on chromosome III, with a significant increase in conversion rates in a vs. α cells if one of the alleles resided within 20 kb of RE on the left arm of chromosome III. We observed similar but less dramatic effects with the arg4 heteroalleles. As evident in Figure 4 and Table 2, recombination between an arg4 allele resident at CHA1 (adjacent to HML) and an allele located at any other site on chromosome III was fivefold higher in a MATa background than in a MATα background. This bias was reduced to less than twofold in a strain for which the RE was deleted. Recombination between heteroalleles at other sites also showed a bias in MATa vs. MATα but not as marked. Thus, this work confirms the previous observation that RE promotes enhanced recombination over the left arm of chromosome III in a cells relative to α cells.
For interchromosomal recombination we created diploid strains with the two arg4 alleles inserted at homologous sites at various positions along the chromosome III homolog (Figure 5A). We then created phenotypic a or α versions of this diploid by deleting one of the MAT loci from the strain and examined recombination rates in the resulting isogenic a/α, a, and α diploid strains. Measurements of spontaneous recombination rates at various sites along the chromosome (Figure 5B) confirmed previous observations that mitotic recombination is higher in a/α strains relative to isogenic a or α strains (Esposito and Wagstaff 1981). Further, we found that recombination rates at all sites were equivalent in a vs. α strains, indicating that the bias for enhanced recombination in a cells during intramolecular recombination is not recapitulated during intermolecular recombination. Finally, while deletion of RE caused a small (approximately threefold) decrease in recombination rates for alleles inserted at the CHA1 locus, such deletions did not bestow an a vs. α recombination bias (Figure 5, B and E). Thus, RE does not confer an apparent cell type-specific recombination bias to intermolecular recombination.
One significant difference between mating-type switching and mitotic intermolecular recombination is that initiation of recombination is rate limiting in mitotic recombination whereas the initiating double-strand break occurs in almost all sensitive cells during mating-type interconversion. To address whether a cell type bias would be revealed in cells in which initiation of mitotic recombination were enhanced, we treated strains carrying arg4 heteroalleles inserted at CHA1 with MMS or γ-irradiation. Both of these treatments induce double-strand breaks and enhance the rate of mitotic recombination. As noted in Figure 5, C and D, both treatments increased recombination rates ∼10-fold. However, in neither case did such treatment reveal a bias in recombination between a and α cell types. Also, while a small reduction in recombination rates was still observed in strains deleted for RE, no cell type bias was manifest. Thus, even when initiation of intermolecular mitotic recombination is enhanced by induction of double-strand breaks, RE does not confer a cell type-specific effect on recombination.
Donor preference is maintained during interchromosomal switching: The results presented in this report show that donor preference is maintained even when donor and recipient loci reside on different chromosomes, indicating that donor selection does not depend on interactions in cis. These results confirm previous work from this lab using a different assay in which the behavior of certain switching events could be only inferred (Weiler and Broach 1992). This also confirms work suggesting that the relative three-dimensional organization of donor and acceptor loci on chromosome III does not play a significant role in donor locus selection (Simonet al. 2002). Rather, our results suggest that donor preference results from cell type differences in the intrinsic ability of the donor loci to participate in gene conversion following HO-initiated cleavage of MAT: in a cells HML is more “accessible” than HMR while in α cells the opposite is true.
Our results differ from those of Wu et al. (1997), who have previously reported that interchromosomal switching results in the decreased ability of MATα resident on chromosome V to utilize either donor locus following HO cleavage. In addition, when these cells did switch, they selected donor loci randomly. There are several possible explanations for this discrepancy. A trivial possibility is that the difference in location of the MAT locus (chromosome II vs. chromosome V), perhaps reflecting different relative positions of donor and acceptor loci within the nuclear landscape, affected donor preference. A second possible explanation lies in the different assays. In the experiments by Wu et al., the MAT locus was flanked by URA3 alleles. Repair of the HO-induced double-strand break was thus a competition between gene conversion using the donor loci and recombination between the flanking URA3 alleles via the single-strand annealing pathway. If the cells could have repaired the double-strand break only by conversion with the donor loci, donor preference might have been maintained. Finally, the switching experiments conducted by Wu et al. were performed with a galactose-induced HO gene. Thus, switching events were initiated in all stages of the cell cycle, not just G1. This might suggest that the bias in donor preference is operative only during switching in G1. In any event, the positive results from our experiments indicate that donor preference can be maintained during intermolecular switching.
One noteworthy phenotype associated with interchromosomal switching was decreased lys2::MATα spore colony size and viability. This phenotype is similar to that observed in intrachromosomal switching in MATα strains deleted for its preferred donor HMR (Wuet al. 1996). In both cases the phenotype is reversed by inactivation of either HO or the mating response pathway. Thus, the cause of the phenotype likely results from a delay in the repair of the double-strand break, the subsequent phenotypic conversion from α to a of the delayed cell, and, finally, mating of cells harboring the break with neighboring unswitched cells. That a similar small phenotype does not arise in MATa cells is likely not due to the fact that switching occurs more efficiently in a cells than in α cells but rather that phenotypic conversion and mating cannot occur in a cells delayed in switching. In fact, our previous data on interchromosomal switching suggest that the delay occurred equally in MATa and MATα cells (Weiler and Broach 1992).
