Linear element formation and their role in meiotic sister chromatid cohesion and chromosome pairing

In the life cycle of sexually reproducing organisms meiosis halves the chromosome number in the germ line cells and produces haploid gametes. This halving is achieved by two consecutive divisions following a single round of DNA replication. The second (equational) division resembles mitotis: sister chromatids segregate into daughter nuclei. However, the first (reductional)division has unique features. During the first meiotic division homologous chromosomes pair, undergo high levels of recombination, and the resulting chiasma formation assists their segregation into daughter nuclei. In most organisms, pairing of homologous chromosomes in meiotic prophase is accompanied by the formation of synaptonemal complexes (SC). The synaptonemal complex is an evolutionary well-conserved, strictly meiosis specific,proteinaceous structure. In early prophase, after DNA replication axial elements (AE) start to connect the sister chromatids. By the pachytene stage of meiotic prophase, chromosome pairing and synaptonemal complex development culminate in the formation of a tripartite structure: the axial elements (now called lateral elements, LE) are connected by a central component (for reviews, see Zickler and Kleckner,1999; Roeder,1997; Kleckner,1996).

The fission yeast Schizosaccharomyces pombe is a haploid,unicellular eukaryote. Naturally, S. pombe cells undergo meiosis directly after mating of two cells of opposite matingtype (zygotic meiosis). However, diploid cells heterozygous for mating-type can be maintained, and synchronous meiosis can be induced by shifting the culture to nitrogen-free medium (azygotic meiosis) (Egel,1973; Egel and Egel-Mitani,1974). Meiosis in fission yeast has unusual features. In prophase,the meiotic nucleus oscillates between the cell poles(Chikashige et al., 1994). These movements confer an elongated shape to the nucleus [horse-tail nucleus(Robinow, 1977)], and are led by the SPB and the attached telomere cluster(Chikashige et al., 1994; Chikashige et al., 1997). Thus,the bouquet structure of chromosomes bundled at the telomeres is maintained during the whole meiotic prophase in fission yeast. Homologous chromosome pairing and recombination occur during horse-tail movements. Mutants impaired in telomere clustering or nuclear movement show decreased homologous pairing and recombination, indicating the importance of these events in homolog juxtaposition (Shimanuki et al.,1997; Cooper et al.,1998; Nimmo et al.,1998; Yamamoto et al.,1999). Fission yeast is highly proficient in meiotic recombination but shows no crossover interference (Munz,1994).

It has been long known that fission yeast does not form synaptonemal complexes. Instead, filamentous structures (linear elements) appear in meiotic prophase. They resemble the axial cores of other eukaryotes(Olson et al., 1978; Hirata and Tanaka, 1982). The adaptation of the nuclear spreading technique to fission yeast made possible a detailed analysis of linear element formation in meiotic time-course experiments (Bähler et al.,1993). Linear elements do not form continuously along the chromosomes and undergo morphological changes during meiotic prophase. Bähler et al. have proposed that the organization of chromatin in the linear elements may facilitate meiotic chromosome functions, such as sister chromatid cohesion, chiasma maintenance, homologous pairing and the resolution of interlocks (Bähler et al.,1993). Analysis of the rec8-110 mutant revealed coincidence of impairment of linear element formation, precocious sister chromatid separation, and decreased homologous pairing for the first time(Molnar et al., 1995). However, rec8 turned out to be a meiotic cohesin(Parisi et al., 1999; Watanabe and Nurse, 1999). As a consequence, direct evidence for the involvement of linear elements in meiotic chromosome functions is still lacking.

Our preliminary observations have shown that linear elements are altered or impaired in several meiotic recombination-deficient mutants, indicating a connection between linear element formation, recombination, and perhaps other meiotic chromosome functions. To learn more about the functions of linear elements we studied linear element formation in those rec mutants that show a strong (rec6, rec12, rec14, rec15, rec16) or intermediate(rec10, rec11) decrease in meiotic recombination(Ponticelli and Smith, 1989; DeVeaux et al., 1992). Because of its important role in meiotic chromosome pairing, meu13, homolog of HOP2 in S. cerevisiae(Nabeshima et al., 2001), was also investigated. In this study we describe several mutants with altered linear element morphology, and discuss the possible reasons for the morphological changes. This investigation of linear element formation has provided a structural explanation for the region-specificity of loss of recombination observed in the rec8, rec10 and rec11 mutants. We show that linear elements are dispensable for sister chromatid cohesion,but contribute to full level homologous pairing of chromosome arms.

