Two novel proteins recruited by synaptonemal complex protein 1 (SYCP1) are at the centre of meiosis

Formation of the synaptonemal complex (SC) is a key step of meiosis. The SC is a zipper-like structure composed of two lateral elements (LEs) that are joined together by transverse filaments (reviewed in Page and Hawley, 2004). In early prophase I, the cohesin core, which is formed by STAG3, SMC1α, SMC1β, SMC3 and REC8, starts to link sister chromatids together, while SYCP2 and SYCP3 begin to form axial elements (AEs) (Eijpe et al., 2003). During the zygotene stage, the paired homologous chromosomes become physically linked through transverse filaments when the AEs synapse to become LEs (reviewed in Page and Hawley, 2004). Transverse filaments are composed of SYCP1 molecules, which bridge the gap between one LE and the central element (CE) (Meuwissen et al., 1992). The organization of the CE, however, has never been fully defined and no protein has been characterised that shows a localisation restricted to the CE. It is believed that, the increased electron-density observed in the CE compared with the adjacent rest of the central region is owing to the arrangement of the N-terminal region of dimers of SYCP1 that interact in a head-to-head fashion, while forming an interdigitating array of dimers (Schmekel et al., 1996; Öllinger et al., 2005). The overall structure of a SYCP1 molecule resembles that of the yeast ZIP-1 protein (Sym et al., 1993), required for normal recombination and crossover interference (Storlazzi et al., 1996) as well as synapsis. SYP-1, a Caenorhabditis elegans SC protein (MacQueen et al., 2002) and crossover suppressor on 3 of Gowen [c(3)G] in Drosophila melanogaster (Page and Hawley, 2001) also have related structures.

Mutations of SC components have shown that failure to assemble this structure leads to meiotic arrest or aneuploidy. In mammals, SMC1β mutants are sterile in both sexes (Revenkova et al., 2004). Male meiosis is blocked at pachytene, whereas female meiosis continues until metaphase II in a highly error-prone fashion. In the case of SYCP3 null mutant mice, males are sterile because of apoptotic cell death during prophase I (Yuan et al., 2000). Mutant females, however, are fertile but oocytes show an increase in aneuploidy rates, which results in increased embryo death (Yuan et al., 2002). Moreover, in humans, there have been reported cases of infertile males where a mutation in the gene encoding SYCP3 was found (Miyamoto et al., 2003). In these patients, spermatogenesis was disrupted at early meiosis, showing severe alteration of the cytoarchitecture of seminiferous tubules.

The study of meiosis in mammals has been performed at a range of levels from the biochemical to the cytogenetic, but has not been easily amenable to the genetic approaches available in model organisms such as Saccharomyces cerevisiae and C. elegans. These genetic approaches have been productive in increasing the number of proteins known to be involved in meiosis but the lack of sequence conservation has made extrapolation to mammals difficult. In a different approach, we hypothesised that expression profiling can be used to find new components of the meiotic machinery (Maratou et al., 2004). Previously, we constructed a microarray from a normalised and subtracted testis library (Maratou et al., 2004) and used this to profile the expression of genes during postnatal testis development in wild-type and in Dazl^{hgutm1/hgutm1} mice (Ruggiu et al., 1997). The mutant mice lose germ cells postnatally and surviving germ cells do not progress beyond leptotene (Saunders et al., 2003). Based on this expression data we selected a number of genes with increasing expression levels between day 7 and day 11 post-partum in normal but not Dazl^{hgutm1/hgutm1} mutant development. Within this group, genes frequently represented as testis expressed sequence tags (ESTs) were selected on the basis of database searches, and a subset of genes of unknown function were selected for verification of expression pattern by in situ hybridization and northern blot. Here, we focus on two genes, Syce1 (synaptonemal complex central element 1, 329aa, GenBank accession number GI12855570) and Cesc1 (central element synaptonemal complex 1, 171aa, GenBank accession number GI21313690), selected on the above described grounds and show that they encode the only proteins known to date that localise exclusively to the CE of the mammalian SC. We also show that SYCE1 and CESC1 form a complex with SYCP1, which appears to be capable of recruiting these proteins to the CE. This suggests a role for SYCE1 and CESC1 in synapsis, and we discuss the implications they could have in assisting recombination.

