VDE-initiated intein homing in Saccharomyces cerevisiae proceeds in a meiotic recombination-like manner


  • Tomoyuki Fukuda,

    1. Department of Integrated Biosciences, Graduate School of Frontier Sciences, University of Tokyo, Bldg. FSB-101, 5-1-5 Kashiwanoha, Kashiwa, Chiba Prefecture, 277–8562 Japan
    Search for more papers by this author
  • Satoru Nogami,

    1. Department of Integrated Biosciences, Graduate School of Frontier Sciences, University of Tokyo, Bldg. FSB-101, 5-1-5 Kashiwanoha, Kashiwa, Chiba Prefecture, 277–8562 Japan
    Search for more papers by this author
  • Yoshikazu Ohya

    Corresponding author
    1. Department of Integrated Biosciences, Graduate School of Frontier Sciences, University of Tokyo, Bldg. FSB-101, 5-1-5 Kashiwanoha, Kashiwa, Chiba Prefecture, 277–8562 Japan
    Search for more papers by this author

  • Communicated by: Masayuki Yamamoto

*E-mail: ohya@k.u-tokyo.ac.jp


Background: Inteins and group I introns found in prokaryotic and eukaryotic organisms occasionally behave as mobile genetic elements. During meiosis of the yeast Saccharomyces cerevisiae, the site-specific endonuclease encoded by VMA1 intein, VDE, triggers a single double-strand break (DSB) at an inteinless allele, leading to VMA1 intein homing. Besides the accumulating information on the in vitro activity of VDE, very little has been known about the molecular mechanism of intein homing in yeast nucleus.

Results: We developed an assay to detect the product of VMA1 intein homing in yeast genome. We analysed mutant phenotypes of RecA homologs, Rad51p and Dmc1p, and their interacting proteins, Rad54p and Tid1p, and found that they all play critical roles in intein inheritance. The absence of DSB end processing proteins, Sae2p and those in the Mre11-Rad50-Xrs2 complex, also causes partial reduction in homing efficiency. As with meiotic recombination, crossover events are frequently observed during intein homing. We also observed that the absence of premeiotic DNA replication caused by hydroxyurea (HU) or clb5Δ clb6Δ mutation reduces VDE-mediated DSBs.

Conclusion: The repairing system working in intein homing shares molecular machinery with meiotic recombination induced by Spo11p. Moreover, like Spo11p-induced DNA cleavage, premeiotic DNA replication is a prerequisite for a VDE-induced DSB. VMA1 intein thus utilizes several host factors involved in meiotic and recombinational processes to spread its genetic information and guarantee its progeny through establishment of a parasitic relationship with the organism.


Inteins and group I introns are remarkable not only as enzymes or ribozymes that catalyse their own splicing, but also as mobile genetic elements. They are parasites on the genome and promote a process termed homing, whereby the intein/intron is efficiently inserted into intein/intron-minus cognates of the intein/intron containing allele (Belfort & Roberts 1997; Chevalier & Stoddard 2001). Homing is initiated with a site specific, double-strand break (DSB) in the intein/intron-minus allele at or near the site of intein/intron insertion caused by an intein/intron-encoded endonuclease (homing endonuclease). Thereafter, intein/intron sequences are copied from the intein/intron-containing allele during DSB repair, resulting in inheritance of the intein/intron. Homing endonucleases have been found in certain bacteriophages and organisms belonging to all of the three biological kingdoms, archaea, bacteria, and eukarya (Belfort & Roberts 1997; Chevalier & Stoddard 2001). In eukaryotic cells, homing occurs both in the mitochondrial and chloroplast genomes, as well as in the nuclear genome (Belfort & Roberts 1997; Chevalier & Stoddard 2001). The homing mechanisms in the nuclear genomes of eukaryotic cells are entirely unknown. On the other hand, those in prokaryote have been well studied mostly in group I intron homing of bacteriophage, providing a viral-prokaryotic system for studying factors influencing intron mobility (Mueller et al. 1996; Huang et al. 1999).

VDE (also called PI-SceI), a homing endonuclease of Saccharomyces cerevisiae is encoded by the intein within VMA1, the yeast nuclear gene whose product is a vacuolar H+-ATPase subunit. It is automatically generated by a post-transcriptional process called protein splicing (Hirata et al. 1990; Kane et al. 1990). VDE introduces a DSB at its recognition sequence (VDE recognition sequence; VRS) specifically observed in the intein-minus VMA1 allele (VMA1(–)) (Gimble & Thorner 1992). DSB repair results in acquisition of the intein precisely at the cleaved site, enabling the intein to spread in the yeast population. A novel characteristic of VMA1 intein homing is that VDE-induced DSBs occur only in meiosis but not in mitosis (Gimble & Thorner 1992), although the gene is also expressed in mitosis. Other yeast homing endonucleases, e.g. I-SceI and I-SceII, act on their recognition sequences immediately after their expression (Zinn & Butow 1985; Wenzlau et al. 1989).

DSBs are produced either spontaneously during normal cellular processes or after exposure to DNA damaging agents and irrespective of their causes potentially lead to lethal lesions. Homologous recombination has emerged as a major DSB repair pathway in yeast. To date, many steps in the pathway of yeast homologous recombination have been studied at the molecular level and their gene products have been identified (Villeneuve & Hillers 2001; Paques & Haber 1999; Smith & Nicolas 1998). Meiotic recombination is induced by DSBs introduced at specific sites called ‘hot spots’ by a topoisomerase-like protein, Spo11p (Keeney et al. 1997; Fig. 1). In addition to SPO11, several other genes are required for DSB formation, although their roles in this process remain obscure (Smith & Nicolas 1998). After the formation of Spo11p-induced DSBs, the 5′ ends of the DSBs are processed to form 3′ single-stranded DNA (ssDNA) tails (White & Haber 1990; Sun et al. 1991). This DSB end processing step is thought to require the Mre11p-Rad50p-Xrs2p (MRX) complex, because the null alleles of any of the genes encoding these proteins result in reduced processing of HO endonuclease-induced DSBs (Ivanov et al. 1994; Tsubouchi & Ogawa 1998). Meiosis specific DSBs mediated by Spo11p are not formed in mre11Δ, rad50Δ, and xrs2Δ null mutants (Alani et al. 1990; Ivanov et al. 1992; Johzuka & Ogawa 1995). However, several separation-of-function alleles of MRE11 (mre11S) and RAD50 (rad50S) have been isolated that allow formation, but not processing, of Spo11p-introduced DSBs (Alani et al. 1990; Nairz & Klein 1997; Furuse et al. 1998; Tsubouchi & Ogawa 1998; Moreau et al. 1999). In such mutants, Spo11p remains covalently bound to the 5′ ends at meiotic break sites and prevents processing while HO-induced DSBs are processed normally (Keeney & Kleckner 1995; Keeney et al. 1997; Tsubouchi & Ogawa 1998; Moreau et al. 1999). The sae2Δ null mutant also exhibits phenotypes similar to mre11S and rad50S mutants, suggesting an interaction of Sae2p with the MRX complex (McKee & Kleckner 1997; Prinz et al. 1997).