S. pombe strains used in this study are listed in Table 1. The rec15::kanMX mutant was created according to the method of Bähler et al. (Bähler et al.,1998). In this construct the whole ORF of rec15 was replaced by the kanMX6 module. Other deletions/disruptions used in this study were described previously: [rec8::ura4⁺(Parisi et al., 1999); rec6-151::LEU2 and rec12-152::LEU2(Lin and Smith, 1994); rec10-155::LEU2 (Lin and Smith,1995); rec11-156::LEU2(Li et al., 1997); rec14-161::LEU2 (Evans et al.,1997); meu13::ura4⁺(Nabeshima et al., 2001); and rep1::ura4⁺. rep1(Sugiyama et al., 1994) is identical to rec16 (Ding and Smith, 1998). To visualize different chromosomal regions, the lacI/lacO system was used. Strains designated his7⁺::lacI-GFP have GFP-tagged lacI inserted at the his7 locus (Nabeshima et al.,1998). The lacO tandem repeats, together with the ura4⁺ and kan^r genes, were integrated at chromosomal loci indicated in Fig. 7F.

YEA (yeast extract agar) and YEL (yeast extract liquid) complete media, and MEA (malt extract agar) sporulation medium were as described previously(Gutz et al., 1974). Diploid strains JB6 and ED1 to ED9 (Table 1) were constructed through interrupted mating(Gutz et al., 1974). Diploid colonies are prototrophic and white on YEA medium as a result of interallelic complementation between the ade6-M216 and ade6-M210mutations (Moreno et al.,1991). PM (S. pombe minimal) and PM-N (PM without NH₄Cl) media used for meiotic time-courses were as described(Beach et al., 1985; Watanabe et al., 1988). Induction of mating and meiosis was done on MEA plates, and living cells were observed microscopically in EMM2-N liquid medium [EMM2 minimal medium without nitrogen source (Moreno et al.,1991)].

Meiotic cultures of diploid strains were prepared as described(Bähler et al., 1993). Shifting a culture to meiotic conditions at highly different cell titers changes the overall progression of the meiotic time-course [e.g. compare Fig. 5B and Fig. 5C in Molnar et al.(Molnar et al., 1995)]. Therefore, in order to compare the different mutants, care was taken to grow each culture to a cell titer of 1×10⁷ to 2×10⁷ in PM medium, before meiosis was induced in PM-N. Samples were taken hourly for DAPI staining of nuclei(Bähler et al., 1993) and for spreading. Nuclear spreads were prepared and silver-stained as described(Bähler et al., 1993),with one modification. To digest the cell walls 1 mg/ml lysing enzyme L2265(Sigma) was used instead of Novozyme 234. Silver-stained nuclei were examined with a Philips EM300 at 60 kV (Philips Eindhoven, The Netherlands). Approximately 100 silver-stained and 200 DAPI-stained nuclei were evaluated at each time point. At least two independent time courses were carried out with each mutant.

Sister chromatid cohesion, chromosome segregation and pairing of homologous chromosomes were monitored in living fission yeast cells carrying the lacI/lacO system. Mating and meiosis were induced by transferring homothallic(h⁹⁰) or crossing heterothallic (h⁺and h^-) strains on MEA and incubating the plates overnight at 26°C. Hoechst 33342, a DNA-specific fluorescence dye, was used to identify the different meiotic stages in living cells. Cells were stained with 5 μg/ml Hoechst 33342 in distilled water for 15 minutes at room temperature, resuspended in EMM2-N, and mounted on a coverslip for microscopic examination. Specimens were observed at 26°C on the CCD microscope system described previously (Molnar et al.,2001a). In order to monitor the position and number of GFP signals, images were taken with an exposure time of 0.5 seconds, in 10 optical sections covering the whole nucleus. Images were analyzed after deconvolution using the Delta Vision program (Applied Precision, Seattle, WA).