A multiple tissue mouse northern blot (Ambion) was hybridised sequentially with cDNA probes for SYCE1, CESC1 and S26 (loading control) (Vincent et al., 1993). Probes were labelled using the Strip-EZ DNA kit (Ambion) and hybridised with ULTRAhyb hybridization buffer (Ambion) according to the manufacturer's recommendations. Signals were detected by fluorography.

Antibodies were raised in rabbits using a mixture of GST fusion protein expressed in Escherichia coli BL21 using the respective cDNAs of Syce1 and Cesc1 in pGEX-4T-1 (Amersham) and synthetic peptides coupled to keyhole limpet haemocyanin (KLH). Peptides used were ⁷¹ENINESRQKDHALM⁸⁴ and ³¹⁶ETAQDQERPSSRT³²⁹ for CESC1 and SYCE1, respectively. Antibodies were purified by affinity purification to peptide immobilised on Sulpholink (Pierce) columns, or to IgG fractions purified on protein G Sepharose (GE Healthcare).

Western blot analysis of testis extracts with affinity purified anti-SYCE1 shows a doublet at 44 kDa in comparison to a predicted size of 38 kDa (Fig. 3). Both bands were competed by the immunising peptide. We have not detected alternative splice forms of the mRNA encoding this protein or alternative transcripts in the databases, and the human and mouse genomes appear to encode only one copy of this gene. Phosphatase treatment did not resolve the doublet, possibly because of a different modification of the protein. Anti-CESC1 antibody did not give a signal on a western blot but was effective on tissue sections (Fig. 1) and spread chromosomes (Fig. 2).

Testes were snap-frozen in isopenthane and processed for cryosectioning by standard methods. Sections were fixed in 1% paraformaldehyde and immunofluorescence was performed as described (Alsheimer et al., 2000). Spread-preparations of meiotic cells were produced (Speed and Chandley, 1982) and then immunostained with antibodies labelled with Alexa Fluor 594 (Molecular Probes) according to the manufacturers instructions or detected with a FITC-labelled secondary antibody (Jackson ImmunoResearch) for light-microscopy and images captured using IPLab software (Scanalytics). Spermatocytes from male Sycp3^–/– mutant mice were fixed in 1% paraformaldehyde and 0.15% Triton X-100. Oocytes from female 18 days post coitum (d.p.c.) Sycp3^–/– mutant embryos were fixed in 0.8% paraformaldehyde and 0.15% Triton X-100 as described previously (Peters et al., 1997). Immunofluorescence was performed using standard methods (Yuan et al., 2000). For pre-embedding-immunogold-localisation of SYCE1 and CESC1, 5-nm cryostat sections of shock-frozen rat testis were prepared, fixed with 2% formaldehyde and incubated with primary and secondary antibodies as described previously (Smith and Benavente, 1992). Fixation and embedding in Epon™ for transmission-electron-microscopy was then performed according to standard protocols.

To express SYCE1, CESC1 and SYCP1 in the culture cell line COS-7 (green monkey kidney), the respective cDNAs were inserted in pEGFP-N vectors (Clontech, Heidelberg, Germany) with which the cells were transfected using the effectene system according to manufacturer's instructions (Qiagene, Hilden, Germany). Analysis of protein distribution was performed after 24 hours by indirect immunofluorescence microscopy as recently described (Alsheimer et al., 2000) using anti-Cesc1 IgG peak to detect Cesc1, affinity purified anti-SYCE1 to detect SYCE1 and antibody VIIId3 (Öllinger et al., 2005) to detect SYCP1. Fluorochrome-coupled secondary antibodies were purchased from Dianova (Hamburg, Germany). DNA was visualised with Hoechst 33258 (Hoechst, Frankfurt a. M., Germany). Digital pictures were processed using Adobe Photoshop 7.0.1.

