Yeast strains:

All S. cerevisiae strains analyzed in this study (Table 1) were derived in the SK1 strain background (Kane and Roth 1974). We constructed strains by standard methods containing reporters of allelic loxP collisions and heteroallelic gene conversion at varying distances from each other (Table 1, Figure 1). Cre recombinase is supplied from a galactose-inducible promoter, GAL1, at the LYS2 locus (Burgess and Kleckner 1999). Starting with SBY1756 and SBY1074 (Table 1 legend), two constructs—one bearing LEU2:GPD1:loxP and the other containing LEU2:loxP:ade2 (Figure 1b; amplified from pSB133 and pSB186, respectively; Peoples et al. 2002)—were used to completely replace the ARG4 locus [Saccharomyces Genome Database (SGD) coordinates: VIII, 140004–141395]. ARG4 was deleted to eliminate sequences homologous to the plasmid-borne arg4 heteroallele constructs. The loxP constructs were inserted such that the GPD1 promoter and ade2 ORF were oriented toward CEN8.

Next, URA3:arg4-bgl/URA3:arg4-nsp heteroallele pairs were inserted (Figure 1, c and d) at the HIS4 locus on chromosome III and the centromere-distal THR1 locus on chromosome VIII (∼25 kb, ∼20 cM from the arg4Δ∷loxP constructs) as previously described (Wu and Lichten 1995; Goldman and Lichten 1996; arg4 heteroallele plasmids were generous gifts from M. Lichten and A. Goldman). For heteroallele insertions at the centromere-distal DED81 locus (3 kb, ∼2 cM from the arg4Δ∷loxP constructs), PCR products from wild-type SK1 genomic DNA (SGD coordinates: VIII, 143513–144010) were TopoTA cloned (Invitrogen, San Diego) and subcloned into the EcoRI site of pMJ113 and pMJ115 to generate pSB397 and pSB398, respectively (arg4-bgl and arg4-nsp), and the BstXI enzyme was used to linearize plasmids prior to yeast integration. Heteroallele inserts at DED81 and THR1 were added in both linkage phases relative to the loxP constructs (Figure 2, a and b), such that the URA3 and arg4 ORFs were directed away from the centromere. For convenience, we use GPD1-bgl and ade2-bgl to refer to the two linkage phases, indicating whether the promoter-containing loxP construct is coupled to arg4-bgl or arg4-nsp, respectively.

To generate strains bearing arg4Δ∷LEU2:GPD1:loxP:ade2 used in linkage control strains, we sporulated SBY2186, SBY2185, SBY2257, and SBY2258 in the presence of 0.03% galactose and identified Ade⁺ spore clones by tetrad dissection (Table 1, Figure 2, c and d). To distinguish crossover- from noncrossover-associated gene conversion at the DED81 locus, we replaced the immediately centromere-distal YHL022c gene with a KanMx antibiotic resistance marker by the method of Wach et al. (1994) to produce Gen^R haploids bearing arg4Δ∷loxP:ade2 alleles. Appropriate pairs of haploids were crossed to produce the experimental diploid strains (Table 1).

Random spore analysis:

Synchronized meiotic time courses were conducted in 10-ml SPM (sporulation media) cultures plus or minus galactose induction of the Cre recombinase, as described (Padmore et al. 1991; Peoples et al. 2002). Galactose was added to a final concentration of 0.03% at t = 1, 2, 3, or 4 hr in SPM for induced samples. Random spore analysis was performed after 24–48 hr by diluting zymolyase-treated sonicated spore suspensions and plating to appropriate media (McKee and Kleckner 1997). Dilutions were targeted to produce ∼200–400 colonies per plate, depending on media type. Arg⁺ prototroph formation (R) was determined by growth on SC −arg plates relative to colony-forming units (CFU) of spores on YPD plates. Ade⁺ prototroph formation (C) was determined by the frequency of white colonies on YPD −ade plates. SC −arg plates were replica plated to YPD −ade to determine the frequency of Ade⁺ Arg⁺ spores (S and L). Where applicable, spores were also replica plated to and from YPD + geneticin media to determine the Gen^R or Gen^S spore genotype, i.e., the presence or absence of the KanMx resistance cassette (for M, N, X, and O). For variables, we use subscripts bgl and nsp to refer to linkage phases GPD1-bgl and ade2-bgl, respectively.

