Identification of novel synaptonemal complex components in C. elegans

Sexually reproducing organisms rely on the correct execution of meiosis, a specialized cell division that produces haploid gametes to transmit genetic information faithfully from parent to progeny. Errors in this process lead to the production of offspring with an abnormal number of chromosomes, or aneuploidy, which is a major cause of miscarriages and congenital conditions such as Down syndrome (Hassold et al., 2007). Separation of homologous chromosomes during meiosis requires that each chromosome pairs with its homologue and forms a physical linkage during meiotic prophase for their biorientation. The physical connections between homologous chromosomes are established as a result of crossover recombination and are visibly shown as chiasmata. Despite the overall conservation of the meiotic program, including the stereotypical changes chromosomes undergo to achieve biorientation, the molecular players carrying out these functions exhibit a surprising degree of divergence among different lineages.

A striking example of this evolutionary pattern is a zipper-like protein structure called the synaptonemal complex (SC) that reinforces homologue pairing in most eukaryotes (Page and Hawley, 2004). First observed by electron microscopy >60 yr ago (Fawcett, 1956; Moses, 1956), the SC is a hallmark of meiotic prophase and acts as a scaffold for meiotic recombination. The SC consists of two parallel axial elements that form between sister chromatids and a central region with transverse filaments that connect the paired chromosomes. The overall appearance of the SC is highly conserved with respect to its basic dimensions and organization (Zickler and Kleckner, 1999), suggesting that its structural features are essential for interhomologue interactions. Interestingly, the proteins comprising the SC have diverged extensively during evolution. Although genetic and cytological studies have identified SC proteins in various organisms, how individual components interact with each other to form the regular, repetitive arrangement of the SC is largely unknown.

Meiotic recombination initiates with programmed DNA double-strand breaks (Szostak et al., 1983). While excess DNA double-strand breaks are generated during meiotic prophase, only few are ultimately selected to become sites of crossovers (Martinez-Perez and Colaiácovo, 2009). These crossovers are nonrandomly distributed far apart from each other, a phenomenon known as crossover interference (Muller, 1916). Since recombination occurs within the context of the SC, it has been proposed that the presence of crossovers could be communicated through the SC to limit crossover formation (Libuda et al., 2013; Sym and Roeder, 1994). The nematode Caenorhabditis elegans represents an extreme case of this regulation where exactly one crossover forms per homologue pair (Hammarlund et al., 2005; Hillers and Villeneuve, 2003; Nabeshima et al., 2004; Yokoo et al., 2012). Recent evidence in C. elegans suggests that the SC has liquid-crystalline properties, which enable long-range signal transduction to mediate a chromosome-wide crossover control (Rog et al., 2017). Further support for crossover control by the SC comes from evidence that the presence of crossover-designated sites also influences the dynamic state of the SC (Libuda et al., 2013; Machovina et al., 2016; Pattabiraman et al., 2017; Woglar and Villeneuve, 2018). However, the molecular mechanisms by which the SC regulates crossover still remain poorly understood.

The SC in C. elegans exhibits a typical tripartite structure (Goldstein and Slaton, 1982), and its assembly is essential for crossover formation (MacQueen et al., 2002). Axial elements in C. elegans comprise meiotic cohesins and four paralogous HORMA domain proteins, High Incidence of Males-3 (HIM-3), Him-Three Paralog-1 (HTP-1), HTP-2, and HTP-3 (Couteau et al., 2004; Couteau and Zetka, 2005; Goodyer et al., 2008; Kim et al., 2014; Martinez-Perez and Villeneuve, 2005; Zetka et al., 1999). Once chromosomes have paired, the central region of the SC assembles. In C. elegans, this structure is known to consist of four coiled-coil proteins, SYnaPsis in meiosis abnormal-1 (SYP-1), SYP-2, SYP-3, and SYP-4, which are mutually dependent for SC assembly (Colaiacovo et al., 2003; MacQueen et al., 2002; Smolikov et al., 2007, 2009). These SYP proteins were identified more than a decade ago by forward genetic or yeast two-hybrid screens, and it has been widely believed that this represents the entire set of SC components in C. elegans.

Here, we report the identification of two novel SC components, SYP-5 and SYP-6, in C. elegans. They transverse the width of the SC, and their recruitment to the SC depends on chromosome axes and other SC components. Interestingly, SYP-5 and SYP-6 are paralogous to each other and play redundant roles in synapsis, which may account for why these proteins have not been identified by previous genetic screens. Using genetic redundancy and structure–function analysis of SYP-5 and SYP-6, we further demonstrate the role of the SC in both limiting and promoting crossover formation. Together, our findings establish SYP-5 and SYP-6 as bona fide components of the SC and shed light on its structure and function.

To gain insight into the organization of the SC, we purified SYP-3–containing protein complexes from a worm strain expressing GFP-SYP-3 (ieSi11; Rog and Dernburg, 2015) and analyzed the SC-associated proteins by mass spectrometry (Fig. 1 A). As expected, the four previously known SYP proteins were specifically purified with GFP-SYP-3 (Fig. 1 B). Few other proteins were also enriched in the GFP-SYP-3 immunoprecipitates (Table S1). Among these, we focused on two previously uncharacterized proteins, Y54E10A.12 and F57B10.4, which are homologous to each other, sharing 29% overall sequence identity (Fig. 1 C). These two proteins are highly expressed in the germline (Kim et al., 2001; Owen et al., 2003) and contain predicted coiled-coil domains (Figs. 1 C and S1 A), a characteristic shared by all known SC components in diverse eukaryotes (de Vries et al., 2005; MacQueen et al., 2002; Page and Hawley, 2001; Sym et al., 1993). Further, they both possess low-complexity sequences in their C-termini, which are conserved in homologues across Caenorhabditis species. Based on these characteristics, we hypothesized that Y54E10A.12 and F57B10.4 are novel components of the SC.

