A suppressor screen in C. elegans identifies a multiprotein interaction that stabilizes the synaptonemal complex

Significance

Gamete production requires tightly regulated interactions between the parental chromosomes, which coalign and exchange information. These events are mediated by the synaptonemal complex—a ladder-like structure that assembles between the parental chromosomes. The molecular interactions that underlie synaptonemal complex assembly remain poorly understood due to rapid sequence divergence and challenges in biochemical reconstitution. Our detailed genetic data suggests a three-component interface in the nematode synaptonemal complex. Destabilization and subsequent restoration of this interface link synaptonemal complex integrity with chromosome alignment and regulation of exchanges. Beyond mechanistic understanding of chromosomal interactions, our work provides a blueprint for genetic probing of large cellular assemblies that are refractory to structural analysis and sheds light on the forces that shape their evolution.

Abstract

Successful chromosome segregation into gametes depends on tightly regulated interactions between the parental chromosomes. During meiosis, chromosomes are aligned end-to-end by an interface called the synaptonemal complex, which also regulates exchanges between them. However, despite the functional and ultrastructural conservation of this essential interface, how protein–protein interactions within the synaptonemal complex regulate chromosomal interactions remains poorly understood. Here, we describe a genetic interaction in the C. elegans synaptonemal complex, comprised of short segments of three proteins, SYP-1, SYP-3, and SYP-4. We identified the interaction through a saturated suppressor screen of a mutant that destabilizes the synaptonemal complex. The specificity and tight distribution of suppressors suggest a charge-based interface that promotes interactions between synaptonemal complex subunits and, in turn, allows intimate interactions between chromosomes. Our work highlights the power of genetic studies to illuminate the mechanisms that underlie meiotic chromosome interactions.

Sexual reproduction requires tightly regulated chromosomal interactions. During meiosis, parental chromosomes (homologs) pair and align along their length. Chromosome alignment sets the stage for reciprocal exchanges called crossovers that physically link the homologs and allow them to correctly segregate during the first meiotic division. Failure to form crossovers, or formation of too many crossovers, can lead to chromosome missegregation, aneuploid progeny, and infertility (1–3).

Crossovers form within, and are regulated by, a dedicated meiotic chromosomal structure called the synaptonemal complex (SC). Upon entry into meiosis, each homolog adopts an elongated morphology organized around a backbone called the axis. The axes of the two homologs are then aligned, end-to-end, concomitant with the assembly of the SC. While the axes are classically considered part of the SC, here we use the term SC to specifically refer to the structure assembled between parallel axes, also known as the central region of the SC. In addition to bringing the homologs, and therefore the DNA sequences that form crossovers, into proximity, the SC directly participates in regulating crossover number and distribution (4, 5).

Our understanding of how the SC brings homologs together and regulates crossovers is limited. While the ultrastructure of the SC is conserved across eukaryotes, appearing as a ~100 nm-wide ladder in electron micrographs (6–11), the protein sequences of its components diverge rapidly (12–14). The poor sequence homology has slowed the identification of SC components, and it is unclear whether the complete set of components has been defined in any organism. Poor sequence homology has also limited the translation of structural findings across model organisms. Finally, we have very limited molecular-genetic information since only a small subset of components has been subjected to systematic mutational analysis or biochemical and structural probing.

The nematode C. elegans has proved a useful model for studying SC structure and function. The SC in worms assembles prior to crossover formation and is essential for all crossovers (8). In addition, the role of the SC in regulating crossover distribution has been directly demonstrated (15). The worm SC includes eight proteins: SYP-1, -2, -3, -4, -5/6, and SKR-1/2 (SYP-5 is partially redundant with SYP-6 and SKR-1 is partially redundant with SKR-2; refs. 8 and 16–21). These components exhibit stereotypical organization within the SC: SYP-1 and -5/6 are organized in a head-to-head fashion to span the distance between the axes, with their N-termini in the middle of the SC and C-termini near the axis (16) and SYP-2, SYP-4, and SKR-1/2 localize to a band in the middle of the SC (21–23). The localization of SYP-3 remains an open discussion, with two studies placing SYP-3 in the middle of the SC (20, 22) and another localizing SYP-3 near the axis (23) (Fig. 1A).

Fig. 1.

Mutagenesis screen identifies suppressors of syp-1^K42E fertility defect. (A, Top) Cartoon of paired homologous chromosomes (blue) with the synaptonemal complex (SC, green) assembled between them. Bottom, cartoon depicting localization of SC proteins relative to the axis. The N- and C- termini of SYP-1 and SYP-5/6 are labeled. Note: SYP-3 is drawn in two possible configurations, either central or near the axis, to reflect studies with differing models for SYP-3 localization. (B) Total self-progeny from syp-1+ and syp-1^K42E animals in a gfp::cosa-1 background at 15 °C and 25 °C. Asterisks indicate statistical significance of the comparison of total progeny between genotypes at 15 °C and 25 °C using an unpaired t test. (C) Schematic of suppressor screen. We mutagenized 50,000 gfp::cosa-1; syp-1^K42E animals (P₀) using ENU and allowed them to recover at 15 °C. After 5 d, when most F₁s were at the L4 developmental stage, we shifted the worms to 25 °C to apply selection. Suppressed plates were able to consume all of the bacterial lawn and starve the plate. We confirmed the suppression phenotype by picking five single worms from each suppressed plate and counting self-progeny at 25 °C (quantified in D). (D) Total self-progeny from worms singled from 23 suppressed plates sorted by degree of suppression. Arrowheads indicate plates with zero to three total self-progeny. These are likely non-suppressed animals, suggesting that when the worms were singled the selective sweep by the suppressors was still incomplete. The incomplete selective sweep also suggests that some of the animals we tested were heterozygous for the suppressors. In B and D, mean values are shown by a horizontal black line and asterisks indicate P-values as follows: *** < 0.001, **** < 0.0001.