Testis extracts were made by homogenising adult mouse testes in 50 mM Tris pH 7.5, 150 mM NaCl, 0.5% Triton X-100, 5 mM EDTA with Complete protease inhibitor (Roche). After sonication and centrifugation lysates were stored at –70°C until used. Prior to use, extracts were incubated with protein G Sepharose and centrifuged at 10,000 g for 10 minutes. Relevant antibodies were added to 500 μl of extract (typically 5 mg total protein) and incubated overnight at 4°C. Twenty microlitres of protein G Sepharose were added and incubated with shaking for one hour. Sepharose beads were harvested by centrifugation at 3000 g for 2 minutes and were washed four times in homogenization buffer. Beads were resupended and analysed by SDS-PAGE and western blot. The following primary antibodies were used: anti-SYCE1 and anti-CESC 1 (this paper), anti-SYCP1 antibodies A2 (Meuwissen et al., 1992) and pep7 (raised against the C-terminal 14 aa of the mouse SYCP1 protein and affinity purified against the same peptide). Secondary antibodies were horseradish-peroxidase-conjugated and purchased from Sigma. GST-fusion proteins were produced by standard methods from cDNA sequences cloned into vector pGEX4T1 (Amersham); they were bound to glutathione-Sepharose (Amersham) using the manufacturers methods. [³⁵S]-labelled proteins were produced in a reticulocyte lysate system (TNT-Stratagene) from PCR products generated from cDNA-containing plasmids. Primers were designed to introduce a T7 polymerase promotor and a Kozak consensus sequence. Beads bound to GST or fusion protein were incubated with the labelled proteins in 50 mM Tris pH 7.5, 150 mM NaCl, 0.5% Triton X-100 and 5 mM EDTA with Complete protease inhibitor overnight at 4°C and were then washed four times in this buffer before loading on a 10% SDS-PAGE gel. Labelled protein was detected by fluorography.

Searches of protein sequences in Swissprot, Trembl and Ensembl, of EST sequences in the sequencing and annotation projects databases of EMBL/Sanger Institute, NCBI and TIGR identified homologous proteins. Each protein is the best reciprocal BLAST hit of the query protein within the appropriate genome. Sequence alignments were generated with ClustalW and displayed using GeneDoc.

According to database searches, Syce1 and Cesc1 genes are mainly expressed in the testis and highly represented as testis ESTs, respectively. A northern blot, containing RNA from mouse tissues probed with cDNAs of these genes confirmed the predominantly testis-confined expression (Fig. 1A). PCR analysis of cDNA from a range of different adult tissues and 14 d.p.c. whole embryo showed that expression of Syce1 was confined to testis (not shown). Cesc1 was expressed in a wider range of tissues at this level of detection, including ovary, male and female whole embryo, spleen, thymus, brain, kidney, epididymis, heart and liver (not shown).

Reciprocal BLAST searches revealed orthologues of both proteins in mammals; for CESC1, additional orthologues were found in chicken, frog, fish and clam (supplementary material Fig. S1). The same search strategy did not reveal orthologues in other completed genomes such as D. melanogaster, C. elegans, S. pombe and S. cerevisiae. This lack of detectable orthologues at the sequence level is common to many synaptonemal complex proteins, e.g. SYCP1, and it is reasonable to expect that structural orthologues will be present in invertebrates. Both genes encode non-globular proteins with a series of loops and helices – including predicted coiled-coil regions – but no significant similarities to known domains were detected. CESC1 contains a predicted coiled-coil region followed by a long α-hairpin fold, which is part of a well conserved novel domain between aa 70-140, and consists of two helices separated by a short loop. SYCE1 has four predicted coiled coil regions but no other domains or folds.