Figure 1.

A schematic representation of Spo11p-induced meiotic recombination and the main proteins involved.

The ssDNA tails made by DSB end processing are potential substrates for Rad51p and Dmc1p (Sugawara et al. 1995; Shinohara et al. 1992, 1997; Bishop et al. 1992). Both Rad51p and Dmc1p are structural and functional yeast homologs of RecA protein, a major bacterial strand exchange protein (Shinohara & Ogawa 1999). They promote strand invasion in which ssDNA tails attack an intact homologous DNA duplex. Although RAD51 is required for mitotic and meiotic recombination, DMC1 is expressed and required only during meiosis. Rad51p physically interacts with other recombination proteins including Rad54p (Shinohara & Ogawa 1999), whose structural and functional relative is Tid1p (Dresser et al. 1997; Klein 1997; Shinohara et al. 1997). Tid1p interacts directly with both Dmc1p and Rad51p, with the Dmc1p–Tid1p interaction being stronger than the Rad51p–Tid1p interaction (Dresser et al. 1997).

In addition to the requirement of Dmc1p and Tid1p, meiotic recombination differs from mitotic recombination in several respects. In meiotic cells, the proportion of recombination events associated with crossovers is high (30–50%), while that in mitotic cells is low (0–20%) (Paques & Haber 1999; Malkova et al. 2000; Villeneuve & Hillers 2001). Moreover, during meiosis, DSB ends are led to choose homologous chromosomes rather than sister chromatids as partner at the strand invasion step (Paques & Haber 1999; Schwacha & Kleckner 1997) and this interhomolog recombination is mainly mediated by Dmc1p and Tid1p (Dresser et al. 1997; Klein 1997; Shinohara et al. 1997; Schwacha & Kleckner 1997; Arbel et al. 1999). This is in contrast to the situation in mitotic cells, where the sister chromatid is the preferred partner.

Available information on VMA1 intein homing has been obtained mostly from in vitro studies on the activity of VDE and its structural characteristics (Christ et al. 2000; Hu et al. 2000; Duan et al. 1997; Mizutani et al. 2002). Additionally, a recent in vivo investigation showed the requirement of trans and cis elements, endonuclease activity of VDE and the existence of VRS, respectively, for intein inheritance (Nogami et al. 2002). Besides these trans and cis elements, host factors involved in the DSB repair process must be needed for intein inheritance. Therefore, host factors required for intein homing must be identified for the complete elucidation of homing reaction in eukaryotic cells. In this study we focused on genes involved in the yeast DNA repair system and investigated the functions of these genes in intein homing. Furthermore, in order to know the involvement of host genes in DSB formation, we examined whether VDE-induced DSBs depend on premeiotic DNA replication.


Kinetics of VDE-induced intein homing in wild-type cells

To investigate the kinetics of VMA1 intein homing, we constructed diploid strains with which we can distinguish VMA1(+) homing products from VMA1(+) donors by restriction site polymorphisms; the BamHI site 1.7 kb upstream of the intein insertion site was replaced by the SalI site in VMA1(+) donors (Fig. 2A). Thus, we can detect the VMA1(+) donor, the VMA1(+) homing product, the VMA1(–) recipient, and the DSB fragment as 10.6 kb, 4.8 kb, 3.4 kb, and 1.7 kb bands, respectively, after digestion of the genomic DNA by BamHI (Fig. 2 A,B).

Figure 2.

Kinetics of VDE-initiated intein homing. (A) A schematic representation of the VMA1 loci indicating the locations of BamHI (designated by Bs) and SalI (designated by Ss) sites, the hybridization probe, and BamHI-digested fragments. VMA1 intein and VDE recognition sequence (VRS) are represented by open rectangles and triangles, respectively. YOC2986 was used for the analysis on homing kinetics as the wild-type strain. VDE produces a 1.7 kb DSB fragment from the 3.4 kb recipient BamHI fragment, and after repair generates a 4.8 kb homing product fragment. YOC3118 was used for the determination of crossover frequencies as the wild-type strain, in which LEU2 and URA3 genes were inserted 1.7 kb upstream and downstream of the intein insertion site, respectively (B) Physical analysis of VMA1 intein homing. DNA was isolated from cells at the indicated times after incubation in SPM sporulation medium and subjected to Southern analysis. (C) Effects of VDE-induced DSBs on meiotic progression. Meiotic nuclear divisions in the homing-positive strain, YOC2986 (circles) and the homing-negative strain YOC2974 (squares) were monitored by fluorescence microscopy of DAPI-stained cells. In YOC2986, VDE-induced DSBs occur and are repaired, resulting in intein homing. YOC2974 carries four nucleotide substitution in the VRS of VMA1(–), resulting in the absence of VDE-induced DNA cleavage and homing. Plotted are the percentages of cells that had undergone first division (MI, left panel) and second division (MII, right panel) at the times indicated. The data are the averages of more than three independent experiments.