MATα influences donor switching independently of its affect on RE: Cell type dictates donor preference strictly through MATα2 (Szeto and Broach 1997). In the absence of α2, RE activates HML for recombination and in the presence of α2 or the absence of RE this enhancement of HML is abrogated. Repression of RE-mediated activation occurs through two binding sites for α2 situated within RE. Data from studies presented in this report indicate that α2 also affects donor preference independently of not only these binding sites but also RE itself. In particular, interchromosomal switching in MATa cells in a strain deleted for RE exhibits random donor selection whereas switching in MATα reΔ strains exhibits a strong bias for HMR. Therefore, α2 either suppresses selection of HML or enhances selection of HMR during intermolecular switching by a mechanism that does not involve RE. Wu et al. (1996) previously showed that HMR inserted at any site in the left half of chromosome III was selected inefficiently in MATα cells. This domain is larger than that influenced by RE and thus their results are consistent with an RE-independent mechanism for suppression of donor selection by α2 of loci located on the left arm of chromosome III. Notably, we find that intramolecular homologous recombination over chromosome III was generally lower in α cells than in a cells, a phenomenon that might be related to the RE-independent donor bias observed. The mechanism for this RE-independent α2 effect could involve some as yet undefined activation of HMR or repression of HML as donor in MATα cells. However, since the effect is seen in intermolecular switching, we can conclude that the position of HML or HMR relative to MAT on chromosome III does not contribute to this process.
RE affects intra- but not interchromosomal homologous recombination: Our results confirmed that intramolecular gene conversion on chromosome III is enhanced in a cells vs. α cells when one of the participating alleles is resident on the left arm of the chromosome. The effects we observed are not as dramatic as those found previously (Wu and Haber 1995), which may reflect a difference in the reporter alleles used for the analysis. Nonetheless, these observations confirm that the recombination enhancer affects homologous recombination and not just mating-type interconversion. Thus, the fact that the recombination enhancer does not confer a similar cell type bias in interchromosomal homologous recombination is surprising and even more so given that it confers such a cell type bias in interchromosomal mating-type switching. One possibility is that the competition for gene conversion leading to prototropy is with sister chromatid repair of initiating lesions, which would not lead to prototrophy (Figure 6). In interchromosomal recombination, the enhancer would activate the sister to an equivalent extent as the homolog, so that the rate of recombination might be increased but most of the events would be with the sister chromatid. In intrachromosomal allele recombination at heterologous sites, events initiated outside the recombination domain would not have an enhanced interaction with the sister chromatid but rather with the heteroallele lying within the activated recombination domain. This possibility would suggest that interchromosomal recombination restricted to G1 might exhibit a cell type bias that is not evident for recombination at other stages of the cell cycle.
The mechanism of recombination enhancement: While our results do not yet provide a molecular mechanism for the function of RE, they can be appreciated in the context of a model positing competition between different sites as a template for repairing a broken DNA molecule, with RE biasing that competition (Figure 6). For mating-type interconversion, the HO-cleaved MAT locus likely samples both HML and HMR repeatedly through a Rad52-dependent homology search that is rapid compared to commitment to recombination. This initial three-strand synapse can either decompose back to the initial broken MAT DNA and the intact donor locus or convert to a stable recombination intermediate consisting, for instance, of a D-loop at the donor loci formed from invasion of a single-stranded region from the resected double-strand break at MAT. In the absence of RE and α2, this conversion to a stable intermediate would be essentially equivalent for both donor locus complexes and thus the relative conversion to MATa or MATα would depend on the relative abundance of the three-strand synapses formed with HML vs. HMR. When MAT is located on a different chromosome than HML and HMR, the relative concentration of the two complexes should be essentially equivalent, and thus the yield of MATa and MATα would be expected to be equal, which is what we observe. When MAT resides on the same chromosome as the donor loci, the proportion of HML and HMR complexes would depend on the frequency of colliding with one vs. the other, making the physically closer locus the predominant donor. Since HMR is closer on the chromosome and in the three-dimensional space of the nucleus (Simonet al. 2002), it becomes the default preferred donor.
In this context we would anticipate that RE would function to stabilize the interaction of HML and MAT, either by retarding the decomposition of the ternary complex (essentially rendering HML and the surrounding DNA more “sticky”) or by accelerating the formation of the stable recombination intermediate. For instance, RE could promote an increase in the localized concentration of specific helicases, which would increase the probability of strand invasion following initial synapsis. This would bias selection in favor of HML. As noted previously, α2 inhibits RE and, as we have observed in this report, α2 also inhibits selection of HML by an RE-independent function that could act at the same step or at a different step from that affected by RE. Thus, in α cells, HMR becomes the preferred donor.
This same mechanism could account for the effects of RE on intramolecular vs. intermolecular gene conversion. Recent evidence suggests that most recombination intermediates arise as a consequence of replication (Terceroet al. 2003), so that repair of the recombinogenic lesion occurs in the presence of a sister chromatid. Thus, for heteroalleles located on the same chromosome, repair of a recombinogenic lesion would involve a competition between the heteroallele on the same chromosome, potentially yielding a prototroph, and the identical allele on the sister chromatid, which would not yield a prototroph. As shown in Figure 6B, the former event but not the latter event could be influenced by RE, by the same mechanism involved in donor preference. This would result in a cell type-dependent change in the rate of prototroph formation, as is observed. In the case of heteroalleles at the same site on homologous chromosomes (Figure 6C), the initiating lesion again could be repaired by interaction with either the heteroallele on the homolog chromatid or the identical allele on the sister chromatid. In this case, RE would affect both reactions, with the consequence that the balance between prototroph-generating and non-prototroph-generating pathways would be unchanged. As observed, RE would not affect the rate of prototroph formation. Thus, this biased competition model can account for all the effects reported here and previously. Further data, though, will be required to pinpoint the precise step affected by RE.
The authors thank Michael Lichten for his generous gift of ARG4 heteroallelic plasmids and Ned Wingreen for thoughtful discussions regarding this work. These studies were supported by National Institutes of Health grant GM48540 to J.R.B.
Communicating editor: B. J. Andrews
- Received September 23, 2003.
- Accepted December 17, 2003.
- Copyright © 2004 by the Genetics Society of America