Ndj1p, a Meiotic Telomere Protein Required for Normal Chromosome Synapsis and Segregation in Yeast

See allHide authors and affiliations

Science  23 May 1997:
Vol. 276, Issue 5316, pp. 1252-1255
DOI: 10.1126/science.276.5316.1252


The Saccharomyces cerevisiae gene NDJ1(nondisjunction) encodes a protein that accumulates at telomeres during meiotic prophase. Deletion of NDJ1 (ndj1Δ) caused nondisjunction, impaired distributive segregation of linear chromosomes, and disordered the distribution of telomeric Rap1p, but it did not affect distributive segregation of circular plasmids. Induction of meiotic recombination and the extent of crossing-over were largely normal in ndj1Δ cells, but formation of axial elements and synapsis were delayed. Thus, Ndj1p may stabilize homologous DNA interactions at telomeres, and possibly at other sites, and it is required for a telomere activity in distributive segregation.

Efficient segregation of chromosomes in meiosis is required for haploidization. In a screen for yeast genes that control meiotic chromosome segregation, we identifiedNDJ1 by its ability to interfere with this process when overexpressed (1). NDJ1 corresponds to the open reading frame YOL104c from chromosome XV (2). The predicted Ndj1 protein comprises 352 amino acids and shows no significant similarities to other proteins. NDJ1 was previously identified as the sporulation-specific gene M45 (3). A 1.4-kb NDJ1 mRNA was detected by Northern blot analysis in yeast cells early in meiosis, but it was not detected in vegetative cells (4).

Deletion of NDJ1 (ndj1Δ) reduced sporulation efficiency (73 versus 88%), increased the number of asci with one or two versus three or four spores (15 versus 7%), and reduced spore viability (Table 1) (3), as compared with the wild type. Furthermore, in ndj1Δ homozygotes, the frequency of spores disomic for chromosome III increased 25-fold from wild-type values of (7.0 ± 0.6) × 10−4, consistent with previous measurements (5), to (1.8 ± 0.2) × 10−2. Thus, NDJ1 is required for efficient segregation of chromosomes in meiosis.

Table 1

Spore viability. Thendj1Δ::TRP1 construct (ndj1Δ), which results in the removal of amino acids 14 to 252 from Ndj1p, was introduced into haploid strains MDY431 and MDY433 (18), which were then allowed to mate and sporulate. Spore viability was determined from dissected tetrads. Vb, viability.

View this table:

Efficient orientation of chromosomes at meiosis I generally requires crossing-over and chiasma formation (6, 7). However, there was no consistent difference between wild-type and ndj1Δcells in the final extent of crossing-over or crossover interference (Table 2), or in the kinetics and yields of gene conversion events (Fig. 1A). Thus, recombination pathways are largely intact in ndj1Δ cells.

Figure 1

Timing of genetic and cytological events. (A) Commitment to intragenic recombination. Wild-type MDY431×433 (18) (closed symbols) and an isogenic derivative homozygous for ndj1Δ:: TRP1, MCY422×423 (open symbols), were shifted into sporulation medium at time zero. After the indicated intervals, samples were plated on selective media to assay gene conversion at heteroallelic loci,HIS7 (circles) and LEU1 (triangles), by production of prototrophs. Values are corrected for viability differences assayed by control platings on complete medium. Similar results were obtained with LYS2 and TYR1. (B and C) Meiotic chromosome structures as analyzed by electron microscopy of silver-stained spread preparations (11). Fifty nuclei were examined per time point, and the percentage of those in or beyond the stage at which axial element formation (B) or synaptonemal complex formation (C) has initiated was determined. Values are slightly inflated by loss of more fragile vegetative cells during treatment with Zymolyase (ICN). (D) Cells that had completed the first meiotic division were identified by staining of whole cells with 4′,6′-diamidino-2-phenylindole. The percentage of cells in or beyond the stage at which chromosomes have divided into two distinct masses was determined. (B through D) Closed symbols, wild type; open symbols, ndj1Δ.

Table 2

Crossing-over and interference.ndj1Δ was introduced into haploid strains MDY431 and MDY433, which were then allowed to mate and sporulate as described in Table 1. Data from tetrads with four viable spores were used to calculate map distances [in centimorgans (cM)] on chromosomes VII, II, and V as indicated (19). Coefficients of crossover coincidence, a measure of interference, are presented between the map distances.

