The Saccharomyces cerevisiae MER3 gene, encoding a novel helicase‐like protein, is required for crossover control in meiosis

Takuro Nakagawa, Hideyuki Ogawa

Author Affiliations

  1. Takuro Nakagawa1,2 and
  2. Hideyuki Ogawa*,1,3
  1. 1 Department of Biology, Graduate School of Science, Osaka University, Toyonaka, Osaka, 560‐0043, Japan
  2. 2 Present address: Ludwig Institute for Cancer Research, University of California San Diego, La Jolla, CA, 92093‐0660, USA
  3. 3 Present address: Iwate College of Nursing, Ohgama, Takizawa, Iwate, 020‐0151, Japan
  1. *Corresponding author. E-mail: hogawa{at}


The MER3 gene is identified as a novel meiosis‐specific gene, whose transcript is spliced in an MRE2/MER1‐dependent manner. The predicted Mer3 protein contains the seven motifs characteristic of the DExH‐box type of helicases as well as a putative zinc finger. Double strand breaks (DSBs), the initial changes of DNA in meiotic recombination, do not disappear completely and are hyperresected late in mer3 meiosis, indicating that MER3 is required for the transition of DSBs to later intermediates. A mer3 mutation reduces crossover frequencies, and the remaining crossovers show random distribution along a chromosome, resulting in a high incidence of non‐disjunction of homologous chromosomes at the first meiotic division. MER3 appears to be very important for both the DSB transition and crossover control.


During meiosis, two successive rounds of chromosome segregation occur, following a single round of DNA replication, producing haploid gametes from diploid cells. The first meiotic division (meiosis I) is unique in that homologous chromosomes (homologs) are synapsed and then distributed to opposite poles. Crossovers and non‐crossovers, two types of recombinants, are generated between homologs during meiotic prophase. Crossovers are associated with reciprocal exchanges of chromosome arms and are needed for faithful segregation of homologs, but non‐crossovers are not (for reviews, see Carpenter, 1988; Kleckner, 1996; Roeder, 1997). Among chromosomes, crossovers are distributed non‐randomly in that every homolog sustains at least one, even though the average number per homolog is very low (1‐3). Along a chromosome, multiple crossovers are further apart than expected on a random basis; this phenomenon is called crossover interference. Although the distribution of crossovers among and along chromosomes is likely to represent different manifestations of the same underlying regulation (Sym and Roeder, 1994; Chua and Roeder, 1998), the mechanism of crossover control is not understood.

In Saccharomyces cerevisiae, meiosis‐specific double strand breaks (DSBs) are resected rapidly to produce 3′ overhanging single strands and are converted to strand exchange intermediates that contain double Holliday junctions (Cao et al., 1990; Sun et al., 1991; Schwacha and Kleckner, 1994, 1995). Rad51 and Dmc1, strand exchange proteins, are required for the generation of both crossovers and non‐crossovers (Bishop et al., 1992; Shinohara et al., 1992, 1997; Bishop, 1994). During or after the transition of DSBs to strand exchange intermediates, homologs are synapsed along their entire lengths forming synaptonemal complexes (SCs) (Padmore et al., 1991; Schwacha and Kleckner, 1994). Zip1, a component of the central region of SC, is required for crossover interference. It has been proposed that, after SC polymerization, Zip1 transmits negative signals from crossover sites to neighbors in order to prevent additional crossovers (Sym et al., 1993; Sym and Roeder, 1994). However, it has also been suggested that, before SC polymerization, Zip1 acts in crossover control, due to the observation that a zip1 mutation in an SC formation‐deficient background further reduces crossing over (Storlazzi et al., 1996). The step of recombination at which the crossover control takes place is as yet unresolved. Intact DNA duplexes containing heteroduplex regions appear shortly before or concomitant with the appearance of mature recombinants (Goyon and Lichten, 1993; Nag and Petes, 1993). This may reflect the coordination between the formation and resolution of strand exchange intermediates.

Mutations in either MRE2, MER1 or MER2 impair DSB formation (Rockmill et al., 1995; Storlazzi et al., 1995; Nakagawa and Ogawa, 1997). MRE2 and MER1 encode RNA‐binding proteins and are required for efficient splicing of the MER2 intron, which contains a non‐canonical 5′ splice site (Engebrecht et al., 1991; Nakagawa and Ogawa, 1997). Elimination of the MER2 intron in an mre2 mutant restores the formation of DSBs and non‐crossovers. However, this mre2 cMER2 (intronless MER2) strain is still defective in crossing over and produces inviable spores (Nakagawa and Ogawa, 1997). Similarly, overexpression of MER2 in a mer1 mutant restores the formation of non‐crossovers, but not crossovers (Engebrecht et al., 1990; Storlazzi et al., 1995). Thus, an unidentified target(s) of MRE2/MER1‐dependent splicing specifically required for crossing over has been suggested.