To study the phenotype of linear elements in meiotic recombination-deficient mutants meiotic time-course experiments were carried out with strains ED1 to ED9, and with JB6 as a control(Table 1). Fig. 1 shows the four regular morphological classes of linear elements(Fig. 1A-D), the quantitation of the different classes (Fig. 1E) and the timing of meiotic events followed by DAPI staining(Fig. 1F) in a time-course of the JB6 strain. A control time-course can be described briefly as follows. First, nuclei with a few, short elements can be detected (class I, Fig. 1A). Later the elements seem to contact each other and become entangled (class IIa, network; Fig. 1B), or in other nuclei several elements align closely (class IIb, bundle; Fig. 1C). Then nuclei with single, long elements appear (class III, Fig. 1D). The degradation of elements occurs through class I morphology. [For a more detailed description of a standard meiotic time course, see Bähler et al.(Bähler et al.,1993).]

A comparison of the morphology of elements and the frequency of their classes in the different meiotic recombination-deficient mutants with the control revealed that the mutants fall into four groups. (1) Some recombination-deficient mutants showed regular phenotype qualitatively(morphology of elements) and quantitatively (frequency of classes). rec6,rec15, and the previously investigated rec7 mutant(Molnar et al., 2001b) belong to this group (data not shown). (2) In the rec12, rec14 and meu13 mutants both the morphology and frequency of certain classes were altered, but the observed morphologies still resembled wild-type. (3) rec11 showed strongly impaired linear element morphology that had been observed before with the other cohesin mutant rec8(Molnar et al., 1995). (4) In the rec10 and rec16/rep1 mutants no traces of linear elements were found.

To investigate the phenotype of linear elements in the rec12mutant meiotic time courses of strain ED5(Table 1) were analyzed. In this mutant two of the observed morphologies of linear elements: class I(short elements, Fig. 2A) and class IIa (network, Fig. 2B)strongly resembled those seen in the control strain. However, class IIb and class III looked different. Instead of having a single, long bundle(Fig. 1C), class IIb nuclei contained several, shorter pieces of bundles(Fig. 2C). The class III nuclei of rec12 can be characterized by an abundance of single long elements, some of which were unusually long(Fig. 2D). The most striking feature of the rec12 mutant was the high frequency of class III nuclei throughout the whole time-course(Fig. 2E). A statistical analysis of the time course (5×2 table test) clearly indicated that the distribution of different classes of linear elements in the rec12mutant is significantly different from that of the wild-type(P<0.005). Class III was the most frequently observed phenotype in statistical sense as well, while class I was rare in the rec12mutant. This suggests that short elements (class I) elongated quickly at the beginning of the time course, and their disassembling at late phases was also quicker in this mutant than in the wild-type. Meiotic divisions occurred early in this mutant (Fig. 2F), and the intensive sporulation prevented the preparation and evaluation of spreads at later than 8 hour time points.

In the rec14 mutant (strain ED6, Table 1) linear element formation began with regular class I morphology(Fig. 3A), but nuclei of this type remained scarce throughout prophase(Fig. 3E). The single short elements developed into networks (class IIa, Fig. 3B), which is the most frequently seen morphology in rec14(Fig. 3E). Many of these networks contained loosely connected elements and showed a sort of`moth-eaten' appearance (Fig. 3B, right). These nuclei may represent impaired network morphology. Alternatively, since the high frequency of networks suggests that rec14 nuclei spend longer time at this stage, the moth-eaten morphology may represent processing or disassembly of networks. Class IIb nuclei were rare, and contained 2-4 short pieces of bundles(Fig. 3C). The processing of linear elements seems to pass through the single long element stage (class III, Fig. 3D), but the quick progression to meiosis I and the heavy sporulation prevented an evaluation of the latest stages (Fig. 3E,F). The difference in the distribution of linear element classes between the rec14 and wild-type strains is statistically significant(P<0.005). Classes I and IIa show the major differences.