Previously, we carried out mutational analysis of the N-terminus of SYP-1 and demonstrated its role in holding the parallel axes together and in regulating crossovers (24). Here, we use a temperature-sensitive single substitution mutant in syp-1 to carry out a suppressor screen that allowed us to suggest an interaction interface in the middle of the SC. This predicted interface ensures efficient assembly of the SC onto chromosomes, and consequently, the formation of tightly regulated crossovers and successful chromosome segregation. Many of the suppressors we isolated did not affect meiosis by themselves despite altering highly conserved residues, illuminating the unusual selective pressures that shape SC protein evolution.

Results

A Suppressor Screen of a Temperature-Sensitive SC Mutant.

We previously identified a lysine-to-glutamic acid substitution, syp-1^K42E, that weakens interactions among SC proteins, causing failures in chromosome synapsis and crossover regulation (24). In syp-1^K42E, the stability of the SC progressively weakens as the temperature increases, leading to worsening meiotic phenotypes (24). At 15 °C, the SC in syp-1^K42E animals is mostly unaffected, and the worms indeed produce many viable self-progeny, although their number is reduced by 53% compared to control animals (Fig. 1B, mean total progeny 15 °C; syp-1+ = 222, syp-1^K42E= 104). At 25 °C, a temperature at which wild-type C. elegans can reproduce without discernable meiotic defects, syp-1^K42E animals are nearly sterile (Fig. 1B, mean total progeny 25 °C; syp-1+ = 145, syp-1^K42E = 1). The SC at this temperature assembles onto unpaired chromosomes, a morphology that is never observed in wild-type animals (24).

We carried out a whole-genome mutagenic screen to identify suppressors of the syp-1^K42E fertility defect. We used the alkylating agent N-ethyl-N-nitrosourea (ENU) to mutagenize ~50,000 P₀ gfp::cosa-1; syp-1^K42E animals grown at 15 °C and divided them among 100 plates. We allowed their F₁ progeny (carrying heterozygous mutations) to grow at 15 °C until they reached the young adult (L4) stage, at which point we shifted them to the nonpermissive temperature of 25 °C. Candidate suppressors of the syp-1^K42E fertility defect quickly outcompeted their non-suppressed siblings and starved the plate (Fig. 1C). Through this approach, we identified 23 suppressed lines that showed a wide range in their ability to suppress the infertility of syp-1^K42E at 25 °C (Fig. 1D).

All Suppressed Lines of syp-1^K42E Contain Mutations in SC Proteins.

To identify the causative suppressor mutations, we employed a combination of candidate Sanger sequencing and whole-genome sequencing followed by variant calling (Fig. 2A). We hypothesized that suppressor mutations could be within SC components, so we sequenced the N-terminus of syp-1 and syp-3. [Two studies place SYP-3 in the middle of the SC (20, 22), and a small truncation in syp-3 exhibits phenotypes similar to those of syp-1^K42E (18).] This approach had the added advantage of confirming the presence of the original syp-1^K42E mutation. Twenty-two out of the 23 suppressed strains had the original syp-1^K42E mutation. The remaining strain had acquired a mutation that resulted in reversion to the wild-type SYP-1 sequence (Fig. 2A, syp-1^E42K revertant). We could distinguish the revertant from wild-type contamination because the suppressed strain used AAA to encode the lysine (K) at position 42, rather than the AAG codon in the reference genome (Fig. 2B). We also identified seven additional candidate suppressor mutations in syp-1 and three candidate suppressor mutations in syp-3. In syp-1, there were three charge-altering mutations, syp-1^E41K and syp-1^E45K (recovered twice independently), three polar to hydrophobic mutations, syp-1^S24L (twice) and syp-1^T35I, and one hydrophobic to hydrophobic mutation, syp-1^V39I. The three mutations in syp-3 were at amino acid position 62 and altered the normally negatively charged aspartic acid (D) to either asparagine (N, polar), valine (V, hydrophobic), or glycine (G, small/hydrophobic).

Fig. 2.

All suppressor strains contain mutations in SC proteins. (A) Pipeline used to identify suppressor mutations. We performed Sanger sequencing on portions of syp-1 and syp-3 in all suppressed lines to confirm the presence of the original syp-1^K42E mutation and to identify possible suppressor mutations. We performed whole-genome sequencing on 14 suppressor strains and controls. We used GATK to call variants unique to each suppressor strain and found an additional 14 unique mutations in SC proteins, some of which occurred multiple times independently. (B) Chromatograph traces from Sanger sequencing of syp-1 in gfp::cosa-1; syp-1^K42E and the revertant strain identified in the screen (gfp::cosa-1; syp-1^E42K) aligned to the reference genome. The reference genome and revertant strain each encode the lysine (K) at position 42 (yellow box) with a different codon (AAG and AAA, respectively). In addition, the silent T->C mutation in the codon encoding a lysine at position 40, which was introduced while creating the original syp-1^K42E mutation, is present in the revertant. (C) Schematic depicting the strategy used for variant calling. We identified SNPs and indels in gfp::cosa-1; syp-1^K42E and in the suppressed lines. Mutations present in gfp::cosa-1; syp-1^K42E but not in the reference genome were considered background mutations. We subtracted these background mutations from the mutations in each suppressed strain to generate a list of putative suppressor mutations. (D) Bar graph showing the number of unique, non-background mutations in each sequenced suppressed line. Average number of mutations per genome = 199. (E) Histogram of Poisson distribution for λ = 10, the average number of ENU-generated mutations per base pair in the screen. The likelihood that substitution in a base pair was not screened for suppression is <5 × 10⁻⁵.