Here, we identify the MER3 gene, the transcript of which is a new target of MRE2/MER1‐dependent splicing. The predicted Mer3 protein has the seven motifs conserved amongst the DExH‐box type of DNA/RNA helicases as well as a putative zinc finger. Meiosis‐specific DSBs do not disappear completely and are hyperresected late in mer3 meiosis, indicating the role of MER3 in the transition of DSBs to later intermediates. A mer3 mutation decreases the frequency and alters the distribution of crossovers, resulting in a high incidence of homolog non‐disjunction at meiosis I. Our results indicate the requirement of MER3, encoding a novel helicase‐like protein, for both the DSB transition and crossover control.


Identification of a novel gene, MER3, which suppresses the post‐initiation recombination defect of an mre2 cMER2 mutant

An mre2 cMER2 mutant is defective in crossing over but is proficient in generating non‐crossovers. To identify a gene specifically required for crossing over, we searched for multicopy suppressors of the crossover defect using the mre2N cMER2 strain. The mre2N allele confers temperature‐sensitive spore formation but impairs crossing over at all temperatures (Nakagawa and Ogawa, 1997). mre2N cMER2 cells were transformed with a yeast genomic library constructed on a multicopy plasmid and induced to undergo meiosis at 23°C. Recombinants were selected based on crossing over in the LEU2‐HIS4 or TRP5‐CYH2 interval as well as on genome haploidization (see legend to Figure 1A). Among ∼9000 transformants, eight displayed increased frequencies of recombinants compared with background. Restriction mapping of plasmids recovered from these eight transformants revealed that two of these contain MRE2 and the remainder contain overlapping inserts. We named the suppressor gene MER3. Multicopy MER3 suppresses the recombination deficiency in mre2N cMER2 and mre2Δ cMER2, but not in mre2N or mre2Δ mutants (Figure 1A), indicating that cMER2 is required for suppression. mre2N cMER2 mutants harboring multicopy MER3 produce higher frequencies of recombinants and viable spores than mre2Δ cMER2 mutants harboring the MER3 plasmid (Figure 1A). This difference may be due to residual activity of Mre2N.

Figure 1.

Suppression of mre2 cMER2 defects by MER3. (A) Assays for crossing over and spore viability. Wild‐type (TNY171), mre2Δ cMER2 (TNY240), mre2N cMER2 (TNY169), mre2Δ (TNY185) and mre2N (TNY170) strains were transformed with a vector (YEp24) or the MER3 plasmid (pTN45), patched on SD‐Ura plates and replica plated to SPM‐Ura to induce meiosis. After 4 days at 23°C, SPM‐Ura plates were replicated to SD‐Ura, −Arg, −Leu, −His, +CYH, +CAN and SD‐Ura, −Arg, −Trp, +CYH, +CAN plates; papillae formed on these plates result from crossing over in the LEU2‐HIS4 and TRP5‐CYH2 intervals, respectively, and from haploidization. The spore viability was examined by colony formation following tetrad dissection. Numbers of viable and total spores are in parentheses. CYH, cycloheximide; CAN, canavanine; ND, not determined. (B) Subcloning of the MER3 gene. A series of deletion constructs were derived from pTN45 and their suppression activities were tested by the plate assay using the mre2N cMER2 strain. YGL251c and MER3 ORFs are shown and their coding regions are numbered below. The positions of primers (priTN1 and 2) used in RT‐PCR and a probe used for Northern blotting are illustrated. H, HindIII; R, EcoRI; Sa, SalI; Sp, SphI.

Splicing of the MER3 transcript depends on MRE2 and MER1

Subcloning and partial DNA sequencing of the MER3 plasmid revealed that the suppression activity resides in a 4.8 kb EcoRI‐SalI fragment (pTN84, Figure 1B), including a hypothetical open reading frame (ORF), YGL251c (Coissac et al., 1996). However, YGL251c with an additional upstream region of 508 bp (pTN66) was not sufficient for the suppression. Given the observation that multicopy MER2 partially suppresses the mre2 or mer1 defect, and that MER2 splicing requires MRE2 and MER1, an intron of MER3 might exist in the region upstream of YGL251c, splicing of which requires MRE2 and MER1. To test this possibility, we designed a pair of primers, priTN1 and priTN2, located in the region upstream (Figure 1B, see Materials and methods), and performed reverse transcription polymerase chain reaction (RT‐PCR) analysis using RNAs prepared from meiotic cells. If there is no intron between the primers, a fragment of 500 bp should be amplified. However, a fragment smaller than 500 bp was amplified exclusively in the wild type (Figure 2A). DNA sequencing of the amplified fragment revealed that the MER3 primary transcript has a 152 nucleotide intron and that the MER3 ORF starts at −575 from YGL251c and includes it (Figure 1B). The 5′ splice site and branchpoint sequences of MER3 differ from both the consensus and those of MER2 (Figure 2B). Furthermore, the 5′ splice site sequence of MER3 is unique among all introns reported in S.cerevisiae. In mre2Δ or mer1Δ mutants, only the unspliced fragment was amplified (Figure 2A), demonstrating that MER3 splicing depends on MRE2 and MER1.