Time-course experiments with strain ED8(Table 1) revealed that linear element formation begins in an unusual way in the meu13 mutant. Instead of class I, small, compact nuclei with 1-2 pieces of bundle-like structures started the process (class IIb early, Fig. 4A). These nuclei were distinguishable from those appearing later (class IIb late, Fig. 4C) by their smaller size. Because the same specimens contained nuclei of different sizes and linear element morphologies (Fig. 4E,see time points 5 to 7 hours), these compact nuclei are not likely to be spreading artifacts. Nuclei designated class IIb late(Fig. 4C) were larger and contained 2 to 4 short bundle pieces. Networks (class IIa) were also altered in the meu13 mutant. They frequently appeared as a kind of combination of the network and regular bundle morphologies(Fig. 4B). Single long elements was the morphology observed latest (Fig. 4D). Divisions in meu13 occurred with a timing similar to the control (Fig. 4F). Because the meu13 mutant showed some obvious differences in LE organization compared to the wild-type (lack of class I nuclei and two types of bundles),we did not carry out a statistical analysis on the distribution of different LE classes. The same applies for the cohesin mutants (see below).

Experiments with strain ED4 (Table 1) revealed that the morphologies of linear elements in the rec11 mutant are virtually indistinguishable from those seen in rec8. Therefore, the same designation of classes was used to describe rec11 as that previously used for the rec8-110 mutant(Molnar et al., 1995). Class A stands for nuclei with a few, very short elements(Fig. 5A). In class B nuclei usually 2 or 3 short and thick elements were visible(Fig. 5B). Class C nuclei appeared last, and contained a single long element(Fig. 5C,D). The above morphologies suggest a severe defect in linear element formation. It is a distinct phenotype, observed so far only in the meiotic cohesin mutants rec8 and rec11 [(Molnar et al., 1995) and this study]. Thus we call it the cohesion-defect phenotype of linear elements. The same morphologies were detected in strain ED2 (Table 1) where both meiotic cohesins are deleted (compare the nuclei in Fig. 5B and C, and data not shown).

Two of the meiotic recombination-deficient mutants, rec16/rep1 and rec10, did not contain any element-like structures. The majority of cells in the rec16/rep1 strain ED9(Table 1) did not undergo meiosis. In two time-course experiments, 18% and 25% final sporulation was measured. Although only a small fraction of cells underwent meiosis, nuclei in meiotic prophase were still detectable in the electron microscope. Before meiotic prophase, the spindle pole body (SPB) locates far from the nucleolus,and consists of a single body or two bodies of equal sizes(Bähler et al., 1993). Normally, meiotic prophase nuclei contain linear elements and have their spindle pole body (SPB) located close to the nucleolus. The SPB consists of a large and two adjacent smaller bodies in meiotic prophase [see Figs 1,2,3,4,5(Bähler et al., 1993)]. At time points 3 to 9 hours, up to 20% of nuclei showed this morphology in the rec16 mutant, but these meiotic nuclei never contained linear elements (Fig. 6A). Sugiyama et al. have shown that rec16/rep1 is deficient in premeiotic DNA synthesis (Sugiyama et al.,1994). This explains the described severe defects.

Meiotic time-courses with strain ED3(Table 1) gave an unexpected result. In the rec10 mutant no trace of linear elements could be observed. The rec10 mutant progressed through meiosis similarly to the JB6 control strain (Fig. 6B). In rec10, empty nuclei with meiotic prophase SPB configuration were observed with a frequency similar to that seen in linear-element-containing nuclei in the control. Horse-tail nuclei appeared with similar dynamics in the two strains. In contrast to rec16/rep1,the rec10 mutant underwent meiosis efficiently(Fig. 6B). Thus rec10can be a useful tool to evaluate the consequences of the lack of linear elements in fission yeast.

