Partner Choice during Meiosis Is Regulated by Hop1-promoted Dimerization of Mek1

Plasmids

Plasmid names, genotypes, and sources can be found in Table 1. Plasmids for this study were constructed by standard procedures using the Escherichia coli strain BSJ72 (Maniatis et al., 1982). Details of plasmid constructions are available upon request. All MEK1 fusions are expressed under the control of the MEK1 promoter. The TAP tag was cloned from plasmid pBS1761, obtained from EUROSCARF. Mutations were introduced directly into pLT11 by site-directed mutagenesis (QuikChange kit, Stratagene, La Jolla, CA). The presence of the mutations was confirmed by DNA sequencing (Research Genetics, Huntsville, AL; Center for the Analysis and Synthesis of Macromolecules at SUNY Stony Brook). For hop1-K593A, hop1-K590A, hop1-564Δ, and hop1-585Δ, the entire sequence of each allele was determined to ensure that no additional mutations were created during the mutagenesis.

Yeast Strains and Media

Strain genotypes can be found in Table 2. NH246 and NH270 are derived from a cross between the slow sporulating BR and A364a genetic backgrounds (Woltering et al., 2000). All other strains are derived from SK1. Details of strain constructions are available upon request. All experiments were conducted at 30°C. Liquid and solid media have been described (Vershon et al., 1992; de los Santos and Hollingsworth, 1999). The inhibitor, 1-NA-PP1, 4-amino-1-tert-butyl-3-(1′naphthyl)pyrazolo[3,4-d]pyrimidine, was diluted from a 10 mM stock purchased from Cellular Genomics (New Haven, CT).

Table 2.

S. cerevisiae strains

Name	Genotypea	Source
NH246		Woltering et al. (2000)
NH270	Same as NH246 only RED1	This work
NH144		Hollingsworth et al. (1995)
YTS3	Same as NH144 only	de los Santos and Hollingsworth (1999)
DW10	Same as NH144 only	de los Santos and Hollingsworth (1999)
DW11	Same as NH144 only	de los Santos et al. (2001)
NH311	Same as NH144 only ade2-bgl	Woltering et al. (2000)
NH217	Same as NH144 only	This work
	Same as NH144 only	Wan et al. (2004)
NH566		This work
NH601		This work
NH624		This work
NH639		This work
		This work
	Same as only	This work

Open in a new tab

^a

Underlines indicate genes on the same chromosome

Antibodies, Immunoprecipitations, and Western Blots

The G/R antibodies, as well as the immunoprecipitation (IP) and Western blot protocols, are described in Wan et al. (2004). The Hop1 antibodies and phosphatase experiment protocol are described in de los Santos and Hollingsworth (1999). To detect phosphorylated Hop1 by SDS-polyacrylamide gel electrophoresis, proteins were fractionated on 8% gels (29:1 acrylamide:bis-acrylamide) 20 cm in length at 15 mAmp for 15 h.

Time Courses

Liquid sporulation conditions used were 2% potassium acetate at 30°C. Sporulation was monitored by phase contrast microscopy of at least 200 cells per strain. DSB and cytological analyses were as described in Woltering et al. (2000) and Loidl et al. (1998). For experiments in which strains were transformed with plasmids, the cells were grown to stationary phase in SD-ura medium instead of rich medium before dilution into YPA. Quantitation of DSB fragments was performed using a Molecular Dynamics PhosphorImager, (Amersham, Piscataway, NJ) and Image Quant 1.11 software. The formula used for quantitation was (DSB_t=x - DSB_t=0)/(DSB_t=x + P_t=x) × 100, where P represents the parental fragment.

