Coordination of Meiotic Recombination, Pairing, and Synapsis by PHS1

See allHide authors and affiliations

Science  02 Jan 2004:
Vol. 303, Issue 5654, pp. 89-92
DOI: 10.1126/science.1091110


Pairing, synapsis, and recombination are prerequisites for accurate chromosome segregation in meiosis. The phs1 gene in maize is required for pairing to occur between homologous chromosomes. In the phs1 mutant, homologous chromosome synapsis is completely replaced by synapsis between nonhomologous partners. The phs1 gene is also required for installation of the meiotic recombination machinery on chromosomes, as the mutant almost completely lacks chromosomal foci of the recombination protein RAD51. Thus, in the phs1 mutant, synapsis is uncoupled from recombination and pairing. The protein encoded by the phs1 gene likely acts in a multistep process to coordinate pairing, recombination, and synapsis.

Meiosis is a specialized type of cell division responsible for accurate segregation of genetic information into gametes. Essential to this process is formation of stable bivalents (pairs of homologous chromosomes), which, in most organisms, involves pairing (alignment of homologous chromosomes), synapsis (zipping up of the paired homologs by the synaptonemal complex proteins), and recombination (exchange of chromosome arms). Although the consequences of specific defects in synapsis and meiotic recombination have been extensively studied (1), the mechanism of homologous chromosome pairing is not understood (2), and little is known about how the three processes are coordinated.

For successful bivalent formation, the chromosome homology recognition mechanism must distinguish between global chromosome homology and local similarities of repetitive sequences and gene families and must allow synapsis only after relevant homology is established. One can predict that mutations in genes required for this mechanism would result in uncoupling of synapsis from pairing and recombination and would lead to chromosomes undergoing synapsis without regard to homology.

Here, we describe the poor homologous synapsis1 (phs1) gene in maize (Zea mays, L.), whose product is required for homologous chromosome pairing and for preventing synapsis between nonhomologous chromosomes. The phs1 mutant was discovered in a screen of a maize population mutagenized with the Mutator transposon system. The mutant plants were male- and female-sterile but exhibited no other obvious phenotypic or developmental abnormalities. Cytological observations of anther squashes from the mutant plants revealed that during meiotic metaphase I, 20 univalents were present in most meiocytes instead of the 10 bivalents typical of wild-type meiocytes (table S1, fig. S1A). Missegregation of chromosomes was observed in anaphase I, which leads to the formation of 37% of tetrads containing microspores with extra micronuclei and 59% of polyads (aggregates containing more than four microspores; fig. S1B), whereas only 4% of tetrads appeared normal. Similar abnormalities were observed in female meiosis.

To study whether the presence of univalents in metaphase I in the phs1 mutant was caused by defects in homologous chromosome pairing and synapsis, we used transmission electron microscopy (TEM) of synaptonemal complex (SC) spreads of meiocyte nuclei (3). In wild-type maize, homologous chromosomes pair and synapse during zygonema in meiotic prophase I. At the beginning of pachynema, chromosomes are completely synapsed and visible in TEM micrographs as aligned lateral elements (Fig. 1A). In phs1 mutant nuclei, very little synapsis was observed in zygonema (encompassing only an estimated 0 to 5% of chromosome length). Although more synapsis was detected at the beginning of pachynema, many of the observed synaptic structures showed improper alignment, i.e., a telomere of one chromosome was synapsed to an interstitial region of another chromosome instead of its telomere (Fig. 1B). Exchanges of synaptic partners, manifested as chromosomes synapsing with one partner in one region along its length and with a different partner in another region, were also apparent. In late pachynema, most chromosomes in phs1 mutant nuclei were completely synapsed but misalignments and exchanges of partners were common (Fig. 1C). These abnormalities indicated the presence of nonhomologous synapsis.

Fig. 1.

Homologous pairing and synapsis defects in the phs1 mutant. (A) Homologous synapsis in a wild-type pachytene nucleus. Completely aligned lateral elements are visible. (B) Limited synapsis in an early pachytene phs1 mutant nucleus. (C) Complete synapsis in a late pachytene phs1 mutant nucleus. Examples of synaptic partner misalignments in (B) and (C) are marked with red arrowheads. Examples of synaptic partner switches are surrounded by green squares and shown schematically in insets. Scale bars, 5 μm. (D) Frequency of pairing of 5S rRNA loci in wild-type (n = 76) and phs1 mutant (n = 75) maize meiocytes.