To determine whether these proteins are indeed SC components, we first inserted small epitopes, 3xFlag and HA, to the N-terminus of Y54E10A.12 and F57B10.4, respectively, using CRISPR-mediated genome editing and examined their localization by immunofluorescence. Y54E10A.12 and F57B10.4 were initially colocalized within SC-like protein aggregates in the nucleoplasm known as polycomplexes (Roth, 1966; Fig. 2, A and B). Both proteins then loaded onto chromosomes as chromosome pairs and were found along the entire length of chromosomes in pachytene, which perfectly mirrors the localization of SYP-2 (Fig. S1, B and C). These data suggest that Y54E10A.12 and F57B10.4 are novel components of the SC; thus, hereafter, we refer to the genes that encode them as syp-5 (Y54E10A.12) and syp-6 (F57B10.4), reflecting their physical order on chromosome I (Fig. 1 D). We note that the level of SYP-6 decreased at the end of pachytene and was no longer detected in diplotene (Fig. S1 C). In contrast, SYP-5 was detected on the “short arm” of the SC in diplotene relative to the crossover site, which defines the site of cohesion loss in meiosis I (Kaitna et al., 2002; Rogers et al., 2002), and persisted until the SC fully disassembled in diakinesis (Fig. S1, B and D), similar to the other SYP proteins in C. elegans (Colaiacovo et al., 2003; MacQueen et al., 2002; Smolikov et al., 2007, 2009).

Interestingly, the C. elegans genome contains another SYP-5/6-related gene, K09H9.1, on chromosome I (Fig. 1 D). The first two exons of K09H9.1 are pseudogenic fragments of syp-5, and the last exon appears to be a pseudogenic fragment of pha-1, a gene predicted to have fucosyltransferase activity. Consistent with this, we did not detect the expression of K09H9.1 in the germline when it was tagged with a small epitope at its N-terminus (not depicted). Thus, we conclude that K09H9.1 is a pseudogene derived from a recent gene duplication event in C. elegans (see Discussion).

To establish SYP-5 and SYP-6 as components of the SC, we visualized Flag-SYP-5 and HA-SYP-6 relative to the axial element protein HIM-3 using nuclear spreading and stimulated emission depletion (STED) microscopy. HIM-3 formed two parallel stretches in pachytene nuclei, which were found to be separated by 135 ± 21 nm (Fig. 2, C and D). While this is larger than the previous measurement using single-molecule localization microscopy (SMLM; Köhler et al., 2017), our data are within the range measured using a similar spreading technique (Woglar and Villeneuve, 2018). The N-terminal epitopes for SYP-5 and SYP-6 were localized between the parallel HIM-3 stainings, demonstrating that SYP-5 and SYP-6 are indeed components of the SC central region.

Given the extent and the predicted length of the coiled-coil domains of SYP-5 and SYP-6, it is possible that a pair of SYP-5/6 spans the width of the SC, similar to SYP-1 (Schild-Prufert et al., 2011; Woglar and Villeneuve, 2018). Thus, we sought to determine their orientations within the SC using superresolution microscopy techniques. To this end, we used CRISPR to generate worm strains expressing a C-terminal HA tag on either SYP-5 or SYP-6 (Fig. S2 A). By STED microscopy, both SYP-5-HA and SYP-6-HA appeared as a single stretch between the chromosome axes (Fig. S2, B and C), similar to the results obtained using the N-terminal HA-tagged proteins. However, the lateral resolution of STED microscopy (Schermelleh et al., 2019) is close to the width of the SC and may not be sufficient to resolve the two ends of SYP-5 and SYP-6.

We thus employed 3D SMLM and computational analyses to precisely map the positions of the SYP-5 and SYP-6 N- and C-termini relative to HIM-3. Similar approaches have recently been used to determine the organization of chromosome axes in C. elegans (Köhler et al., 2017). We estimated the resolution of our SMLM to be 40–60 nm by Fourier ring correlation (Nieuwenhuizen et al., 2013; Fig. S3, A–D), which is sufficient to resolve substructures within the SC. Averaging multiple SC stretches from SMLM images revealed that the HIM-3 axes were separated by ∼128 nm (Fig. 3 Cand Fig. S3, F and G), which is consistent with the distance measured by STED. However, a clear distinction emerged between the positions of HA-SYP-5/6 and SYP-5/6-HA. While the N-termini of SYP-5/6 were localized at the center of the SC (Figs. 3 Aand S3 G), two distinct stretches of SYP-5-HA and SYP-6-HA were found adjacent to HIM-3, separated by 84 nm in the frontal view (Fig. 3, B and C; and Fig. S3, F and G). Therefore, the C-termini of SYP-5/6 are separated by ∼42 nm from their N-termini and are positioned ∼21 nm away from the HIM-3 axis. We also note that both ends of SYP-5/6 were confined to a narrow z plane in the cross-sectional view (Fig. 3, A and B, right), with half width at half maximum (HWHM) of 40∼50 nm (Fig. S3 F). Together, our results demonstrate that both SYP-5 and SYP-6 span the width of the SC in a head-to-head manner, with the two N-termini positioned at the center and the C-termini facing the axes (Fig. 3 D).

To further investigate the behavior of SYP-5 and SYP-6, we raised polyclonal antibodies against the last 14 amino acids of SYP-5 and SYP-6. The affinity-purified antibodies robustly stained the SC in the hermaphrodite germline, which mirrors the localization of Flag-SYP-5 and other SYP proteins (Fig. 4 A). To determine the specificity of the antibodies, we generated null alleles for syp-5 and syp-6 by targeting their first exons using CRISPR (Fig. S4 A). The antibody raised against the SYP-5 peptide was specific to SYP-5, as the signal on the SC was abolished in syp-5 mutants, but not in syp-6 mutants (Fig. S4 B). On the other hand, the antibody raised against the SYP-6 peptide was able to stain the SC in syp-5 or syp-6 single mutants, while the signal was completely eliminated in syp-5 syp-6 double mutants (Fig. S4 C). Thus, the putative SYP-6 antibody recognizes both SYP-5 and SYP-6, and we hereafter refer to it as the SYP-5/6 antibody. Interestingly, the signal for SYP-6, as visualized by the SYP-5/6 antibody in syp-5 mutants, was diminished at the end of pachytene (Fig. S4 D), confirming our earlier observations with the HA-tagged SYP-6 (Fig. S1 C).