To examine the kinetics of VMA1 intein homing, we induced synchronous sporulation and performed Southern analysis. As shown in Fig. 2B, DSBs appeared as early as 2–3 h after incubation in SPM sporulation medium and began to disappear after 5–7 h. At 5 h, the intein homing product appeared and began to accumulate. FACS and DAPI staining analyses revealed that premeiotic DNA replication was almost complete within 3–4 h and cells entered meiosis I (MI) at 5 h (Fig. 2C and data not shown). These results suggested that intein homing takes place between premeiotic DNA replication and MI, coinciding with the time when Spo11p-mediated DSBs and following recombination events occur at hot spots (Kleckner 1996). Although DSBs are potentially lethal lesions and occasionally disturb cell cycle progression, DAPI staining experiments showed that there was no obvious difference in meiotic progression between cells of the homing-positive and negative strains (Fig. 2C), suggesting that the DSBs produced by VDE are efficiently repaired with no deleterious effects.

Inefficient homing in mutants defective in DSB end processing

To explore the yeast host factors involved in intein homing, we focused on the DSB repair system and meiotic recombination. First, we examine the involvement of Spo11p in VMA1 intein homing by synchronous sporulation and Southern analysis. In the spo11Δ mutants, VDE-induced DSBs and intein homing efficiently occurred (Fig. 3), indicating that homing can occur independently of meiotic recombination at hot spots.

Figure 3.

Kinetics of VMA1 intein homing in mutants deficient in DSB end processing. (A) DNA was isolated from cells at the indicated times after incubation in SPM sporulation medium and subjected to Southern analysis. Southern blotting was performed as in Figure 2. (B) The intensities of the bands corresponding to the homing products were quantified and expressed as values relative to those corresponding to the donor.

Next, to examine the involvement of DSB end processing proteins in VMA1 intein inheritance, we studied the kinetics of intein homing in the mre11Δ, rad50Δ, and xrs2Δ mutants by Southern analysis. DSBs appeared in these mutants at 3 h and their homing products began to accumulate at 5 h exactly as in the wild-type strain, but the amounts of their homing products were reduced (Fig. 3). The bands of unrepaired DSB signals were stronger than those in the wild-type strain and were detectable to the end of the experiment (Fig. 3A). These results suggested that the absence of the MRX complex reduces processing efficiency of VDE-induced DSB ends and it may prevent effective intein homing. It is known that Spo11p-mediated DSBs do not form at hot spots in the mre11Δ, rad50Δ, and xrs2Δ mutants. Together with the result of the spo11Δ mutant, it was suggested that meiotic-DSB forming machinery is not required for the VDE-mediated DSB formation.

To examine whether Sae2p is involved in the repair pathway of VDE-induced DSBs as well as that of Spo11p-induced DSBs, we assessed the kinetics of homing in the sae2Δ mutant. The DSB bands observed in the sae2Δ mutant were not so strong as those of the mre11Δ, rad50Δ, and xrs2Δ mutants, implying less involvement of Sae2p in the end processing of VDE-induced DSBs as compared with the MRX complex (Fig. 3). However, the sae2Δ mutant showed a delay in disappearance of DSBs and a partial reduction in the generation of the homing product compared with the wild-type strain (Figs 2B and 3). As these data suggested that Sae2p as well as the MRX complex is required for efficient VMA1 intein homing, we used the mre11Δ sae2Δ double mutant to find the relationship between the functions of Sae2p and the MRX complex in homing. The mre11Δ sae2Δ double mutant did not show any detectable differences from the mre11Δ mutant by Southern analysis (Fig. 3), indicating that MRE11 is epistatic to SAE2 in VMA1 intein homing. These results implied that Sae2p functions in the DSB repair pathway involving the MRX complex (see Discussion).

Strand invasion proteins are indispensable for intein homing

Next, we induced mutant cells lacking the gene of a yeast RecA homolog, Rad51p, and those with a deletion in the gene for a meiosis-specific RecA homolog, Dmc1p, to sporulate synchronously and examined them by Southern analysis. As shown in Fig. 4, both mutants displayed a severe reduction in the homing products and persistence of unrepaired DSBs. The dmc1Δ strain exhibited a severer defect in intein inheritance than rad51Δ; in the dmc1Δ mutant little increase in the homing product was detected, while in the rad51Δ mutant some homing product appeared at 9 h and the amount continued to increase (Fig. 4). A small amount of the homing product at 0 h was sometimes observed in the dmc1Δ mutant and also in the wild-type cells but never in the rad51Δ mutant (Figs 4A and 2B, and data not shown), suggesting that there is minor pathway in which VDE introduces DSBs during mitosis and that Dmc1p is dispensable for this mitotic homing. On the other hand, the homing product was totally absent for the entire period of observation in the rad51Δ dmc1Δ double mutant strain (Fig. 4A). These results suggested that during meiosis, Dmc1p-mediated strand invasion plays a major role in the repair of VDE-induced DSBs and that efficient intein inheritance requires both of the RecA homologs.

Figure 4.

Kinetics of VMA1 intein homing in the mutants deficient in strand invasion. (A) DNA was isolated from cells at the indicated times after incubation in SPM sporulation medium and subjected to Southern analysis. Southern analysis was carried out as Figure 2. (B) The intensity of the band corresponding to each homing product was quantified and expressed as a value relative to the intensity of the band corresponding to the donor.

To examine the function of RecA homolog-interacting proteins in intein homing, we performed Southern blotting of the rad54Δ and tid1Δ mutants. In the rad54Δ mutant, the appearance of the homing product was delayed (Fig. 4). The tid1Δ mutant exhibited a further delay in the accumulation of the homing product, thereby showing the severer phenotype than the rad54Δ mutant. Like the dmc1Δ mutant, the tid1Δ mutant sometimes showed a small amount of the homing product at 0 h, while the rad54Δ tid1Δ double mutant hardly formed any homing product till the end of observation (Fig. 4 and data not shown). Thus, the effects on homing kinetics by the rad54Δ and tid1Δ mutations were similar to those of the rad51Δ and dmc1Δ mutations, respectively. These phenotypic similarities are consistent with the specific interactions between Rad51p and Rad54p and between Dmc1p and Tid1p (Dresser et al. 1997).