We identified the remaining suppressor mutations through whole-genome sequencing. We identified all homozygous SNPs and indels in the suppressed strains relative to the parental strain and the reference genome, WBcel235 (Fig. 2C). In addition to the 2,036 SNPs present in gfp::cosa-1 relative to the reference genome, each suppressed genome had 199 homozygous mutations on average (Fig. 2D). Using this number, plus the number of mutagenized P₀ animals (50,000), the number of F₁s per P₀ at 15 °C (Fig. 1A, gfp::cosa-1; syp-1^K42E average total progeny = 104), and the haploid genome size of C. elegans (1 x 10⁸ base pairs), we estimate that each base pair in the genome was mutagenized 10 times in our screen. Using a Poisson distribution, the probability that a base pair was not screened is 4.5 × 10⁻⁵ (Fig. 2E, λ = 10).

Out of the 2,789 homozygous mutations in the 14 independent suppressed lines we sequenced, 1,968 were in intergenic regions and UTRs, 159 were synonymous site mutations and 96 mutations were in gene introns (SI Appendix, Fig. S1A and Dataset S1), in line with the genomic abundance of these elements. We found 537 missense mutations and 29 mutations that were likely to abolish gene function through the loss of a start or a stop codon, a nonsense or frameshift mutation or a mutation of a splice acceptor/donor site (SI Appendix, Fig. S1A). The distribution of SNPs matched the expected mutagenic profile of ENU with an overall bias toward GC > TA transitions, although all transitions and transversions were represented (SI Appendix, Fig. S1B and ref. 25).

To prioritize suppressor mutations, we sorted for genes whose RNAi phenotype, allele phenotype, or gene ontology included the words “meiosis” or “meiotic” (Dataset S2). Each genome had between one and four missense mutations in meiotic genes out of 34 missense mutations per genome on average. Strikingly, every genome had a missense mutation in an SC subunit. These included three mutations in syp-1 [syp-1^S24L (identified twice), syp-1^T35I, and syp-1^V39I] and three mutations in syp-4 [syp-4^A81T, syp-4^E90G, and syp-4^E90K (identified eight times)]. Given the number of ENU-generated missense mutations per genome, the probability of uncovering 21 independent mutations in SC proteins by chance is less than 1 × 10⁻⁴².

The location of the suppressors in narrow regions of the same SC proteins and the independent isolation of multiple instances of the same mutation, like syp-4^E90K, suggest that the mutations in SYP-1, -3, and -4 are the causative suppressor mutations.

Suppression Strength Correlates with Amino Acid Charge.

We noticed that many of the suppressors change a negatively charged residue to a positively charged residue (e.g., syp-1^E41K, syp-1^E45K, and syp-4^E90K) or neutralize a negatively charged residue (e.g., syp-3^D62N, syp-3^D62V, syp-3^D62G, and syp-4^E90G). Accordingly, we divided the suppressors into three categories and carefully analyzed their effects; one, suppressor mutations that neutralized a negative charge (syp-3^D62N, syp-3^D62V, and syp-4^E90K), two, suppressor mutations that altered a polar residue (syp-1^T35I and syp-1^S24L), and three, suppressor mutations that altered a hydrophobic residue (syp-4^A81T and syp-1^V39I). Rather than solely rely on homozygous suppressed animals recovered from the screen, we engineered three of the strongest suppressor mutations (syp-3^D62N, syp-3^D62V, and syp-4^E90K) using CRISPR/Cas9 in the gfp::cosa-1; syp-1^K42E background. (In the figures, we designated the genotype of the suppressed strains, as opposed to the edited strains, with a gray background.) The suppression we observed in these engineered strains (see below) suggests that non-SC mutations likely have limited effects on the meiotic phenotypes and further confirms the identity of the suppressors in SYP-3 and -4.

The original mutation, syp-1^K42E, has very low total progeny and a high incidence of male self-progeny at 25 °C (Fig. 1B and ref. 24), caused by meiotic nondisjunction of the X chromosomes. Animals suppressed with syp-3^D62N, syp-3^D62V, or syp-4^E90K showed complete suppression of defects in total progeny and percent males at 25°C (Fig. 3 A and B). Like wild-type animals, they had large brood sizes (Fig. 3A, blue-filled circles, gfp::cosa-1; syp-1^K42E; syp-3^D62N = 223, gfp::cosa-1; syp-1^K42E; syp-3^D62V = 210, gfp::cosa-1; syp-1^K42E; syp-4^E90K = 189) and rare male self-progeny (Fig. 3B, blue-filled circles, all strains <1% male progeny). Animals suppressed with syp-1^T35I and syp-1^S24L showed an intermediate phenotype with average total progeny of 178 and 99, respectively, and a slightly elevated percentage of male self-progeny (Fig. 3 A and B, teal-filled circles, gfp::cosa-1; syp-1^K42E+T35I = 1.4% and gfp::cosa-1; syp-1^K42E+S24L = 4% male progeny). Finally, animals suppressed with syp-4^A81T and syp-1^V39I provided the weakest suppression and had relatively low total progeny and a high percentage of male offspring, albeit significantly suppressed compared to gfp::cosa-1; syp-1^K42E (Fig. 3 A and B, yellow filled circles, gfp::cosa-1; syp-1^K42E; syp-4^A81T average total progeny = 25, average percent males = 9.5%, gfp::cosa-1; syp-1^K42E+V39I average total progeny = 31, average percent males = 14.8%).