Lacking bioinformatic clues to function, we raised antibodies against these proteins. They were raised against a mixture of GST-fusion protein and synthetic peptide, and affinity-purified either with the immunizing peptide or with IgGs that had been purified using protein G sepharose; antibodies were then tested in PFA-fixed mouse-testis sections. Both methods of purification produced antibody preparations which gave the same signal (data not shown), which was also reproduced in frozen tissue-sections of testis from adult mice (Fig. 1). SYCE1 and CESC1 are present only in meiotic cells, specifically in the synaptonemal complex. This signal is abolished when the antibody is pre-incubated with the peptide or the GST-fusion protein (Fig. 1C,I). Squash preparations of mouse seminiferous tubules incubated with anti-CESC1 or anti-SYCE1 confirmed this localisation (data not shown). To further establish antibody specificity under conditions used in immunocytochemistry, we have stained cells transfected with plasmids expressing SYCP1, SYCE1 or CESC1 with the cognate or the two non cognate antibodies. No cross-reactivity was evident in these experiments (data not shown). Using spread preparations, visualisation of the AEs and or LEs of the chromosomes with antibodies against STAG3, a meiotic cohesin that maintains sister chromatid pairing (Prieto et al., 2001) were performed in combination with antibodies to SYCE1 or CESC1 (Fig. 2). Both SYCE1 and CESC1 localise only to the synapsed part of the synaptonemal complex, while STAG3 stains the length of the AEs and or LEs. This localisation of SYCE1 and CESC1 to paired regions of the SC was replicated in female meiotic chromosomes (supplementary material, Fig. S2). In both male and female spreads, CESC1 shows a more punctate pattern of staining than SYCE1. Staining of SCs with antibodies to both proteins was first evident in zygotene spermatocytes and was progressively lost as synapsis decreased in diplotene spermatocytes.

The distribution of SYCE1 and CESC1 we describe here is indistinguishable from that previously found for SYCP1 (Meuwissen et al., 1992). SYCP1 is a known SC component, which also only occurs in synapsed regions of the SC. To examine the relationship between the localisation of SYCP1, SYCE1 and CESC1 we have used immuno-gold electron microscopy. Fig. 2P and Q show images where gold particles from both antibodies are confined to the CE. By contrast, the control of an antibody directed to the C-terminus of SYCP1 shows the predicted LE staining (data not shown). No signal on the LEs or transverse filaments was evident showing that SYCE1 and CESC1 are entirely located in the CE, unlike SYCP1.

The data presented here show that SYCE1 and CESC1 are two novel proteins highly enriched or specific to gonadal tissue and that localise specifically to meiotic cells. Within these cells, we demonstrate that these proteins are confined to the synapsed regions of the SC. They are the first proteins known to exclusively localise to the CE.

Given the restricted distribution of SYCE1 and CESC1 to the CE, and the known localisation of the N-terminus of SYCP1 to the same region of the SC, we decided to test whether the three proteins are part of the same complex and attempt to map the interaction between them. Some interactions of SYCP1 are already known. RAD51, a recombinase present in early recombination nodules, has been shown to interact with SYCP1 in yeast-two-hybrid and in vitro pull-down assays (Tarsounas et al., 1999). Also, FKBP6 has recently been shown to be part of a complex with SYCP1 in mouse testis-extracts but has not been shown to interact directly (Crackower et al., 2003). We have explored the interactions of SYCP1 and SYCE1 using immunoprecipitation. When incubated with a testis extract, antibodies directed against SYCP1 recognise a complex containing SYCE1 (Fig. 3A). Within this complex the interaction between SYCP1 and SYCE1 may be indirect, and dependent on other proteins. We have looked for further interactions relevant to the assembly of the complex in an invitro system, using GST-SYCE1 and GST-CESC1 fusion proteins in a pull-down assay with ³⁵S-labelled proteins synthesised in a reticulocyte lysate system. As predicted from electron-microscopic localisation, the N-terminal 200 amino acids (aa) of SYCP1 interacts with both SYCE1 and CESC1 (Fig. 3B). The GST-CESC1 fusion protein interacts with labelled SYCE1 protein (Fig. 3C) and both SYCE1 and CESC1 show homotypic interaction in this assay (Fig. 3C). In this system CESC1 is not pulled down by GST-SYCE1 although this fusion protein can interact with ³⁵S-labelled SYCE1. The reciprocal interaction of GST-CESC1 is detectable, suggesting that the structure of the GST-SYCE1 fusion protein is altered compared to the native protein. These interactions are consistent with the proteins containing coiled-coil motifs, which are thought to be favourable for protein-protein interactions and with our two hybrid experiments. Half the number of clones from yeast two-hybrid experiments designed to look for proteins that interact with CESC1 are themselves derived from CESC1 (data not shown). Similar two-hybrid experiments with SYCE1 gave a small number of SYCE1-derived interactors. We propose that this set of binary interactions is capable of building a major part of the CE.