Figure 2.

Splicing and meiotic induction of the MER3 transcript. Total RNAs were prepared from wild‐type (TNY058), mre2Δ (TNY060), mer1Δ (TNY305) and mer3Δ (TNY286) cells immediately before the induction of meiosis and 5 h later. (A) RT‐PCR assay for the MER3 splicing. Meiotic RNA samples were subjected to RT followed by PCR, using priTN1 and priTN2 primers. Products were separated on a 1.2% agarose gel and visualized by staining with ethidium bromide. The positions of unspliced and spliced products are indicated on the left and the sizes of molecular weight standards are shown on the right. ACT1 splicing was observed in mre2Δ and mer1Δ, as well as in wild type (data not shown). (B) The three conserved elements in introns. The 5′ splice site, branchpoint and 3′ splice site sequences are shown for MER3, MER2 and the consensus (Rymond and Rosbash, 1992). The same branchpoint sequence of MER3 has been reported in yeast (Myslinski et al., 1990). The positions of these elements are also indicated. (C) Northern blot analysis of MER3 transcripts. MER3 (top panel) and ACT1 (bottom panel) transcripts were detected using the same membrane. ACT1 was used as a standard for the amount of RNA loaded on the gel. The positions of 18S and 25S rRNAs detected by ethidium bromide staining are indicated on the right. We do not know the significance of a 3.0 kb RNA on the MER3 blot that is shorter than the MER3 ORF.

Using mitotic and meiotic RNAs, Northern blotting was carried out to see the MER3 transcript. In the wild type, the transcripts of ∼4.2 and 3.0 kb were observed only in meiosis with a probe located within the pTN84 insert (Figures 2C and 1B). In mer3Δ (see below), those transcripts were detected in neither mitosis nor meiosis (Figure 2C). Consistent with the meiosis‐specific transcripts of MER3, URS1 elements (Steber and Esposito, 1995) were identified around the first ATG of MER3 (TCGGCGGGT, position −132 to −124; AGCCGCCAA, position 260‐268). Even in the absence of MRE2 or MER1, MER3 transcripts were detected in meiosis, although at 60% of the wild‐type level (Figure 2C). The reduction of the amount of RNA may be due to the instability of unspliced RNAs.

To confirm that MER3 pre‐mRNA is a target of MRE2/MER1‐dependent splicing, the intron was eliminated from the genomic locus by substituting an intronless MER3, cMER3, constructed from the RT‐PCR product (Figure 2A). As was seen for cMER2, introduction of cMER3 did not change meiotic division, sporulation or spore viability significantly in the wild type (Table I). While only slight changes in the meiotic properties were observed when cMER3 was introduced into mre2Δ and mer1Δ mutants, cMER3 greatly improved the spore viability of mre2Δ cMER2 and mer1Δ cMER2 mutants (Table I). However, neither sporulation nor spore viability reach wild‐type levels in the case of either mre2Δ cMER2 cMER3 or mer1Δ cMER2 cMER3, suggesting another target(s) of MRE2/MER1‐dependent splicing.

View this table:
Table 1. Meiotic properties of mre2Δ and mer1Δ derivatives

The Mer3 protein contains helicase motifs and a putative zinc finger

The spliced MER3 mRNA encodes a 1187 amino acid polypeptide. The predicted Mer3 protein contains the seven motifs characteristic of the DExH‐box type of DNA/RNA helicases (Gorbalenya and Koonin, 1993) and a putative zinc finger (CFHSCKDKTQCRHLCC), which may participate in protein‐DNA or protein‐protein interactions (Figure 3A). The BLASTP 2.0.7 search using the entire sequence of Mer3 shows that the Mer3 helicase domain has significant homology (E value <4e−5) to yeast Brr2 and its human homolog U5‐200kD, and to yeast Sgs1 and its human homolog Blm, whose helicase activity has been shown by biochemical assays (Lu et al., 1996; Karow et al., 1997; Laggerbauer et al., 1998; Raghunathan and Guthrie, 1998) (Figure 3B). Brr2 and U5‐200kD are RNA splicing factors. On the other hand, mutations in SGS1 or BLM cause a genomic instability, and individuals with Bloom's syndrome (BLM is mutated) show a predisposition to cancer. The sequence similarity of Mer3 to these known helicases suggests that the biochemical function of Mer3 is to unwind nucleic acid helices.