View this table:

To determine the type of segregation errors in ndj1Δcells, we studied the segregation of homologous or heterologous pairs of yeast artificial chromosomes (YACs) (8). In wild-type cells, the homologous YAC pair undergoes recombination and disjunction at a frequency similar to that for genuine yeast chromosomes, whereas the heterologous YAC pair does not recombine but nevertheless undergoes disjunction at meiosis I (Table 3) (8, 9) by distributive segregation, which in yeast is less efficient than chiasmate segregation (8, 10). Disjunction of heterologous YACs was markedly reduced in ndj1Δ cells (Table 3) (random segregation would be 50%), establishing that NDJ1 is required for their distributive segregation. Increases in precocious sister separation (PSS) and double precocious sister separation (PSS×2) as well as in meiosis II errors of homologous YACs were also observed in ndj1Δ cells (Table 3). In contrast to the behavior of linear YACs, distributive segregation of a pair of circular chromosomes was not impaired in ndj1Δ cells (Table 3).

Table 3

Segregation of tester chromosomes. Data are the percentage of tetrads per class and, in parentheses, the actual number scored. The assays for segregation errors were performed as described with yeast strains containing homologous and heterologous linear human YACs (pairs A and F) (8). Each YAC contains a specific centromere marker whose segregation can be monitored relative to the segregation of a centromere marker on chromosome VI. Circular centromere plasmids were pMCB248, a randomly cloned 8.1-kb Bam HI fragment from chromosome XIII in pRS314, and pMCB249, a random 6.8-kb Bam HI fragment from chromosome XIV in pRS316. Their segregation was assayed in the same strain background as for the linear YACs. The classes of segregation errors are (i) nondisjunction, both YACs or plasmids go to the same pole at the first division; (ii) PSS (precocious sister separation), sister chromatids of one YAC or plasmid segregate to opposite poles at meiosis I; (iii) PSS×2 (double precocious sister separation), sister chromatids from both YACs or plasmids separate prematurely; and (iv) abnormal meiosis II, chromosome loss or nondisjunction at the second division results in one pair of sister spores with normal segregation and the other pair with one member missing a YAC or plasmid. Significant effects (Gtest) of ndj1Δ were detected for the disjunction of heterologs [G = 9.97; 1 degree of freedom (df);P < 0.005], combined PSS and PSS×2 of homologs (G = 4.53; 1 df; P < 0.05), and meiosis II errors of homologs (G = 3.75; 1 df; P = 0.05).

View this table:

Multiple cytological abnormalities were apparent in ndj1Δcells. Initiation of the formation of axial elements (Fig. 1B) and of synapsis (synaptonemal complex formation) (Fig. 1C) was delayed, even though the kinetics of spindle pole body (SPB) duplication were similar to those of wild-type cells (4). Completion of synapsis was further delayed, resulting in abnormal nuclei with fully duplicated SPBs but incomplete synapsis (Fig. 2A) and, finally, in nuclei with apparently complete synapsis and polycomplexes (accumulations of synaptonemal complex material common in nuclei delayed in meiotic prophase) (Fig. 2B). Cells completing the first meiotic division appeared after a total delay of 4 to 6 hours (Fig.1D). There was no abnormal accumulation of cells in the first division, and the second division was completed after the normal delay between the first and second divisions (4).

Figure 2

Electron micrographs of silver-stained spread preparations (11). (A) Accumulation of unsynapsed axial elements (single lines) in anndj1Δ cell, even though synapsis (paired lines separated by a uniform spacing) is complete for some chromosome pairs in this zygotene nucleus (6 hours after shift to sporulation medium). Complete duplication of the SPB, as shown here (small arrowhead), is normally accompanied by complete synapsis in wild-type cells. (B) Nuclei in ndj1Δ cells with apparently complete synapsis (pachytene) have polycomplexes (one of six indicated by larger arrowhead) (10 hours after shift to sporulation medium). Small arrowhead as in (A). Bar, 2 μm.

To determine the location of Ndj1p, we replaced the wild-typeNDJ1 gene with a version that encoded the hemagglutinin (HA) epitope at the 3′ end of the coding sequence. The extents of sporulation and spore viability for an NDJ1-HA homozygote were similar to those of the wild type. Spread preparations of meiotic nuclei (11) from an NDJ1-HA homozygote were stained with a monoclonal antibody to the HA epitope and visualized by indirect immunofluorescence. Ndj1p is abundant at telomeres (Fig.3A), suggesting a telomere function even though it may function at other chromosomal locations at concentrations not readily detected by immunocytology, as do telomere proteins Rap1p, Sir3p, and Sir4p (12, 13).