REC8, the gene encoding a meiosis-specific member of the Rad21 cohesin family, is needed for meiotic intersister cohesion to connect homologous chromosomes and to ensure their proper segregation (Klein et al. 1999). During meiosis, a rec8Δ mutant accumulates recombination intermediates that are also seen in strand invasion deficient mutants and reduces interhomolog recombinants, suggesting that cohesion ensures DSBs to find their target sequences for strand invasion (Klein et al. 1999; Sjogren & Nasmyth 2001). To investigate the function of meiotic cohesion in VMA1 intein homing, we examined the kinetics of homing in the rec8Δ mutant. The rec8Δ mutant showed a delay in the accumulation of the homing product and a persisting level of DSBs (Fig. 4). These phenotypes are similar to those seen in the strand invasion deficient mutants, suggesting the possibility that meiotic cohesion mediated by Rec8p facilitates efficient homing at strand invasion.

Intein homing during RTG

Beside Southern analysis, VMA1 intein homing can be assessed by the return-to-growth (RTG) assay (Nogami et al. 2002). In this assay, cells withdrawn from synchronous sporulation cultures at 5 h and 11 h were returned to mitotic growth on YPD plates and the intein inheritance was tested on the resultant colonies with PCR (Fig. 5A). In the mre11Δ, rad50Δ, and xrs2Δ mutants, partial reduction in homing frequency was detected by the RTG assay (Fig. 5B), supporting the results of Southern analysis. This assay also showed the sae2Δ mutant to be defective in homing and SAE2 to be involved in the same epistasis group as MRE11 (Fig. 5B). In the rad51Δ mutant, homing colonies hardly appeared, indicating that Rad51p is necessary for homing in the RTG assay (Fig. 5B). The rad54Δ mutant also exhibited a severe defect in homing during RTG (Fig. 5B). On the contrary, little or no obvious defects in intein inheritance were seen in the dmc1Δ and the tid1Δ mutants although they exhibited severe phenotypes in Southern analysis (Figs 4 and 5B). Southern blotting revealed that the high homing frequencies detected in the dmc1Δ and the tid1Δ mutants are not due to a loss of the unrepaired chromosome, but are due to their homing ability (Fig. 5C and data not shown). No homing colonies were detected in the rad51Δ dmc1Δ and rad54Δ tid1Δ mutants, suggesting that homing detected in the dmc1Δ and tid1Δ mutants was mediated by Rad51p and Rad54p, respectively (Fig. 5B). Thus we concluded that VMA1 intein homing during RTG is mediated mainly by the mitotic repair pathway, in which the Rad51p-Rad54p complex plays a central role in the repair of VDE-induced DSBs. This conclusion is supported by the report that Spo11p-induced DSBs are repaired mainly by the mitotic, rather than meiotic, strand invasion proteins during RTG (Zenvirth et al. 1997).

Figure 5.

Commitment to VMA1 intein homing. (A) A schematic diagram of PCR analysis for detection of intein homing. Arrows indicate the positions and the directions of PCR primers. If homing does not occur, two DNA fragments corresponding to the donor and the recipient are amplified. If homing takes place, on the other hand, only the longer fragment corresponding to the donor and the homing product is amplified. (B) The frequencies of cells in which homing had occurred during RTG. Cells were sampled from cultures that had been synchronously sporulating in SPM medium for 5 h and 11 h and spread on plates of YPD vegetative growth medium. After 2 or 3 days, genomic DNA was purified from more than 48 colonies formed and subjected to PCR analysis. Bars indicate the percentages of colonies in which intein homing had occurred. Values over the bars indicate surviving fractions (%) of each strain, which were obtained from colony counts produced by samples that were plated at each time point, relatives to the number of colonies plated at 0 h. Open bars, 5 h in SPM; filled bars, 11 h in SPM. (C) Kinetics of VMA1 intein homing during RTG. Cells were incubated in SPM for 5 h and transferred to YPD medium. DNA was isolated from cells at the indicated times and subjected to Southern analysis as in Figure 2.

VMA1 intein homing is accompanied by crossover

In yeast, the fraction of crossovers accompanied by homologous recombination is small at mitosis. (Paques & Haber 1999; Malkova et al. 2000; Villeneuve & Hillers 2001). In contrast, Spo11p-induced meiotic recombinations associate more highly with crossovers than observed in mitotic cells (Paques & Haber 1999; Malkova et al. 2000; Villeneuve & Hillers 2001). To assess the crossover rate of VMA1 intein homing, we constructed a tester strain harbouring the VMA1(+) allele marked with the LEU2 and URA3 genes 1.7 kb upstream and downstream of the VMA1 intein insertion site, respectively (LEU2:: VMA1 (+)::URA3). We carried out a tetrad analysis on asci from the LEU2::VMA1(+)::URA3/VMA1(–) heterozygous diploid strain and determined crossover rates by analysing the segregation of the flanking markers. In 4VMA1(+): 0VMA1(–) segregation tetrads, the proportion of homing associated with crossovers was 35.8% (53/148) (Table 1), suggesting that meiotic frequency of crossovers accompanies intein homing. Crossovers between LEU2 and URA3 were dependent on the intein homing event because no crossovers were observed either in the 2VMA1 (+):2VMA1(–) segregation tetrads or in cells containing an endonuclease-deficient form of VDE (Table 1).

Table 1.  Frequency of crossovers associated with VMA1 intein homing. For the determination of crossover frequencies, the LEU2::VMA1(+)::URA3/VMA1(–) heterozygous diploid strain was used, in which LEU2 and URA3 genes were inserted 1.7 kb upstream and downstream of the intein insertion site, respectively (Fig. 2A). Parental ditype (PD), tetra type (TT), and non-parental ditype (NPD) asci were identified with respect to the LEU2 and URA3 markers flanking VMA1(+). VMA1-104 is VMA1(+) but has a mutation in the VDE coding region that results in a loss of homing ability. Only four-spore-viable tetrads were used for analysis. The frequencies of crossovers were examined by map distances between LEU2 and URA3 expressed in terms of centimorgans (cM), which were calculated from the following formula: cM = 100/2(TT + 6NPD)/(PD +TT +NPD)
StrainVMA1(+):VMA1(–)Number of tetradsPDTTNPDTetrads with coconversion (s)cM
 4 : 0882643 51449
3 : 1 6 3 2 0 1ND
2 : 2 8 8 0 0 0ND
LEU2::VMA1-104::URA3/VMA1(–)4 : 0 0
3 : 1 0
2 : 23737 0 0 0 0
exo1Δ4 : 0693321 21329
3 : 1 2 0 2 0 0ND
2 : 21010 0 0 0ND
msh4Δ4 : 0762729 21835
3 : 1 2 0 2 0 0ND
2 : 2 4 4 0 0 0ND
hdf1Δ4 : 0993446 41542
3 : 1 0
2 : 2 0

