The Arabidopsis SYN1 cohesin protein is required for sister chromatid arm cohesion and homologous chromosome pairing

The faithful transmission of chromosomes during mitosis and meiosis is essential for the survival of eukaryotic organisms. A critical aspect of chromosome segregation is sister chromatid cohesion, which is required for proper attachment of chromosomes to the spindle and the faithful segregation of sister chromatids to opposite poles of the cell during anaphase (reviewed by Orr-Weaver, 1999; van Heemst and Heyting, 2000). Sister chromatid cohesion is mediated, in part, by a group of highly conserved proteins, referred to as the cohesin complex. Four proteins (SMC1, SMC3, SCC1 and SCC3) form the core of the mitotic cohesin complex, which is utilized by a wide range of organisms. In S. cerevisiae, the cohesion complex is found on chromosomes from S phase to anaphase, with preferential binding in centromeric regions (Michaelis et al.,1997; Toth et al.,1999; Uhlmann and Nasmyth,1998). The release of chromosome cohesion at the metaphase to anaphase transition and the subsequent separation of sister chromatids is triggered in most organisms by separase, a cysteine protease, which specifically cleaves SCC1 (Ciosk et al.,1998; Hauf et al.,2001; Uhlmann et al.,1999; Uhlmann et al.,2000).

Sister chromatid cohesion also serves a critical role in meiosis; however,there are several important differences between its roles in meiosis and mitosis. In the first meiotic division, which is a reductional division,homologous chromosomes segregate. Attachment of sister kinetochores to spindles occurs with the same polarity, termed monopolar attachment, ensuring that homologs and not sister chromatids segregate in this first division. Therefore, whereas sister chromatids are attached to microtubules emanating from opposite poles during mitosis, they attach to microtubules from the same pole during meiosis I. Also, with few exceptions, recombination between homologous chromosomes occurs during meiotic prophase to form chiasmata, which hold maternal and paternal chromosomes together. In order for homologs to separate during anaphase I, chiasmata between homologous chromosomes must be resolved and sister chromatid cohesion released along the arms. Therefore,meiotic divisions require sister chromatid cohesion to be released in two steps. In the first step cohesion is released along the arms to facilitate resolution of chiasmata while it is maintained at the centromeres. Destruction of centromeric cohesion at anaphase II then allows the separation of sister chromatids in an equational division.

Sister chromatid cohesion during meiosis is mediated by cohesin complexes that are similar to their mitotic counterparts and share many of the same subunits (reviewed in Lee and Orr-Weaver,2001; Orr-Weaver,1999). However, meiotic cohesin contains at least one, and in some instances more, meiosis-specific forms of the core cohesin proteins, including REC8 for SCC1 and in animal cells SMC1β for SMC1 and STAG3 for SCC3(Prieto et al., 2001; Revenkova et al., 2001). All meiotic cohesin complexes studied to date contain the meiosis-specific REC8 cohesin. Mutations in REC8 have been identified and characterized in a number of organisms (Bai et al., 1999; Bhatt et al., 1999; Klein et al., 1999; Lin et al., 1992; Parisi et al., 1999; Pasierbek et al., 2001; StoopMyer and Amon, 1999; Watanabe and Nurse, 1999). In S. cerevisiae and S. pombe, rec8 mutations result in reduced recombination, alterations in synaptonemal complex formation and premature separation of sister chromatids (Molnar et al., 1995; Klein et al.,1999; Watanabe and Nurse,1999). In C. elegans, depletion of REC8 using RNAi resulted in the formation of univalents and chromosome fragmentation at diakinesis (Pasierbek et al.,2001). Finally, Arabidopsis plants containing mutations in the REC8/SCC1 ortholog, referred to as SYN1/DIF1, exhibit alterations in chromosome condensation and cohesion that lead to chromosome fragmentation at metaphase I (Bai et al., 1999; Bhatt et al., 1999).