Throughout eukaryotes, SC assembly requires chromosome axes (Colaiacovo et al., 2003; Couteau et al., 2004). As expected, SYP-5 and SYP-6 failed to load onto chromosomes in mutants lacking an axis component HIM-3 and instead formed polycomplexes together with the other SYPs (Fig. 4 B). Moreover, SYP-5 and SYP-6 were no longer detected in syp-2 mutants (Fig. 4 B), indicating that SYP-5 and SYP-6 depend on other SC proteins for localization, as is the case for the other SYP proteins (Smolikov et al., 2007, 2009). Taken together, our data demonstrate that SYP-5 and SYP-6 share characteristics found in other SC components.

In their coding regions, SYP-5 and SYP-6 share 52% sequence identity at the nucleotide level and 29% sequence identity at the protein level. Protein basic local alignment search tool (BLAST) search of C. elegans using SYP-5 as a query yields SYP-6 as the top hit, other than SYP-5 itself, suggesting that SYP-5 and SYP-6 are paralogs. To determine the evolutionary origins of SYP-5 and SYP-6, we performed translated nucleotide BLAST (tBLASTn) searches of 10 additional Caenorhabditis species using C. elegans SYP-5 and SYP-6 as queries. All queried species contain a single, high-confidence, BLAST hit. Examination of the syntenic locus of each syp-5/6-related orthologue revealed that these genes are present in the same locus as C. elegans syp-5 (Fig. 5 A). Thus, we refer to each of these syp-5/6-related genes as syp-5. We also generated a maximum-likelihood phylogeny based on a protein alignment of SYP-5 and SYP-6 sequences from 11 Caenorhabditis species. C. elegans SYP-5 and SYP-6 were grouped together with high bootstrap support (Fig. 5 B), indicating that they are likely paralogs born from a recent gene duplication event. Therefore, we conclude that syp-5 was likely present in the common ancestor of the Elegans and Japonica species groups, making it >30 million years old (Kumar et al., 2017), and that syp-6 is likely to have arisen from a gene duplication event within the C. elegans lineage.

We next examined the effects of deleting syp-5 and syp-6. Self-progeny of hermaphrodites homozygous for the syp-5 null mutation showed significantly reduced viability (34% vs. 100% in N2), and 15% of surviving progeny were males (compared with 0.1% for wild-type; Fig. S5 A). syp-6 single mutants, however, did not show obvious phenotypes in meiosis, producing viable progeny (100%; Fig. S5 A). Males in C. elegans arise from X-chromosome nondisjunction during meiosis (Hodgkin et al., 1979), and the high incidence of males (Him phenotype) is indicative of errors in meiotic chromosome segregation. In contrast to the single mutants, syp-5 syp-6 double mutants exhibited a striking drop in viability of progeny to 3%, with an increase in the Him phenotype to 45% among surviving progeny (Fig. S5 A), which is comparable to the phenotypes reported in syp-1–null mutants (MacQueen et al., 2002). These results indicate that, while SYP-6 itself is dispensable, both SYP-5 and SYP-6 contribute to proper chromosome segregation during meiosis.

Indeed, SYP-5 and SYP-6 were found to play partially redundant roles in SC assembly. In syp-5 mutants, completion of synapsis was delayed, and levels of SYP-2 on chromosomes were reduced compared with wild type (Fig. 6 A). Recent evidence has established that defects in synapsis or crossover formation delay meiotic progression by prolonging the kinase activity of Checkpoint kinase 2 (CHK-2; Kim et al., 2015; Rosu et al., 2013; Stamper et al., 2013), which governs homologue pairing, synapsis, and meiotic recombination in C. elegans (MacQueen and Villeneuve, 2001). In response to the synapsis defects, the activity of CHK-2 was extended in syp-5 mutants, as measured by the percentage of germline-length positive for phosphorylation of CHK-2 substrates, HIM-8 and its paralogs (Zinc finger InMeiosis-1 [ZIM-1], -2, and -3), relative to the length from meiotic onset to the end of pachytene (79% in syp-5 vs. 54% in wild type; Fig. S5, B and C). Consistent with the egg count Fig. S5 A), animals lacking SYP-6 showed no obvious defects in SC assembly and displayed normal meiotic progression (53% “CHK-2 active zone” in syp-6). However, synapsis failed completely in mutants that lacked both SYP-5 and SYP-6 (Fig. 6 A), which further extended CHK-2 activity (84% in syp-5 syp-6 vs. 54% in wild type; Fig. S5, B and C).

C. elegans has six pairs of chromosomes. Thus, six bivalents are typically found in wild-type oocytes at diakinesis. Oocytes in syp-5 mutants exhibited an average of seven DAPI-staining bodies (Fig. 6, B and C), indicative of a failure to form crossover in one homologue pair. In syp-6 mutants, crossovers formed normally for all chromosomes, and six DAPI-staining bodies were visible. However, the number of achiasmate chromosomes dramatically increased to 12 in syp-5 syp-6 double mutants (Fig. 6, B and C), revealing a complete failure to form crossovers. Quantitative Western blot analysis of an HA tag fused to both SYP-5 and SYP-6 showed that SYP-6 is expressed at only 25–30% of the level of SYP-5, which may provide an explanation for the greater functional contribution by SYP-5 (Fig. 6 D).

The SC mediates robust crossover interference in C. elegans, yielding precisely one crossover event per chromosome (Libuda et al., 2013). Thus, we asked whether the redundancy between SYP-5 and SYP-6 could provide a unique opportunity to study the role of the SC in crossover control. Using the protein CrossOver Site-Associated-1 (COSA-1) as a cytological marker (Yokoo et al., 2012), we examined how crossovers are designated in the absence of SYP-5 and/or SYP-6. Compared with the invariant 6 COSA-1 foci in the wild-type nucleus, syp-5 mutants exhibited a wider distribution of four to eight COSA-1 foci (Fig. 6, E and F), supporting a role of the SC in tightly regulating the number of crossovers to one per homologue pair (Hayashi et al., 2010; Libuda et al., 2013). As expected, deletion of syp-6 did not affect the number of COSA-1 foci, while the syp-5 syp-6 double mutants displayed two to six COSA-1 foci despite the absence of chiasmata, consistent with recent evidence shown for other mutants completely lacking the SC (Cahoon et al., 2019; Woglar and Villeneuve, 2018). Therefore, recombination intermediates in the syp-5 syp-6 double mutants are capable of recruiting crossover-promoting enzymes such as COSA-1; however, those sites cannot mature into crossovers without the SC, and no chiasmata are observed. Taken together, our data support the role of the SC in both limiting and promoting crossover formation.