The frequency of crossovers accompanying Spo11p-induced meiotic recombination has been reported to be reduced in the absence of Msh4p and Exo1p (Kirkpatrick et al. 2000; Tsubouchi & Ogawa 2000; Khazanehdari & Borts 2000). In addition to intrinsic hot spots, artificial introduction of the HO endonuclease recognition sequence results in recombination repair with MSH4-dependent efficient crossovers during meiosis (Malkova et al. 2000). In homing, the crossover rate observed in the exo1Δ mutant and in the msh4Δ mutant was reduced (Table 1). These data suggested that VDE-induced intein homing is accompanied by Exo1p- and Msh4p-dependent crossovers as in the case of Spo11p-induced meiotic recombinations.

Intein homing is not mediated by an NHEJ protein, Hdf1p

As shown above, the efficiency of VMA1 intein homing is reduced in mutants deficient in strand invasion or end processing, indicating that homing is mediated by the yeast homologous recombination system. To examine the involvement of an alternative DSB repair pathway, the non-homologous end joining (NHEJ) pathway, in intein homing, we prepared a strain with deletion in the HDF1 (also called YKU70) gene whose product is involved in DSB repair along the NHEJ pathway (Milne et al. 1996). As compared with the wild-type strain, the hdf1Δ mutant exhibited a higher homing frequency although the crossover frequency remained unchanged (Table 1). The homing rate in hdf1Δ tetrads was 100% (198/198). This result suggested that intein homing occurs extremely efficiently in the NHEJ deficient mutant and raises the possibility that NHEJ competes with homologous recombination for VDE-induced DSBs.

Coupling between premeiotic DNA replication and homing initiation

Direct coupling between premeiotic DNA replication and recombination-initiating DSBs was recently reported (Borde et al. 2000; Smith et al. 2001). Motivated by the idea that VMA1 intein homing shares the same molecular machinery with meiotic recombination at the DNA cleavage stage, we examined the possibility that VDE-induced DSBs is dependent on premeiotic DNA replication. Wild-type cells were incubated in SPM supplemented with 100 mm hydroxyurea (HU) to block premeiotic DNA replication. FACS analysis confirmed the absence of premeiotic DNA replication (data not shown). Interestingly, cells with premeiotic DNA replication blocked showed a significant reduction both in VDE-induced DSBs and in the homing product (Fig. 6A). Moreover, in the clb5Δ clb6Δ double mutant, in which meiotic replication is prevented with no induction of the MEC1 block (Stuart & Wittenberg 1998 and Fig. 6B) and Spo11-induced DSBs fail to form (Smith et al. 2001), the appearance of VDE-induced DSBs and homing products was extremely delayed and the amounts are reduced (Fig. 6B). These results suggested that DSB production by VDE depends on premeiotic DNA replication. The DNA cleavage by VDE is not regulated by the expression level of VDE since Western analysis revealed that VDE was expressed with no regard to premeiotic DNA replication (Fig. 6C), suggesting the presence of another regulation (see Discussion).

Figure 6.

Blocking of premeiotic DNA replication causes defects in the VDE-induced DSB and VMA1 intein homing. (A) Kinetics of VMA1 intein homing in the absence of the premeiotic DNA replication by HU addition or the clb5Δ clb6Δ double mutation. DNA was isolated from cells at the indicated times after incubation in SPM sporulation medium and subjected to Southern analysis as described in the legend to Figure 2. (B) Meiotic divisions in the wild-type cells with (squares) or without (circles) HU or in the clb5Δ clb6Δ cells (triangles) were monitored by fluorescent microscopy of DAPI-stained cells. Plotted are the percentages of cells that had undergone first division at the indicated times. (C) Western analysis of VDE on cells with or without premeiotic DNA replication. For analysis, 40 µg of cell extracts was prepared at the indicated times after transfer to SPM and detected by the anti-VDE antibody.


In this study, we performed genetic analysis of genes involved in DNA repair and meiotic recombination and, for the first time, presented evidence that gives insights into the requirements of host factors for VMA1 intein homing in eukaryotes. VDE-induced DSBs and the subsequent repair process occur in the same period as that of Spo11p-initiated meiotic recombinations. The intein inheritance is highly dependent on the host homologous recombination system and is accompanied by frequent crossovers. Finally, VDE-induced DSBs depend on premeiotic DNA replication. In the viral-procaryotic system, homing has been demonstrated to require recombination proteins involved in strand invasion and end processing (Mueller et al. 1996; Huang et al. 1999 and references therein). Our findings on the VMA1 intein generalize the requirement of homologous recombination for intein/intron homing, crossing the boundary between the prokaryotes and the eukaryotes. Most of our results are consistent with the idea that VMA1 intein homing resembles Spo11p-induced meiotic recombination, which is known to require recombination repair system with supplementation of meiosis-specific functions that channel and regulate repair to produce an outcome specific for meiosis. In summary, we would like to propose the following model analogous to that of Spo11p-induced meiotic recombination: (1) DSBs are induced by VDE after premeiotic DNA replication (2) DSBs are subjected to end processing mediated by the MRX complex (3) ssDNA tails lead to the intein-containing donor sequence on the homologous chromosome (4) strand invasion is promoted mainly by the Dmc1p-Tid1p complex with the assistance of meiotic cohesion, and (5) finally DSBs are repaired with frequent crossovers.