Inhibition of Mek1 Kinase Activity after DSB Formation Allows DSB Repair and Produces Viable Spores in a dmc1 spo13 Diploid

Previously we showed that inhibition of Mek1 kinase activity after DSB formation in a dmc1 mutant suppressed the arrest triggered by the meiotic recombination checkpoint, caused DSBs to disappear, and produced inviable spores (Wan et al., 2004). This result was interpreted to mean that Mek1 kinase activity was necessary to prevent DSB repair using sister chromatids, but this experiment did not rule out alternative explanations. For example, degradation of the ends of the DSBs would also result in the loss of detectable DSBs and dead spores. In these two scenarios, however, the spores are dead for different reasons: the lethality that occurs in spores that have undergone DSB repair using sister chromatids is a result of nondisjunction occurring because homologues are not physically connected before MI. In contrast, the spore death arising from hyper-resection is due to the irreversible loss of genetic information. These two possibilities can be distinguished assaying spore viability in the absence of SPO13. Spore inviability arising due to a lack of interhomologue recombination should be suppressed by the single division meiosis conferred by spo13 (Malone and Esposito, 1981). If, however, spore death is due to DNA degradation, the spores will be dead even in a spo13 background.

Mek1 kinase activity can be specifically inhibited by the addition of 1-NA-PP1 to strains containing the mek1-as1 mutant (note that the mutation in mek1-as1 is Q241G and not Q247G as originally reported in Wan et al. (2004)). A mek1-as1 dmc1 spo13 diploid was transferred to sporulation medium, a final concentration of 1 μM 1-NA-PP1 was added to 1-ml aliquots at 0, 2, 4, and 6 h, and the cells returned to the 30°C incubator. After a total of 24 h, sporulation was monitored by phase contrast light microscopy. In the absence of inhibitor, the diploid behaved like a dmc1 mutant, with only 4.5% of the cells forming dyads, all of which were immature. In contrast, all of the time points to which inhibitor was added exhibited >35% sporulation, similar or better than the 36% sporulation observed for spo13 alone (Table 3). In addition, at the 4-h time point, 1 μM 1-NA-PP1 was added to one-half of the sporulating culture, and the cells were returned to the 30°C shaker. Cells were then fixed at 2-h intervals for analysis of DSBs at the naturally occurring YCR048w and HIS2 DSB hotspots (Wu and Lichten, 1994; Bullard et al., 1996). In the absence of inhibitor, DSBs failed to be repaired and became hyper-resected with increasing time in spo medium. Within 2 h after the addition of inhibitor, however, the DSBs were no longer detectable (Figure 1, A and B; unpublished data).

The behavior of the DSBs after addition of inhibitor in the mek1-as1 dmc1 spo13 diploid is highly similar to what was previously observed for the mek1-as1 dmc1 SPO13 strain (Wan et al., 2004). A major difference in the two experiments is the spore viability. Although only 3% of the spores were viable in the SPO13 experiment, 46.7% of the spores were viable in the mek1-as1 dmc1 spo13 mutant (Table 3; Wan et al., 2004). The fact that 32.6% of the dyads contained two viable spores provides strong support for the argument that DSBs are disappearing as a result of repair rather than degradation.

mek1-as1 dmc1 spo13 Mutants Are Reduced for Interhomologue Recombination and Increased for Intersister Recombination

If Mek1 kinase activity is acting to prevent sister chromatid repair in the mek1-as1 dmc1 spo13 diploid, then inactivation of Mek1 should result in dyads which exhibit reduced levels of interhomologue recombination and increased levels of intersister recombination. To monitor both interhomologue and intrachromosomal recombination, a sister chromatid recombination reporter, SCR::URA3, was introduced between LEU2 and HIS4 on one of the chromosome III homologues (Figure 1C; Kadyk and Hartwell, 1992). Interhomologue recombination between URA3 and MAT can be detected by a change in the coupling relationship between the two genes, whereas intersister events can be detected by the generation of a full-length ADE3 gene (see below).

