ABSTRACT
The goal of meiosis is to produce haploid gametes from diploid progenitor cells. While meiosis was likely present in the last eukaryotic common ancestor (LECA), diversity in meiotic mechanisms has long been observed among sexually reproducing eukaryotes. Here we describe a new, comparative model system for molecular analysis of meiosis, the nematode Pristionchus pacificus, a distant relative of the widely studied model organism Caenorhabditis elegans. Despite superficial similarities in germline organization and meiotic progression between P. pacificus and C. elegans, we identify fundamental differences in the molecular mechanisms underlying homolog pairing, synapsis, and crossover regulation. Whereas C. elegans has lost the meiosis-specific recombinase Dmc1, P. pacificus expresses both DMC-1 and RAD-51, which localize sequentially to meiotic chromosomes during prophase. We find that Ppa-spo-11 and Ppa-dmc-1 are required for stable homolog pairing, synapsis, and crossover formation, while Ppa-rad-51 is dispensable for these key processes during early prophase and plays a supporting role in meiotic double-strand break repair. Additionally, we show that elevated crossover recombination in P. pacificus likely arises through a Class II pathway normally inactive in C. elegans, shedding light on crossover control and the evolution of recombination rates.
INTRODUCTION
All sexually reproducing organisms rely on the specialized cell division process of meiosis to generate haploid gametes from diploid precursors. Upon fertilization, haploid gametes fuse and restore the diploid chromosome complement in the zygote. Thus, meiosis is essential for the survival of sexually reproducing species.
A defining feature of meiosis is a “reductional” segregation in which homologous chromosomes are separated, usually during the first of two nuclear divisions. A highly choreographed series of chromosome transactions precedes this division and ensures faithful homolog segregation: (1) pairing, in which chromosomes contact and recognize their homologous partners; (2) synapsis, defined as the assembly of a protein ensemble called the synaptonemal complex (SC) between homologs, which leads to their lengthwise alignment; and (3) crossover (CO) recombination, which creates physical linkages between chromosomes that promote proper bi-orientation during anaphase I. Failure to form at least a single CO between paired homologs results in nondisjunction and aneuploid gametes (Zickler and Kleckner, 1999).
Although pairing, synapsis and CO recombination are nearly ubiquitous among eukaryotes, details of these mechanisms show remarkable diversity among different lineages. In most model fungi, plants, and animals, stable homologous chromosome pairing and synapsis depend on early steps of the recombination pathway. Meiotic recombination is initiated by a conserved topoisomerase-like enzyme called Spo11, which catalyzes programmed DNA double-strand breaks (DSBs) across the genome, a subset of which are ultimately processed into COs (Keeney, 2008; Keeney et al., 1997). DSBs are resected to form 3’ single-stranded DNA overhangs. Dmc1, a meiosis-specific recombinase, forms filaments along these ssDNA segments and promotes interhomolog strand invasion. Spo11-dependent induction of DSBs and Dmc1-dependent strand invasion are crucial for proper pairing and synapsis in the budding yeast Saccharomyces cerevisiae, the flowering plant Arabidopsis thaliana, and mammalian model Mus musculus (Bishop et al., 1992; Couteau et al., 1999; Grelon, 2001; Pittman et al., 1998; Rockmill et al., 1995; Yoshida et al., 1998).
In contrast, recombination-independent mechanisms of pairing and synapsis have been characterized in other prominent model systems, including the dipteran Drosophila melanogaster and the nematode Caenorhabditis elegans. While recombination is essential for successful execution of meiosis in C. elegans and in female fruit flies, homolog pairing and synapsis can be uncoupled from early recombination events. During female meiosis in D. melanogaster, pairing initiates in proliferating germline stem cells before oocytes enter meiosis (Christophorou et al., 2013; Joyce et al., 2013; Rubin et al., 2016) and is stabilized by SC formation. D. melanogaster males lack recombination and SCs, and have apparently evolved a distinct mechanism to stabilize homolog pairing and enable reductional segregation (McKee et al., 2012). In C. elegans, pairing and synapsis are driven by Pairing Centers, specialized sites on each chromosome bound by a family of zinc-finger proteins that mediate nuclear envelope attachment and chromosome dynamics (MacQueen et al., 2005; Phillips et al., 2005; Phillips and Dernburg, 2006; Sato et al., 2009). While nuclear envelope attachment and chromosome movement play important roles in meiotic pairing and synapsis across eukaryotes, in C. elegans they have acquired a critical role in coupling homolog pairing to synapsis initiation (Penkner et al., 2009; Sato et al., 2009).