Some of our findings shed new light on the DNA repair system during intein homing. A first interesting observation is that intein homing occurs even in the dmc1Δ and tid1Δ mutants during RTG. It is noted that the dmc1Δ and tid1Δ mutants show a drastic decrease in Spo11p-induced interhomolog recombinant at hot spots in the RTG assay. This difference may be caused by an unusually unstable condition that the two sister chromatids experience during intein homing. In normal meiotic recombination initiated by Spo11p, two sister chromatids are rarely cleaved at one hot spot. Therefore, Spo11p-induced DSBs can be repaired using either the sister chromatid or the homologous chromosome as a template, and most of them are repaired by Rad51p-Rad54p mediated intersister recombination in the absence of Dmc1p or Tid1p, resulting in a reduced incidence of interhomolog recombinants during RTG (Dresser et al. 1997; Klein 1997; Shinohara et al. 1997; Schwacha & Kleckner 1997; Zenvirth et al. 1997; Arbel et al. 1999). In the case of VMA1 intein homing, however, two sister chromatids are mostly cleaved by VDE. The two sister chromatids then become unstable, allowing the DSBs to be repaired only by interhomolog recombination. Therefore, despite natural preference for intersister recombination, the Rad51p-Rad54p complex mediates intein homing by interhomolog recombinations during RTG in the absence of the Dmc1p-Tid1p complex.

A second stimulating observation is that the sae2Δ mutant shows inefficient homing. The effect of the sae2Δ mutation seems small compared with that caused by the absence of the MRX complex. The sae2Δ, rad50S, and mre11S mutations are thought to affect the processing of Spo11p-induced DSB ends in meiosis, but not in that of HO-induced DSB ends in either meiosis or mitosis (Alani et al. 1990; Nairz & Klein 1997; Furuse et al. 1998; Tsubouchi & Ogawa 1998; Moreau et al. 1999). These results together further support the idea that Sae2p is not deeply involved in the end processing of DSBs produced by endonucleases such as HO and VDE. However, the absence of Sae2p certainly causes a partial reduction in intein inheritance. MRX complex plays multiple roles in chromosome metabolism and DNA repair (Paques & Haber 1999). Since MRE11 is epistatic to SAE2 in VMA1 intein homing, Sae2p may be involved in the process mediated by MRX complex, which is required for efficient repair of VDE-produced DSBs. To explain the homing defect seen in the sae2Δ mutant, the following possibilities can be raised. First, once VRS is cleaved, VDE may remain bound to the cleaved products, protecting against the end processing and strand invasion and Sae2p may be required for the efficient removal of VDE from the DSB products. Second, Sae2p may have additional functions in DSB repair, for example, strand invasion after DSB ends are processed, recruitment of DSB repair proteins to the DSB sites, or establishment of a chromosome structure that connects homologous chromosomes. Third, Spo11p-bounded DSBs, which accumulate in the sae2Δ mutant, may trap the DSB repair machinery, preventing VDE-induced DSBs from gaining full access to repair proteins (Neale et al. 2002; Usui et al. 1998).

One surprising outcome of our study is the finding that there is a coupling between premeiotic DNA synthesis and VDE-induced DSBs. The blockage of premeiotic DNA replication by HU addition or the clb5Δ clb6Δ double mutation caused severely reduced or delayed cleavage by VDE despite the expression of VDE (Fig. 6A,C). VDE is a novel homing endonuclease in that it produces DSBs exclusively during meiosis (Gimble & Thorner 1992), while other homing endonucleases reportedly introduce DSBs to their recognition sequences immediately after their expression (Zinn & Butow 1985; Wenzlau et al. 1989). Therefore, our finding provides a new example that host factors affect homing at the DNA cleavage stage. Moreover, it is interesting that VMA1 intein homing is similar to meiotic recombination at the formation stage as well as at the repair stage of DSBs. The molecular mechanism of direct coupling between Spo11p-induced DSBs and premeiotic DNA replication is not completely understood in spite of accumulating information. The formation of DSBs at hot spots requires an appropriate chromatin structure; hot spots are located in regions of accessible chromatin that are hypersensitive to DNaseI and micrococcal nuclease (MNase) in mitotic and meiotic cells and the hypersensitivity to nuclease increases during meiosis, before the appearance of DSBs (Ohta et al. 1998; Smith et al. 2001; and references therein). This change in chromatin structure during meiosis is also coupled with premeiotic DNA replication (Smith et al. 2001). The chromatin structure at or near the VRS possibly regulates meiosis-specific cleavage by VDE. Since some homing products were sometimes detected at 0 h probably resulting from mitotic homing and the clb5Δ clb6Δ double mutant displayed VDE-induced DSBs albeit with a delay, the coupling between premeiotic DNA synthesis and VDE-induced DSBs is not so tight as in the case of Spo11p-mediated DSBs at hot spots. One explanation for the regulation is that VDE-induced DSBs are dependent on the chromosome structure that changes at around premeiotic DNA replication. In the clb5Δ clb6Δ mutant, nuclear division does occur despite the absence of premeiotic DNA replication (Smith et al. 2001), and then nuclear division may also cause a certain change in chromosome structure, allowing the appearance of some VDE-induced DSBs. Another explanation is that other host protein(s) or modification(s) of VDE regulate its endonuclease activity or that other host protein(s) protect the VRS. The activation or accessibility of VDE is subjected to premeiotic DNA replication-dependent regulation, which is blocked by the MEC1-induced checkpoint when replication is inhibited. In any case, coupling between premeiotic DNA synthesis and VDE-induced DSBs appears to ensure that VDE-induced DSBs occur coincidentally with Spo11p-induced DSBs and are similarly repaired.

The results presented here suggest that VDE-induced intein homing resembles Spo11p-induced meiotic recombinations at hot spots. The utilization of the meiotic recombination machinery for intein homing appears to be favourable for VMA1 intein as a genetic parasite, since the expression of genes involved in recombinational repair is highly induced during meiosis and DSBs are preferentially repaired by interhomolog recombinations mediated by Dmc1p and Tid1p. In addition to these two proteins, there are some parallel regulations that connect homologous chromosomes (Peoples et al. 2002 and references therein). Therefore, the meiosis specificity of VDE-induced DSBs may guarantee an efficient repair of the DSBs by interhomolog recombination, resulting in efficient intein inheritance. The whole process is favourable to the host because DSBs, if left repaired, threaten its genomic stability and viability. We theorize that this relationship between the parasite and the host contributes to the efficient propagation and preservation of VMA1 intein in the yeast genome. It is interesting to note that homing endonucleases of eukaryotes are mostly found in the mitochondrial and chloroplast genomes but rarely in the nuclear genomes. The yeast genome contains a homing endonuclease, VDE, and another kind of homing endonuclease, HO, which shares homology with VDE and introduces a DSB to a 24-bp sequence in the mating type (MAT) locus (Hirata et al. 1990; Gimble & Thorner 1992). HO activity is known to be subjected to tight transcriptional and protein degradational regulations (Cosma et al. 1999; Kaplun et al. 2000). Utilization of many host factors may be necessary for homing endonucleases to survive in the nuclear genome of eukaryotes.