It has been proposed that linear elements have a role in meiotic chromatin organization and that they may be necessary for the proper completion of meiotic chromosome functions (Bähler et al., 1993). In order to test the involvement of linear elements in sister chromatid cohesion, rec10 and control heterothallic strains bearing the lacI-GFP/lacO recognition system were crossed with strains lacking GFP labeling (heterozygous cross for GFP). Deletion strains of the meiotic cohesins rec8 and rec11(Table 1) were also investigated. Sister chromatid cohesion was checked at the centromere and three different loci along the right arm of chromosome II(Fig. 7F). Cells in meiotic prophase (horse-tail nuclei) were identified after Hoechst 33342 staining, and the number of GFP signals was determined in living cells (see Materials and Methods). In a heterozygous cross a single GFP signal indicates regular sister chromatid cohesion. Appearance of two separated GFP signals is an obvious sign of the loss of sister chromatid cohesion. Sometimes nuclei with two closely associated but still unseparated signals were detected. Because it was frequently seen in the cohesin mutants, this doubling of the GFP signal probably indicates a loosening of sister chromatid association. Nuclei with separated and doubled GFP signals were scored separately(Fig. 7).

Sister chromatids were rarely separated at the centromere in prophase in the rec8 and rec11 mutants(Fig. 7A). In contrast, an increased impairment of sister chromatid cohesion was detected along the chromosome arm in both mutants (Fig. 7B-D). The degree of impairment was fairly constant at all loci examined in rec8. In rec11, a slight increase was observed towards the telomere. This is in contrast to the rec10 mutant where only slight aberrancies were detected at each locus(Fig. 7A-D).

χ² test showed no statistical difference in sister chromatid cohesion between the wild-type and the rec10 mutant at any of the chromosomal loci. A comparison of rec8 and rec11 showed that there clearly was no difference between them either. To get an idea about the difference between the wild-type and the cohesin mutants, we compared the combined data of the wild-type strains (wt and rec10) to the combined data of the cohesin mutant strains (rec8 and rec11). The reduction in sister chromatid cohesion in the cohesin mutants was statistically not significant at the centromere (0.05<P<0.1). In contrast, cohesion is significantly reduced at all the other loci(his2, ade1, ade8) examined in the cohesin mutants(P=0.005). We conclude that the cohesion of sister chromatids is basically intact despite the lack of linear elements in meiotic prophase of rec10.

Precocious separation of sister chromatids may occur at the first meiotic division. It is conceivable that cytologically invisible pieces of linear elements remain at the chiasmata and support the proper segregation at meiosis I. Therefore, we examined chromosome segregation at the first meiotic division in each mutant. Cells having two nuclei were identified after Hoechst 33342 staining and the number of GFP signals was determined. When both nuclei carried a GFP signal in a heterozygous cross, sister chromatids segregated prematurely. To examine segregation of homologous chromosomes, homothallic strains (homozygous cross for GFP, Table 1) were also analyzed. In a homozygous cross both homologous chromosomes have GFP labeling. If one of the sister nuclei after meiosis I had no GFP signal, nondisjunction of homologous chromosomes had occurred. In accordance with previous studies (Watanabe and Nurse, 1999; Molnar et al., 2001a), a high level of precocious sister chromatid separation was detected in rec8 (PSSC; Fig. 7E). Precocious sister chromatid separation was rare in rec10 and rec11, but both mutants showed high level of chromosomal nondisjunction at the first division(NDJI; Fig. 7E). A comparison of the combined data of the rec10 and rec11 strains to the combined data of the wt and rec10 strains showed a statistically highly significant difference in NDJI (P=0.005). In summary, our observations with GFP-labeled chromosomes in living cells suggest that linear elements are dispensable for sister chromatid cohesion in fission yeast.

Next we asked whether linear elements are necessary for homologous chromosome pairing. Homothallic rec10 (strains 119, 148, 149 and 166)and control (CT2111-2, JW555, JW558 and 159) strains(Table 1) bearing GFP labeling at different chromosomal loci (Fig. 7F) were examined. Horse-tail stage cells were identified after Hoechst 33342 staining and the GFP signals were analyzed as described in Materials and Methods. Homologous chromosomes were scored as paired when their GFP signals touched each other or only a single signal was visible. The results are summarized in Table 2.

