A Synaptonemal Complex Protein Promotes Homology-Independent Centromere Coupling

See allHide authors and affiliations

Science  06 May 2005:
Vol. 308, Issue 5723, pp. 870-873
DOI: 10.1126/science.1108283


We describe a process in meiotic cells of budding yeast in which chromosomes become joined together in pairs at their centromeres independent of chromosomal homology. These centromeric interactions depend on the synaptonemal complex component Zip1. During meiosis in wild-type diploids, centromere couples are initially nonhomologous and then undergo switching until all couples involve homologs. This transition to homologous coupling depends on Spo11, a protein required for the initiation of meiotic recombination. Regions of synaptonemal complex assembled early in meiosis are often centromere-associated. We propose that centromere coupling facilitates homolog pairing and promotes synapsis initiation.

Segregation of chromosomes at the reductional division of meiosis depends on a series of interactions between homologous chromosomes, including pairing, assembly of the synaptonemal complex (SC), genetic recombination, and formation of chiasmata. Pairing relies on both recombination-independent and recombination-dependent mechanisms. Although recombination is required for full levels of pairing, a substantial amount of homolog alignment occurs in the absence of recombination (1). Recombination-independent pairing is believed to depend, at least in part, on telomere clustering leading to bouquet formation (2, 3).

The Zip1 protein is a component of the SC in budding yeast. Zip1 is a coiled-coil protein that bridges the space between the cores of homologous chromosomes (4). Although formation of the SC serves to stabilize pairing interactions between homologs, the complex is generally not assumed to play a role in pairing per se.

Meiotic recombination initiates with the formation of developmentally programmed double-strand breaks in the DNA (5). In budding yeast, mutants defective in the initiation of meiotic recombination, such as spo11, fail to make mature SCs (1). However, the Zip1 protein does localize to discrete foci on chromosomes in a spo11 background, as detected by immunostaining of surface-spread meiotic nuclei (Fig. 1A) (6, 7). Unlike the polymerization of Zip1 along the lengths of chromosomes (8, 9), the localization of Zip1 to foci in spo11 strains is independent of the Zip2 and Zip3 proteins (10), components of the synapsis initiation complex (11). Formation of Zip1 foci in spo11 also does not require the meiosis-specific chromosomal core protein Red1 (10, 12).

Fig. 1.

Centromeres are coupled in a Zip1-dependent manner in the spo11 mutant. (A) A single meiotic nucleus that has been surface-spread and stained for Zip1 and Ctf19-myc and with the DNA-binding dye 4′,6′-diamidino-2-phenylindole (DAPI). Quantification of Ctf19 foci in (B) spo11, (C) zip1 spo11,(D) and ndj1 spo11 diploids. A representative Ctf19-stained nucleus is shown for each strain. Indicated in each graph is the average (Avg.) number of Ctf19 foci with standard deviation. Numbers of nuclei scored are, for spo11, 49; zip1 spo11, 45; and ndj1 spo11, 46. Arrowheads indicate the positions of polycomplexes. Bars indicate 2 μm.

We found that the chromosomal Zip1 foci present in spo11 strains are often located at or near centromeres. Figure 1A shows a meiotic nucleus stained with antibodies to both Zip1 and an epitope tag fused to Ctf19, a component of the yeast kinetochore (6, 13). Most Zip1 foci (78.1 ± 11.7%) overlap with Ctf19 foci (47 nuclei scored).

Almost all nuclei from spo11 strains contain a polycomplex, which is an aggregate of Zip1 proteins unassociated with DNA (1). Ctf19 and the centromere protein Ndc10 are often found in the polycomplex (Fig. 1 and fig. S1), indicating that the interaction between these proteins and Zip1 does not require an intact centromere.

A diploid yeast cell contains 32 chromosomes representing 16 pairs of homologs. The number of centromere (i.e., Ctf19) foci per meiotic prophase nucleus averages 17 in the spo11 mutant and shows little variation among nuclei (Fig. 1B). Because the number of centromere foci is about half the number of chromosomes, these observations suggest that centromeres become joined together in groups of two. This association will hereafter be referred to as “coupling” or “centromere coupling,” although we cannot distinguish between coupling at the centromere per se versus coupling involving centromere-proximal sequences.