In spo13 meioses, chromosomes may segregate either reductionally (homologues segregate to opposite poles), equationally (sisters segregate to opposite poles), or aberrantly (three chromatids go to one pole and one to the other pole; Klapholz and Esposito, 1980; Hugerat and Simchen, 1993). Defects in interhomologue recombination result in improved spore viability and dyads that display predominantly equational segregation (Wagstaff et al., 1982; Hollingsworth and Byers, 1989). The dyads formed by the mek1-as1 dmc1 spo13 diploid in the presence of inhibitor displayed both of these properties. The spore viability of the spo13 strain was only 25.5%, whereas the mek1-as1 dmc1 spo13 spore viabilities ranged from 38.5 to 61.2%, depending on when the inhibitor was added (Table 3). The highest viability was observed when inhibitor was added at 0 h, suggesting that there is a deleterious effect in allowing DSBs to form in the presence of Mek1 kinase activity when DMC1 is absent. As for segregation, 72.7% of the spo13 dyads could be unambiguously defined as equational segregants for chromosome III, increasing to 98.4% for mek1-as1 dmc1 spo13 plus inhibitor. (To discriminate between reductional and equational segregants in recombinant dyads, a centromere linked marker is necessary but was not available in this strain. Because MAT and URA3 are on different arms of chromosome III, nonrecombinant dyads produce distinctive patterns for reductional and equational segregation; Table 3). These results indirectly support the hypothesis that interhomologue recombination is reduced in mek1-as1 dmc1 spo13 diploids lacking Mek1 kinase activity.

To directly examine interhomologue recombination, changes in the coupling relationship between URA3 and MAT were examined. Because the poor spore viability of the spo13 diploid resulted in a low yield of two viable spore asci, the analysis used spores from both one and two viable spore dyads (Table 3). Out of 499 dyads from the mek1-as1 dmc1 spo13 strain to which inhibitor was added, only 4 (0.8%) were recombinant for MAT and URA3. In contrast, 35 of the 236 spo13 dyads were recombinant (14.8%). Therefore interhomologue recombination is significantly reduced in the absence of DMC1 and Mek1 kinase activity (χ² analysis; p < 0.0001).

Recombination between truncated ade3 alleles in the SCR::URA3 reporter was used to determine whether sister chromatid recombination was elevated under these conditions (Figure 1C; Kadyk and Hartwell, 1992). To prevent ectopic interactions, ADE3 was deleted from its normal chromosomal position. ADE3 is required for the biosynthesis of histidine and the diploid is therefore His^-. Two types of intersister events are detectable by the formation of His⁺ recombinants: unequal reciprocal recombination and gene conversion (Kadyk and Hartwell, 1992). The frequency of intersister recombination events for the mek1-as1 dmc1 spo13 diploid was 2.2% (11/499; Table 3). In two viable spore dyads it is possible to discriminate between gene conversion and reciprocal exchange events. All of the His⁺ two viable spore dyads resulted from gene conversion—no reciprocal recombinants were detected. In the spo13 diploid, no His⁺ spore colonies were detected out of 236 spo13 dyads, although five would be expected if the frequency on intersister recombination were the same as mek1-as1 dmc1 spo13 (Table 3). Although the sample size is small, these data suggest there may be a bias toward intersister gene conversion events in the absence of DMC1 and Mek1 kinase activity.