REC8 localization studies on meiotic chromosomes have been conducted in several organisms. S. pombe Rec8 is present from the time of premeiotic DNA synthesis until after meiosis I(Parisi et al., 1999). It is localized as foci throughout chromosomes, with the highest concentration at the centromeres (Watanabe and Nurse,1999). Likewise, S. cerevisiae Rec8 is found as punctate foci along chromosomes in early prophase I. It subsequently localizes to centromeric regions where it persists until approximately anaphase II(Klein et al., 1999). REC8 localization patterns in C. elegans were similar to those observed in yeast. Specifically, REC8 was partially lost along chiasmata-distal portions of the arms at anaphase I and at the centromeres at metaphase II(Pasierbek et al., 2001).

Differences have also been reported concerning the release of sister chromatid cohesion during meiosis. In yeast and C. elegans the anaphase promoting complex (APC)-activated separase pathway is required for the release of cohesin at the onset of anaphase I(Buonomo et al., 2000; Siomos et al., 2001). In contrast, experiments in Xenopus have suggested that chromosome segregation at meiosis I takes place in the absence of APC activity and in the presence of high levels of securin, the separase inhibitor(Peter et al., 2001; Taieb et al., 2001). This suggests that removal of cohesin from the arms of Xenopus chromosomes during meiosis I may occur by a mechanism similar to that observed for the removal of cohesin from other vertebrate arms during mitotic prophase(Darwiche et al., 1999; Losada et al., 1998; Sumara et al., 2000; Waizenegger et al., 2000; Warren et al., 2000). Finally,phenotypic differences have also been observed in cells containing mutations in cohesin subunits. For example, in rec8 mutants of S. cerevisiae, chromosomes segregate randomly at meiosis I(Klein et al., 1999), whereas in S. pombe rec8 mutants, sister chromatids segregate equationally at anaphase I (Watanabe and Nurse,1999). Therefore, a number of differences in the distribution and release of REC8 as well as the effect of rec8 mutations have been identified in the relatively few organisms studied to date. This suggests that, while the general nature and properties of meiotic cohesin complexes are similar, differences probably exist in the way cohesion is controlled in different organisms.

As part of studies to better understand sister chromatid cohesion in plants, we have further characterized the role of a putative Arabidopsis REC8 ortholog, SYN1, which we previously identified in a T-DNA-tagged, meiotic mutant (syn1) of Arabidopsis(Bai et al., 1999; Peirson et al., 1997; Peirson et al., 1996). Preliminary cytological studies indicated that syn1 plants exhibit defects in chromosome cohesion and condensation that result in fragmentation of the chromosomes and the formation of polyads(Bai et al., 1999; Peirson et al., 1997). In the experiments described below, we show that SYN1 encodes a protein that localizes to arms of meiotic chromosomes from approximately S phase to anaphase I. The protein is not detected at the centromeres or after metaphase I. Furthermore, fluorescence in situ hybridization (FISH) experiments on microsporocytes from syn1 plants demonstrate that the mutation eliminates arm cohesion as early as leptotene whereas centromere cohesion is maintained until approximately anaphase I.

Seeds of wild-type and syn1 Arabidopsis thaliana, ecotype Wassilewskija (WS) were grown on a commercial potting mix in a growth chamber at 22°C with a 16:8 light:dark cycle. The syn1 mutation is a T-DNA-tagged mutation that has been previously described(Bai et al., 1999). Approximately 15-18 days after germination, buds with lengths between 0.3 and 0.7 mm were collected from prebolting plants, fixed and analyzed as described below.

A portion of the SYN1 cDNA containing amino acids 178 to 353 was cloned into pET24b (Novagen), transformed into BL21(DE3)pLysS cells and overexpressed as a histidine-tagged protein. Upon induction the overexpressed protein accumulated in the insoluble fraction. Inclusion bodies were collected from overexpressing cells, washed, solubilized in PBS containing 8 M urea and purified using nickel chromatography. The isolated protein was further purified by SDS polyacrylamide gel electrophoresis and used to inject New Zealand White rabbits using standard procedures(Harlow and Lane, 1988). The antibody was affinity purified using the E. coli expressed protein and found to be specific for SYN1; it did not cross-react with the three other SYN1-like Arabidopsis proteins expressed in E. coli.