Since mammalian meiosis cannot be simply assayed in culture, we have used a mitotic system to confirm the nature of the interactions between SYCE1, CESC1 and SYCP1. Interactions between the axial element proteins SYCP2 and SYCP3 have been established using this approach (Pelttari et al., 2001). Significantly, when SYCP1 is expressed in somatic cells it forms cytoplasmic aggregates (Yuan et al., 1996) and fibres or polycomplexes with a structure at the EM level similar to stacks of SCs (Öllinger et al., 2005). Polycomplexes, of unknown composition but of similar morphology, are found in the germ cells of some species (Raveh and Ben Ze'ev, 1984). They have been thought to represent storage deposits for SC proteins. Expression of SYCE1 in COS-7 cells produces cytoplasmic aggregates and nuclear particles (Fig. 4A). When both SYCP1 and SYCE1 are expressed in the same cell, the SYCE1 signal uniformly decorates the fibres and particles formed by SYCP1 (Fig. 4B-D). This indicates that SYCP1 is recruiting SYCE1 to the polycomplexes. An enhanced-green-fluorescence-protein (EGFP)-SYCE1 fusion protein coexpressed with SYCP1 shows the same localisation of the EGFP signal, ruling out the possibility that this association could be an artifact owing to antibody cross-reactivity (data not shown). This not only confirms the interaction between SYCP1 and SYCE1 found in testis extracts, but suggests that in meiotic cells SYCP1 could be directly responsible for recruitment of SYCE1 to the CE of the synaptonemal complex. Parallel experiments, co-transfecting SYCP1- and CESC1-expression constructs (Fig. 4F-G), gave a similar result. CESC1 alone gives a diffuse pattern of localisation in both nucleus and cytoplasm (Fig. 4E). When coexpressed with SYCP1, the cytoplasmic fibres formed are decorated with the CESC1 protein, again suggesting that SYCP1 could be recruiting CESC1 to the CE.

Mutation of the Sycp3 gene in mouse has been shown to disrupt completion of meiosis. In males, cells are arrested at the zygotene stage of prophase I (Yuan et al., 2002). In females, oocytes progress through meiosis, although at the expense of increased rates of aneuploidy (Yuan et al., 2002). Spermatocytes enter meiotic prophase, but the assembly of the SC is impaired. Labelling of SYCP1 in spermatocytes revealed that although SYCP1 still localised to fibrillar structures resembling SCs (Liebe et al., 2004), these were generally shorter than normal SC structures, showed gaps and did not associate with the centromeres (Yuan et al., 2000). The same pattern was also observed in females, except for the length of the fibrillar structures; they were longer in the mutants compared to the wild-type animals (Yuan et al., 2002).

To gain further insight on the association of SYCE1 and CESC1 with SYCP1 and the role of SYCP1 in recruiting these proteins to the CE, immunofluorescence analysis of SYCE1 or CESC1 together with SYCP1 was performed in Sycp3 null spermatocytes and oocytes. Male and female germ cells were collected from adult testes and foetal ovaries (18 d.p.c.) and used to prepare spread preparations. These preparations were then stained for SYCP1 and SYCE1 or CESC1. As observed in wild-type spermatocytes, the staining for SYCP1 was mimicked by that of SYCE1 and CESC1, with full colocalisation of the proteins (Fig. 5A-C). Short SC structures with gaps as described for SYCP1 were also observed. The same colocalisation was observed in 18 d.p.c. oocytes, where long SC structures with gaps were visible (data not shown).

Co-transfection of SYCP1 and SYCE1 or CESC1 in COS-7 cells previously suggested a role for SYCP1 in the recruitment of these proteins to the CE of the SC. Evidence that in meiotic Sycp3-null cells, defective localisation of SYCP1 coincides with a defective localisation of SYCE1 and CESC1 further implies the dependence of these proteins from SYCP1 in such event. In Sycp1 null animals SYCE1 is delocalised from the synaptonemal complex (C. Heyting, personal communication), confirming this conclusion.

We have described SYCE1 and CESC1, two new proteins that localise exclusively to the CE of the synaptonemal complex. These proteins do not have any specific motifs with known function, but contain coiled-coils and in the case of CESC1, an α-hairpin fold, the type of secondary structures normally associated with the mediation of protein-protein interactions. SYCP1, the partner of these novel proteins, also contains multiple coiled-coils in its secondary structure.