Fig. 3.

Strength of suppression correlates with amino acid charge. Total progeny (A), and percent male self-progeny (B), for seven representative suppressors at 25 °C. Phenotype of the heterozygous state of the three strongest suppressors is also shown in A and B. Number of replicates are listed in parentheses. Throughout the figure, suppressor strains are colored according to the characteristic of the residue that was mutated in the screen. In addition, genotypes with gray highlighting are homozygous suppressor strains isolated from the screen. Genotypes without this label were engineered using CRISPR in the gfp::cosa-1 background. (C) Bar graphs showing percentage of early pachytene nuclei with partially or completely asynapsed chromosomes at 25 °C. (D) Number of GFP::COSA-1 foci per nucleus in suppressed strains at 25 °C. Values in parentheses indicate the number of nuclei examined. Note that only the two weak suppressors, syp-1^K42E; syp-4^A81T and syp-1^{K42E + V39I}, have an elevated SD (SD = 2.38 and 3.11, respectively) compared with syp-1+ (SD = 0). (E) Immunofluorescence images of pachytene nuclei showing the extent of synapsis and GFP::COSA-1 foci for representative suppressed genotypes at 25 °C. The number of foci is indicated in nuclei with more or less than the six expected GFP::COSA-1 foci. (Scale bar, 5 µm.) (F) Example images of whole gonads stained with DAPI used to measure transition zone length. Transition zones are marked with a yellow line, pachytene is marked with a blue line. (Scale bar, 10 µm.) (G) Quantification of transition zone length as percent of meiosis at 25 °C. For panels A–D and G, asterisks (black) indicate statistical significance compared to gfp::cosa-1; syp-1^K42E and horizontal black lines show mean values. Asterisks are representative of P-values as follows: * < 0.05, ** < 0.01, *** < 0.001, **** < 0.0001. See methods for detailed description of statistical analyses.

We hypothesized that if the suppressor sites in SYP-3 and -4 physically interact with SYP-1^K42E, then even one copy of a suppressor mutation may allow the suppressor proteins to assemble into the SC and restore interactions among some of the SC proteins. We therefore also tested three of the strongest suppressors and found that heterozygous syp-3^D62N, syp-3^D62V, and syp-4^E90K provided an intermediate suppression phenotype (Fig. 3 A and B, blue open circles). This also implies that the selection in our screen began in the F₁ rather than the F₂ generation since the fertility of F₁s containing a heterozygous suppressor mutation was boosted when they were shifted to the nonpermissive temperature (Fig. 1C).

Suppressor Mutations Restore SC Assembly, SC Stability, and Crossover Regulation.

We next verified that the suppressors restored the SC defects in female meiosis in syp-1^K42E animals. At 20 °C, syp-1^K42E nuclei in mid-pachytene fail to synapse homologs end-to-end and typically have at least one asynapsed and one partially synapsed (forked) chromosome pair; at 25 °C, the SC assembles onto unpaired chromosomes (24). We found that strains suppressed by syp-3^D62N, syp-3^D62V, syp-4^E90K, syp-1^T35I, and syp-1^S24L achieved complete synapsis by mid-pachytene at 25 °C (Fig. 3 C and E). We observed almost no forked chromosomes or unpaired chromosomes lacking SC (Fig. 3 C and E and SI Appendix, Fig. S2). In contrast, strains suppressed by syp-4^A81T and syp-1^V39I failed to achieve full synapsis and only had a few synapsed chromosomes per nucleus (Fig. 3 C and E and SI Appendix, Fig. S2), likely accounting for their failure to form crossovers between all homolog pairs and, consequently, their smaller brood sizes and high percentage of male self-progeny.

We next assessed the ability of the suppressed strains to carry out accurate crossover regulation. syp-1^K42E exhibits weaker crossover interference, with more than one crossover observed on the same chromosome and the same stretch of SC (24). We leveraged a cytological marker of crossovers, GFP::COSA-1, which localizes to each of the designated crossovers, numbering six in wildtype C. elegans (one per chromosome) (26). Mutations that impact the integrity of the SC, including syp-1^K42E (24), syp-1 RNAi (15), and syp-4^CmutFlag (ie25) (22), reduce crossover interference so that there are more than six crossovers per nucleus. We found that strains suppressed by syp-3^D62N, syp-3^D62V, syp-4^E90K, syp-1^T35I, and syp-1^S24L all had six crossovers per nucleus (Fig. 3 D and E), indicating restoration of crossover interference. In contrast, the weak suppressors syp-4^A81T and syp-1^V39I showed a wide distribution of crossover number, with some nuclei having only one or two crossovers and some having more than 10 (Fig. 3 D and E). This is likely not just a reflection of the incomplete synapsis in these mutants, since both harbored instances of chromosomes (i.e., a stretch of SC) with more than one crossover (SI Appendix, Fig. S3).