The DMC1-independent DSB Repair Observed in the Absence of MEK1 Requires RAD54

To further test the hypothesis that sister chromatid recombination is responsible for DSB repair in mek1Δ dmc1Δ, the dependence of this repair on RAD54 was analyzed. The Rad54 protein stimulates Rad51 activity in vitro and is involved primarily in intersister recombination during meiosis (Arbel et al., 1999; Bishop et al., 1999; Petukhova et al., 1999). Therefore if the disappearance of DSBs in mek1Δ dmc1Δ strains is due to recombination between sister chromatids, this repair should not occur in a rad54Δ dmc1Δ mek1Δ diploid. To compare isogenic strains, a rad54Δ mek1Δ dmc1Δ diploid was transformed with RAD54, MEK1, DMC1 or vector to generate mek1Δ dmc1Δ, rad54Δ dmc1Δ, mek1Δ rad54Δ and rad54Δ mek1Δ dmc1Δ diploids, respectively. The wild-type strain was also included as a control. DSBs at the naturally occurring YCR048w hotspot were monitored by Southern blot analysis (Wu and Lichten, 1994). DSBs appeared in the wild-type diploid by 3 h and the bulk of the breaks were gone by 9 h (Figure 2). The kinetics of DSB appearance and disappearance were similar in the mek1Δ rad54Δ strain, although there appeared to be significant hyper-resection as indicated by the long smear in the DSB region of the gel (Figure 2A). This repair is presumably being mediated by Dmc1, indicating that MEK1 is not required to promote Dmc1 function. DSBs persisted in the rad54Δ dmc1Δ diploid and became hyper-resected (Figure 2). Deletion of RAD54 from the mek1Δ dmc1Δ diploid blocked DSB repair, with the DSBs exhibiting even more hyper-resection than the rad54Δ dmc1Δ strain (Figure 2). There was a delay in the onset of break formation in the triple mutant, raising the possibility that repair might also be delayed. Extending the time course to 15 h gave the identical result, however, making this possibility unlikely (Figure 2B). RAD54 is therefore necessary for DMC1-independent repair in the absence of MEK1.

Suppression of dmc1Δ by Overexpression of RAD51 Is Dependent on Mek1 Kinase Activity

Overexpression of RAD51 largely suppresses the interhomologue recombination and sporulation defects of dmc1Δ during meiosis (Tsubouchi and Roeder, 2003). If the ability of Rad51 to promote meiotic interhomologue recombination is due to a BSCR, then RAD51 suppression of dmc1Δ should be dependent on Mek1 kinase activity. This hypothesis was tested by transforming a dmc1Δ mek1-as1 diploid with either vector, DMC1 or a plasmid overexpressing RAD51 and testing the transformants for a variety of meiotic phenotypes in the presence or absence of the 1-NA-PP1 inhibitor. To eliminate any possible negative effects arising from the overexpression of RAD51 in vegetative cells, RAD51 was expressed under the control of the meiosis-specific HOP1 promoter (Hollingsworth et al., 1990).

Meiotic time courses were performed on four independent mek1-as1, mek1-as1 dmc1Δ, and mek1-as1 dmc1Δ/2μ RAD51 cultures. The culture overexpressing RAD51 was split immediately after transfer to sporulation medium and a final concentration of 1 μM 1-NA-PP1 was added to half. The cells were then incubated at 30°C for 24 h. As expected in the absence of inhibitor, the mek1-as1 diploid sporulated well (96.0 ± 0.9% asci) and produced viable spores (90.0 ± 5.2%, 72 asci dissected), in contrast to the mek1-as1 dmc1Δ strain which failed to sporulate. Consistent with the results of Tsubouchi and Roeder (2003), overexpression of RAD51 partially rescued the sporulation defect of mek1-as1 dmc1Δ (36.6 ± 5.1% asci). The high spore viability of the tetrads produced in the absence of inhibitor in this strain (70.0 ± 13.2% viable spores, 102 asci dissected) indicates that interhomologue recombination is occurring. Measurements of plasmid stability at the time of transfer to spo medium showed that ∼70% of the mek1-as1 dmc1Δ cells contained the RAD51 plasmid. Therefore the observed suppression of sporulation underestimates the amount of possible suppression. Addition of inhibitor to the RAD51 overexpressing strain resulted in even higher levels of sporulation (72.6 ± 10.6% asci) but inviable spores (0.8 ± 1.0%, 101 asci dissected).