SYP-5 and SYP-6 contain negatively charged regions in their disordered C-termini, which are highly conserved across Caenorhabditis species (Fig. 7 A). To determine the significance of these regions in SC assembly, we generated three truncation alleles of syp-5 and syp-6 that sequentially remove acidic patches from their C-termini by inserting premature stop codons (Fig. 7 A). Truncated SYP-5 proteins were expressed at similar levels to the full-length protein, except in the strain with the largest deletions of SYP-5 and SYP-6 (SYP-5^{Δ50 aa} SYP-6^{Δ53 aa}). SYP-5^{Δ50 aa} was not detectable by Western blot but nevertheless was visible in the germline by immunofluorescence (Figs. S5 E and 7 C). Truncations led to commensurate reduction in egg viability from 99% in wild type to 7% in syp-5^{Δ50 aa} syp-6^{Δ53 aa}, and the percentage of males among surviving progeny also increased from 2.1% in wild type to 30% in syp-5^{Δ50 aa} syp-6^{Δ53 aa} (Figs. 7 B and S5 D).

The degree of truncations correlated with the severity of synapsis and crossover defects. SC assembly was greatly delayed, even in syp-5^{Δ2 aa} syp-6^{Δ2 aa} mutants missing just the last two aspartic acid residues from SYP-5 and SYP-6, and extending the deletions further exacerbated synapsis defects (Figs. 7 C and S5 F). In particular, syp-5^{Δ50 aa} syp-6^Δ53 aa mutants exhibited severe phenotypes such that only one or two SC stretches were found in the nucleus, demonstrating that the C-terminal tails in SYP-5/6 are critical for SC assembly. The number of crossover-designated sites was greatly reduced in all three truncation mutants, and COSA-1 foci were found specifically on synapsed chromosomes (Fig. 7, D and E), reflecting the role of SC in promoting crossover recombination. Surprisingly, more than one COSA-1 focus was frequently detected on a given SC stretch in syp-5^{Δ50 aa} syp-6^{Δ53 aa} mutants (Fig. 7, D and F), suggesting that crossover control is compromised when the C-terminal acidic patches are removed entirely from SYP-5/6.

To further investigate the contribution of the C-terminal tails of SYP-5/6 on crossover control, we examined the number and distribution of COSA-1 foci along the two-chromosome fusion mnT12 (IV;X) in the syp-5^{Δ50 aa} syp-6^{Δ53 aa} background (Fig. 8 A). The increased length of the mnT12 chromosome frequently permits the formation of double crossovers, which enables the analysis of crossover interference (Hillers and Villeneuve, 2003; Libuda et al., 2013). While one or two COSA-1 foci were observed on mnT12 in wild type as previously reported (Libuda et al., 2013; Yokoo et al., 2012), syp-5^{Δ50 aa} syp-6^{Δ53 aa} mutants displayed 2–5 COSA-1 foci along synapsed mnT12 (Fig. 8, B and C). 3D tracing revealed that the average distance between COSA-1 foci on mnT12 was greatly reduced in syp-5^{Δ50 aa} syp-6^{Δ53 aa} animals (3.6 µm), compared with the wild type (7.9 µm). Even in instances where mnT12 had two COSA-1 foci in syp-5^{Δ50 aa} syp-6^{Δ53 aa} mutants, the average distance between them was significantly smaller (4.6 µm) than that in controls (Fig. 8 D). To determine the effect of deleting SYP-5/6 C-termini on crossover interference, we divided the mnT12 chromosome into 10 segments (A–J, from the X-chromosome pairing center protein, HIM-8 (Phillips et al., 2005) and binned the position of COSA-1 foci into these segments. In wild-type animals, either a single COSA-1 focus was detected in the middle of mnT12, or two COSA-1 foci were observed far apart from each other at opposite ends of the chromosome (Fig. 8 E), consistent with strong crossover interference (Libuda et al., 2013). By contrast, COSA-1 foci were randomly distributed along mnT12 in syp-5^{Δ50 aa} syp-6^{Δ53 aa} mutants (Fig. 8 F), indicating that interference is impaired in the absence of SYP-5/6 C-terminal tails.

We also determined whether the additional COSA-1 foci on mnT12 in syp-5^{Δ50 aa} syp-6^{Δ53 aa} mutants are resolved into multiple chiasmata, based on the morphology of HTP-3 and the number of SYP-enriched domains in diakinesis nuclei (Martinez-Perez et al., 2008). In wild-type nuclei, either cruciform structures (one chiasma) or ring-like structures (two chiasmata) were observed in mnT12 (Fig. 8 G). By contrast, the majority of mnT12 in syp-5^{Δ50 aa} syp-6^{Δ53 aa} mutants displayed ring-like structures with two chiasmata (58%) or distorted structures with three or more chiasmata (37%; Fig. 8, G and H). Thus, the excess COSA-1 foci observed in pachytene nuclei in the syp-5^{Δ50 aa} syp-6^{Δ53 aa} background indeed correspond to increased numbers of chiasmata.

Here we identify two paralogous proteins, SYP-5 and SYP-6, as components of the SC central region in C. elegans. Remarkably, these two genes have evaded identification by genetic screens for more than a decade (Colaiacovo et al., 2002; Srayko et al., 2000), perhaps because loss-of-function mutations of syp-5 or syp-6 alone do not cause strong phenotypes. Our biochemical approach has overcome these challenges and enabled the discovery of these two proteins, which function redundantly to promote synapsis.