To quantify the extent of homologous and nonhomologous synapsis, we monitored the progression of chromosome pairing using fluorescent in situ hybridization (FISH) with the 5S ribosomal RNA (rRNA) locus probe (3). In wild-type maize meiocytes, two unpaired, 5S rRNA loci, visible as two separate foci were present at the beginning of zygonema. By the beginning of pachynema, only one focus, representing the 5S rRNA loci homologously paired, was detectable in each nucleus (Fig. 1D). In the phs1 mutant, no pairing of 5S rRNA foci was detected in zygonema (Fig. 1D and fig. S2A). At the beginning of pachynema, only about 5% of mutant meiocytes had the 5S rRNA loci homologously paired, whereas in most meiocytes the 5S loci stayed unpaired (Fig. 1D). In late pachynema, although the fraction of phs1 mutant meiocytes with homologously paired 5S loci did not change (Fig. 1D and fig. S2B), 5S rRNA loci in nearly all of the remaining mutant meiocytes formed associations with nonhomologous partners and were visible as two distinct foci per nucleus, one or both of them located on apparent bivalents (Fig. 1D, and fig. S2, C, D, and E).

These data indicate that 95% of synapsis in the phs1 mutant occurs between nonhomologous loci, whereas the extent of chromosome synapsis overall is almost the same as in wild-type plants. Because synapsis is initiated at chromosome ends (1) and the 5S rRNA locus in maize is cytologically close to the telomere, the 5% of homologous synapsis, which we observed between the 5S rRNA loci in the phs1 mutant, may simply reflect the frequency of two chromosome ends synapsing with each other by chance. This would suggest that homologous pairing is completely abolished in the phs1 mutant.

In maize, telomeres cluster in a single region of the nuclear envelope during early meiotic prophase and form a telomere bouquet, a structure thought to facilitate chromosome pairing (4). The phs1 mutation has little effect on bouquet formation. In the mutant, normal telomere bouquets were present in about 50% (9 of 18 analyzed) of the zygotene meiocytes, whereas the remaining meiocytes exhibited only mild disruptions of bouquet formation (fig. S3), which cannot account for the severe pairing defect observed in this mutant.

To determine whether installation of the meiotic recombination machinery is disrupted in the phs1 mutant, we analyzed the nuclear distribution of the recombination protein RAD51 (3), a RecA homolog involved in meiotic recombination (5). During leptonema, RAD51 was diffuse within the nucleus in both wild-type and phs1 mutant meiocytes. In zygonema, RAD51 foci formed on the chromosomes in wild-type meiocytes, reaching a peak of 500 ± 47 per nucleus in mid-zygonema (Fig. 2A). In late zygonema and in pachynema, the number of RAD51 foci decreased in wild-type meiocytes to an average of about 12 ± 3 per nucleus. In contrast, mid-zygotene nuclei of phs1 mutant meiocytes showed dramatically lower numbers of RAD51 foci of about 3 ± 1 per nucleus (Fig. 2A). In pachynema, even fewer foci were found, on average less than one per nucleus (Fig. 2A). However, a Western blot analysis (3) showed that the amount of RAD51 protein in phs1 mutant anthers was roughly the same as in wild-type anthers (Fig. 2B), which indicates that the reduction in the number of foci was not caused by the absence of the RAD51 protein.

Fig. 2.

Installation of the RAD51 recombination protein on meiotic chromosomes. (A) RAD51 immunolocalization in meiocytes. Red, chromatin; green, RAD51. Scale bar, 10 μm. (B) Western blot analysis of the RAD51 protein in wild-type and phs1 mutant anthers at mid-zygonema shows normal levels of RAD51 in the phs1 mutant.

To determine whether the dramatic decrease in the number of RAD51 foci in phs1 nuclei was caused by the lack of double-strand breaks (DSBs), which are sites of RAD51 foci installation (1), we devised a modification of the transferase-mediated dUTP nick-end labeling (TUNEL) assay to increase its sensitivity to detect meiotic DSB formation (3). In wild-type maize, the modified TUNEL assay produced a strong signal on meiotic chromosomes in leptonema through early pachynema but no signal in nuclei in premeiotic interphase or late pachynema (Fig. 3). In phs1 mutant meiocytes, we observed strong TUNEL staining in leptonema, zygonema, and pachynema (Fig. 3). In contrast to wild-type cells, the staining was also present in late pachytene nuclei and only disappeared just before the end of pachynema (Fig. 3). These results indicated that meiotic DSBs are present in the phs1 mutant, although their repair is delayed.

Fig. 3.

Cytological detection of meiotic DSB formation in wild-type and phs1 mutant meiocytes. The modified TUNEL assay shows that DSBs are present during meiotic prophase I in both wild-type (right side) and phs1 mutant (left side) meiocytes. However, DSBs in the phs1 mutant remain until the very end of pachynema, whereas in wild-type cells they disappear in early pachynema. DNA breaks can be artificially generated in wild-type meiocytes in late pachynema by treating the fixed meiocytes with deoxyribonuclease I (+DNaseI, center). In late pachytene phs1 mutant meiocytes, chromosomes do not fill the entire nucleus, as in wild-type meiocytes, but collapse into a small, tight ball. This appearance is most likely caused by chromosomes being attached to several partners (due to synaptic partner switches) while they keep condensing and shortening. Scale bars, 5 μm.