Previous work has shown that meiotic nuclei sense unpaired chromosomes, and, in response, spend a longer time in the so-called transition zone, where homology search and SC assembly occur (corresponding to the classically defined leptotene and zygotene stages of meiosis; ref. 27). The extended transition zone could allow for more time to complete SC assembly. We found that most suppressed strains failed to suppress the extended transition zone of syp-1^K42E animals (Fig. 3 F and G and SI Appendix, Fig. S4). In the case of the weak suppressors, that is likely due to partial synapsis (Fig. 3 C and E). Interestingly, the strong and intermediate suppressors exhibited different transition zone lengths—extended in the strong suppressors and shorter in the intermediate suppressors—even though almost all nuclei in these strains achieved complete synapsis by mid-pachytene (Fig. 3 C and E). The mechanism that senses and responds to asynapsed chromosomes depends on axis proteins, likely involving exposed axes not associated with the SC (28, 29). A possible reason for the difference in transition zone length between the strong and intermediate suppressors might be their different propensities to associate with the axes, either in the context of assembled SC or with aberrant conformations, such as forked chromosomes. Alternatively, the difference between the strong and intermediate suppressors may reflect slower kinetics of synapsis or defects that cannot be discerned at the resolution of confocal microscopy.

Finally, we tested whether the suppressor mutations restore the unstable SC in syp-1^K42E animals (24). We measured the proportion of GFP::SYP-3 on chromosomes in mid-pachytene relative to the total nucleoplasmic amount in syp-1^K42E and suppressed (syp-1^{T35I + K42E}) live animals. We identified interchromosomal regions using HTP-3::wrmScarlet, a component of the axis that assembles independently of the SC. Consistent with our published findings, we found that less GFP::SYP-3 is recruited to chromosomes in syp-1^K42E compared to a wild-type control (Fig. 4 A and B and ref. 24). The suppressor mutation (syp-1^T35I) restored the amount of SC recruited to chromosomes to wild-type levels (Fig. 4 A and B). Notably, since we analyzed animals at the semipermissive temperature of 20 °C, where mostly normal synapsis occurs in all tested strains, we likely underestimated the full destabilizing effect of syp-1^K42E.

Fig. 4.

Suppressor mutation restores SC recruitment to chromosomes. (A) Images of chromosome volumes generated using HTP-3::wrmScarlet fluorescence signal and live fluorescence images of GFP::SYP-3 taken at 20 °C. (B) Dot plot showing percent SC signal localized to chromosomes relative to total nucleoplasmic fluorescence in a single nucleus at 20 °C. Data in B was acquired from five animals from each genotype.

In summary, the suppressors of the fertility defect of syp-1^K42E also suppressed its other meiotic phenotypes: incomplete synapsis, unstable SC, and weakened crossover interference. Furthermore, weaker suppressors consistently exhibited incomplete suppression. Our data thus suggest that SYP-1 physically interacts with SYP-3 and SYP-4 via charge–charge interactions between the N-terminus of SYP-1, position 62 in SYP-3, and position 90 in SYP-4. SYP-1^K42E weakens this interaction and, as a result, the SC is less stable and fails to fully assemble onto chromosomes and to confer robust crossover interference (24). The suppressors restore this predicted charge interface and consequently rescue the chromosomal and meiotic defects.

Suppressors Alter Conserved Residues in SC Proteins.

We next wished to address the molecular mechanism of suppression. Unfortunately, atomic structures for C. elegans SC proteins have not been solved, in a complex or in isolation, preventing us from localizing the mutated sites relative to one another [beyond their colocalization in the middle of the SC (22, 23)]. It has also been challenging to derive structural insight from other model organisms through modeling, due to limited sequence homology between clades (12–14). Our attempts to model a docking interface between the regions surrounding the mutations using AlphaFold did not yield a high-confidence interface.

Lacking tertiary structural information, we examined the potential impact of the suppressor mutations on coiled-coil domains—a conserved feature of SC proteins throughout eukaryotes (12). The suppressor mutations in SYP-1 occur before the start of the coiled-coil domain (Fig. 5 A and B). In SYP-3 and SYP-4, the suppressor mutations lie within extended coiled-coil domains but do not alter their predicted structure (Fig. 5 C–F and SI Appendix, Fig. S5 A and B). In SYP-3, the site of the suppressor mutation is not contained within a heptad repeat—the distinguishing feature of coiled-coils. So, even though one of the mutations introduces a glycine, which can disrupt alpha helices (30), the suppressors are not likely to impact the overall coiled-coil structure of the protein (Fig. 5 C and D and SI Appendix, Fig. S5A). In SYP-4, the suppressor mutations are found within perfect heptad repeats, where the “a” and “d” (first and fourth) positions are occupied by hydrophobic residues, and the “e” and “g” (fifth and seventh) positions are polar. The suppressor mutations do not change the underlying heptad structure: They either alter a residue in the nonsignificant “c” position (A->T) or swap one polar residue for another in the “e” position (E->K; Fig. 5 E and F and SI Appendix, Fig. S5B).

Fig. 5.