Figure 3.

The Mer3 protein contains helicase motifs and a putative zinc finger. (A) The seven conserved motifs (I, Ia, II, III, IV, V, VI) of DNA/RNA helicases and a putative zinc finger in Mer3. The first and last amino acid positions are shown. (B) The amino acid sequences of the helicase motifs. Amino acids identical in at least three cases are shaded.

The mer3Δ mutant is defective in the transition of DSBs to later recombination intermediates

A mer3Δ strain was constructed, in which three‐quarters of the MER3 ORF including the first ATG and all the helicase motifs are deleted (Materials and methods). Under vegetative conditions, no growth defect or altered sensitivity to the DNA‐damaging agent, methyl methanesulfonate, was observed in mer3Δ cells (data not shown). To see whether MER3 has a role in early steps of meiotic recombination, meiosis‐specific DSBs in the HIS4::LEU2 region were examined by Southern blotting (Figure 4A). DNA was prepared from synchronous cultures, digested with PstI and separated on an agarose gel. In wild type, DSBs were prominent at 4 h after the induction of meiosis (t = 4 h), and were much less prominent thereafter (Figure 4B). DSB signals were smeared downwards, indicating the processing of DSB ends. In mer3Δ, DSBs were first observed at t = 4 h and were still seen at t = 12 h at both 30 and 23°C (Figure 4B). A fraction of the DSBs were hyperresected at late meiosis, although less extensively than those in rad51 and dmc1 mutants (data not shown). These results indicate that the transition of DSBs to later recombination intermediates is partially blocked in the mer3Δ mutant. However, it appears that some of the late DSBs are not hyperresected, suggesting that there is an additional defect (e.g. DSB formation at late meiosis). Note that the elevation in the steady‐state levels of DSBs was much more pronounced at 30 than at 23°C (Figure 4B and C). Interestingly, the mer3Δ mutant displays a defect in meiotic cell cycle progression, the severity of which parallels that of DSB accumulation (Figure 4D). At 30°C, only 20% of cells underwent meiotic nuclear division after a delay, while the remaining 80% arrested permanently. At 23°C, in contrast, all cells exhibited a delay in progression but 78% did eventually undergo meiosis I.

Figure 4.

DSB formation and nuclear division in meiosis. (A) The positions of major (site I) and minor (site II) DSB sites and PstI (Ps) restriction sites in the HIS4::LEU2 region are shown. (B) DNA was prepared from wild‐type (NKY1551) and mer3Δ (TNY286) cells caused to undergo meiosis at 30 and 23°C, digested with PstI, separated by agarose gel electrophoresis and transferred to a nylon membrane. A probe prepared from pNKY291 was used to detect fragments of interest by Southern hybridization. Fragments indicated by * may result from ectopic gene conversion between his4::LEU2 and leu2::hisG loci. P, parental fragments; DSB I, DSB fragments at site I; DSB II, DSB fragments at site II. (C) The steady‐state levels of DSBs at site I observed in (B) were measured by phosphoimager. The percentage of DSBs in the total DNA in each lane is shown. (D) Meiotic nuclear divisions were examined by 4′,6‐diamidine‐2‐phenylindole (DAPI) staining as described in Table I at 30 and 23°C. Plotted is the percentage of cells that had undergone one or both nuclear divisions (MI ± MII) at various times throughout sporulation. Sporulation frequencies in mer3Δ at t = 24 h were 4 and 24% at 30 and 23°C, respectively. Spore viabilities for mer3Δ were 22% (44/200 spores) and 27% (53/200 spores) at 30 and 23°C, respectively; spore viability for the wild type was 97% (194/200 spores) at both temperatures.

The MER3 gene is required for normal crossing over and for faithful segregation of homologs at meiosis I