To determine whether or not there were specific biases in the measured frequency of Arg⁺ Ade⁺ spores due to the direction of selection, we performed the reciprocal selection on strain SBY2074 by screening and patching white Ade⁺ colonies from YPD −ade to SC −arg to determine the frequency of Ade⁺ Arg⁺ spores. We were unable to directly select for Ade⁺ spore clones, due to hypomorphic expression of the ADE2 gene when loxP sequences reside in the 5′-UTR (our unpublished observations). The frequency of coincident Arg⁺ Ade⁺ spores was in strong agreement when selection (or screening) was carried out in either direction (data not shown), so we report coincidence data normalized to Arg⁺ frequency to facilitate comparisons between strains and reduce sampling error.

All heteroallele reporter strains were subjected to random spore analysis on at least two different occasions from independent meiotic time courses. Errors were calculated as the standard deviation of the mean from three independent cultures sporulated in parallel for each experiment. Measured genotypic frequencies with a given strain were consistent between independent experiments, and spore viability was consistently near 100% in all strains and experiments (data not shown). Student's t-tests (two tails) on transformed frequency data were used to evaluate differences in the data at a confidence threshold of P ≤ 0.05.

Calculation of collision frequency per meiotic gene conversion:

We measured the collision frequency among gene convertants as S(obs) = Ade⁺/Arg⁺ spores in collision tester strains. We calculated the expected frequency of Ade⁺/Arg⁺ spores as S(exp) = 2LC, where C is collision frequency (Ade⁺/CFU in the collision tester strains) and L is the coupling constant between the GPD1:loxP:ade2 construct and the coupled heteroallele upon selection for Arg⁺ spores (Ade⁺/Arg⁺ in the linkage control strains). If L = 0.5, then there is no effect of Arg⁺ selection on the recovery of the GPD1:loxP construct. If L = 1, then all Arg⁺ spores are coupled to GPD1. If L = 0, then no Arg⁺ spores contain GPD1.

If gene conversion and collision events occur independently and are unlinked, we expect the collision frequency to be the same between unselected and gene conversion-selected spores; i.e., S(exp) = C (where L = 0.5). But if the collision and conversion events occur independently and are linked, the effects of coupling and repulsion between the two assays can become evident if there are differential recombination parameters between the two heteroalleles. For example, if all Arg⁺ spores were the result of arg4-bgl → ARG4 gene conversion with no associated crossing over, then L_bgl = 1 and L_nsp = 0, and thus S_bgl(exp) = 2C and S_nsp(exp) = 0. Restated, if both a collision and a gene conversion happen in the same meiotic cell, the Mendelian expectation is that GPD1:loxP:ade2 and URA3:ARG4 would occupy the same spore one-quarter of the time; i.e., S = C. But if the Arg⁺ gene conversion recipient in one linkage phase always ended up on a GPD1-containing chromatid, then GPD1:loxP:ade2 and URA3:ARG4 are expected to occupy the same spore one-half of the time, so S(exp) = 2C, whereas in the other linkage phase S(exp) = 0 (Figure 2, a and b).

We estimated the values of L_bgl and L_nsp for heteroallele insertions at DED81 and THR1, using derivative linkage control strains in which all loxP inserts possessed the loxP-proximal ade2 ORF, while only one homolog carried the loxP-distal GPD1 promoter (Figure 2, c and d). In this way, the GPD1-containing chromatid could be followed independently of collisions. L was measured as the frequency of Ade⁺ prototrophs among Arg⁺-selected spores (Ade⁺/Arg⁺). To evaluate potential cis effects of the GPD1 promoter on biases between the heteroalleles in generating Arg⁺ spores and/or associated crossover frequencies, we measured the sum L_bgl + L_nsp, which indicates the effect of the arg4Δ∷loxP heterozygosity on heteroallelic gene conversion parameters. If L_bgl + L_nsp = 100%, then we would conclude that there is no cis effect of loxP heterozygosity on the recombination parameters of the arg4 heteroalleles.