Suppressors alter conserved, charged residues in SYP-1, -3, and -4, but do not disrupt coiled-coils. (A) Coiled-coil scores per position for SYP-1. The location of original and suppressor mutation is indicated by vertical red and gray lines, respectively. A higher score implies that the region is more likely to form a coiled-coil. (B) Impact of suppressor mutations on the heptad repeats that underlie coiled-coils in SYP-1. Continuous heptad repeats are indicated by their positional nomenclature, a–g, and by a gray coil. Important positions in the heptad repeats are highlighted. Position a and d (blue) are typically hydrophobic, position e and g (yellow) are typically charged or polar. Note: in SYP-1, the original and suppressor mutations occur outside the coiled-coil domain. The heptad structure is shown for SYP-1+ and is aligned to SYP-1 mutants. Dots indicate the position has the same heptad structure in the mutants. (C–F) Same as A and B except for SYP-3 and SYP-4. In B, D, and F, the sequence shown is from the wild-type protein and suppressor and original mutations are colored in bold gray or red letters, respectively. (G) Partial alignment of SYP-1, -3, and -4 from 18 Caenorhabditis species. Suppressor mutations are indicated above the alignment. Amino acids are highlighted accordingly: negatively charged = red, positively charged = blue, serine or threonine (polar) = cyan, valine, isoleucine or alanine (hydrophobic) = lilac, glycine = yellow. Note that suppressor residues are either completely conserved (e.g., SYP-1^S24 SYP-3^D62 and SYP-4^E90) or have been sampled by evolution [e.g., SYP-1^T35 – C. panamensis, C. nouraguensis and C. becei all encode isoleucine (I) at position 35].

The assembly of the SC through condensation (31) suggests a potential function for intrinsically disordered regions that can drive phase-separation (32). However, the mutations do not lie within conserved intrinsically disordered regions (12), arguing that the suppressors do not act by altering these domains.

To obtain hints as to the functional importance of the residues mutated in the suppressors, we analyzed their conservation in an alignment from 18 Caenorhabditis species, estimated to share a common ancestor ~20 Mya (Fig. 5G and ref. 33). The suppressors are significantly enriched for positions that are completely conserved in Caenorhabditis, like the original syp-1^K42E mutation [Fig. 5G, e.g., syp-1^S24L, syp-1^E45K, syp-4^E90K, chi-square test, X² (1, N = 22) = 15.65, P < 0.0001]. While 16/22 suppressor mutations altered completely conserved residues, only 5% of the amino acids in SYP-1, 19% of the amino acids in SYP-3 and 16% of amino acids in SYP-4 are completely conserved. Interestingly, in many cases where the residue was not completely conserved, evolution has previously sampled the suppressor mutation: e.g., the glutamic acid to lysine substitution in position 41 in SYP-1 is also present in C. tribulationis and C. zanzibari (Fig. 5G; see also syp-1^T35I, syp-1^V39I and syp-4^A81T). Many of the positively charged and polar residues mutated in the suppressors were conserved throughout Caenorhabditis, although two were not—threonine 35 and glutamic acid 41 in SYP-1. Notably, the hydrophobic residues mutated in the suppressors were both not conserved.

Consistent with the conservation of the mutated sites, we found almost no variation at suppressor sites among 550 wild isolates of C. elegans (34) and no variation at homologous sites in 14 wild isolates of C. remanei (35) (Datasets S3–S6). In C. elegans wild isolates, only two substitutions overlapped with suppressed sites, SYP-1^V39A and SYP-1^E45D, both of which are conservative replacements that do not alter charge. Together, these findings suggest that the SYP-1/-3/-4 interface has coevolved to maintain interactions among SC proteins.

Suppressor Mutations alone Exhibit No Meiotic Phenotype.

We next investigated the effects of the suppressor mutations alone in female meiosis. Since the change from a positive to a negative charge in syp-1^K42E causes severe synapsis and crossover regulation defects, we hypothesized that the charge-reversing suppressors would exhibit similar phenotypes by themselves. We analyzed the three strongest suppressors, syp-3^D62N, syp-3^D62V, and syp-4^E90K as well as a strain that combined two suppressor mutations, syp-3^D62V; syp-4^E90K. In most conditions, there was not a significant decrease in the number of progeny or increase in the percentage of male self-progeny in the suppressors when compared to the gfp::cosa-1 control at 15 °C, 20 °C, and 25 °C (Fig. 6 A–C). The only exception was the double suppressor, syp-3^D62V; syp-4^E90K, which demonstrated a mild (31%) but statistically significant reduction in total progeny at 25 °C (Fig. 6C). We also quantified crossover events and asynapsis and found no evidence of defects in either syp-3^D62N, syp-3^D62V, syp-4^E90K or syp-3^D62V; syp-4^E90K (Fig. 6 D–F).

Fig. 6.

Suppressors alone do not affect the SC. Total progeny and percent male progeny for suppressors alone at 15 °C (A), 20 °C (B) and 25 °C (C). Note that (C) also contains a strain with two suppressor mutations combined (gfp::cosa-1; syp-3^D62V, syp-4^E90K). We used an ordinary one-way ANOVA with Dunnett’s test for multiple comparison to test for a difference of means for total progeny (A–C, Top), and we use a Kruskal–Wallis test with Dunn’s test for multiple comparison to test for differences in percent males among genotypes (A–C, Bottom). (D) Number of GFP::COSA-1 foci per nucleus. (E) Immunofluorescence images of mid-pachytene nuclei showing GFP::COSA-1 foci and extent of synapsis. (F) Bar graphs showing percentage of early pachytene nuclei with partially or completely asynapsed chromosomes. For panels A–D, asterisks (blue) indicate statistical significance compared to gfp::cosa-1 and horizontal black lines indicate mean values. Asterisks are representative of P-values as follows: * < 0.05, ** < 0.01, *** < 0.001, **** < 0.0001. See methods for detailed description of statistical analyses.

While the single mutants do not appreciably affect meiosis at typical growth conditions, we wondered whether they might affect the response to extreme environments. Meiosis in general, and the SC in particular, are highly temperature-sensitive, even in wild-type plants and animals (36, 37). In C. elegans, a shift in temperature from 25 °C to 26.5 °C causes SC aggregation in early pachytene, although SC that is already assembled seems protected from this fate (36). We found no difference between syp-4^E90K; gfp::cosa-1 and the gfp::cosa-1 control in their response to elevated temperature (SI Appendix, Fig. S6). In both strains, there were SC aggregates in early pachytene but not late pachytene after a 16-h exposure to 26.5 °C, suggesting that the SC in gfp::cosa-1; syp-4^E90K is neither sensitized nor resistant to high temperatures.