This experiment provides a unique opportunity to compare the kinetics of meiotic progression in a culture in which Rad51 is being used to repair DSBs either using homologues or sister chromatids as templates. When interhomologue recombination utilizes only Rad51, the onset of MI is delayed, supporting the idea that Dmc1 facilitates this process (Figure 3). In contrast, inhibition of mek1-as1 in the RAD51 overexpressing dmc1 diploid allowed even more rapid meiotic progression than wild type (Figure 3). Comparison of DSBs in the RAD51 overexpression strain showed that in the presence of inhibitor, DSBs at the YCR048w hotspot were repaired by 6 h, whereas in the absence of inhibitor the DSBs did not disappear until 12 h (some of these persisting DSBs are most likely due to those cells that lost the RAD51 plasmid before the initiation of meiosis). Similar results were obtained looking at the ARG4 DSB hotspot (unpublished data). These results demonstrate that removal of the BSCR in dmc1 mutants results in a rapid, efficient repair of DSBs. Furthermore the ability of RAD51 to mediate interhomologue recombination when overexpressed in a dmc1 mutant is dependent on the BSCR being present.

Mutagenesis of the C-terminal Tail of Hop1 Identifies Two Lysines, K590A and K593A, Which Are Important for HOP1 Function

Intragenic complementation studies using various mutant alleles of HOP1 indicated that the 605 amino acid protein contains at least two discrete functional domains (Friedman et al., 1994). One domain consists of the last 20 amino acids of the protein (the C domain), whereas the other domain is comprised of the rest of the Hop1 protein (the N domain). One of the alleles used by Friedman et al. (1994), hop1-R6Δ, resulted not only in the deletion of the last 41 amino acids of Hop1, but also the addition of 18 amino acids as a result of translation of downstream vector sequences. Therefore it was not clear whether the spore inviability observed for hop1-R6Δ was due to the loss of the Hop1 C domain or to the insertion of extra amino acids at the end of the protein. To address this question, stop codons were introduced into HOP1 immediately after codon 564 (hop1-564Δ) or 585 (hop1-585Δ), thereby truncating Hop1 by 41 and 20 amino acids, respectively. Diploids carrying either hop1-564Δ or hop1-585Δ produced 0 viable spores out of 22 tetrads dissected, indicating both mutations create null alleles with regard to spore viability. The truncations do not appear to destabilize the mutant proteins, however, because their protein levels are similar to wild-type Hop1 (measured 3 h after transfer to sporulation medium; Figure 4A). The conclusion therefore, is that the last 20 amino acids of Hop1 are essential for its function.

Assuming that the C domain of Hop1 represents a discrete functional module, it should be possible to isolate point mutations that abolish the function of this domain without affecting activities that are mediated by the rest of the protein. Null alleles of HOP1 exhibit a number of mutant phenotypes, including low spore viability, reduced levels of interhomologue recombination and DSBs, as well as a defect in chromosome synapsis (Hollingsworth and Byers, 1989; Woltering et al., 2000; Wan et al., 2004). In addition, Hop1 is a DNA binding protein that physically interacts both with itself and with Red1 (Kironmai et al., 1998; de los Santos and Hollingsworth, 1999). To generate separation of function alleles, charged residues located between amino acids 567 and 605 were mutated to alanine and assayed for defects in spore viability. Out of 15 amino acids that were mutated, two resulted in a reduction in spore viability. The most severe mutant, hop1-K593A, produced <1% viable spores in the SK1 background (36 asci dissected), equivalent to a deletion of HOP1. In addition, changing K590 to alanine or methionine reduced spore viability to 42 and 59%, respectively (hop1-K590A, 49 asci; hop1-D584A K590M, 49 asci). The amino acid substitutions present in hop1-K590A and hop1-K593A do not appear to decrease protein stability (Figure 4A).

hop1-K593A Mutants Exhibit Increased Levels of Crossovers and DSBs Compared with a Null Allele of HOP1