We cloned the phs1 gene using a Mutator1 (Mu1) element (3), whose insertion caused the mutant phenotype (fig. S4, A to C; fig. S5). The phs1 gene is expressed in anthers from premeiotic interphase through the pollen stage, with a peak at early- and mid-zygonema (fig. S4D). In addition, lower levels of expression are detected in some somatic tissues, such as mature leaf and the shoot apical meristem. Somatic expression of phs1 is not surprising; meiotic genes in plants frequently show expression in somatic tissues, even genes that have strictly meiotic functions (68). No phs1 transcript was detected in zygotene anthers of the phs1 mutant, which indicates that phs1-O is a null allele.

The phs1 gene encodes a putative 347–amino acid protein with a predicted molecular mass of 38 kD. BLAST searches did not reveal any significant homology to known proteins or any known functional motifs in the putative PHS1 protein sequence. However, using more sensitive database searching methods (3), we found restrained local similarities with two families of helicases from fungi.

Amino acid sequence comparison of putative phs1 homologs from several monocot and dicot plants reveals a relatively low level of evolutionary conservation of the PHS1 protein. Nevertheless, several short conserved domains can be identified in the PHS1 sequence. Two conserved domains are located in a 50–amino acid–long stretch in the N-terminal portion of the protein and contain about one-half of all conserved amino acid residues found in PHS1 (Fig. 4). Because of their conservation, these two domains are likely to be essential for the PHS1 function. One of the two domains is absent from a splice variant found among phs1 transcripts, which suggests that the protein may have two distinct functions.

Fig. 4.

Alignment of partial protein sequence of phs1 homologs from monocot and dicot plants. Conserved amino acid residues are shaded.

Two features of the phs1 mutant phenotype make this mutant unique among meiotic mutants in all organisms: (i) a nearly complete replacement of homologous synapsis by synapsis between nonhomologous chromosomes, and (ii) an early recombination defect resulting in a dramatic decrease in the number of RAD51 foci to less than 1% of the normal number. The extent of synapsis in the phs1 mutant is at the wild-type level, which shows that the mutation completely uncouples synapsis from pairing and recombination.

The mutant phenotype suggests that the phs1 gene plays a role in the homology search and in the coordination between meiotic pairing, synapsis, and recombination. This proposed role is similar to the proposed function of the budding yeast Hop2 protein (9). However, the hop2 mutant, in contrast to the phs1 mutant, still exhibits a significant level of homologous pairing and shows only about 60% nonhomologous synapsis, which suggests that homology recognition is not completely abolished. The hop2 mutant also exhibits a recombination failure as it does not complete DSB repair (9). However, it shows a normal number of DMC1 foci, which suggests that the recombination pathway defect occurs after the DMC1/RAD51 complexes are loaded on chromosomes. Our results indicate that PHS1 acts earlier during recombination than Hop2, although they both affect pairing in similar ways. This suggests continuous coordination between the progression of pairing and recombination. It also implies that homology recognition is linked to this coordination and is a multistep process. PHS1 and Hop2 most likely represent different steps in the coordination process. hop2 homologs are found in fission yeast and mammals (10), as well as plants (11). Although the short length of the conserved regions in the PHS1 protein makes identification of potential homologs difficult, using profile hidden Markov models (12), we found putative proteins in human and budding yeast showing some similarity to PHS1, which could represent potential homologs.

Although it is possible that the PHS1 protein is primarily a component of the recombination pathway, we believe that this possibility is unlikely. First, neither spo11 knockouts, which abolish meiotic DSB formation, nor knockouts of either rad51 or dmc1, which disrupt DSB repair, show extensive nonhomologous synapsis in plants, yeast, or mouse (6, 9, 1317). Second, the low level of evolutionary conservation of PHS1 argues against its being directly involved in DNA recombination, because recombination proteins are usually highly conserved (18). We also do not believe that the PHS1 protein is an SC component because extensive SC formation has been observed in meiosis in haploid yeast (19) and plants (20), which indicates that homology is not essential for synapsis. Instead, we propose that PHS1 plays a unique function in meiosis that impacts both pairing and recombination.

Chromosome homology recognition mechanisms in complex eukaryotic genomes must operate over long stretches of DNA because repetitive sequences, which are shared between nonhomologous chromosomes, are often longer than several kilobases (21). RecA homologs, such as RAD51 and DMC1, which have been proposed to play roles in homology recognition (15, 2224), may be able to promote pairing of DNA substrates that are a few kilobases long (25). However, this may fall short of the length of many repetitive DNA elements. Additional proteins may, therefore, be required for efficient homology recognition. The PHS1 protein may be directly involved in such a function. Alternatively, given its severe mutant phenotype, it is also plausible that PHS1 is involved in loading the homology searching proteins, including RAD51 and DMC1, onto chromosomes.

Supporting Online Material

Materials and Methods

Table S1

Figs. S1 to S5

References and Notes

View Abstract

Stay Connected to Science

Navigate This Article