Our phylogenetic analysis shows that syp-5 has been duplicated in C. elegans to yield syp-6 and a pseudogene, K09H9.1. Duplication of SYP-5 is expected to be unfavorable, as this would lead to stoichiometric imbalances among the SC components. Indeed, the majority of duplicated genes experience strong purifying selection, especially when the gene product is involved in macromolecular complexes (Lynch and Conery, 2000; Papp et al., 2003). The pseudogene K09H9.1 is likely to have resulted from silencing of a recent syp-5 duplicate and has acquired degenerative mutations to become nonfunctional. We speculate that several adaptations have helped preserve syp-6 in C. elegans and minimize the dosage imbalance caused by gene duplication. First, SYP-5 and SYP-6 have distinct expression levels. By Western blotting the identical epitope fused to SYP-5 and SYP-6, we have shown that the level of SYP-6 is greatly reduced compared with SYP-5 (Fig. 6 D). In addition, the expression of SYP-6 decreases abruptly in late pachytene, while SYP-5 persists on the SC short arm until its complete disassembly in diakinesis (Figs. S1 and S4). The mechanism underlying the distinct regulation of SYP-6 is unclear. Although SYP-6 itself is dispensable for meiosis, it nevertheless contributes to synapsis and crossover formation in the absence of SYP-5. Thus, duplication of syp-5 may buffer against the effect of harmful mutations and provide a selective advantage to retain both copies, providing an example of subfunctionalization of duplicated genes (Force et al., 1999; Lynch and Force, 2000).

Using superresolution microscopy, we demonstrate that SYP-5 and SYP-6 span the width of the SC with their N-termini localized at the center and the C-termini facing the chromosome axes (Fig. 3). This suggests that the C. elegans SC is unique in containing multiple transverse filament proteins: SYP-1 and SYP-5/6 (Schild-Prufert et al., 2011; Woglar and Villeneuve, 2018). SYP-1 and SYP-5/6 show no homology other than short stretches (<10 amino acids) within their coiled-coil domains. Interestingly, the extent of coiled-coils and the overall size of SYP-1 and SYP-5/6 are significantly smaller than in other transverse filament proteins (e.g., yeast Zip1, Drosophila C(3)G, Arabidopsis ZYP1, and mammalian SYCP1), despite the conserved appearance of the C. elegans SC on electron micrographs (Colaiacovo et al., 2003; Rog et al., 2017). Thus it is unclear whether SYP-1 and SYP-5/6 interact with each other to form a homo- or heterotetramer through their N-terminal domains, as shown for SYCP1 (Dunce et al., 2018; Liu et al., 1996), or form tetramers through their coiled-coils, as suggested for Zip1 (Dong and Roeder, 2000). Alternatively, the N-termini of SYP-1 and SYP-5/6 could be connected indirectly by other SYP proteins comprising the central element. Nevertheless, the head-to-head orientation has been observed for all known transverse filament proteins from diverse eukaryotes (Anderson et al., 2005; Cahoon et al., 2017; Dong and Roeder, 2000; Schücker et al., 2015), suggesting that this architecture of the SC is widely conserved.

Our work reveals conserved motifs within SYP-5/6 that are crucial for SC assembly. Truncating the C-terminal acidic sequences of SYP-5/6 causes severe synapsis defects, and extending the deletion results in progressively fewer SC stretches (Fig. 7), suggesting that the overall charge of SYP-5/6 C-termini might be critical for proper synapsis. Previous evidence from yeast and Drosophila has shown that the C-terminal domain of transverse filament proteins is essential for tethering the SC to chromosome axes (Jeffress et al., 2007; Tung and Roeder, 1998). However, robust SC stretches do form in our C-terminal truncation mutants of SYP-5/6, which could be attributed to the presence of the additional transverse filament SYP-1 in C. elegans. Interestingly, our SMLM data show that the C-termini of SYP-5/6 are separated by ∼21 nm from the most proximal axis component, HIM-3 (Köhler et al., 2017), suggesting that SYP-5/6 do not tightly associate with the chromosome axis. Consistent with this, we did not find any axis components in our SYP-3 immunoprecipitates and vice versa from our previous purification of the chromosome axis (Kim et al., 2014). Thus, how the SC interfaces the chromosome axis and preferentially assembles between the axes remains unknown. Moreover, whether and how the C-terminal tails of SYP-5/6 are involved in protein–protein interactions with other SYP proteins will be important questions for the future.

Using genetic redundancy and structure–function analyses to truncate C-terminal tails of SYP-5/6, we have provided evidence supporting the role of SC in crossover regulation. Increased incidence of crossover designation in syp-5 mutants is very much in line with the phenotypes observed when SYP-1 levels are reduced by partial RNAi (Hayashi et al., 2010) and supports the role of SC in limiting crossovers to one per homologue pair (Libuda et al., 2013). In syp-5 syp-6 double mutants, crossover-promoting enzymes are found to be recruited to multiple nuclear foci, despite the fact that no crossovers form in these mutants (Fig. 6). This is in agreement with recent evidence that the SC directs crossover-promoting activities and facilitates interhomologue recombination (Cahoon et al., 2019; Rog et al., 2017), potentially by confining crossover-designated sites in a bubble-like structure (Woglar and Villeneuve, 2018). Synapsis defects caused by the C-terminal truncations of SYP-5/6 correlate with a reduced number of crossover designations, and crossover factors are always found on synapsed chromosomes (Fig. 7 D), reflecting the role of the SC in promoting crossover formation (Cahoon et al., 2019). However, deletion of the entire acidic C-terminal tail causes stronger defects in SC assembly and increases the incidence of multiple crossover intermediates per SC stretch, indicating that the chromosome-wide signaling to control crossover number is impaired in these mutants.

Recent evidence suggests that the SC has liquid crystalline-like properties, thereby serving as a phase-separated compartment to mediate a chromosome-wide crossover control (Rog et al., 2017). Many examples of phase-separated cellular bodies are characterized by the presence of intrinsically disordered proteins with stretches of low-complexity sequences (Shin and Brangwynne, 2017). Indeed, SYP-5 and SYP-6 contain highly conserved low-complexity sequences in their C-termini, which include the acidic patches that are shown here to play important roles in SC assembly (Fig. 7). It remains to be determined whether these regions within SYP-5/6, as well as disordered regions in other SYP proteins, contribute to the liquid crystalline properties of the SC. Future mapping of protein–protein interactions among the SYP proteins and analysis of structural changes within the SC upon crossover designation will provide crucial insights into their roles in crossover regulation.