The highest level of chromosome `pairing' was measured at the centromere in the control strain. [Actually, in fission yeast a clustering of all the centromeres occurs in prophase (Scherthan et al., 1994).] Along the chromosome arm, an increase of pairing was detectable towards the telomere. The rec10 mutant showed wild-type level of clustering of centromeres but decreased pairing at all other loci examined. A comparison of the homologous pairing in rec10to the control (Fig. 8)revealed that in rec10 the impairment was slight (statistically not significant) at the his2 and ade8 loci. These loci are located towards the centromere and telomere ends of the chromosome arm,respectively (Fig. 7F). At the ade1 locus, which is situated in the middle of the right arm of chromosome II, significantly decreased pairing (0.01<P<0.025)was detected. These results suggest that linear elements contribute to regular homologous chromosome pairing and their role is especially important in the interstitial regions of chromosome arms.

A distinguishing feature of meiosis is the appearance of proteinaceous structures that connect homologous chromosomes. In most of the examined eukaryotes synaptonemal complexes (SC) develop during the first meiotic prophase. The functions of this complicated structure are not really understood (Zickler and Kleckner,1999; Roeder,1997; Kleckner,1996). Fission yeast is an exception. It does not form SC but instead linear elements, which have been proposed to be minimal structures required for proper chromosome functions during meiotic prophase and at the first meiotic division (Bähler et al.,1993; Kohli, 1994; Kohli and Bähler, 1994; Scherthan et al., 1994). The elaboration of the function(s) of linear elements may contribute to a better understanding of the mechanism of meiosis and may help to understand the functions of the synaptonemal complex as well.

Three of the recombination deficient mutants, rec12, rec14 and meu13 showed altered linear element formation, but the morphology patterns were close to those seen in wild-type. Because the prophase stages in other organisms are defined by the development of the synaptonemal complex, a similar subdivision of meiotic prophase in fission yeast is not possible. A correlation of recombination events other than DSBs, as detected by physical analysis, with the cytological changes is still missing in fission yeast. Nevertheless, it is noteworthy that the three mutations caused different changes of morphologies and their frequencies, at different stages of meiotic prophase. This argues for a biological significance of the different linear element morphologies.

Rec12 is a homolog of Spo11, a protein that forms recombination-initiating double-strand breaks in S. cerevisiae(Cervantes et al., 2000; Davis and Smith, 2001; Keeney et al., 1997). The conservation of the catalytic site suggests an identical function in S. pombe (Keeney et al.,1997; Cervantes et al.,2000). In the rec12 mutant of fission yeast no detectable meiosis-specific breakage of chromosomes occurs(Cervantes et al., 2000; Young et al., 2002). Therefore, the rec12 mutant is an appropriate tool to address a possible interdependence of formation of double-strand breaks and linear elements. Linear elements, although with altered morphology, were observed in the rec12 mutant (Fig. 2). Consequently, the formation of linear elements is not dependent on the occurrence of double-strand breaks. Analysis of the rec6,rec7 (Molnar et al.,2001b) and rec15 mutants confirmed this conclusion. They are defective in DSB formation (Cervantes et al., 2000; Davis and Smith,2001), but show normal linear element morphology. In addition,meiotic breakage of chromosomes does not occur in strains lacking the rec14 gene product (Cervantes et al., 2000; Davis and Smith,2001), but it showed altered linear elements(Fig. 3) that differed from those in rec12 (Fig. 2).

Conversely, although LE formation does not depend on DSB formation, some early prophase proteins influence both processes and recombination. The rec8 and rec10 mutants were shown to form meiotic breaks with reduced efficiency (Cervantes et al.,2000). In the cohesin mutants rec8(Molnar et al., 1995) and rec11, as well as in the double mutant rec8rec11(Fig. 5), identical and, from wild-type, strongly deviating LE structures were observed. In rec10the linear elements were missing completely(Fig. 5). Obviously cohesins and rec10 are involved in LE and DSB formation (see below). rec12 and rec14 are also involved in both processes. The most striking feature of the rec12 mutant was the abundant occurrence of nuclei with single long elements at the expense of networks and bundles(Fig. 2E). Networks and bundles are the most complicated linear element morphologies. Their rarity indicates that linear element processing is altered in the rec12 mutant. Thus fission yeast Rec12 is similar to its budding yeast homolog in the sense that both proteins are involved in the initiation of homologous recombination and also in proper chromosome organization. However, their role in chromosome organization might be somewhat different: spo11 null mutants are capable of forming axial elements, but exhibit severe homolog synapsis defects(Loidl et al., 1994).