When expressed at high levels, this protein can assemble into structures similar to SCs, the so-called polycomplexes (Öllinger et al., 2005). Polycomplexes share a very similar ultrastructure with naturally occurring SCs, forming LE-like and CE-like structures and suggesting the importance of SYCP1 in the establishment of synapsis. As shown in Fig. 4, coexpression of SYCP1 in a heterologous system with either SYCE1 or CESC1 indicates that SYCP1 directly recruits SYCE1 and CESC1. In vivo, SYCP1 also appears to be playing the same role, as suggested by analysis for SYCP3 null mice. So what could be the sequence of events leading to the assembly of the CE and hence to synapsis? It is possible that, as the transverse filaments are assembled and start forming the characteristic ladder-like structure found in the central region, they directly recruit SYCE1 and CESC1, which themselves are highly expressed at this stage of meiosis. Since all three proteins interact with themselves and each other, we suggest a model to describe their distribution in the CE of the SC (Fig. 6). The interdigitation of the N-terminal globular domain of SYCP1 is visible in the CE of the SC; in our model a complex formed by a dimer of CESC1 and two dimers of SYCE1 links each multimer of SYCP1 dimers. Stoichiometry still needs to be determined for these proteins, so this is just one of the several models that could be proposed. We hypothesise that SYCE1 and CESC1 are involved in the formation and stabilisation of the SC. In this model SYCE1 and CESC1 serve to separate and reinforce the transverse filaments formed by SYCP1, conferring added stability to the entire structure of the SC. Mechanical stress may be important in regulating recombination processes and might require a more robust and/or flexible SC than provided by SYCP1 alone. Borner and colleagues (Borner et al., 2004) have recently proposed a model for SC-assisted recombination. In yeast, full length SCs were shown to occur simultaneously with the third transition in the crossing-over pathway, from single-end invasion (SEI) to double Holliday junction (dHJ) (Hunter and Kleckner, 2001). In the model proposed by Borner and co-workers, mechanical forces or stress generated by chromatin-loop-expansion would cause the SC to twist, bringing the axes of the chromatids involved in recombination into close proximity. Twisting of the SC has been frequently reported in mammals and other organisms (Moens, 1978; Borner et al., 2004; Blat et al., 2002). This would then allow SC-associated recombination complexes – the `weak points' along the structure – to target and transduce axis stress into local changes. In essence, SC-twisting could be coordinating local changes at both the DNA level (the SEI to dHJ transition) and between chromatid axes (axis exchange) (Borner et al., 2004; Moens, 1978; Moens, 1974). We speculate that, within the context of this model, SYCE1 and CESC1 could be primarily involved in assembling the SC as a robust structure that endures the mechanical stress but is also sufficiently flexible to twist at the sites of recombination.

Reciprocal BLAST searches of all completed eukaryotic genomes (including human, rat, dog, pig, cow, chicken, frog, zebrafish, trout, pufferfish, C. elegans, Drosophila, S. pombe and S. cerevisiae) primarily reveal orthologues in vertebrates. Why are these CE proteins only conserved in vertebrates? We do not yet know the answer but it is clear from a comparison of other SC components, such as SYCP1, SYCP2 and SYCP3, that there is little sequence conservation between them and their functional homologues in lower eukaryotes. Instead, the overall secondary structure of the proteins is conserved. In the case of SYCP1, although at sequence level there is very little similarity, the functional homologues in Drosophila, C. elegans and S. cerevisiae all form extensive coiled coils (Sym et al., 1993; Page and Hawley, 2004; MacQueen et al., 2002). Proteins structurally equivalent to SYCE1 and CESC1 are likely to exist in a wide range of organisms. Despite this lack of obvious homologues in invertebrate model organisms, SYCE1 and CESC1 are probably important for mammalian recombination and fertility.

We thank Christa Heyting, Rolf Jessberger and Peter Moens for the gift of antibodies; Mary Taggart, Paul Perry and Sandy Bruce for assistance with mice stocks and images and members of all labs for helpful discussions. This work was supported by the Medical Research Council and a DFG grant to R.B. (Be 1168/6-1). Y.C. is supported by the Portuguese grant PRAXIS XXI (BD/21802/99).