Figure 3

Immunocytological localization of Ndj1p and analysis of telomere behavior. (A) Pachytene nucleus from an NDJ1-HA homozygote labeled with monoclonal antibody 12CA5 (Babco) to the HA epitope (yellow-green staining). No fluorescence signal is apparent in nuclei not containing the HA epitope under these conditions. (Bthrough D) Pachytene nuclei from wild-type (B) andndj1Δ (C and D) cells labeled with antibodies to Rap1p (red staining). Blue staining in (A) through (D) represents 4′,6′-diamidino-2-phenylindole–stained DNA. Bar in (D), 1 μm.

We examined telomeres directly with antibodies to Rap1p, which binds to sites in the telomeric DNA repeat sequence (12,13). Bivalent ends in wild-type pachytene nuclei (11) normally show a single spot of Rap1p (Fig. 3B). We detected 35.4 ± 4.0 spots per nucleus (n = 11), close to the expected value of 32 if the four telomeres at each bivalent end lie in close proximity. However, in ndj1Δ pachytene nuclei, ends were often marked by fainter or multiple smaller spots (Fig. 3, C and D) and nuclei averaged 68.9 ± 16.6 spots (n = 21), which was significantly different from the wild-type value (t test, P < 0.001) and indicates abnormal telomere organization. This observation is consistent with a failure in the formation or the persistence of stable associations between homologous or sister telomeres in ndj1Δ cells.

The ndj1Δ phenotype thus includes (i) delayed formation of axial elements, (ii) longer delays in the initiation of synapsis and in completion of the first meiotic division, (iii) a defect in telomere localization of Rap1p, (iv) an increased frequency of nondisjunction and PSS of homologs, (v) reduced levels of sporulation and spore viability, and (vi) defective distributive segregation of linear, but not circular, heterologs. Induction of meiotic recombination is similar to that in the wild type and the extent of crossing-over is largely normal. Localization of Ndj1p to the telomeres and the immunocytological and distributive segregation defects inndj1Δ cells suggest that Ndj1p functions at telomeres, but it is possible that it also acts interstitially. Some or all of the defects in ndj1Δ cells may result indirectly from the delay that begins before or at the time at which axial elements form, but we favor a different model.

Taken together, our results suggest that Ndj1p stabilizes homology-dependent interactions. These interactions may be similar in properties to the paranemic joins proposed by Weiner and Kleckner (14) and be telomere-specific, or, if Ndj1p also functions at nontelomeric sites, they might correspond to the same paranemic joins and simply be more abundant at the telomeres than interstitially. Given that NDJ1 influences segregation of heterologs, which presumably are nonrecombinant, the proposed stabilization would occur independently of recombination. The concentration of this activity at the telomeres could account for the retention of association of telomere-adjacent regions even though interstitial interactions are lost in a recombination-defective mutant (14).

Although nondisjunction of homologs in ndj1Δ cells could result from relatively subtle defects in crossing-over (perhaps in specific regions or on specific chromosomes), the same phenotype would result from defective chiasma function as a result of decreased cohesion of sister chromatids (15). The increase in PSS and in meiosis II abnormalities in ndj1Δ cells could result from a failure of sister chromatid cohesion (16) specifically at the telomeres. In addition, Ndj1p could stabilize sister interactions interstitially, perhaps at hotspots of double-strand breakage, which have been demonstrated to reduce PSS (9). By holding sister chromatids together or initiating interactions, Ndj1p might facilitate axial element formation, although the precise mechanism for such an effect is unclear. Stabilization of interstitial homolog interactions could facilitate synapsis. Stabilization of telomere interactions would bring subtelomeric sequences into proximity (17) and hold chromosomes in register, which could also facilitate synapsis.

Ndj1p is required for linear heterologs to segregate distributively with the same efficiency as circular heterologs, which do not require this protein. Thus, in the absence of Ndj1p, telomeres interfere with the distributive segregation pathway.

  • * To whom correspondence should be addressed. E-mail: dresserm{at}


View Abstract

Navigate This Article