In budding yeast, the Ndj1 protein is required for bouquet formation and efficient homolog pairing (3). To determine whether Ndj1 is important for centromere coupling, we measured the number of Ctf19 foci in diploid nuclei from an ndj1 spo11 double mutant (Fig. 1D). The average number of centromere foci is 15, indicating that centromere coupling is not dependent on bouquet formation.

Does centromere coupling require the Zip1 protein? In a zip1 spo11 double mutant, the number of foci increased to about 32 (Fig. 1C), indicating that centromeres are not coupled in the absence of Zip1.

The observation that there are ∼16 centromere foci per nucleus suggested that the centromeres of homologous chromosomes are paired in a spo11 background. To test this hypothesis, we used the lacO/LacI-green fluorescent protein (GFP) system (14) to visualize centromere-proximal regions of a homolog pair (either chromosome III or XI) (6). The centromeres of homologous chromosomes are unpaired in most nuclei (Fig. 2, A and B). The frequency of pairing is 23.9% for CENIII and 15.3% for CENXI. Thus, the majority of centromere couples in a spo11 strain represent interactions between the centromeres of two nonhomologous chromosomes.

Fig. 2.

Nonhomologous chromosomes couple in the absence of Spo11 and in haploids. (A and B) Nonhomologous centromere coupling in spo11. Diploid spo11 cells with LacO insertions near (A) CENIII or (B) CENXI were spread and stained for LacI-GFP and Ctf19-myc. Arrows point to the locations of homologous chromosomes. Numbers of nuclei scored: CENIII, 88, and CENXI, 59. Quantification of Ctff19 foci in (C) spo11, (D) zip1 spo11, and (E) wild-type (WT) haploids. Representative Ctf19-stained nuclei are shown. Indicated in each graph is the average number of Ctf19 foci with standard deviation. Numbers of nuclei scored: spo11, 56; zip1 spo11, 38; and WT, 50. Bars, 2 μm.

To obtain independent evidence of nonhomologous centromere coupling, we counted centromere foci in a spo11 haploid strain engineered to enter meiosis (6). Such a strain contains 16 chromosomes with no potential for homolog pairing. In a spo11 haploid, the average number of Ctf19 foci per nucleus was eight (Fig. 2C), indicating that nonhomologous chromosomes are coupled. As is the case in diploids, the Zip1 protein colocalizes with centromeres in spo11 haploid cells (87.2 ± 11.0% of Zip1 foci colocalize with Ctf19 foci; 48 nuclei scored), and centromere coupling requires the Zip1 protein (Fig. 2D). We observed about eight centromeric foci in nuclei from a wild-type haploid (Fig. 2E) and 16.3 ± 1.8 foci in a zip1 haploid (37 nuclei scored), demonstrating that Zip1-dependent coupling of nonhomologous centromeres is not unique to the spo11 mutant (10).

To determine whether Zip1-mediated centromere coupling takes place during normal meiosis, we examined centromere coupling throughout meiosis in wild-type diploids. On the basis of the Zip1 staining pattern, nuclei were classified into three categories, I, II, and III, representing progressively later stages in SC assembly (6). Nuclei in category I are similar to spo11 nuclei in terms of the number of Ctf19 foci (Fig. 3D), the extent of Zip1 colocalization with centromeres (Fig. 3A), and the degree of homologous centromere pairing (Fig. 3G). As synapsis progresses, the number of Ctf19 foci remains fairly constant (Fig. 3, D to F), whereas the frequency of homologous centromere coupling increases until full pairing is achieved (Fig. 3, G to I).

Fig. 3.