Inflorescences of 15-18 day old plants were fixed for 2 hours in Buffer A(Dernburg et al., 1996)containing 4% paraformaldehyde, washed twice and stored at 4°C in Buffer A until needed. Buds were staged by squashing and staining an individual anther in acetic orcein. The remaining anthers were squashed between two perpendicular poly-L-lysine slides. Male meiocytes were covered with a thin layer of agarose and treated with β-glucuronidase(Peirson et al., 1997). After washing in 1× PBS, the slides were blocked in 1× PBS containing 5%BSA for 60 minutes and then incubated overnight at 4°C in a moist chamber with anti-SYN1 antibody, diluted 1:250 in blocking buffer. After washing, the slides were treated with Alexa-488-labeled goat anti-rabbit secondary antibody for 2 hours at room temperature. After washing, the DNA was stained with 2μg/ml propidium iodide (PI) and the slides mounted in DABCO antifade mounting media. Samples were viewed with a Nikon PMC-2000 Confocal Microscope System. Individual optical z-sections were captured, the three-dimensional data were stacked (maximum intensity) using Image Pro Plus and were represented as two-dimensional images.

Inflorescences were fixed in acetic alcohol (ethanol:glacial acetic acid,3:1) for 2 hours at room temperature and stored at -20°C after replenishing the fixative. Staged buds were subjected to FISH using previously published procedures (Caryl et al.,2000; Fransz et al.,1996). The following probes were used in this study: (1) pAL1 containing a pericentromeric 180 bp repeat(Martinez-Zapater et al.,1986); and (2) BAC probes F15E21 and MFG13 corresponding to the lower arms of chromosomes one and five, respectively. Southern blotting demonstrated that the BAC clones hybridized to a single copy region of the genome. The pAL1 probe was generated by primary PCR amplification using the M13 forward and reverse primers followed by random primer labeling in the presence of biotin-labeled dUTP (Roche). The BAC probes were digested with EcoR1 to fragment the DNA followed by random primer labeling in the presence of biotin-labeled dUTP. Biotin-labeled probes were used in hybridization solution at 10 μg/ml and detected with 10 μg/ml fluorescein-labeled streptavidin. Slides were counterstained with PI, mounted and viewed as above.

Buds were fixed and spread as described above for immunolocalization experiments. In situ hybridization using DNA probes was conducted essentially as described previously (Dernburg et al.,1996). Specifically, agarose-covered, poly-L-lysine slides containing spread PMCs were washed twice in Buffer A, twice in 20% deionized formamide/2× SSC, twice in 40% deionized formamide/2× SSC/0.1%Tween-20 and twice in 50% deionized formamide/2× SSC/0.1% Tween-20. Hybridization solution (50% deionized formamide/2× SSC/0.1% Tween-20/10%Dextran Sulfate) containing a PCR fragment (15 μg/ml) corresponding to the pericentromeric 180 bp repeat labeled with digoxygenin-labeled dUTP (Roche)was added and the specimen covered with a coverslip and sealed with rubber cement. The slides were incubated at 40°C for 30 minutes, denatured at 96°C for 6 minutes and then hybridized overnight at 37°C. After hybridization the slides were washed in 50% deionized formamide/2× SSC at 37°C and 20% deionized formamide/2× SSC/0.1% Tween-20, 2×SSC/0.1% Tween-20 and 2× SSC/5% BSA, all at room temperature. They were then incubated in binding solution (2× SSC/0.1% Tween-20/5% BSA)containing 20 μg/ml mouse anti-digoxygenin (Roche) for 1 hour at 37°C. After washing, the slides were incubated in binding solution containing 20μg/ml Texas Red goat anti-mouse IgG (Jackson Labs) for 1 hour at room temperature, washed in 2× SSC/0.1% Tween-20 and incubated overnight at 4°C in binding solution containing anti-SYN1 antibody (1:250 dilution). After washing the slides were incubated in 20 μg/ml Alexa-488-labeled goat anti-rabbit IgG (Molecular Probes) in binding solution at 37°C for 1 hour,washed, mounted and viewed as above.