Calculation of collision frequency among crossover and noncrossover recombinants:

For experiments with DED81∷arg4 heteroalleles that also contained a segregating yhl022cΔ∷KanMx allele in repulsion to the GPD1:loxP (or GPD1:loxP:ade2) construct (Table 1), we measured the observed and expected frequency of collisions associated with crossover and noncrossover recombination. For collisions among crossovers, X(exp) = 2MC, where M is the frequency of Ade⁺ Gen^R/Arg⁺ spores in linkage control strains (SBY3155 and SBY3156), whereas X(obs) is Ade⁺ Gen^R/Arg⁺ and C is Ade⁺/CFU in collision tester strains (SBY3153 and SBY3154). For collisions among noncrossovers, O(exp) = 2NC, where N is the frequency of Ade⁺ Gen^S/Arg⁺ spores in linkage control strains and O(obs) is this value in collision tester strains.

For the collision tester strains (SBY3153 and SBY3154), only Arg⁺ Ade⁺ spores are informative of a collision and gene conversion occurring in the same meiosis, where a Gen^S genotype indicates a collision and NCR-associated gene conversion and Gen^R indicates a collision plus CR-associated gene conversion. The two linkage phases are informative for different classes of recombination events occurring with a collision (Arg⁺ Ade⁺ spores).

For the linkage control strains (SBY3155 and SBY3156), we could evaluate all four detectable classes of recombination event among Arg⁺ gene conversions independent of collision: Ade⁺ Gen^S and Ade⁻ Gen^R indicate NCR-associated gene conversion, while Ade⁺ Gen^R and Ade⁻ Gen^S indicate CR-associated gene conversion. When GPD1 is coupled to arg4-bgl, the Arg⁺ Ade⁺ Gen^S genotype indicates a bgl recipient with no crossover (N_bgl), while Arg⁺ Ade⁺ Gen^R indicates either a nsp recipient with a proximal CR or a bgl recipient with a distal crossover (M_bgl). In the opposite linkage phase (when GPD1 is coupled to arg4-nsp), a reciprocal set of events is reported: Arg⁺ Ade⁺ Gen^S indicates a nsp recipient and no crossover (N_nsp), but Arg⁺ Ade⁺ Gen^R indicates either a nsp recipient and a distal CR or a bgl recipient and a proximal crossover (M_nsp). Furthermore, if the presence of the loxP heterozygosity has no effect on heteroallele recombination parameters, we expect that (M_bgl + N_bgl) + (M_nsp + N_nsp) = 1.

A two-locus genetic assay to measure allelic pairing interactions and meiotic recombination during budding yeast meiosis:

To address whether or not allelic interactions occur at higher levels for allelic sites adjacent to meiotic recombination events, we used Cre/loxP site-specific recombination to analyze the relative magnitude of allelic interactions at the ARG4 locus in relation to gene conversion events occurring at sites on the same or different chromosomes (Figure 1).

• Download figure
• Open in new tab
• Download powerpoint

Figure 1.—

Two-locus assay system for measuring collisions and gene conversions. (a) The frequency of Cre-mediated loxP recombination (or collisions) indicates the relative spatial proximity of two loxP sites engineered into the genome. (b) Structure of the heterozygous loxP constructs inserted at the arg4Δ locus. Collisions are measured as the frequency of Ade⁺ prototroph formation upon Cre induction from a GAL1 promoter in cells synchronized to enter meiosis. Only one of the two interacting homologous chromatids expresses ADE2 following Cre-mediated recombination. (c) Structure of the arg4-bgl/arg4-nsp heteroallele pair. Either the proximal nsp mutation or the distal bgl mutation can be gene converted using wild-type information from the other allele to generate an ARG4 chromatid. Double-strand DNA break (DSB) hotspots are located immediately proximal and distal to the arg4 gene. (d) Schematic map (not to scale) indicating the relative positions of the loxP collision reporter at arg4Δ (black vertical bar) and the URA3:arg4 heteroallele insertion sites (black triangles). DED81, THR1, and HIS4 are 3 kb, 25 kb, and unlinked to the loxP constructs, respectively.