Discussion

The molecular interfaces that allow the SC to mediate interactions between the homologs remain a major gap in our understanding of meiotic chromosome dynamics. The identity and nature of these interfaces are poorly understood, largely because the SC has proven a difficult candidate for traditional structural approaches. An exception is the mammalian SC, where multiple recombinant SC proteins have been subjected to structural analysis (38–43). However, even these pioneering studies have not yet systematically explored the interfaces regulating SC assembly.

Here, we present genetic evidence for direct interactions between specific residues in SYP-1, -3, and -4 in C. elegans. We leveraged a previously identified temperature-sensitive mutation, syp-1^K42E (24), to perform a saturating forward suppressor screen. All suppressor mutations mapped to three short stretches in SC proteins, which likely represent a protein–protein interaction interface. This proposed interface is likely in the middle of the SC, based on the locations of SYP-4 and the N-terminus of SYP-1 (22, 23), thereby supporting prior cytological and in vitro data that placed SYP-3 in the middle of the SC (20, 22). Several lines of evidence support direct physical interaction. First, mutations were in conserved residues. Second, most mutations altered charged residues, compensating for the charge swap in the original mutation. Third, the suppressors restored not only the fertility defect that formed the basis of the screen but also other aspects of SC dysfunction like incomplete SC assembly, SC instability, and weaker crossover interference. Fourth, the degree of suppression correlated with the nature of the mutation, with charge-neutral mutations conferring the weakest suppression. The extensive coverage in our screen further suggests that we isolated most of the possible suppressors and that other SC proteins are unlikely to participate directly in this proposed interface.

The simplest model to explain our findings is a charge–charge interaction interface involving the three proteins, SYP-1, -3, and -4. Such a model readily accounts for the suppression of the original charge-swap mutation in SYP-1 position 42 by compensatory charge mutations in SYP-1, -3, and -4. The SYP-1/-3/-4 interface could in turn stabilize large-scale interactions between SC subunits to allow the SC to hold the axes of the two homologs together and/or to spread along chromosomes to synapse them end-to-end (44). However, without additional structural information, it would be challenging to test this model. It is also possible that some of the suppressor residues interact with other SC proteins (SYP-2/5/6 or SKR-1/2) or with regulators of the SC. However, we find this unlikely since we did not uncover suppressor mutations in other SC proteins.

Different model systems offer limited help in substantiating this predicted interface due to the rapid sequence divergence of SC proteins (12–14). For instance, while the polymerization of the mammalian SC was proposed to involve interactions between tetramers of the human functional homolog of SYP-1, SYCP1 (39), these are mediated by hydrophobic residues, not charge–charge interactions as we propose here. Excitingly, the recent discovery of SKR-1/2 as a component of the SC in C. elegans allowed for the purification of recombinantly expressed worm SC (21). Future work in this system will potentially test the impact of the original and suppressor mutations on SC structure.

Recent studies in worms and plants demonstrated that the SC directly regulates crossover distribution (15, 45–49). The data in worms relied on perturbations to SC components, including partial depletion of SC proteins (15), the syp-1^K42E mutant used here (24), and a serendipitous alteration in the C-terminus of SYP-4 [syp-4^CmutFlag(ie25); ref. 22]. These perturbations allow homologs to synapse and crossovers to form but reduce crossover interference so that each chromosome undergoes more than one crossover. Here, we observe multiple suppressors of the homolog synapsis function of the SC that also restore crossover interference. Strikingly, weak suppressors of the former conferred only partial suppression of the latter phenotype. Our findings, therefore, strengthen the idea that crossover interference is mediated by a mechanism that depends on the integrity of the SC (45, 46, 48–51). Destabilization of the SC by syp-1^K42E (24) and restoration of stability by the suppressors (Fig. 4) further suggest that the biophysical properties of the SC control the strength of crossover interference.

Although the protein sequences of SC components are overall highly diverged in Caenorhabditis (12), the sites of the original and suppressor mutations tend to be well conserved. Sixteen out of 22 suppressor mutations occurred at sites that are 100% conserved in the Elegans and Japonica groups (Fig. 5G), including the strong suppressors SYP-3^D62V and SYP-4^E90K. Similarly, we observed very little variation at suppressor sites in wild Caenorhabditis isolates. This conservation across and within species supports the importance of these positions for SC function.

Given the conservation of many of the suppressor sites, it is surprising that the suppressor mutations alone do not exhibit meiotic phenotypes (Fig. 6). While the suppressor sites may not be important for SC function in the absence of syp-1^K42E, their lack of phenotype is also consistent with a multivalent interaction interface. If SC assembly relies on multiple points of contact among its subunits (each likely composed of multiple interacting residues), altering one such interaction interface may not have a discernable effect. Consistent with this idea, combining two suppressors exposed a mild fertility defect (Fig. 6C). Nonetheless, redundancy provides a poor mechanism to account for the purifying selection these residues seem to be subjected to, and it does not provide a good explanation for the strong phenotype exhibited by the single substitution in syp-1^K42E. Alternatively, the suppressor sites may play a more important role in natural populations that are exposed to many biotic and abiotic stressors that sensitize meiotic fitness, and where minor fertility defects play out over many generations. We have tested one such stressor—transient exposure to high temperature—and did not observe a significant effect. However, we cannot rule out minor effects below our detection level or a role for other environmental stressors. Finally, it is worth noting that the suppressors could impact male meiosis (not assessed in this study), which could explain the purifying selection on these residues.