The formation of recombination products, which are either associated with crossing over or not, was examined by Southern blotting (Figure 5). DNA prepared from synchronous cultures was digested with XhoI and MluI, and separated on an agarose gel. Two parental (P1‐2) and four recombinant fragments (R1‐4) from the HIS4::LEU2 recombination hot spot can be detected using the probe indicated in Figure 5A. In wild type, R1 and R2 are known to be correlated with events that, in tetrads, are associated with crossing over of flanking markers; R3 is not, and R4 results from both types of recombination (Storlazzi et al., 1995). The distribution between crossovers and non‐crossovers was different in mer3Δ and wild‐type strains (Figure 5B); crossovers (R2) were present at 50‐60% of the wild‐type level (Figure 5C), while non‐crossovers (R3) were present at normal levels at t = 8 h and continued to increase to ∼2.5 times the wild‐type level by 24 h (Figure 5D). Although the physical analysis reveals the kinetics of recombinant formation, assignment of fragments that arise in the mutant to the crossover and non‐crossover classes is based on the assumption that the relationships are the same as in wild type. Thus, we further examined the crossover frequency using mer3Δ tetrads formed at 23°C (Figure 4D). The tetrad analysis showed reduction of crossover frequencies in five intervals on two different chromosomes (Table II). The average decrease in crossover frequencies was 2.4‐fold. In contrast, the frequency of 1:3 or 3:1 aberrant segregation of genetic markers, which can occur with or without crossing over, was increased at all three loci examined (Table III).

Figure 5.

Physical analysis of crossovers and non‐crossovers. (A) The positions of polymorphic XhoI (X) and MluI (M) restriction sites in the HIS4::LEU2 region. Restriction fragments (P1‐2 and R1‐4) produced by XhoI and MluI digestion are illustrated. The two DSB sites are shown by arrows. (B) Southern blot analysis of crossovers and non‐crossovers. Wild‐type (NKY1551) and mer3Δ (TNY286) cells were incubated at 30 and 23°C and sampled at the times indicated after induction of meiosis. Restriction fragments of interest were detected using a probe prepared from pNKY155. DSBs detected in this assay are formed at site I on the chromosome containing his4X::LEU2‐URA3. CR, crossovers; NCR, non‐crossovers. (C and D) Amounts of crossovers and non‐crossovers produced at 23°C. (C) R2, crossovers. (D) R3, non‐crossovers. The percentages are the mean values obtained from four blots started from two independent cultures. Similar results were obtained at 30°C.

View this table:
Table 2. The mer3Δ mutation reduces crossing over
View this table:
Table 3. The mer3Δ mutation increases aberrant segregation

There is a tendency for the frequency of double crossovers in an interval to be lower than that predicted for single crossovers; this phenomenon is called interference. We tested whether MER3 is also required for crossover interference. Non‐parental ditypes (NPDs, Table II) are indicative of double crossovers in a given interval. The NPD ratio is the frequency of NPDs observed divided by that expected in the absence of interference (Materials and methods). Thus, no interference results in a ratio of 1.0. While all NPD ratios in wild type ranged from 0.20 to 0.44, those in mer3Δ were close to 1.0, from 0.74 to 1.32 (Figure 6). To analyze further the distribution of the crossovers along the chromosome, we examined the pattern of zero, one or two crossover events in wild‐type and mer3Δ strains (Table IV). In the wild type, the patterns in all three intervals examined were significantly different from those predicted by a Poisson distribution. On the other hand, the patterns in mer3Δ were not significantly different from those predicted.

Figure 6.

Distribution of crossovers along a chromosome. NPD ratios in three intervals for wild type (open bars) and mer3Δ (filled bars). The number of NPDs observed was compared with that expected. In the wild type, CAN1‐URA3 and URA3‐HOM3, P <0.005; HOM3‐TRP2, P <0.1. In mer3Δ, CAN1‐URA3 and HOM3‐TRP2, P >0.5; URA3‐HOM3, P >0.2.

View this table:
Table 4. Patterns of zero, one and two crossover events

The spore viability of the mer3Δ mutant was 20‐40%, while that of the wild type was ∼97%. In mer3Δ, the proportion of four spore viable tetrads was decreased, and that of two or zero spore viable tetrads was increased compared with the wild‐type levels (Table V). Three spore viable tetrads were not predominant among zero to three spore viable tetrads (Table V). These results suggest that non‐disjunction of homologs occurs in mer3Δ. This prompted us to monitor chromosome segregation during mer3Δ meiosis in a strain background (TNY374) where homologous CENIIIs can be distinguished by URA3 and TRP1 markers. Examination of 408 two‐spore‐viable tetrads from mer3Δ diploids revealed that 388 (95%) contained pairs of sister spores (i.e. both spores were Ura+/Trp, Ura/Trp+ or Ura+/Trp+), and 68 (17%) were disomic for chromosome III (i.e. both spores were Ura+/Trp+). In contrast, all five two‐spore‐viable tetrads from wild‐type diploids were pairs of non‐sister spores (i.e. Ura/Trp+ and Ura+/Trp sets of spores). These results indicate that the mer3Δ mutation causes non‐disjunction of homologs at meiosis I. No crossovers were seen in either the MAT‐CENIII or the CENIII‐HIS4 intervals in the 68 pairs of chromosome III disomes, while 11 crossovers were expected based on the crossover frequency seen in four‐spore‐viable tetrads from mer3Δ (Table II; see Materials and methods). In addition, among 111 one‐spore‐viable tetrads from mer3Δ, 23 (21%) were disomic for chromosome III (i.e. Ura+/Trp+) and none of them were recombinant. Thus, the homolog non‐disjunction is likely to be due to the mer3Δ defect in crossing over causing some pairs of homologs not to have any crossovers. No evidence of precocious separation of sister chromatids (i.e. Ura+/Trp+, Ura/Trp+ and Ura/Trp+ sets of spores) was observed among 81 three‐spore‐viable tetrads from mer3Δ.