The most striking feature of the rec14 mutant was the high frequency of networks (Fig. 3). rec14 is a functional homolog of REC103 of S. cerevisiae(Evans et al., 1997). Both genes were shown to be involved in early stages of meiotic recombination(Evans et al., 1997; Gardiner et al., 1997; Cervantes et al., 2000), but the known phenotypes of their mutants do not allow a clear conclusion about the function of rec14 in linear element formation. However, it should be noted that the expression pattern of rec14 is different from that of other recombination genes in fission yeast: its transcript is present both in meiotic and mitotically dividing cells, and the mutant shows a slow mitotic growth phenotype (Evans et al.,1997). Thus it is conceivable that the observed alteration in linear element formation is a consequence of a disturbance of basic chromosome structure not directly related to the initiation of recombination.

meu13 participates in homologous chromosome pairing in fission yeast in a recombination-independent mechanism and shows significant sequence homology to Hop2 (Nabeshima et al.,2001), a protein that ensures synapsis between homologous chromosomes in S. cerevisiae (Leu et al., 1998). Double-strand breaks form in meu13deletion strains and their repair is retarded(Shimada et al., 2002). The typical morphological change in this mutant was the frequent occurrence of network-and bundle-like structures (Fig. 4), some of which rather resembled an unspecific deposition of linear element material than a functional structure(Fig. 4A). Meu13p is localized to meiotic chromatin during the horse-tail nuclear movement stage(Nabeshima et al., 2001). Because linear elements were most aberrant at early prophase stages in the meu13 mutant (Fig. 4A), it is suggested that Meu13 might serve as a loading site for linear element proteins. In turn, linear elements are needed to achieve the full level of meiotic chromosome pairing(Table 2; Fig. 8). Thus the decreased meiotic pairing and recombination observed in the meu13 mutant might be a consequence of imperfect formation of linear elements.

DeVeaux and Smith have observed first a regional specificity of loss of meiotic recombination in the rec8, rec10 and rec11 mutants(DeVeaux and Smith, 1994). They have found that meiotic recombination was impaired most severely in a∼2 Mb region surrounding the ade6 locus of chromosome III, while other loci examined were less severely affected. Parisi et al. and Krawchuck et al. extended their study and showed that meiotic recombination is decreased severely in the centromeric region of each chromosome in these mutants(Parisi et al., 1999;Krawchuck et al., 1999). Based on epistasis analysis and classical chromosome segregation studies, Krawchuk et al. proposed that Rec8, Rec10 and Rec11 are involved in a `meiotic sister chromatid cohesion pathway' and promote homologous chromosome pairing in the centromer proximal regions of chromosomes(Krawchuk et al., 1999). Our results largely confirm this hypothesis and suggest that the underlying structural reason for the regional specificity is a defect in the formation of linear elements in the rec8, rec10 and rec11 mutants.

Rec8 and Rec11 are meiotic cohesins(Watanabe and Nurse, 1999; Parisi et al., 1999; Davis and Smith, 2001). Rec8 has two different functions correlating with distinguishable localization. It was found to locate to the centromeres of chromosomes(Watanabe and Nurse, 1999) and to ensure reductional segregation at the first meiotic division(Watanabe and Nurse, 1999; Molnar et al., 2001a). In addition, Rec8 as well as Rec11 are involved in sister chromatid cohesion along the chromosome arms (Fig. 6). We have found a severe defect in linear element formation in rec8 and rec11 mutants(Molnar et al., 1995)(Fig. 5). Deletion of either of the meiotic cohesins resulted in a typical morphological change observed so far only in these mutants. Moreover, the double mutant showed the same phenotype (Fig. 5). The most straightforward interpretation of these observations is that Rec8 and Rec11 work in a complex, and that functional cohesin complexes are indispensable for proper linear element formation. Cohesin complexes may serve as loading sites for linear element polymerization. A similar function of S. cerevisiae Rec8 for axial element formation has been suggested(Klein et al., 1999). The rec10 mutant does not form linear element material at all, but it undergoes meiosis with similar timing and efficiency to that of a wild-type strain (Fig. 6). Direct investigation of the sister chromatid cohesion along chromosome II in the rec10 mutant has shown that linear elements are basically dispensable for sister chromatid cohesion in fission yeast(Fig. 7).