All strains were maintained on nematode growth medium (NGM) plates at 20°C under standard conditions. N2 Bristol was used as the wild-type strain. The mutations and strains used in this study are summarized in Table S2, Table S3, and Table S4. To score egg viability and male progeny, L4 hermaphrodites were picked onto individual plates and counted immediately after each laying period, and surviving progeny were scored when worms reached adult stage.

C. elegans strain expressing GFP-SYP-3 (ieSi11; Rog and Dernburg, 2015) was synchronously grown in liquid culture at 20°C until worms reach the young adult stage. Animals were harvested by sucrose flotation, snap frozen in liquid nitrogen, and disrupted using a mixer mill (Retsch). To enrich for nuclear fractions, ground worms were incubated in hypotonic buffer (5 mM Tris-HCl, pH 8.0, 40 mM KCl, 1 mM EDTA, 0.3 mM spermidine, 0.15 mM spermine, and Roche protease inhibitor cocktail). Cells were homogenized in a Dounce homogenizer and spun at 100 g for 3 min to separate nuclei from large debris. The supernatant was spun at 2,000 g for 15 min to pellet the nuclei, and the pellet was resuspended in nuclei buffer (10 mM Tris-HCl, pH 8.0, 80 mM KCl, 2 mM EDTA, 0.75 mM spermidine, 0.3 mM spermine, and Roche protease inhibitor cocktail). The nuclei were further purified by centrifugation over a sucrose cushion (30% wt/vol in nuclei buffer) at 2,000 g for 15 min. The pellet containing nuclei was resuspended in lysis buffer (50 mM Hepes, pH 7.4, 1 mM EGTA, 1 mM MgCl₂, 100 mM KCl, and 10% glycerol) containing 5 mM CaCl₂ and treated with 200 units/ml of micrococcal nuclease (NEB) for 1 h at room temperature to digest chromatin. Cleared extracts were used for immunoprecipitation of SYP-3–containing protein complexes using a GFP-binding protein coupled to a HiTrap NHS-activated column (GE Healthcare). Bound proteins were eluted with 0.1 M glycine, pH 2.5, and eluted fractions were immediately neutralized by addition of Tris base. Trypsin-digested samples were analyzed for protein identification by MudPIT.

To generate strains expressing 3Flag-SYP-5, HA-SYP-5, SYP-5-HA, HA-SYP-6, and SYP-6-HA, N2 worms or worms expressing GFP-COSA-1 were injected with 16 µM Cas9 protein complexed with 16 µM trans-activating CRISPR RNA (tracrRNA)/crRNA oligonucleotides (IDT), pCFJ104 (5 ng/µl), and pCFJ90 (2.5 ng/µl), together with a single-stranded DNA oligonucleotide (100 ng/µl; IDT) as a repair template (see Table S2 and Table S3), using an Olympus IX51 inverted microscope fitted with a Narishige micromanipulator and an Eppendorf Femtojet microinjector. F1 progeny were singled to individual plates to self-propagate and were screened for successful edits via PCR.

For generating null alleles of syp-5 and syp-6, N2 worms were injected with 16 µM Cas9 protein complexed with 16 µM tracrRNA/crRNA oligonucleotides (IDT), together with pCFJ104 (5 ng/µl) and pCFJ90 (2.5 ng/µl), to induce DNA double-strand breaks within the first exons of syp-5 and syp-6, respectively.

Polyclonal HIM-3 antibodies were raised against a full-length recombinant HIM-3 protein (Kim et al., 2014) in chicken (Pocono Rabbit Farm) and were purified from the immune serum using the HIM-3 protein coupled to a HiTrap NHS-activated column (GE Healthcare). For SYP-5 and SYP-6 antibodies, synthetic phospho-peptides corresponding to the C-terminus of SYP-5 (CEEGDADQpSIWGSDD; Biomatik) and SYP-6 (CDDGGLDQpTIWDGDD) were coupled to maleimide-activated keyhole limpet hemocyanin (Thermo Fisher Scientific) and used for immunization of rabbits (Pocono Rabbit Farm). Polyclonal SYP-5 and SYP-5/6 antibodies were affinity purified from the immune sera using HiTrap NHS-activated columns (GE Healthcare) coupled to the same SYP-5 and SYP-6 peptides, respectively.

Chromosome spreads were prepared from 20–40 adult hermaphrodites dissected in 5 µl dissecting solution (85% vol/vol HBSS [Gibco] and 0.1% vol/vol Tween-20) on an ethanol-washed 22 × 22–mm coverslip. 50 µl of spreading solution (2.5% wt/vol PFA, 2% wt/vol sucrose, 0.32% vol/vol Lipsol, and 0.04% wt/vol sarcosyl) was added, and dissected gonads were immediately dispersed across the coverslip using a pipette tip. Coverslips were dried at room temperature for >5 h, washed in methanol at −20°C for 20 min, and rehydrated with PBS with Tween (PBST). Samples were then processed for immunofluorescence.

STED microscopy was performed using a Leica TCS SP8 microscope equipped with a 100× oil-immersion, 1.4-NA STED white objective and a Lecia HyD detector. Single z-plane images were acquired at room temperature using Leica LAS-X software. Fluorescent secondary antibodies were systematically tested for optimal multicolor STED imaging: Alexa Fluor 488 was depleted with STED 592-nm laser, and Alexa Fluor 594, CF 633, and Alexa Fluor 647 were depleted with STED 775-nm laser. 100 nm TetraSpeck microsphere beads (Thermo Fisher Scientific) were used to validate the STED efficiency and adjust for thermal drift during image acquisition. Slides were mounted using ProLong Diamond (Invitrogen). Image processing was performed using the Scientific Volume Imaging Huygens Professional software package.