View this table:
Table 5. Distributions of tetrad types


Here, we have identified the MER3 gene as a multicopy suppressor of the crossover defect in a mre2 cMER2 mutant. The predicted Mer3 protein contains the seven helicase motifs as well as a putative zinc finger. In a mer3 mutant, DSBs appear at normal timing but do not completely disappear thereafter, and a fraction of DSBs are hyperresected late in meiosis. Crossovers are reduced and distributed randomly along and among chromosomes, resulting in a high incidence of homolog non‐disjunction at meiosis I. Thus, a mutation in MER3 encoding a novel helicase‐like protein impairs both the transition of DSBs to later recombination intermediates and crossover control.

MRE2/MER1‐dependent splicing of MER3 pre‐mRNA

The MER3 gene is transcribed only in meiosis and has an intron. MER3 splicing depends on MRE2 and MER1, which are also required for MER2 splicing. The result that elimination of both MER3 and MER2 introns improves the spore viability of mre2 and mer1 mutants confirms that MER3 as well as MER2 pre‐mRNA is a target of MRE2/MER1‐dependent splicing. Non‐canonical 5′ splice sites are the prominent feature shared by MER3 and MER2 introns (Figure 2B). It has been shown that MER1 is no longer required when the 5′ splice site of MER2 or U1 snRNA is mutated to increase base pairing between them (Nandabalan et al., 1993). In addition, Puig et al. (1999) have shown recently that MRE2 (also called NAM8) facilitates pre‐mRNA splicing if the 5′ splice site is manipulated to be non‐canonical. Thus, MER3 splicing may be regulated at the interaction of the 5′ splice site with the splicing complex by Mre2 and Mer1. However, the meiotic phenotype of mre2 mutants is more severe than that of mer1 mutants (Table I), suggesting different roles for MRE2 and MER1 in RNA splicing.

Meiotic cell cycle checkpoint

The severity of DSB accumulation correlates with the degree of cell cycle arrest; both are more pronounced at 30 than at 23°C in a mer3 mutant. This correlation is consistent with the notion that a meiotic checkpoint monitors recombination intermediates (Bishop et al., 1992; Lydall et al., 1996; Xu et al., 1997). At both temperatures, however, a mer3 mutation reduces crossovers, but increases non‐crossovers. Thus, the severity of the cell cycle arrest does not correlate with that of the crossover defect. In addition, crossing over in tetrads that have completed both meiotic divisions is reduced in the mutant. Thus, it is unlikely that a mer3 mutation simply delays the progression of meiotic events; rather, the mutation directly causes a recombination defect.

A role for MER3 in crossover interference

In addition to reduced frequencies of crossovers, a mer3 mutant shows random distribution of crossovers along and among chromosomes, resulting in a high incidence of homolog non‐disjunction at meiosis I. These results indicate that MER3 has an essential role in crossing over occurring on every pair of homologs. In contrast to crossovers, non‐crossovers and aberrant segregation at some loci are increased. This raises the possibility that a mer3 mutation impairs the crossover control that is imposed at an early step of recombination, before the differentiation of intermediates into crossovers or non‐crossovers. However, it is also possible that there is a default pathway in the mutant that gives non‐crossovers only, because non‐crossovers are at almost the same level as in wild type at the time point when the wild‐type level reaches the maximum (t = 8 h, Figure 5D), and increase further thereafter. Interestingly, immunostaining experiments using anti‐Mer3 antibodies showed that the Mer3 protein localizes at discrete sites on meiotic chromosomes (T.Nakagawa and H.Ogawa, unpublished data). This result suggests that Mer3 functions at sites of recombination to impose crossover interference. It has been proposed that Zip1 transmits negative signals from crossover sites that prevent crossovers nearby (Sym and Roeder, 1994). From this point of view, Mer3 might radiate or receive the negative signal. Alternatively, Mer3 might be required for the polymerization of Zip1 along entire lengths of homologs, which is suggested to be required for Zip1 to function in interference. However, it is also proposed that there are geometrically two distinct types of double Holliday junctions, one of which is subject to the crossover control regardless of SC polymerization (Storlazzi et al., 1996). Thus, it is also possible that Mer3 affects the geometric conformation of Holliday junctions.