Experimental procedures


All the plasmids used in this study were constructed by standard procedures (Sambrook et al. 1989). Integration of the 1.1 kb fragment upstream of VMA1 (amplified by PCR with the primers, 5′-ACCTTAAGAAGTTTGCTCTAGTTACAGCCC and 5′-TATCTAGGTACCTGACAGGAATAGG using yeast genomic DNA as template) at the KpnI-XbaI site of pRS306 integrating vector (Sikorski & Hieter 1989) yielded pYO2494. A SalI linker was inserted into the blunted BamHI site of pYO2494 to produce pYO2495. The 1.7 kb DNA fragment downstream of VMA1 was amplified by PCR with the primers, 5′-AATTGTTGAGCTCTATGCAAGAAAG and 5′-TGGATCGTTGAGAAATCAGC using yeast genomic DNA as template, and cloned into the SacI-SmaI site of pRS306 to result in pYO2496. A BamHI linker was inserted into the blunted XbaI site of pYO2496 to generate pYO2497. pYO2498 and pYO2500 were constructed by inserting the SacI-KpnI fragment from pYO2495 and pYO2497, respectively, to pBluescript SK (Stratagene) at the SalI-KpnI site. The SalI fragment of pJJ283 (Jones & Prakash 1990) containing the LEU2 gene was ligated to the SalI site of pYO2498 to make pYO2499. The URA3-containing PvuII fragment of pJJ242 (Jones & Prakash 1990) was ligated with a BamHI linker, digested with BamHI, and inserted to the BamHI site of pYO2500 to give rise to pYO2501.

Media, growth conditions, and DNA manipulations

Yeast cells were grown vegetatively in YPD (1% yeast extract, 2% peptone, 2% glucose), YPG (1% yeast extract, 2% peptone, 3% glycerol), or SD (0.67% yeast nitrogen base without amino acids and 2% glucose), supplemented with necessary amino acids (Kaiser et al. 1994). Agar (2%) was added to the above media to produce solid media. Yeast growth, tetrad analysis, mating-type determinations, and yeast transformation were performed as previously described (Kaiser et al. 1994). Standard procedures were followed for all DNA manipulations and Escherichia coli transformations (Sambrook et al. 1989).


The Escherichia coli strain SCS1 was used as the plasmid host. The yeast strains used in this study are listed in Table 2. All the strains were derivatives of the SK1 strain, which enters meiosis in a highly synchronous manner and are often used in studies of meiotic recombination in yeast (Kleckner 1996). VMA1-101, VMA1-103, and VMA1-201 alleles have complete deletion of VMA1 intein (VMA1(–)). In addition, the VRS in VMA1-103 carry four mutations (G844T, T845C, C846T, and G849C) in the VMA1 gene. The VMA1, VMA1-104 and VMA1-202 alleles contain VMA1 intein (VMA1(+)). VMA1-104 has a mutation (A1826T) in the VDE coding region that results in a loss of homing ability (Nogami et al. 2002). For experimental convenience, the XbaI site located 728-bp downstream of the VMA1 stop codon is replaced with the BamHI site in the VMA1-201 and VMA1-202 alleles. To distinguish VMA1(+) homing products from VMA1(+) donors, the BamHI site located 866-bp upstream of VMA1 start codon was replaced with the SalI site in the VMA1-202 allele. Replacement of the XbaI and BamHI restriction sites was achieved by two-step procedures using SplI-digested pYO2497 and BglII-digested pYO2495, respectively. To measure the frequency of crossovers accompanying homing, we constructed the LEU2::VMA1::URA3 allele by one-step procedure with pYO2498 and pYO2500.

Table 2.  Yeast strains used in this study
YOC2813a ho::hisG leu2 ura3 lys2 TRP1 VMA1-101Nogami et al. (2002)
YOC2814αho::LYS2 leu2 ura3 lys2 trp1 VMA1Nogami et al. (2002)
YOC2968a ho::hisG leu2 ura3 lys2 TRP1 VMA1This study
YOC2969αho::LYS2 leu2 ura3 lys2 trp1 VMA1-101This study
YOC2970a ho::hisG leu2 ura3 lys2 TRP1 VMA1-103Nogami et al. (2002)
YOC2971αho::LYS2 leu2 ura3 lys2 trp1 VMA1-104Nogami et al. (2002)
YOC2974a/αho::LYS2/ho::hisG leu2/″ ura3/″ lys2/″ TRP1/trp1
Nogami et al. (2002)
YOC2986a/αho::LYS2/ho::hisG leu2/″ ura3/″ lys2/″ TRP1/trp1
This study
YOC3118a/αho::LYS2/ho::hisG leu2/″ ura3/″ lys2/″ TRP1/trp1
This study
YOC2988a/αho::LYS2/ho::hisG leu2/″ ura3/″ lys2/″ TRP1/trp1
This study
YOC2989a/α YOC2986 but rad51Δ::hisG::URA3::hisGThis study
YOC2990a/α YOC2986 but rad54Δ::hisG::URA3::hisGThis study
YOC2991a/α YOC2986 but rad50Δ::hisG::URA3::hisGThis study
YOC2992a/α YOC2986 but mre11Δ::hisG::URA3::hisGThis study
YOC2993a/α YOC2986 but xrs2Δ::LEU2This study
YOC2994a/α YOC2986 but dmc1Δ::cgLEU2This study
YOC2995a/α YOC2986 but tid1Δ::cgLEU2This study
YOC2996a/α YOC2986 but sae2Δ::cgLEU2This study
YOC2997a/α YOC2986 but sae2Δ::cgLEU2 mre11Δ::hisG::URA3::hisGThis study
YOC2998a/α YOC2986 but clb5Δ::KanMX4 clb6Δ::CgURA3This study
YOC3001a/α YOC2986 but dmc1Δ::cgLEU2 rad51Δ::hisG::URA3::hisGThis study
YOC3119a/α YOC2986 but rec8Δ::cgLEU2This study
YOC3120a/α YOC3118 but hdf1Δ::KanMX4This study
YOC3121a/α YOC3118 but exo1Δ::KanMX4This study
YOC3122a/α YOC3118 but msh4Δ::KanMX4This study