Caenorhabditis syp-5 and syp-6 orthologues were identified by using C. elegans SYP-5 and SYP-6 to query 10 Caenorhabditis species genomes using tBLASTn (Altschul et al., 1997) implemented in WormBase or Caenorhabditis.org genome databases. All queries produced a single, high-confidence hit to an annotated gene. The syp-5 coding sequence in C. afra required manual editing of one intron-exon boundary that caused the resulting protein sequence to be artificially truncated. The C. afra syp-5 sequence was corrected based on alignment to a closely related species, C. sulstoni, and produced a full-length SYP-5 protein after curation. We annotated the syntenic locus of each identified syp-5/6-related gene according to orthologous genes in C. elegans, and syp-5/6-related genes were named according to their shared syntenic location. SYP-5 and SYP-6 protein sequences were aligned using ClustalW (Larkin et al., 2007) implemented in Geneious Prime 2019.0.4. Alignments were further refined manually, including removal of gaps. Maximum likelihood phylogenetic trees of SYP-5 and SYP-6 protein sequences were generated using the LG substitution model in PhyML, implemented in Geneious (Guindon et al., 2010), using 100 bootstrap replicates for statistical support. Phylogenies are rooted on the common ancestor of the Elegans and Japonica species groups.

Germlines were dissected from 24-h post-L4 adults in egg buffer (25 mM Hepes, pH 7.4, 118 mM NaCl, 48 mM KCl, 2 mM EDTA, 5 mM EGTA, 0.1% Tween-20, and 15 mM NaN₃) and fixed in 1% formaldehyde before freezing in liquid nitrogen. Dissected germlines were further fixed in methanol at −20°C for 1 min and rehydrated with PBST. Samples were then blocked with blocking reagent (Roche) for 1 h and incubated with primary antibodies overnight at 4°C. Primary antibodies were used at the following concentrations: FLAG (mouse, 1:500; Sigma F1804), HA (rat, 1:500; Chromotek, 7C9), SYP-2 (rabbit, 1:500; Colaiacovo et al., 2003), SYP-1 (goat, 1:500; MacQueen et al., 2005), HTP-3 (guinea pig, 1:500; MacQueen et al., 2005), HIM-3 (chicken, 1:500; this study), GFP Booster (1:200; Chromotek), HTP-1/2 (rabbit, 1:500; Martinez-Perez et al., 2008), phospho-HIM-8/ZIMs (rabbit, 1 µg/ml; Kim et al., 2015), HIM-8 (rat, 1:500; Phillips et al., 2005), SYP-5 (rabbit, 1:500; this study), and SYP-5/6 (rabbit, 1:500; this study). Slides were washed with PBST three times for 10 min each and incubated with secondary antibodies for 1–2 h. Secondary antibodies labeled with Alexa Fluor 488, 555, 594, or 647 or CF633 were purchased from Invitrogen or Sigma and used at 1:200 dilution. Washed slides were mounted using ProLong Gold (Invitrogen). Images were acquired at room temperature using a DeltaVision Elite system (GE) equipped with a 100× oil-immersion, 1.4 NA objective and a scientific complementary metal–oxide–semiconductor camera (PCO). 3D image stacks were collected at 0.2-µm intervals, processed by iterative deconvolution (enhanced ratio, 20 cycles), and projected using the SoftWoRx package. Composite images were assembled and colored using Adobe Photoshop.

For quantification of GFP-COSA-1 foci, COSA-1 foci were counted in nuclei at least three rows after their first appearance in the germline. Only the foci colocalizing with HTP-3 were counted. mnT12 chromosomes were identified by HIM-8 staining, and continuous SC stretches (visualized by Flag-SYP-5 immunofluorescence) were traced in three dimensions using the Imaris FilamentTracer tool (Bitplane). The positions of COSA-1 foci were recorded along the length of synapsed mnT12 and normalized by dividing the measured length of each mnT12 chromosome into 10 evenly spaced intervals and binning the foci positions into these intervals.

For quantification of chiasmata formation, mnT12 chromosomes were identified in diakinesis nuclei based on their size, and only the ones that were unambiguously distinct from other DAPI bodies were scored for the number of chiasmata (1, 2, or ≥3) based on morphology of HTP-3 staining and the number of FLAG-SYP-5-enriched domains.

For SMLM, coverslips (Precision coverslips, 24 mm, Carl Roth) were cleaned by a 10-min wash in ethanol, rinsed in milli-Q water, and plasma cleaned (PlasmaPrep2, GaLa Instrumente). Clean slides were incubated with 0.01% poly-l-lysine (30,000–70,000 D, Sigma-Aldrich) for 10 min. After incubation, coverslips were rinsed in milli-Q water and dried at room temperature. Poly-l-lysine–coated coverslips were stored at 4°C until use.

Immunofluorescence samples for SMLM imaging were prepared as described above with a few modifications. Worms were immobilized in 0.02% of tetramisole during dissection and fixed with 1% PFA on poly-l-lysine–coated coverslips. Primary antibodies used for SMLM were: anti-HA (1:200, mouse monoclonal 2-2.2.14, Thermo Fisher Scientific), anti-HIM-3 (1:200 or 1:100, rabbit polyclonal 53470002, Novus Biologicals). Secondary antibodies were Alexa Fluor 647 anti-mouse (1:1,000 or 1:500, donkey polyclonal, Jackson Immunoresearch) and CF680 anti-rabbit (1:1,000 or 1:500, goat F(ab)′ fragment, Sigma-Aldrich). The samples were mounted in a custom sample holder and imaged in blinking buffer (50 mM Tris HCl, pH 8, 10 mM NaCl, 10% (wt/vol) d-glucose, 500 µg/ml glucose oxidase, 40 µg/ml catalase, and 35 mM mercaptoethylamine).