The function of Mer3

The Mer3 protein contains the seven motifs characteristic of DNA/RNA helicases and shows significant homology to several known helicases including Sgs1 and Blm (Figure 3). An alanine substitution for a highly conserved lysine in the helicase motif I, a putative nucleotide‐binding region, decreases crossing over and spore viability, predominantly at a low temperature (23°C) (T.Nakagawa and H.Ogawa, unpublished data). These results indicate the importance of the helicase domain for Mer3 function. Hyperresected DSBs are seen late in mer3 meiosis. In addition, the strand exchange proteins Rad51 and Dmc1 transiently localize as foci on meiotic chromosomes in wild type, but they persist in a mer3 mutant as shown in a zip1 mutant (Bishop, 1994; T.Nakagawa and H.Ogawa, unpublished data). These results are consistent with the possibility that MER3 functions in the DSB transition to later recombination intermediates. Interestingly, the RecQ protein, which is believed to be an Escherichia coli homolog of Sgs1 and Blm, has been shown to possess a dual role in vitro, promoting the formation of joined DNA molecules catalyzed by the E.coli RecA and SSB protein and dissociating the joined molecules (Harmon and Kowalczykowski, 1998).

It has been proposed that crossover control is imposed during the DSB transition, from the observation that a few normally resected DSBs are detected late in meiosis in a zip1 mutant, which also have a defect in crossover control (Sym et al., 1993; Storlazzi et al., 1996; Xu et al., 1997). The requirement for MER3 for both the DSB transition and crossover control supports this hypothesis. However, we cannot rule out the possibility that MER3 affects the expression of other genes and thus is required for different steps of recombination, as some proteins containing helicase domains are known to regulate gene expression (Eisen and Lucchesi, 1998).

Materials and methods

Strains and media

Yeast strains are listed in Table VI. All are of the SK1 strain background (Kane and Roth, 1974), except for the TNY367 and TNY368 strains which are SK1 congenic and derived from MY263 (Sym and Roeder, 1994). his4::hisG and trp5::hisG strains were constructed by replacement of 1.3 kb SnaBI‐BglII and 0.7 kb SpeI‐BglII regions, respectively, with a 1.2 kb hisG fragment. The mer1::LEU2 strain was derived from NKY2204 (Storlazzi et al., 1995). To make mer3::hisG and cMER3 strains, a 4.6 kb SacI fragment from pTN105 and a 5.6 kb EcoRI fragment from pTN149, respectively, were introduced into a yeast diploid strain. In his4::hisG, trp5::hisG, mer3::hisG and cMER3 constructions, uracil auxotrophs were selected by plating cells on SD plates supplemented with 5‐fluoro‐orotic acid. DNA integration was carried out by lithium acetate transformation (Ito et al., 1983) and verified by Southern blot analysis.

View this table:
Table 6. Yeast strains

Yeast media were prepared according to Treco and Lundblad (1992). MYPD, MYPL, YPA, SPM and synthetic medium were prepared as described earlier (Nakagawa and Ogawa, 1997). Cycloheximide and canavanine were added to the media at final concentrations of 3 and 60 μg/ml, respectively.

Induction of meiosis

For induction of meiosis at 23°C, cultures in both pre‐sporulation medium (YPA) and sporulation medium (SPM) were incubated at 23°C. Synchronous meiotic cultures were obtained as described previously (Nakagawa and Ogawa, 1997). Tetrad dissection was carried out using spores produced on SPM plates.


Plasmids were constructed by standard methods (Sambrook et al., 1989). The original MER3‐containing plasmid, pTN45, has an ∼9 kb fragment of yeast genomic DNA in YEp24 (New England Biolabs). To create the mer3::hisG‐URA3‐hisG plasmid, pTN105, a 4.8 kb NcoI‐SalI fragment from pTN45 was introduced into the SacII‐SalI sites of pBluescriptII KS+ (Stratagene) to give pTN97, and then a 3.8 kb hisG‐URA3‐hisG fragment (Alani et al., 1987) was substituted for a 3.5 kb AflII‐BstXI MER3 region (−15 to 3466) of pTN97. To create the cMER3‐URA3‐MER3 plasmid, pTN149, a 2.5 kb EcoRI‐ClaI fragment containing the cMER3 N‐terminal region and a 3.0 kb BamHI‐SalI fragment containing the MER3 C‐terminal SphI‐SalI region, in which the SphI site had been destroyed, were introduced into the EcoRI‐ClaI and BamHI‐SalI sites of YEp24, respectively. In pTN149, a 0.7 kb SphI‐ClaI MER3 region is directly duplicated and flanking URA3. A 0.3 kb AflII‐SpeI cMER3 fragment prepared from the RT‐PCR product was used for DNA sequencing and construction of the cMER3 gene.