Deletion mutations of DMC1, TID1, SAE2, REC8, and CLB6 were constructed by the PCR-mediated gene disruption method (Sakumoto et al. 1999). The CgLEU2 and CgURA3 genes were amplified with a tag with upstream and downstream sequences of each gene. The disruption of EXO1, HDF1, and MSH4 was carried out by one-step procedure with a PCR fragment of the KanMX4 gene flanked by upstream and downstream sequences of the corresponding gene. To disrupt RAD51, RAD54, RAD50, MRE11, and XRS2, we transformed cells with pΔRAD51, p54-HUH-B, pNKY83, pKJ1112-S, and pEI140, respectively, after linearization with appropriate restriction enzymes (Shinohara et al. 1992; Ogawa unpublished; Alani et al. 1989; Johzuka & Ogawa 1995; Ivanov et al. 1994). All constructions were confirmed either by PCR or by digestion of PCR fragments with appropriate restriction enzymes.

Sporulation and return-to-growth assay

Synchronous sporulation and return-to-growth experiments were performed essentially as described before (Nogami et al. 2002; Ohta et al. 1998). Cells were grown at 30 °C in SPS presporulation medium (0.5% yeast extract, 1% peptone, 0.17% yeast nitrogen base without ammonium sulphate and amino acids, 0.05 m potassium phthalate, 1% potassium acetate and 0.5% ammonium sulphate, pH 5.0) to a concentration of 1–4 × 107 cells/ml. Cells were then pelleted, washed in water, and resuspended at the same density in SPM sporulation medium (1% potassium acetate and required amino acids at one-fifth the standard concentrations) and were cultured with vigorous aeration at 30 °C. For return-to-growth assay, synchronous meiotic cultures sampled at various times were spread on YPD plates and incubated at 30 °C or were pelleted, resuspended in YPD, and cultured at 30 °C. Homing was detected by PCR using the primers, 5′-GGCCTGTTCGTGTTCCAAGACCAGTTACTG and 5′-TTGATCAGCAGGCATCTCACCCAAACGACC.

Physical analysis of intein homing

Genomic DNA was extracted by the zymolyase method (Cao et al. 1990), except that purification of the DNA on Sephadex was replaced by ethanol precipitation and the purified DNA was digested with BamHI. The digested DNA fragments were separated by electrophoresis in a 0.7% agarose gel and blotted on to a nylon membrane (Hybond-N+, Amersham) under vacuum (VacuGene XL, Amersham). Sequences were probed with the 859-bp PCR fragment amplified with primers, 5′-TCTTGATGGAATTCCCAGAG and 5′-ATGCTTAACGTCACCAGTAG using yeast genomic DNA as template. Probes were labelled by random priming (Readyprime Kit, Amersham) according to the manufacturer's instructions. Unincorporated nucleotides were separated by filtration through a ProbeQuant G-50 Micro column (Amersham). Hybridization was performed with Rapid-hyb buffer (Amersham) as recommended by the manufacturer. Signals were quantified with BAS5000 phosphoimager (Fuji-film). The same analysis was carried out more than five times using independent cultures and the average value for each strain is presented in each figure.

Analysis of DNA contents and meiotic divisions

Cells from meiotic cultures were harvested and resuspended in 40% ethanol, 0.1 m sorbitol and stored at −20 °C until FACS and microscopy analysis. Cells were harvested and resuspended in 0.3 ml of 50 mm Tris-HCl at pH 7.5 containing 1 mg/ml RNAse and incubated at 37 °C for 3 h. After addition of 0.1 ml propidium iodide solution (38 mm sodium citrate, 10 mm NaCl, 50 µg/ml propidium iodide), the cells were incubated at room temperature for an additional 30 min 0.9 ml of 0.2 m Tris-HCl and 10 µl of 10 µg/ml propidium iodide solution were subsequently added. The samples were analysed by a FACScan analyser (Becton-Dickinson) after brief sonication. Occurrence of meiotic divisions was monitored by fluorescence microscopy of DAPI-stained cells.

Immnoblot analysis of VDE

Cell extracts were prepared as follows. Cells were resuspended in lysis buffer (50 mm Tris-HCl, pH 7.5, 10% glycerol, 0.1% SDS, 150 mm NaCl, 50 mm NaF, 5 mm EDTA, 1% TritonX-100, 1 mm sodium orthovanadate, 50 mmβ-glycerol phosphate, 5 mm sodium pyrophosphat), to which protease inhibitors (1 mm phenylmethylsulphonylfluoride, and 25 µg/ml each of tosylphenylalanine chloromethyl ketone, tosyllysine chloromethyl ketone, leupeptin, pepstatin A, anti-pain, and aprotinin) were added. After vortexing with glass beads, supernatants from cell suspensions were prepared. Protein samples (40 µg) were loaded on an SDS-polyacrylamide gel, electrophoretically separated, transferred to a PVDF membrane (Amersham), and blotted with an anti-VDE antibody and developed with the ECL detection kit (Amersham) using an anti-rabbit HRP antibody.


We express our appreciation to H. Ikeda, K. Ohta, and N. Kleckner for providing yeast strains and plasmids. We are also grateful to K. Homma for a critical reading of the manuscript. This work was supported in part by a grant in aid for Scientific Research from the Ministry of Education, Culture, Sports, Science and Technology of Japan. This work was also assisted by Research Fellowships of the Japan Society for the Promotion of Science for Young Scientists to T.F.