Previously we identified alterations in microsporocytes of syn1plants that begin just prior to leptonema(Bai et al., 1999). In contrast to wild-type microsporocytes, in which chromosomes appear as long thin threads that loop out of a dense synizetic knot, chromosomes in syn1 plants typically appeared tangled, and the synizetic knot was darkly stained. In wild-type plants, paired homologous chromosomes were readily apparent during zygonema, whereas at approximately zygonema in the mutant, a series of intertwined knots was observed; chromosomes rarely appeared as thread-like structures, and paired chromosomes were not observed. From approximately anaphase I, approximately 15 to 20 chromosomes and chromosome fragments were observed in syn1 plants. These results suggested that the syn1 mutation caused defects in chromosome cohesion, condensation and possibly homologous chromosome pairing, which ultimately result in chromosome fragmentation. In order to evaluate this possibility further, FISH experiments were conducted using chromosome arm and centromere-specific probes. Furthermore, antibodies to SYN1 were generated and used in immunolocalization experiments to determine the distribution and timing of the association of the protein with chromosomes.

Sister chromatid cohesion and homologous chromosome pairing was investigated using Alexa-488-labeled BAC clones, which correspond to the arms of chromosomes one and five. Similar results were obtained with both probes. In wild-type plants two FISH signals were typically observed in cells during meiotic interphase and early leptonema when probes corresponding to chromosome arms were used (Fig. 1A). During zygonema in wild-type meiocytes, the number of arm signals was reduced from two to one as homologous chromosomes paired(Fig. 1B). One signal was typically observed through diakinesis (Fig. 1C-D). In contrast, when probes corresponding to chromosome arms were used in FISH against syn1 microsporocytes, four distinct signals were normally observed from meiotic interphase to metaphase(Fig. 1E-H). The presence of four arm signals was consistent with a lack of sister chromatid cohesion. Occasionally cells with two or three signals were observed; however, most cells contained four (average=3.9, Table 1). This indicated that in addition to a lack of sister chromatid arm cohesion, the arms of homologous chromosomes failed to pair in syn1 plants.

Cohesion and pairing of centromeric regions was also examined by FISH using PAL1, a 180 bp tandemly repeated sequence that localizes to the central domain of the pericentromeric heterochromatin of all 10 Arabidopsischromosomes (Fransz et al.,1998). In wild-type plants approximately 10 centromere signals(eight to 10) were observed during meiotic interphase and leptonema(Table 1). As expected, during zygotene and pachytene between two and five centromere signals were observed,with averages of 4.2 and 4.9 respectively. During diplotene/diakinesis in wild-type meiocytes, five to six signals were typically observed. These results are consistent with those previously observed in wild-type Arabidopsis (Armstrong et al.,2001). Syn1 meiocytes resembled wild-type plants during meiotic interphase and leptotene, exhibiting approximately 10 (six to 10)centromere signals (Fig. 2A-B). This result contrasts with results obtained with arm-specific probes and suggests that sister chromatid cohesion at the centromeres of syn1meiocytes was maintained during prophase. Consistent with a general lack of pairing, syn1 meiocytes continued to exhibit approximately 10 centromere signals throughout prophase(Fig. 2C-D; Table 1). In a number of cells,fewer than 10 centromere signals (six to nine) were observed(Table 1), suggesting that some pairing of centromeric regions may have occurred. However, we believe that it is more likely that the reduced number of signals in these cells is not due to chromosome pairing, but rather because of the generally intertwined and sticky nature of chromosomes in syn1 meiocytes, and possibly general centromere clustering.