Cells employ several mechanisms to effect chromosome pairing. The contribution of horse-tail movements and telomere clustering to homologous chromosome pairing is well-demonstrated in fission yeast (for a review, see Yamamoto and Hiraoka, 2001). The rec10 mutant shows regular horse-tail movements (M.M.,unpublished). Because mutants impaired in telomere clustering perform aberrant nuclear movement (Cooper et al.,1998; Nimmo et al.,1998; Hiraoka et al.,2000), the above observation indirectly indicates that telomere clustering is regular in rec10. Therefore, the decrease in chromosome pairing in the rec10 mutant is likely to be attributable to the lack of linear elements.

The highest level of homologous chromosome contact was detected at the centromere of chromosome II, both in the control and the rec10 mutant strains (Table 2). A similar result was obtained in a wild-type strain by fluorescence in situ hybridization experiments performed on nuclear spreads: Scherthan et al. has shown that all the centromeres form a cluster and maintain this state throughout meiotic prophase in fission yeast(Scherthan et al., 1994). This observation suggests that cells have a mechanism to accomplish the clustering of centromeres independently from linear element formation or rec10function. In the control strain, a gradual increase in chromosome pairing was observed towards the telomere (Table 2). This can be explained by the effect of telomere clustering,which increases the possibility of chance contacts primarily at the telomere proximal regions of chromosomes.

In the rec10 mutant a decrease in homologous chromosome pairing was measured at each locus along the chromosome arm(Table 2). The reduction was slight at the his2 and ade8 loci(Fig. 8). Among the examined loci, ade8 is closest to the telomere and his2 is to the centromere. Telomere clustering may promote pairing of homologous chromosomes most efficiently in the ade8 region in absence of linear elements. The efficient clustering of centromeres in rec10 might exert a similar supporting effect for pairing of centromere proximal loci. An alternative explanation is based on the fact that the his2 locus is near the mating-type locus, which contains heterochromatin. Thus, fission yeast may use the heterochromatin of the mating-type and centromeres to achieve linear element independent pairing of the centromere-proximal region in this chromosome arm. The ade1 locus is situated in the middle of the chromosome arm, and the most severe defect in chromosome pairing was detected at this locus (Fig. 8). Linear elements are thus likely to be most important for chromosome pairing at interstitial arm regions. Clustering of telomeres and centromeres provide for alignment of the telomere and centromere proximal regions, respectively. This probably ensures frequent contact of the corresponding regions, and homologous sequences may have a better chance for pairing despite the lack of chromatin organization by linear elements.

rec10-155::LEU2 is a partial deletion lacking residues 683 to 791(Lin and Smith, 1995). This mutant undergoes meiosis similarly to wild-type, but lacks linear elements completely. This provided a good opportunity to analyze the role of linear elements in meiosis. We have found that linear elements are basically dispensable for sister chromatid cohesion, but contribute to homologous paring of chromosomes. Although we cannot rule out the possibility that rec10 promotes homologous chromosome pairing independently from its function in linear element formation, the most straightforward interpretation of our data is that Rec10 exerts its effect on homologous chromosome pairing through its function in linear element formation. What is the role of rec10 in linear element formation? rec10 might encode a structural protein of linear elements. However, a regulatory role is also plausible. rec10 may regulate linear element formation directly or through more general processes, for example, through the regulation of chromatin structure.

We thank Gerald R. Smith, Masayuki Yamamoto and Hiroshi Nojima for strains,Toni Wyler and Karl Babl for technical assistance, and Jürg Bähler and Josef Loidl for critical reading of the manuscript. This work was supported by the Swiss National Foundation (to J.K.) and the Japan Science and Technology Corporation (Y.H.).