The epistasis we observe between syp-1^K42E and the suppressor mutations has implications for the evolutionary trajectories of SYP-1 in Caenorhabditis. The SYP-1 fitness landscape in the presence or absence of a suppressor mutation is distinctly different, with only the former allowing a subsequent syp-1^K42E mutation. In this regard, the suppressor mutations that are not conserved (e.g., syp-1^T35I) are particularly interesting as they change the mutational landscape that could be sampled by SYP-1 (52, 53).

Our work demonstrates the power of genetic analysis in general, and suppressor screens in particular, to illuminate molecular mechanisms in metazoans. Several factors assisted in the success of the screen: the temperature sensitivity of the original mutation which allowed us to grow a large number of worms at the permissive temperature; the near complete sterility at the nonpermissive temperature, conferring very tight selection; and the use of ENU as a mutagen, rather the more commonly used EMS (54, 55), which likely helped in isolating substitutions, since EMS mostly generates premature stop codons (56, 57). In the future, similar screens may prove particularly important in probing protein assemblies that are refractory to biochemical studies, as well as for studying cellular structures that assemble through condensation, as is the case for the SC (31), due to their reliance on weak multivalent interactions (58).

Summary of Materials and Methods

See SI Appendix for a complete description of the materials and methods used in this study.

Worm Strain Maintenance.

All strains were grown according to standard methods (59) and maintained at 20 °C, with the exception of gfp::cosa-1 (meIs8) II; syp-1^K42E(slc11) V which was maintained at 15 °C. For a complete list of strains used in this study, see Dataset S7.

Mutagenesis with N-ethyl-N-nitrosourea (ENU) and Screen for Suppressors.

We mutagenized synchronized L4 worms in 50 mM ENU and then let them recover for 5 d at 15° C. We then moved the plates to 25° C and monitored for the ability of the worms to proliferate.

Whole-Genome Sequencing.

We used the Zymo Quick-DNA Miniprep Plus Kit (cat#D4068) to extract genomic DNA for whole-genome sequencing. For library preparation, we used the Nextera DNA Flex library prep kit. Samples were sequenced on an Illumina NovaSeq on a 2 × 150 bp run to sequencing depth of 30× coverage.

Genome Assembly and Variant Calling.

We used BWA MEM (60) to align reads to the C. elegans reference genome (version WBcel235 from wormbase.org). We used GATK to call single nucleotide variants and indels relative to the reference genome (61, 62). We filtered the ENU induced mutations for homozygous variants that had a read depts of greater than 15 and a quality score greater than or equal to 200 using SnpSift (63). We annotated each of the remaining variants on the C. elegans genome and predicted its impact with SnpEff (64). All genome sequences generated in this study are available at www.ncbi.nlm.nih.gov under BioProject PRJNA1006752.

CRISPR Genome Editing.

We generated an RNA mix by combining 4 μL 200 μM tracrRNA (IDT Alt-R® CRISPR-Cas9 tracrRNA), 3 μL 200 μM crRNA matching our gene of interest (IDT Alt-R® CRISPR-Cas9 crRNA), and 1 μL dpy-10 crRNA. We heated the RNA mix at 95 °C for 5 min and then let it cool on the benchtop for 5 min. We next made an injection mix by combining 3.5 μL of the RNA mix with 1 μL Cas9 (IDT, Alt-R™ S.p. Cas9 Nuclease V3, 10 μg/μL) 0.5 μL 200 μM dpy-10 repair template, and 3 μL repair template to our gene of interest (200 μM). We injected 24 h post-L4 animals (P₀s) and screened for Dpy and Rol F₁s 4 to 5 d following injection. We singled Dpy and Rol F₁s and genotyped them by extracting DNA from their progeny, performing PCR and a digest to distinguish the desired CRISPR allele from a wild-type allele. All crRNAs, repair templates and oligos used for genotyping are listed in Dataset S8.

Immunofluorescence.

Immunofluorescence was carried out essentially as previously described (65).

Live Imaging.

We dissected 24 h post-L4 animals grown at 20 °C on a coverslip in modified embryonic culture medium (ref. 66, 84% Leibowitz L-15 medium, 9.3% Fetal calf serum, 1% 2 mM EGTA). We transferred dissected animals to a 2% agarose pad on a glass slide and sealed the coverslip with valap sealant (equal parts petroleum jelly, lanolin, and paraffin). We imaged intact gonads within 20 min of dissecting on a Zeiss LSM880 confocal microscope with Airyscan and a 63 × 1.4 NA oil objective. We took z-stacks (step size 0.4 μm) that spanned a depth of approximately one nucleus (~15 slices). All images took around 30 s to collect. The power for the 488 argon laser and the 561 nm diode laser was kept constant at 2.6 for all experiments and genotypes.

Supplementary Material

Appendix 01 (PDF)

Dataset S01 (XLSX)

Dataset S02 (XLSX)

Dataset S03 (XLSX)

Dataset S04 (TXT)

Dataset S05 (TXT)

Dataset S06 (TXT)

Dataset S07 (XLSX)

Dataset S08 (XLSX)

Dataset S09 (XLSX)

Data, Materials, and Software Availability

C. elegans genome sequencing data have been deposited in NCBI (PRJNA1006752) (67).