hop1Δ and red1Δ mutants reduce, but do not eliminate, meiotic recombination (Hollingsworth and Byers, 1989; Rockmill and Roeder, 1990; Mao-Draayer et al., 1996). hop1Δ displays a more severe recombination phenotype than red1Δ, with hop1Δ red1Δ resembling hop1Δ alone, suggesting that HOP1 has a function in recombination independent of RED1 (Rockmill and Roeder, 1990). To determine whether the C-terminus is required for this recombination function, the effect of hop1-K593A on interhomologue crossing over was measured in spo13 diploids heterozygous for markers on two different chromosomes. A hop1Δ mutation was used as the null control. The hop1Δ diploid exhibited a mean 60-fold reduction in crossing over measured in four intervals. In contrast, hop1-K593A decreased crossing over on average only 11-fold (Table 4). This phenotype is highly similar to that observed for hop1-R6Δ, supporting the idea that the K593A mutation abolishes the function of the C domain. Both hop1-K593A and hop1-R6Δ are phenotypically similar to an isogenic red1Δ diploid, which also reduced crossing over on average 11-fold (Table 4). These results argue that HOP1 contains a recombination function that is partially intact in the hop1-K593A mutant.

hop1Δ and red1Δ mutants exhibit decreased steady state levels of DSBs (Mao-Draayer et al., 1996; Xu et al., 1997; Woltering et al., 2000). In the SK1 background, red1Δ mutants exhibit a higher frequency of DSBs compared with hop1Δ, even when resection of the ends is prevented by mutation of SAE2/COM1 (Figure 5; Woltering et al., 2000; Pecina et al., 2002). hop1Δ is epistatic to red1Δ with regard to DSBs, as a red1Δ hop1Δ sae2Δ diploid produces a level of DSBs equivalent to hop1Δ sae2Δ (unpublished data). The amount of DSBs in the hop1-K593A sae2Δ diploid resembled that of red1Δ sae2Δ, representing a sixfold increase over the hop1Δ sae2Δ strain (Figure 5). Similar results were obtained in the SAE2 strains, although the levels of DSBs in these strains were lower than those in the sae2Δ diploids, presumably because DSBs in the SAE2 diploids are repaired and therefore do not accumulate (Figure 5). These results argue that the N domain either promotes initiation of DSBs or prevents DSB ends from being degraded.

The Hop1-K593A Protein Localizes to Chromosomes but hop1-K593A Mutants Are Defective in Chromosome Synapsis

Null mutants of HOP1 result in the formation of AEs but no SCs (Hollingsworth et al., 1990; Loidl et al., 1994). Electron microscopic analysis of spread chromosomes from hop1-K593A revealed a similar phenotype (Figure 6). The synapsis defect was also manifested by the failure of Zip1, a component of meiotic chromosomes frequently used as an indicator of synapsis (Sym et al., 1993; Sym and Roeder, 1995), to localize along the lengths of chromosomes from both hop1-K593A and hop1Δ diploids. Localization of the Hop1-K593A protein to unsynapsed chromosomes was unaffected, however, indicating that although the C domain is necessary for synapsis, it is not required for Hop1 to interact with chromosomes (Figure 6).

The C Domain Is Not Required for Interaction with Red1 or Hop1

The fact that overexpression of RED1 specifically suppresses a mutation at codon 595 in the HOP1 C domain had suggested that Red1 might physically interact with the Hop1 C-terminus (Hollingsworth and Johnson, 1993). This idea was tested by examining the ability of Red1 to IP various Hop1 tail mutants. For these experiments a polyclonal antibody generated against a fragment of the Red1 protein was used (Wan et al., 2004). Co-IP of Red1 with Hop1 is readily detectable using these antibodies (Figure 4B). Soluble extracts from isogenic diploids containing various alleles of HOP1 were used to precipitate either Hop1 or Red1 by addition of the appropriate antibody. After the IPs were fractionated by SDS-PAGE, the filters were probed with α-Hop1 antibodies. As expected, no Hop1 was observed in the strains deleted for HOP1 (Figure 4A). Hop1, Hop1-K590A, Hop1-K593A, Hop1-564, and Hop1-585 all co-IPed with Red1 (Figure 4A). The C-terminus of Hop1 is therefore dispensable for Red1 binding.