Calculation of interference and statistical analysis

The frequency of NPDs expected was calculated from the Papazian equation (Papazian, 1952), NPD = 1/2[1 − T − (1 − 1.5T)2/3], where T is the frequency of tetratypes shown in Table II. Because the T value for the CAN1‐URA3 interval in wild type was >3/2, the expected frequency of NPDs in that interval only was determined as follows: NPD = T2/8(1 + 2T/3) (Papazian, 1952).

Since only one class gives NPDs among four types of two‐crossover (CR) tetrads, the numbers of zero, one and two CR events shown in Table IV were calculated as follows, assuming no chromatid interference: 0‐CR = PD − NPD; 1‐CR = TT − 2NPD; 2‐CR = 4NPD.

Data sets were analyzed using the χ2 test. Values of P <0.05 were considered significant.

Calculation of recombination frequencies among disomes

From the crossover frequency seen in four‐spore‐viable tetrads of mer3Δ (Table II), 9.7 and 12.3 crossovers among 68 disomes are expected to occur in the MAT‐CENIII and CENIII‐HIS4 intervals, respectively. However, crossovers in the MAT‐CENIII or CENIII‐HIS4 interval followed by homologous non‐disjunction will generate MATa/MATa and MATα/MATα sets or His/His and His+/His+ sets of spores in half of all meioses, owing to random segregation of meiosis II. Thus, the expected number of recombinants in 68 disomes is (9.7 + 12.3)/2 = 11.

Northern blotting and RT‐PCR analysis

RNA of yeast cells was prepared by glass bead and phenol extraction (Treco, 1989a). For Northern blotting, total RNAs were separated on a 0.7% agarose gel in MOPS/formaldehyde buffer (Sambrook et al., 1989), soaked in a 0.05 M NaOH buffer for 20 min for partial digestion of RNA and transferred to NYTRAN nylon membranes (Schleicher & Schuell) in a 10× SSC buffer. For the detection of MER3 and ACT1 RNAs, a 1.0 kb BstBI‐ClaI fragment from pTN45 and a 0.6 kb ClaI fragment from pTN7 (Nakagawa and Ogawa, 1997), respectively, were 32P‐labeled by the random primer method (Sambrook et al., 1989) and used as hybridization probes.

A 2.5 μg aliquot of total RNA was treated with RNase‐free DNase I FPLCpure™ (Pharmacia) to eliminate contaminating DNA and subjected to reverse transcription with 16 U of M‐MuLV reverse transcriptase (New England Biolabs) using 3 pmol of priTN2 (5′‐CGCCTCTTCATCAGGTGTCTGCTCTAAATCG‐3′; position 437‐467). PCR (Saiki et al., 1988) was performed using 20 pmol each of priTN1 (5′‐GGTGGATTTGACAACTTAAGAGGCGTCG‐3′; position −33 to −6) and priTN2 under the following conditions: 1 min at 94°C and then 30 s at 94°C, 10 s at 54°C and 30 s at 74°C for 35 cycles. A total of 2.5 U of KOD dash DNA polymerase (Toyobo) was used for each PCR.

Physical detection of meiotic recombination events

DNA was prepared as described by Treco (1989b). Detection of restriction fragments of interest was performed as described earlier (Storlazzi et al., 1995). Digested DNA samples were separated by electrophoresis on a 0.7% agarose gel and transferred to NYTRAN nylon membrane (Schleicher & Schuell). A 1.5 kb PstI‐EcoRI fragment from pNKY291 (Cao et al., 1990) or a 1.6 kb PstI‐SacI fragment from pNKY155 (Cao et al., 1990) labeled with 32P by the random primer method were used as probes for Southern hybridization. Southern and Northern blot signals were quantified with a Fuji BAS2000 phosphoimager.


We are grateful to A.Shinohara, N.Kleckner, N.Hunter, M.Lichten, R.D.Kolodner and C.Chen for comments on the manuscript, and to members of the Ogawa laboratory for helpful discussion. We also thank N.Kleckner, G.S.Roeder and C.Yanofsky for providing strains and plasmids. This work was supported by a Grants‐in‐Aid for Specially Promoted Research from the Ministry of Education, Science, Sports and Culture of Japan, the Howard Hughes Medical Institute, CREST of Japan Science and Technology and by the Human Frontier Science Program. T.N. was supported by a fellowship of the Japan Society for the Promotion of Science.