First, heterozygous loxP constructs were used to replace the endogenous ARG4 locus on chromosome VIII (Figure 1, b and d). One loxP insert bears a centromere-distal constitutive promoter (GPD1:loxP), while the other bears a centromere-proximal promoterless reporter (loxP:ade2). Cre is supplied under the control of the galactose-inducible promoter (GAL1), and Cre-mediated crossovers between loxP sites (loxP “collisions”) are detected by Ade⁺ prototroph formation (Figure 1, a and b). The collision frequency reports the relative spatial proximity of pairs of loxP sites engineered into the yeast genome (Burgess and Kleckner 1999; Peoples et al. 2002; Peoples-Holst and Burgess 2005; Lui et al. 2006). Cre-mediated loxP collisions are detected only between homologous chromatids; sister–sister loxP recombinants are undetectable, but their occurrence may influence the frequency of events between the homologs.

Second, arg4-bgl/arg4-nsp heteroallelic gene conversion reporters (Wu and Lichten 1994; Goldman and Lichten 1996) were added at one of three locations: a site immediately adjacent to the loxP site at DED81 (3 kb, ∼2 cM), a more distal linked site at THR1 (25 kb, ∼20 cM), and an unlinked site at HIS4 (chromosome III, 50 cM) (Figure 1, c and d). Strains heterozygous for arg4-bgl and arg4-nsp can generate wild-type ARG4 information by gene conversion, and Arg⁺ spores produced during meiosis represent recipients of meiotic DSB repair initiated near one of the heteroallele sites (Figure 1c; Nicolas and Petes 1994).

For each gene conversion reporter located at sites linked to loxP (i.e., at DED81 and THR1), four diploid strains were constructed to control for the effects of linkage phase between the heterozygous loxP constructs and the heterozygous arg4 constructs (Figure 2). If Arg⁺ gene conversions and/or crossing over are biased between the arg4-bgl and arg4-nsp heteroalleles, then the effects of coupling and repulsion between the heteroallele pair and the loxP constructs will become evident. The heteroallele pair, arg4-bgl/arg4-nsp, was inserted in both linkage phases relative to the collision tester constructs, GPD1:loxP/loxP:ade2 (Figure 2, a and b), and also derivative linkage control constructs, GPD1:loxP:ade2/loxP:ade2 (Figure 2, c and d). In the former strains Ade⁺ spores are produced by Cre-mediated loxP collisions, whereas in the latter strains the Ade⁺ phenotype segregates with GPD1-containing chromatids, independent of collisions. For convenience, we refer to the two linkage phases as GPD1-bgl (Figure 2, a and c) and ade2-bgl (Figure 2, b and d). For variables, we use subscripts bgl and nsp, respectively, to refer to the two different linkage phases.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2.—

The linkage phase of the arg4 heteroalleles and the loxP constructs affects the coincidence of Arg⁺ gene conversion and Ade⁺ collision in spores. (a and b) Collision tester strains are illustrated (SBY2186 and SBY2185, respectively, for DED81 inserts; SBY2257 and SBY2258, respectively, for THR1 inserts; Table 1). (c and d) Control strains are illustrated (SBY2709 and SBY2710, respectively, for DED81 inserts; SBY2711 and SBY2712, respectively, for THR1 inserts). In the linkage control strains, the homolog bearing the GPD1 promoter always expresses the ADE2 open reading frame, independent of allelic loxP collisions. Asterisks denote chromatids that are Ade⁺ Arg⁺. In a, where arg4-bgl is coupled to the GPD1:loxP allele, coincident Ade⁺ Arg⁺ spores are generated by a collision and a gene conversion of arg4-bgl associated with no crossover (NCR) between arg4 and loxP or a distal crossover (dCR) beyond arg4 or alternatively by conversion of arg4-nsp with an associated proximal crossover (pCR) between arg4 and loxP. In b, where arg4-bgl is coupled to the loxP:ade2 allele, the situation is reversed, with collisions accompanied by arg4-nsp conversions associated with NCR or dCR and arg4-bgl pCR conversions generating Ade⁺ Arg⁺ spores. In c, where arg4-bgl is coupled to the GPD1:loxP:ade2 allele, coincident Ade⁺ Arg⁺ spores are generated by gene conversion of arg4-bgl with NCR between arg4 and loxP or a dCR beyond arg4 or alternatively by conversion of arg4-nsp with an associated pCR. In d, where arg4-bgl is coupled to the loxP:ade2 allele, the situation is reversed, with arg4-nsp NCR/dCR conversions and arg4-bgl pCR conversions generating Ade⁺ Arg⁺ spores. Note that only two-strand double events are illustrated; one of the two classes of three-strand doubles can also generate coincident Ade⁺ Arg⁺ spores, whereas the four-strand double and the other class of three-strand double do not generate coincident spores.