Purified Hop1 protein exists as a dimer in solution (Kironmai et al., 1998). To test whether the C domain affects Hop1 dimerization, Hop1 and Hop1-585 were purified after expression in E. coli and analyzed by gel filtration analysis under nondenaturing conditions. Both proteins exhibited the same elution profile, indicating that Hop1 can dimerize in the absence of the C domain (unpublished data). These results are consistent with genetic data showing that mutations in the C domain can intragenically complement mutations in the N domain (Friedman et al., 1994).

C Domain Mutants in HOP1 Can Be Bypassed by Ectopic Dimerization of Mek1

Given that the Hop1 C domain functions downstream of DSBs as well as Red1 and Hop1 binding, we wanted to test whether the C-terminal tail of Hop1 is needed for Mek1 activation. Our standard Mek1 kinase assay uses Gst-Mek1 partially purified from meiotic extracts and monitors Mek1 autophosphorylation (de los Santos and Hollingsworth, 1999). A hop1-K593A GST-MEK1 diploid was therefore constructed (in these experiments all of the HOP1 and MEK1 alleles are integrated into the chromosome in single copy, unless otherwise stated). Given that a hop1-K593A MEK1 diploid produces inviable spores, the finding that hop1-K593A GST-MEK1 produces nearly the same high level of viable spores as an isogenic HOP1 GST-MEK1 diploid was unexpected (Table 5). Suppression of hop1-K593A requires Mek1 kinase activity, as a catalytically inactive mutant, GST-mek1-K199R, fails to suppress (Table 5). Zip1 staining demonstrated that GST-MEK1 also suppresses the synapsis defect of hop1-K593A (unpublished data). These results suggest that the presence of GST in the Mek1 protein is sufficient to bypass the requirement for the Hop1 C domain during meiosis.

It has previously been shown that GST dimerizes in solution (Lim et al., 1994; Vargo et al., 2004). This observation raised the possibility that the function of the Hop1 C domain is to promote dimerization of Mek1. In this case, the presence of a dimerization domain such as GST in Mek1 could bypass the requirement for the Hop1 C domain by providing an alternative means for dimerization. Two experiments were performed to test this hypothesis. In the first experiment, alternative N-terminal fusions to MEK1 were assayed for their ability to complement the spore inviability of mek1Δ in a HOP1 diploid as well as for their ability to suppress hop1-K593A. Similar to GST, the lexA protein has been shown to form dimers in solution (Mohana-Borges et al., 2000). In contrast, the TAP tag contains protein A sequences as well as a calmodulin-binding domain, neither of which is known to dimerize (De et al., 1997; Puig et al., 2001). The TAP-MEK1 and lexA-MEK1 fusions were transformed into mek1Δ HOP1 and mek1Δ hop1-K593A diploids and assayed for spore viability. In single copy, both lexA-MEK1 and TAP-MEK1 complemented well, producing >85.0% viable spores (Table 5). TAP-MEK1 failed to improve the spore viability of hop1-K593A, but lexA-MEK1 conferred partial suppression, producing 23.3% viable spores (Table 5). Overexpression of lexA-MEK1 exhibited better suppression of hop1-K593A than single copy lexA-MEK1, producing 46.5% viable spores (Table 5), suggesting that the partial suppression exhibited by lexA-MEK1 may be due to inefficient dimerization within the cell.

The second experiment to test the dimerization hypothesis was to introduce amino acid substitutions into GST that are likely to disrupt dimerization. Using the crystal structure of dimerized GST as a guide for mutagenesis (Lim et al., 1994; Vargo et al., 2004), proline was substituted for an arginine at position 72 and an arginine was substituted for an aspartic acid at position 76 to disrupt hydrophobic and hydrophilic interactions required for dimerization (gst-RD-MEK1). gst-RD-MEK1 complemented well in the mek1Δ HOP1 diploid, indicating that the MEK1 in this fusion is functional (Table 5). Suppression of the spore viability defect of hop1-K593A was greatly reduced in the gst-RD-MEK1 background, however, dropping from 87.9% for GST-MEK1 hop1-K593A to 6.7% (Table 5). Furthermore, Zip1 staining revealed that chromosome synapsis was reduced in the hop1-K593A gst-RD-MEK1 strain, although some nuclei showed partial SC formation indicating that the phenotype was somewhat better than hop1-K593A alone (unpublished data). Attempts to confirm Mek1 dimerization by coimmunoprecipitation of differentially tagged Mek1 proteins have thus far been unsuccessful. Given the strong genetic evidence for dimerization, this negative result seems likely to be due to technical issues. One possibility is that only a fraction of Mek1 is dimerized during meiosis. If this fraction is small relative to the total amount of Mek1 protein, then detecting the interaction by biochemical methods may be difficult.