Approximately eight (five to 10) centromere signals were detected in optical sections of syn1 meiocytes at metaphase I(Fig. 2E). In contrast, 12 to 18 signals were present in the 43 cells observed at telophase I (average=14.8; Fig. 2F). Because of the highly condensed nature of the chromosomes at telophase I, we believe that this number may actually be an under-representation of the true number of centromere signals. Nonetheless, the presence of more than 10 centromere signals indicates that although sister chromatids remain associated at their centromeres up to metaphase I, centromere cohesion is lost by telophase I. During metaphase I, centromere signals were always found in the condensed chromosome mass at the center of the cell(Fig. 2F), suggesting that the presence of a centromere was sufficient for movement of the chromosome to the spindle assembly. Consistent with this is the observation that centromere signals were always found at the spindle poles during anaphase I. In contrast,acentric chromosome fragments, present in syn1 meiocytes, failed to attach to the spindles, although they too were often found in the center of the cell (Fig. 2F). The absence of centromere sequences confirmed that they represented chromosome fragments rather than univalents. Taken together these results confirm that syn1 meiocytes are defective in sister chromatid cohesion and homologous chromosome pairing; however, sister chromatids remain attached at their centromeres up to metaphase I.

To examine the distribution of SYN1 on chromosomes during meiosis,antibodies were raised to the central portion (amino acids 176-353) of SYN1. This region was chosen because it is the least conserved portion of the protein. Arabidopsis contains four SCC1/REC8 paralogues. Like all SCC1/REC8 proteins, the greatest similarity is found at the N- and C-terminal regions of the proteins. In contrast the central portions of the proteins show very little (less than 15% identity) sequence conservation(Dong et al., 2001). Consistent with this observation, the SYN1 antibodies did not crossreact with E. coli expressed protein for the three other Arabidopsiscohesin proteins, SYN2, 3 and 4 (data not shown).

Immunolocalization experiments on wild-type microsporocytes with SYN1 antibody revealed a strong signal in the nucleus beginning at approximately meiotic interphase. Meiocytes at interphase displayed diffuse chromatin and SYN1 labeling (Fig. 3A). Although some labeling was observed in the centrally located nucleolus, the SYN1 signal was clearly stronger in the surrounding nucleoplasm. During early leptotene, the SYN1 signal associated with the condensing chromatin(Fig. 3B). As meiocytes proceeded through leptotene, SYN1 labeling went from a relatively diffuse pattern (Fig. 3C) at early stages to approximately 100 large foci at the leptotene/zygotene transition(Fig. 3D). During zygotene and pachytene, the SYN1 signal was distributed over most of the chromosomes(Fig. 3E-G). As the chromosomes began to condense during diplotene, SYN1 labeling was reduced and began to shift from the chromosomes into the nucleoplasm(Fig. 3H). As the cells proceeded through diplotene and diakinesis, labeling in the nucleoplasm became progressively stronger until it completely filled the nucleus(Fig. 3I,J). No labeling was detected in the nucleolus. By the time the nuclear envelope had broken down SYN1 was only detected on prometaphase chromosomes, suggesting that after release the protein is degraded (Fig. 2K). As cells proceeded through metaphase I the SYN1 signal became progressively weaker until it was no longer detectable by the beginning of anaphase I (Fig. 3L-N). SYN1 was never observed on chromosomes after the onset of anaphase I. However, we were able to detect very weak SYN1 staining in the nucleus of interphase II cells (Fig. 3O). Staining of cells at this stage was always weak and short-lived.

SYN1 labeling was never observed in somatic cells of the anther or in Arabidopsis cell cultures (data not shown). This was consistent with results from in situ hybridization experiments, in which SYN1 transcripts were only detectable in the locules of stage 8 and 9 anthers (data not shown). Likewise, SYN1 labeling was not detected in meiocytes of syn1 plants(data not shown), which confirms that the antibody is specific for SYN1. Taken together these results support phylogenetic evidence(Bai et al., 1999), indicating that SYN1 is the Arabidopsis REC8 ortholog.