We performed random spore analysis for this set of nine strains (Table 1) to determine the gene conversion frequency R (Arg⁺/CFU), the unselected collision frequency C (Ade⁺/CFU), the relative effect of selecting for Arg⁺ on the recovery of GPD1-containing chromatids L, and the conversion-selected collision frequency S(obs) (Ade⁺/Arg⁺) (Table 2).

Endogenous meiotic recombination aids Cre-mediated recombination at nearby loxP sites:

We sought to distinguish whether recombination locally aids loxP collisions or the converse by inducing the expression of Cre at different time points in synchronized meiotic cultures. If loxP collisions facilitate gene conversion at nearby sites, then induction of the Cre recombinase at later time points in early prophase I should cause a decrease in S_bgl(obs) toward that expected, since cells at later time points would have already initiated meiotic recombination. Alternatively, if recombination aids collision, we would not observe a decrease in coincident spores at late Cre induction times.

We added galactose to synchronized meiotic cultures of SBY2186 and SBY2709 (DED81 heteroallele inserts in the GPD1-bgl linkage phase) to induce Cre expression at 1, 2, 3, and 4 hr after transfer to sporulation media and evaluated the frequency of Ade⁺, Arg⁺, and Ade⁺ Arg⁺ spores after sporulation was complete (Figure 3). Heteroallelic gene conversion at DED81∷arg4 was unchanged for different induction times in both the tester and the control strains (data not shown). Varying induction time also had no effect on Ade⁺ segregation in the linkage control SBY2709; for all four induction times, control Ade⁺ spores arose at a 50% frequency, and Ade⁺ constituted 80% of Arg⁺ spore clones (L), as expected. These results indicate that inducing Cre at later time points does not affect gene conversion frequency or the relative bias in heteroallele repair between arg4-bgl and arg4-nsp.

• Download figure
• Open in new tab
• Download powerpoint

Figure 3.—

Gal-induction time course shows increased coincidence of gene conversions and allelic loxP collisions for tightly linked sites in late prophase I. Cre was induced with galactose at t = 1, 2, 3, or 4 hr after transfer to sporulation medium for SBY2186 and SBY2709, as indicated on the x-axis. On the primary y-axis, yellow bars indicate the expected frequency of Ade⁺ among Arg⁺ spore clones, S_bgl(exp) = 2LC, while dark blue bars indicate the observed frequency of Ade⁺ among Arg⁺ spore clones, S_bgl(obs). On the secondary y-axis, the red line shows the ratio S_bgl(obs)/S_bgl(exp).

Allelic loxP collisions were less frequent when galactose was added at later points in meiosis. This result was expected, since the length of time Cre had to act on the loxP sites before the meiosis I division was decreased. In the collision tester SBY2186, the frequency of Ade⁺ collision was 3.9 ± 0.3% of spores for induction at t = 1 hr, decreasing to 1.9 ± 0.5% and 2.0 ± 0.6% for t = 2 and t = 3 hr induction, respectively, and was further reduced to 1.0 ± 0.1% with a t = 4 hr induction time. However, the coincidence of allelic collisions and nearby gene conversions showed a much less appreciable change: S_bgl(obs) was 17.9 ± 3.2% for induction at t = 1 hr and 11.7 ± 1.7% for the t = 4 hr induction time. Thus, S_bgl(obs) was threefold greater than S(exp) for early Cre induction, increasing to sevenfold for late Cre induction (Figure 3). These data strongly argue that Cre-mediated loxP crossovers do not affect meiotic recombination in cis. Instead, ongoing meiotic recombination adjacent to the loxP sites increases the probability of a Cre-mediated event.