GST-MEK1 not only suppresses the K593A point mutant in the HOP1 C-domain, it also partially suppresses a version of Hop1 in which the tail is deleted (hop1-585Δ; Table 4). This result argues that the primary role of the Hop1 C domain is to actively promote dimerization of Mek1.

Dimerization of Mek1 Is Necessary for Preventing DMC1-independent Repair of Meiotic DSBs

We infer that the spore inviability of hop1-K593A results from missegregation of chromosomes at Meiosis I due to a failure to prevent recombination between sister chromatids. If true, then the absence of the BSCR in hop1-K593A should allow DSBs in dmc1Δ diploids to be repaired and dmc1Δ hop1-K593A diploids should sporulate. A hop1Δ dmc1Δ diploid was constructed and various alleles of HOP1 and MEK1 introduced by transformation. Addition of both HOP1 and DMC1 creates a wild-type diploid that sporulates well and exhibits high spore viability (Table 6). In the HOP1 dmc1Δ diploid, sporulation was reduced to <0.2% as previously reported (Bishop et al., 1992). Deletion of HOP1 in the dmc1Δ background allowed the cells to sporulate, consistent with a role for HOP1 in the BSCR. The hop1-K593A dmc1Δ diploid also sporulated well and produced dead spores, indicating that the Hop1 C domain is required for preventing DMC1-independent repair (Table 6). The possibility that hop1-K593A does not generate sufficient DSBs to trigger the meiotic recombination checkpoint is ruled out by the fact that GST-MEK1, while having no effect on sporulation in the hop1Δ dmc1Δ strain, restores the meiotic arrest of the hop1-K593A dmc1Δ strain (Table 6). As with spore viability, the ability of GST-MEK1 to restore the arrest to the hop1-K593A dmc1Δ diploid requires Mek1 kinase activity (Table 6).

Hop1 Is a DSB-dependent Phosphoprotein

Because of the genetic interactions between the Mek1 kinase and Hop1, we tested to see whether Hop1 is a phosphoprotein. Our previous work had detected only a single band on protein gels for Hop1 (e.g., Figure 4). We found, however, that running the gels for a much longer period of time enabled the detection of slower migrating species (Figure 7A). The slower migrating forms are eliminated by treatment of IPed Hop1 with λ protein phosphatase, demonstrating that Hop1 is a phosphoprotein. The absence of MEK1 did not affect the Hop1 mobility shift, indicating that Mek1 is unlikely to be the kinase that phosphorylates Hop1 (Figure 7A). Hop1-585, which is deleted for the Hop1 C domain, exhibits only a single, phosphatase-insensitive form, indicating either that the C-domain is phosphorylated directly or that its presence is required for phosphorylation elsewhere on Hop1 (Figure 7A).

To test whether Hop1 phosphorylation is regulated by DSBs, the gel mobility of Hop1 IPed from a rec104 diploid was examined. REC104 is one of several meiosis-specific genes required for generating meiotic DSBs (Pecina et al., 2002). Although the phosphorylated form of Hop1 was present at both 5 and 7 h after the induction of meiosis, the bulk of the Hop1 remained unphosphorylated in the rec104 diploid at both time points (Figure 7B). Similar results have been obtained with spo11 mutants (unpublished data). Therefore the majority of phosphorylated Hop1 protein present in meiotic cells is dependent on the formation of DSBs.