Results shown in Fig. 3H-Jindicated that most SYN1 was lost from the chromosomes during diplotene/diakinesis. Fig. 4A-Dclearly shows that during diplotene SYN1 labeling associated with the condensing chromosomes is dramatically reduced. By diakinesis most of the SYN1 labeling is not associated with the chromosomes, rather it appears to be free in the nucleoplasm. At these stages signal was never associated with the nucleolus. During metaphase I, SYN1 was detectable on the chromosomes, but not in the nucleoplasm (Fig. 4E,F). In contrast to its localization during zygotene and pachytene, SYN1 labeling was more narrowly focused on metaphase chromosomes. By early anaphase I SYN1 was clearly no longer detectable (Fig. 4H). SYN1 signal was never observed in the approximately 50 anaphase I cells that were examined. However, during meiotic interphase II,SYN1 was briefly detected in the nucleus of meiocytes. This signal was very transient and disappeared before metaphase II (data not shown). Therefore,SYN1 was detectable in meiocytes from approximately interphase I to interphase II. Most of the protein appeared to disassociate from the chromosomes at diplotene/diakinesis, and labeling of the chromosomes was not detected after metaphase I.

Results from our immunolocalization studies suggested that SYN1 was localized preferentially along the arms of meiotic chromosomes and not at the centromeres. In order to investigate this possibility further we conducted dual SYN1 immunolocalization/centromere FISH experiments. Centromere labeling patterns and the distribution of SYN1 in the dual immunolocalization/centromere FISH experiment resembled the results obtained for the individual experiments. During leptotene, zygotene and early pachytene(Fig. 5A-C), SYN1 labeling was clearly evident on chromosome arms. However, the dispersed nature of the chromosomes made it difficult to determine if the centromeres were also labeled with the SYN1 antibody. Beginning at late pachytene(Fig. 5D), as the centromeres started to become distinguishable from the chromosome arms, SYN1 labeling was not detected at centromeric regions (verified by 3D analysis; however only 2D projections are shown). Likewise, as overall SYN1 levels decreased during diplotene, diakinesis and metaphase, labeling was found primarily along the chromosome arms and not at the centromeres(Fig. 5E-H). In the approximately 350 centromeres examined in >70 cells observed at these stages, overlap between the SYN1 and centromere signals was observed 38 times(11%). In cells that were oriented such that the centromeres were clearly distinguishable from the arms, no overlap in labeling was detected between SYN1 and the centromere repeat clone. Therefore, SYN1 is generally not detectable at the centromeres.

Components of the cohesin complex and general features associated with the establishment of cohesion and its subsequent removal are generally conserved amongst different organisms and between mitosis and meiosis. However,differences in the distribution of cohesin proteins, the mechanism of their removal and the phenotype of cells containing mutations in cohesin subunits have also been observed. This suggests that subtle differences exist in the ways that sister chromatid cohesion is controlled. In order to investigate meiotic cohesins in plants we have characterized the effect of the syn1 mutation on homologous chromosome pairing and sister chromatid cohesion and investigated the distribution of the protein on meiotic chromosomes of Arabidopsis, which contains four SCC1/REC8-like genes. Results presented here confirm previous studies, which suggested that SYN1 is a meiotic cohesin(Bai et al., 1999). SYN1 transcripts are detectable only in locules of stage 8 and 9 anthers (data not shown), which contain meiocytes at premeiotic interphase and meiosis respectively (Armstrong and Jones,2003). Furthermore, SYN1 is detected in meiocytes from approximately meiotic interphase to metaphase I(Fig. 2); it is not detected in somatic cells. Finally, inactivation of SYN1 disrupts sister chromatid arm cohesion beginning at approximately interphase(Fig. 1). Therefore, SYN1 is an Arabidopsis meiotic cohesin.