To investigate how the meiotic program is modified during evolution, we have established tools to investigate meiosis in the free-living nematode Pristionchus pacificus. Like its distant relative C. elegans, P. pacificus is an androdioecious species, characterized by a population of mostly self-fertilizing hermaphrodites (XX) and a low frequency of males (XO) (Sommer et al., 1996). Like C. elegans, P. pacificus has a short life cycle of 3.5 days, produces large broods of about 200 progeny by self-fertilization, and is easily cultured in the lab (Hong and Sommer, 2006). Although C. elegans and P. pacificus diverged an estimated 200-300 million years ago (Pires-daSilva, 2004), they share the same number of chromosomes (2n=12) and, with the exception of one major chromosomal translocation, macrosynteny is maintained between the two species (Dieterich et al., 2008; Rödelsperger et al., 2017). P. pacificus has been established as a model for comparative studies in development, evolution and ecology (Sommer, 2015). Recent improvements in the genome assembly (Rödelsperger et al., 2017) and advances in genome editing (Lo et al., 2013; Namai and Sugimoto, 2018; Witte et al., 2015) have facilitated investigation of cell biological processes at a more mechanistic level.
In addition to these general features that make P. pacificus a tractable model system, previous studies revealed interesting variations from C. elegans. First, genome sequencing revealed the presence of the Dmc1 gene, which is absent from the entire Caenorhabditis clade (Figure 2 - Supplement 1 and Dieterich et al., 2008). Loss of Dmc1 correlates with the adaptation of recombination-independent mechanisms for pairing and synapsis in Drosophila and Caenorhabditis (Villeneuve and Hillers, 2001). Therefore, it was of great interest to us to examine how homologous chromosomes pair and synapse in the presence of Dmc1. Second, genetic linkage maps have revealed that multiple crossovers typically occur per chromosome pair during meiosis in P. pacificus (Srinivasan et al., 2003, 2002). This is in striking contrast to C. elegans, which exhibits complete crossover interference; only a single Class I CO normally occurs per chromosome pair (Martinez-Perez and Colaiácovo, 2009). These observations suggest a major difference in the mechanism of crossover control between these two species.
By employing CRISPR/Cas9-mediated genome editing techniques, genetics, immunocytochemistry and microscopy, we describe here the early events of meiotic prophase in P. pacificus. We show that homolog pairing, synapsis and CO recombination are dependent on Ppa-spo-11 and Ppa-dmc-1, while Ppa-rad-51 is not essential for meiosis. We also provide evidence that a single Class I crossover occurs per homolog pair in P. pacificus, implying that the higher recombination rate is due to Class II crossovers. Our work establishes tools for future investigation and highlights the flexibility of the meiotic program within the nematode lineage.
RESULTS
P. pacificus as a comparative model system for meiosis
The morphology and organization of the P. pacificus germline is superficially very similar to that of C. elegans. Hermaphrodites have two gonad arms in which sperm and ova are produced sequentially, while males have a single arm (Rudel et al., 2005). DAPI staining of adult hermaphrodite gonads reveals a cylindrical monolayer of cells; the distal tip is populated by proliferating germline stem cells (Figure 1A-C). As germ cells enter meiosis, their nuclear morphology abruptly becomes asymmetrical, as the chromosome mass adopts a conspicuous, crescent-shaped morphology (Figure 1A). Immediately proximal to this “transition zone,” DAPI staining reveals parallel tracks, indicative of paired and synapsed homologous chromosomes in the pachytene stage. Oocyte chromosomes in P. pacificus undergo a dramatic decondensation between diplotene and diakinesis, a stage that has been referred to as the “growth zone” (Rudel et al., 2005) or “diffuse stage” (Zickler and Kleckner, 1999). Similar diffuse chromosome morphology is observed during late prophase in other eukaryotes, but is rarely seen in C. elegans. As oocytes mature in the proximal region of the hermaphrodite gonad, they form a single row of large cells, and chromosomes condense dramatically. Six bivalents can be detected as compact DAPI-staining bodies during diakinesis, the stage preceding the first meiotic division, during oogenesis (Figure 1A, Sommer et al., 1996, Rudel et al., 2005).
It has been assumed that chromosomes in P. pacificus are holocentric, as in C. elegans, but we are unaware of direct evidence to support this idea. We thus identified a gene encoding CENP-C (hcp-4 in C. elegans), a conserved kinetochore protein, in the P. pacificus genome and inserted a V5 epitope tag at its 3’ end using CRISPR/Cas9. The appearance of mitotic chromosomes in embryos and in the premeiotic germline confirmed their holocentric organization (Figure 1C). Ppa-CENP-C::V5 forms linear structures, a hallmark of holocentric chromosomes, instead of discrete foci as observed for monocentric chromosomes.