A role for crossover-bound recombination intermediates in juxtaposing nearby allelic sites:

Taken together, the results of the gene conversion experiments described above demonstrate a cis effect of a subset of recombination intermediates on the frequency of allelic loxP collisions. We hypothesized that CR recombination intermediates play a specific cis role in affecting loxP collisions, whereas NCR recombination intermediates do not: If CR recombination intermediates and/or products aid loxP collisions locally, then collisions may be more likely when crossovers occur nearby, yet less likely when crossovers occur at a more distant linked site, due to genetic interference. Selection for gene conversion at a particular site effectively enriches for spores with associated crossover events, while chromosomal interference simultaneously depletes the occurrence of incidental crossovers at linked sites. NCR recombination intermediates are unlikely to repress loxP collisions at a distance, since NCR products do not show chromosomal interference (Hurst et al. 1972; Mortimer and Fogel 1974; Malkova et al. 2004).

To distinguish the contributions of CR and NCR recombination to the allelic loxP collision frequency, we added a heterozygous KanMx antibiotic marker (which confers a Gen^R phenotype) at the YHL022c locus, immediately distal to DED81, to produce strains SBY3153–SBY3156, representing both collision tester and linkage control strains for recombination at DED81 in both linkage phases (Table 1, Figure 4). We conducted random spore analysis with these new strains, using galactose induction at t = 1 hr in SPM (Tables 3 and 4). The genotypic frequencies of Ade⁺, Arg⁺, and Ade⁺ Arg⁺ spores were comparable to those previously observed in the absence of KanMx for all four DED81∷arg4 diploids (data not shown), and 50% of random spores were Gen^R, as expected. In the linkage control strains, 9.0 ± 1.9% and 9.1 ± 1.0% (for GPD1-bgl and ade2-bgl, respectively) of Gen^R spores were Ade⁺, indicating a 9-cM map distance between YHL022c and arg4Δ∷loxP. Galactose induction had no effect on spore genotypic frequencies, except that in the collision testers all spores were Ade⁻ (data not shown).

• Download figure
• Open in new tab
• Download powerpoint

Figure 4.—

Schematic of linkage control strains for measuring relative CR and NCRs per Arg⁺ gene converted spore. (a) SBY3155; (b) SBY3156. Data are summarized in Table 3. Annotations are as in Figure 2, except that blue boxes indicate a KanMx marker (conferring a Gen^R phenotype) inserted immediately distal to DED81 at YHL022c. Below each schematic are shown the recombinant classes sorted by genotype. For the collision tester strains, the situation is the same, except only Ade⁺ spores generated by loxP collision are informative regarding the class of associated Arg⁺ gene conversion.

Meiotic recombination facilitates homologous chromosome interactions in cis:

The importance of genes required for early steps of meiotic recombination in the coalignment and juxtaposition of homologous chromosomes in many eukaryotes led us to wonder whether homolog interactions occurred at higher levels for sites in cis to meiotic recombination events. We used a two-locus assay in budding yeast to measure the coincidence of meiotic recombination (gene conversion of arg4 heteroalleles yielding Arg⁺ spores) and Cre-mediated recombination between allelic loxP sites (allelic loxP collisions yielding Ade⁺ spores). We found that the frequency of coincident Ade⁺ Arg⁺ meioses exceeded random expectations when the loxP and heteroallele constructs were at tightly linked positions, suggesting that meiotic recombination can bring nearby loxP sites into close proximity for Cre to act.

An alternate possibility is that Cre-mediated loxP crossovers increase the chance of nearby heteroallelic gene conversion, but several lines of evidence argue against this. First, the frequency of heteroallelic gene conversions was not affected by the presence of nearby loxP constructs or by expression of Cre. Second, the coincidence of gene conversions and allelic loxP collisions did not decrease—but rather increased—when Cre was induced at progressively later time points in synchronized sporulations. These data suggest that ongoing recombination serves to locally juxtapose nearby chromosomal regions, rather than Cre-mediated loxP crossovers influencing nearby recombination.