Varying phenotypes have been observed for rec8 mutants in other organisms. For example, when REC8 is depleted using RNAi in C. elegans, meiotic chromosomes undergo presynaptic alignment, but not synapsis (Pasierbek et al.,2001). Separation of sister chromatids is observed as early as leptotene but is typically not wide spread until diakinesis. In contrast,approximately 70% of S. cerevisiae cells carrying a rec8mutation exhibit FISH labeling patterns consistent with defects in both cohesion and chromosome pairing/alignment(Klein et al., 1999). Interestingly, in S. pombe cells containing the rec8-110mutation, pairing of interstitial and centromeric chromosome regions was strongly impaired, whereas pairing at chromosome ends was less impaired(Molnar et al., 1995).

Our results indicate that the phenotype of syn1 meiocytes is similar to that observed in S. cerevisiae; specifically, SYN1 is required for sister chromatid arm cohesion and homologous chromosome pairing. Results from FISH experiments on syn1 plants revealed, in general,four arm-specific signals throughout prophase consistent with both a lack of sister chromatid arm cohesion and homologous chromosome pairing. Approximately 10 (six to 10) centromere signals were observed from meiotic interphase to anaphase I, when on average 15 (12 to 18) signals were observed(Table 1). These results are also consistent with a lack of homologous chromosome pairing but suggest that sister chromatids remain attached at their centromeres until anaphase I. At this point it is not clear why we see differences in cohesion between chromosome arms and the centromeres. One possibility is that the centromeres remain topologically intertwined during prophase. It is also possible that another protein helps link the centromeres prior to anaphase I in Arabidopsis.

SYN1 is clearly detectable in the nuclei of meiocytes at approximately interphase. It is likely that SYN1 is present as early as S phase, as appears to be the case in S. pombe(Parisi et al., 1999). However, because it is difficult to accurately assess the stage of cells within this period, we can not conclude with certainty that SYN1 is present during S phase. As meiocytes proceed through leptotene, SYN1 labeling progresses from a relatively diffuse pattern to approximately 100 large foci at the leptotene/zygotene transition (Fig. 2D). During zygotene and pachytene, chromosome arms are completely covered with SYN1 (Fig. 2E-G). During diplotene and diakinesis, much of the SYN1 signal is released and moves into the nucleoplasm as the chromosomes condense(Fig. 2H). By prometaphase I,unbound SYN1 is no longer detected in the cell. It is, however, still localized to discreet regions of the chromosomes where it persists until the metaphase/anaphase transition. By the beginning of anaphase I, SYN1 was no longer detectable (Fig. 2M). These results for SYN1 were generally consistent with those observed in other organisms.

S. cerevisiae Rec8 was initially found as punctate foci during leptotene and zygotene (Klein et al.,1999). By pachytene it formed a continuous line along the longitudinal axes of the chromosomes. By anaphase I, Rec8 labeling was dramatically reduced but still detectable as a number of small foci that remained visible in the vicinity of the centromeres until shortly after the onset of anaphase II. In S. pombe, Rec8 was highly enriched in centromeric regions from pre-meiotic S phase through metaphase II(Parisi et al., 1999). Finally, C. elegans REC8 was found as dots along unsynapsed chromosome axes during leptotene/zygotene(Pasierbek et al., 2001). During pachytene, REC8 antibodies associated with the SC. From anaphase I through metaphase II, REC8 signals were less intense and restricted to the region between the centromere and the chiasmata. Therefore, in both S. cerevisiae and C. elegans REC8 appeared to localize to the SC during pachytene, and REC8 was localized to the centromeres of all three organisms from anaphase I to metaphase II. Although the resolution of our experiments does not allow us to say with certainty that SYN1 is associated with the SC, on the basis of its distribution and the results from other systems, this is highly likely.