SMLM data were acquired at room temperature on a custom-built microscope (Deschamps et al., 2016) equipped with a high-NA oil-immersion objective (160×/1.43 NA; Leica). The sample was excited by a Luxx 638-nm laser coming from a laser combiner (LightHub; Omicron-Laserage Laserprodukte) and an additional booster laser (iBEAM-SMART-640-S, Toptica Photonics). The two lasers were combined by a polarizing beam splitter, focused on a speckle reducer (LSR-3005-17S-VIS; Optotune), and coupled into a multimode fiber (M105L02S-A; Thorlabs). The output of the multimode fiber was first magnified by an achromatic lens and then focused onto the sample. The beam was cleaned from fiber-generated fluorescence with a laser cleanup filter (390/482/563/640 HC Quad; AHF). Emitted fluorescence was separated by a 4× dichroic mirror and split ratiometrically by a 665-nm longpass dichroic mirror (Chroma, ET665lp). After the 665LP dichroic mirror, the transmitted and reflected photons were filtered with a 685/70 (Chroma ET685/70m) and a 676/37 (Semrock, FF01-676/37-25) filter, respectively. The fluorescence emission was recorded by an electron-multiplying charge-coupled device camera (Photometrics, Evolve512D). For 3D SMLM, an astigmatic lens (f = 1,000 mm, Thorlabs) was placed in the emission beam path. The focus was stabilized by totally internally reflecting an infrared laser on the coverslip and detecting it on a quadrant photodiode. The quadrant photodiode was in a closed feedback loop with the piezo objective positioner (Physik Instrumente). A field-programmable gate array (Mojo; Embedded Micro) was used to control the lasers, switch filters, and stabilize the focus. The field programmable gate array is controlled by a custom-written plugin for µManager (Edelstein et al., 2010). Typically, we acquired 100,000–150,000 frames with an exposure time of 15 ms and laser power densities of ∼6 kW/cm².

All postprocessing of raw data was performed using the custom-written software SMAP (Superresolution Microscopy Analysis Platform, https://github.com/jries/SMAP). Before the superresolution experiment, bead stacks were acquired to generate a model of the astigmatic point spread function (Li et al., 2018). To localize the molecules, peaks are first detected by a difference of Gaussian algorithm. These peaks are then fitted globally across both channels with the previously generated point spread function. The z positions of the emitters are linked between the two channels, whereas the x and y coordinates are taken only from the transmitted channel. Subsequently, localizations are filtered by their localization precision and their z position to reject dim and out-of-focus emitters. Emitters that were localized within 35 nm in consecutive frames were grouped into one displayed localization with a maximal dark time of 1 frame. Sample drift was corrected by a custom algorithm based on redundant cross-correlation. For this, the localizations were binned in time, and the superresolved image for each bin was reconstructed. Then, the pairwise cross-correlation of each image against each other image was calculated and fitted by a spline to correct the x and y coordinates. The axial drift was corrected by redundant 1D cross-correlations. The color of the individual fluorophores was assigned according to their relative brightness in the two channels. The superresolved image was reconstructed by plotting Gaussians at the fitted positions with a width proportional to their localization precision.

To map the localization of protein domains within the SC, individual regions of interest corresponding to SCs in frontal view were selected from dual-color SMLM images of late-pachytene nuclei (identified by the presence of bright GFP-COSA-1 foci). To this end, regions of interest containing two clearly separated, parallel, and straight stretches of HIM-3 localizations, corresponding to axes in frontal view, were selected. The total length and number of individual stretches are summarized in Fig. S3 A. Individual stretches were rotated and aligned using a custom script in R (version 3.6.0). To determine the positions of individual protein domains, their localization densities in frontal (x) and cross-sectional (z) views were fitted by unimodal or bimodal Gaussian distributions to obtain peak-to-peak distances (d, for bimodal distributions only) and widths (HWHM). SDs of these parameters were determined by a bootstrapping approach (Köhler et al., 2017).

Statistical analysis was performed using Prism 8 (GraphPad). For the analysis of the distance between COSA-1 foci in mnT12 fusion chromosomes (Fig. 8 D), P values were calculated using a two-tailed Mann–Whitney U test, which is a nonparametric statistic ideal for such datasets. For the analysis of CHK-2 active zone shown in Fig. S5 C, normality test was performed using the D’Agostino–Pearson method, and P values were calculated using an ordinary one-way ANOVA. The number of data points (n) is indicated in the figure legends.

Fig. S1 shows the coiled-coil prediction and germline expression of SYP-5 and SYP-6. Fig. S2 shows STED microscopy images demonstrating that the C-termini of SYP-5 and SYP-6 also localize between the chromosome axes. Fig. S3 shows the supporting data and measurements for SMLM analyses. Fig. S4 shows the schematics of null alleles generated for syp-5 and syp-6 and validation of the SYP-5 and SYP-5/6 antibodies. Fig. S5 shows the impact of SYP-5/6 deletion on meiotic progression and the analysis of C-terminal truncation mutants of syp-5/6. Table S1 shows additional hits from the mass spectrometry analysis of SYP-3–interacting proteins. Table S2 lists the alleles generated in this study. Table S3 lists the crRNAs, repair templates, and genotyping primers for mutant alleles generated in this study. Table S4 lists the worm strains used in this study.

We thank Bob Cole in the Mass Spectrometry and Proteomics Facility at Johns Hopkins School of Medicine for mass spectrometry analysis, Abby Dernburg (University of California, Berkeley, Berkeley, CA) for antibodies, and members of the Y. Kim laboratory and John Kim for comments on the manuscript.

Some strains were provided by the Caenorhabditis Genetics Center, which is funded by the National Institutes of Health Office of Research Infrastructure Programs (P40 OD010440). This work was supported by funding from the European Molecular Biology Laboratory to I. Čavka, S. Köhler, P. Hoess, and J. Ries, the European Research Council to J. Ries and P. Hoess (ERC CoG-724489), the Human Frontier Science Program to J. Ries (RGY0065/2017), and the National Institutes of Health to J.G. Gall (R01GM33397), O. Rog (R35GM128804), and Y. Kim (R35GM124895).

The authors declare no competing financial interests.

Author contributions: M.E. Hurlock and Y. Kim designed and performed most experiments. I. Čavka, P. Hoess, J. Ries, and S. Köhler acquired and analyzed the SMLM data. L.E. Kursel performed the evolutionary analysis of SYP-5 and SYP-6. M.E. Hurlock and J. Haversat performed STED microscopy, assisted by M. Wooten, Z. Nizami, and J.G. Gall. R. Turniansky performed the biochemical purification of SYP-3. M.E. Hurlock, I. Čavka, L.E. Kursel, O. Rog, S. Köhler, and Y. Kim wrote and revised the manuscript.