The extent to which meiotic recombination elevates nearby allelic interactions is likely underestimated by our measurements, since the total number of meiotic recombination events that initiate near the assayed site exceeds the levels of Arg⁺ gene conversions. Many recombination events initiated within or near the heteroalleles may not include either arg4 mutation in heteroduplex DNA. Furthermore, the outcome of mismatch repair can lead to either restoration or conversion of ARG4/arg4 heteroduplex. Indeed, there are ∼10 times more DSBs per chromatid around the arg4 heteroallele constructs than Arg⁺ spores, independent of the recombination position effect (Borde et al. 1999).

Crossover recombination plays a special role in juxtaposing nearby homologous sites:

Only some recombination-mediated interactions exerted a cis effect on allelic loxP collisions, whereas others had no detectable effect. In one linkage phase, gene conversion near the loxP site (at DED81) significantly increased the probability of a collision. When gene conversion occurred further away from loxP (at THR1), there was significantly less probability of collision. In the other linkage phase, we observed no significant deviation from random expectations at either site. From these data, we inferred a unique role for CR recombination intermediates in tethering nearby allelic sites, which predicted a strong difference in the probability of informative Arg⁺ spores associated with CRs between the two linkage phases. We verified our prediction for gene conversions at DED81 and further found that CR-associated gene conversion significantly elevated nearby allelic loxP collisions in both linkage phases, while NCR-associated gene conversion did not.

While the formation of recombinant products may occur synchronously for CRs and NCRs, their respective joint molecule intermediates likely have different stabilities and/or persistence times (Merker et al. 2003; Maloisel et al. 2004). Recent evidence suggests that strand invasion recombination intermediates bound for an NCR fate are indeed considerably less stable than those bound for a CR fate (McMahill et al. 2007). Our data suggest that the greater stability of CR-bound joint molecules increases the time available for Cre to act on nearby loxP sites located at allelic sites on homologous chromatids. The phenomenon of genetic interference between CRs then explains the decreased collisions for conversions 25 kb away at THR1, as well as the lack of effect of NCR recombinants. We suspect that NCR-bound SDSA recombination intermediates still make contributions to the end-to-end coalignment of homologous chromosomes, but individual events have short-lived joint molecules that make minor cis contributions to stable interactions between allelic sites.

Crossover interference and homologous chromosome interactions:

Future sites of CRs are marked by “synapsis initiation complexes” (SICs) containing Zip3p, formed at discrete sites along chromosomes during prophase I, and these SIC foci show chromosomal interference (Fung et al. 2004). Zip3p interacts with a number of other proteins including Zip2p and Mer3p, all of which play important structural or biochemical roles in the formation of CR recombination products, perhaps by stabilizing joint molecule recombination intermediates (Borner et al. 2004).

The zip2Δ, zip3Δ, and mer3Δ mutants have all been shown to have intermediate defects in allelic loxP collisions during meiosis, presumably due to a decrease in recombination intermediate stability in the absence of SICs (Peoples-Holst and Burgess 2005). In this study using wild-type yeast, selecting for gene conversions nearby the loxP contructs (at DED81) may effectively enrich for the local presence of SICs. Since SICs display chromosomal interference, selection for recombinants at a more distant site (THR1) may deplete the presence of SICs immediately adjacent to the loxP sites. In effect, this latter case locally mimics a zip2Δ, a zip3Δ, or a mer3Δ mutant around the loxP sites.

Crossover product formation is not required for SIC interference or wild-type allelic interactions. Null mutants of ZIP1 and MSH4 cause defects in crossover formation and genetic interference, yet still display interfering Zip3p foci (Fung et al. 2004). The zip1Δ and msh4Δ mutants also have wild-type or nearly wild-type levels of allelic interactions as measured by the Cre/loxP collision assay (Peoples-Holst and Burgess 2005). These data indicate that stabilized joint molecule recombination intermediates conducive to CR formation, rather than CR products per se, are responsible for increasing local allelic interactions during meiosis and imposing interference. Furthermore, Cre-mediated loxP crossovers are insufficient to impose crossover interference on the immediately adjacent interval, consistent with other evidence that the “substrate” for interference signaling depends on meiotic recombination intermediates present prior to crossover formation (Bishop and Zickler 2004).