Chromosome coupling transitions from nonhomologous to homologous in wild-type. (A to C) Wild-type cells with LacO insertions near CENXI were spread and stained for Zip1, Ctf19-myc, and LacI-GFP. The arrows indicate CENXI locations. Examples of nuclei with (A) unpaired and (B and C) paired CENXI are shown. Quantification of Ctf19 foci in nuclei of categories I (D), II (E), and III (F). Representative Ctf19-stained nuclei are shown. Indicated in each graph is the average number of Ctf19 foci with standard deviation. Quantification of CENXI pairing in nuclei of categories I (G), II (H), and III (I). The number of nuclei used to measure Ctf19 foci and CENXI pairing was 69, 67, and 49 for categories I, II, and III, respectively. Bar, 2 μm.

In category II nuclei, linear stretches of Zip1 are often associated with a centromere (Fig. 3B), suggesting that synapsis (i.e., the elongation of Zip1 along chromosomes) initiates at centromeres. To quantitate this association, we focused our attention on short linear stretches of Zip1 staining in nuclei at early stages of synapsis initiation (Fig. 4) (6). Of 129 linear stretches examined, 98 (76%) were associated with a centromere.

Fig. 4.

Early synapsis initiates at centromeres. (A) Examples of nuclei at early stages of synapsis initiation. Wild-type spreads were stained for Zip1 and Ctf19-myc. Arrows and an arrowhead indicate Zip1 stretches used to quantitate centromere associations. Of the seven stretches shown, only one (arrowhead) is not associated with a centromere. Bar, 2 μm. (B) Additional examples of Zip1 stretches used to measure the frequency of synapsis initiation at centromeres. Bar, 0.5 μm.

Our data demonstrate that meiotic chromosomes become joined together in pairs at their centromeres and that this coupling is both dynamic and independent of chromosomal homology. In wild-type cells, the number of centromere couples (i.e., Ctf19 foci) remains fairly constant throughout meiotic prophase, whereas the fraction of homologous couples steadily increases. Thus, centromeres that become uncoupled must immediately form new associations. Perhaps dissociation of an existing couple is triggered only after a new partner is recognized.

What function(s) might Zip1-mediated centromere coupling perform? We propose that coupling facilitates homolog pairing by holding two chromosomes together in a stable configuration while homology is being assessed. However, centromere coupling is not essential for pairing, because homologs do align correctly in a zip1 mutant (15). Centromere coupling, bouquet formation, and recombination might all contribute in different ways to the efficacy of pairing.

The transition from largely nonhomologous centromere coupling to homologous coupling during meiotic prophase depends on the Spo11 protein, presumably through its role in the initiation of recombination (5). Coupling occurs in haploids even when Spo11 is present, indicating that Spo11 does not function to prevent associations between nonhomologous chromosomes. Instead, Spo11 probably plays a role in the recognition of homologous partners. Once two chromosomes are recognized as homologs, they may become irreversibly locked together at their centromeres and thus excluded from subsequent switching events.

Our data strongly suggest that homologously coupled centromeres serve as sites of synapsis initiation. Homolog recognition may trigger the polymerization of Zip1 at centromeres, providing the locking mechanism proposed above. Previous studies (of pachytene nuclei) have indicated that synapsis initiates at noncentromeric locations (8, 9, 11). These observations can be easily reconciled if one supposes that synapsis initiates predominantly at centromeres at early times, but noncentromeric sites make an increasingly important contribution to synapsis initiation as meiosis progresses.

Another function for centromere coupling might be the distributive disjunction system, which ensures the segregation of chromosomes that failed to cross over. Centromere regions in female flies have been shown to be important for distributive disjunction (16). In yeast, nonexchange chromosomes associate at their centromeres before segregation, and interfering with this association increases the rate of nondisjunction (17).

Centromeres appear to contribute to homolog alignment and segregation in other species as well, although the underlying mechanisms may be somewhat different. Nonhomologous centromere coupling is observed before homolog pairing during meiosis in wheat, but it is also observed in premeiotic tissues, arguing that centromere coupling does not depend on an SC protein (18, 19). In fission yeast, only centromeric regions are able to pair independently of meiotic recombination (20). The contributions of centromeres and SC proteins to homolog pairing in higher eukaryotes remain to be elucidated.

Supporting Online Material

Materials and Methods

Fig. S1


References and Notes

Stay Connected to Science

Navigate This Article