The greatest difference between our results and those obtained in S. cerevisiae, S. pombe and C. elegans is our observation that SYN1 antibody does not localize to centromeric regions or to meiotic chromosomes after metaphase I. This raises the interesting possibility that SYN1 is not involved in maintaining centromeric cohesion and that one of the other three Arabidopsis cohesin proteins (SYN2, SYN3 or SYN4) may be responsible for centromeric cohesion. If true, then this would suggest that meiotic chromosome cohesion is controlled differently in plants than in other organisms. However, several observations suggest that this is not the case. First, SYN1 staining was detected in the nucleus of interphase II cells(Fig. 3O). Although this staining was always weak and short-lived, it was reproducible. This suggests that low levels of the protein may be present on the chromosomes of anaphase I cells but that it is inaccessible to the antibody. This is similar to results obtained with antibodies to vertebrate SCC1, which failed to detect the protein on metaphase chromosomes (Darwiche et al., 1999; Losada et al.,1998; Sumara et al.,2000). Through the use of myc-tagged SCC1 Waizenegger et al. were,however, able to demonstrate that SCC1 is present on metaphase I chromosomes and that it localizes to the centromeres(Waizenegger et al., 2000). Our observation that the SYN1 signal is relatively weak during diplotene is consistent with the theory that chromosome conformation and possibly the location of the protein on the chromosomes has a major effect on the observed signal. During pachytene the SYN1 signal is very strong(Fig. 3F), whereas during diplotene the total level of SYN1 signal is dramatically reduced as the chromosomes condense (Fig. 3H). SYN1 signal is again relatively high at late diakinesis when most of the signal is no longer associated with the chromosomes(Fig. 3J). Finally, similar to other rec8 mutants, meiocytes in syn1 plants exhibit defects in centromere cohesion beginning at approximately anaphase I. This indicates that SYN1 plays an important role in centromere cohesion. Therefore, we believe that SYN1 is present at the centromeres of meiotic chromosomes but that it is not detectable with our antibody. We are currently investigating this question further through the use of GFP- and epitope-tagged versions of SYN1 and detailed localization studies for SYN2, SYN3 and SYN4.

Most SYN1 crossreactive material is released from Arabidopsischromosomes beginning at approximately diplotene. From diplotene to prometaphase I SYN1 labeling is very strong in the nucleoplasm. This is in contrast to results obtained in C. elegans where REC8 was observed along the axes of diplotene/diakinesis chromosomes. It was not until the cells entered metaphase I that REC8 labeling appeared to become progressively less intense. The release of SYN1 from the chromosomes coincident with chromosome condensation during diplotene/diakniesis resembles the situation in mitotic cells of animals where the bulk of cohesin is removed from chromosomes during prophase (Losada et al.,1998). This early release of cohesin from chromosome arms is referred to as the prophase pathway and is separase independent(Waizenegger et al., 2000). The segregation of chromosomes at meiosis I in Xenopus also appears to take place in the absence of APC activity and in the presence of high levels of securin, the separase inhibitor(Peter et al., 2001; Taieb et al., 2001). In contrast, in S. cerevisae and C. elegans, the resolution of chiasmata along meiotic chromosome arms depends on the cleavage of REC8 by separase at the onset of anaphase I(Buonomo et al., 2000; Siomos et al., 2001). At this time it is not known if the removal of SYN1 from meiotic chromosome arms is separase dependent. The Arabidopsis genome does, however, contain a putative separase homologue, suggesting that the APC-mediated pathway is utilized for the release of centromeric and/or possibly arm cohesion. Experiments are underway to investigate this question.

We are grateful to Sue Armstrong, Chris Franklin, Gareth Jones and Hong Ma for critical reading of the manuscript, to Greg Copenhaver for the pAL1 clone and the Electron Microscopy Facility at Miami University for providing resources for this work. BAC clones were obtained from the Ohio State Stock Center. This research was supported by a grant from the NSF (MCB-9982367).