Stable homolog pairing requires early recombination factors
BLAST searches of the P. pacificus genome revealed an open reading frame encoding an unambiguous ortholog of Dmc1, a meiosis-specific paralog of Rad51 (Table S1). An ortholog of the Dmc1 cofactor Mnd1 was also readily identified; although its partner Hop2 was not apparent among the predicted proteins or published nucleotide sequence (Figure 2 – Supplement 1). By contrast, Dmc1/Mnd1/Hop2 are absent from both C. elegans and D. melanogaster, two model organisms that have evolved recombination-independent mechanisms of homolog pairing and synapsis (Villeneuve and Hillers, 2001). We analyzed the genome sequences of other nematodes to determine the evolutionary history of these genes within the nematode lineage (Bolt et al., 2018; https://parasite.wormbase.org). This analysis revealed that Dmc1/Mnd1/Hop2 have been lost several times during the evolution of nematodes, including the entire Caenorhabditis genus and all sequenced members of Clade IV (Figure 2 – Supplement 1). Not surprisingly in light of its essential function in DNA repair, the recombinase Rad51 was detected in almost all genomes with the exception of Haemonchus contortus, Parascaris univalens and Parascaris equorum (data not shown). It is likely that the absence of Rad51 orthologs reflects incomplete genome assembly and/or annotation.
Our analysis also identified homologs of four related zinc finger proteins required for pairing center activity in C. elegans (HIM-8, ZIM-1, ZIM-2, and ZIM-3) in most of the genome sequences from Clade V. Thus, many species within this clade have genes encoding both pairing center proteins and Dmc1/Mnd1/Hop2, but Caenorhabditids have lost the latter while Pristionchus apparently diverged before the emergence of the former.
We tested whether homolog pairing in P. pacificus is dependent on these early recombination factors. To generate spo-11, dmc-1 and rad-51 null mutants, we employed TALEN-mediated gene disruption, and later CRISPR/Cas9 genome editing techniques (this study; Lo et al., 2013; Witte et al., 2015). Methods such as co-CRISPR that facilitate detection of genome editing events have not worked consistently in our hands (data not shown). Nevertheless, we were able to isolate mutant alleles by screening a large number of F1 progeny from injected hermaphrodites (see Materials and Methods). Independent alleles isolated from either TALEN or CRISPR-mediated genome editing resulted in identical mutant phenotypes. All data presented here were based on alleles generated by CRISPR/Cas9. Because balancer chromosomes are not currently available for P. pacificus, most mutations described here were maintained in unbalanced heterozygotes, which were identified by PCR-based genotyping. Self-fertilization of heterozygotes results in broods with 25% homozygous mutant animals.
As expected, disruption of either spo-11 or dmc-1 resulted in the detection of 12 DAPI-staining univalent chromosomes at diakinesis, indicative of a failure in CO recombination (Figure 5). Surprisingly, rad-51 mutants showed only mild meiotic defects (see below), and we thus confirmed the loss of RAD-51 function in mutant animals by generating multiple alleles, and confirming the absence of RAD-51 by immunofluorescence using a polyclonal antibody against Ppa-RAD-51 (Figure 4 – Supplement 1A, see Materials and Methods). Mutations in spo-11 and dmc-1, but not rad-51, also resulted in an obvious extension of the region of the germline displaying the crescent-shaped nuclear morphology characteristic of early meiosis (Figure 2 – Supplement 2). A similar “extended transition zone” phenotype is seen in C. elegans mutants that fail to synapse their chromosomes during meiosis, suggesting that spo-11 and dmc-1 might be required for synapsis in P. pacificus.
To visualize and quantify homolog pairing, we generated FISH probes against two short tandem repeats found on Chromosome X and IV (Figure 2A). We measured the distance between pairs of homologous FISH signals in individual nuclei for each genotype. To analyze pairing kinetics, we divided the distal gonads into five zones of equal length. In zone 1, which mostly consists of proliferating germ cells, pairs of FISH signals remained far apart, with an average distance of 2.4 ± 1.0 µm (SD) and 2.5 ± 0.8 µm for Chrom X and IV, respectively (Figure 2B, D). In zone 2, which spans the transition zone, the average distances decreased significantly in wild-type animals (1.2 ± 1.1 µm and 1.1 ± 1.1 µm for Chrom X and IV, respectively), and homologous signals remained closely associated in the subsequent meiotic stages. Notably, homologous FISH probes localized closer together, on average, in the two zones immediately following meiotic entry, and showed some separation in zones 4 and 5 (Figure 2B, D). By contrast, in wild-type C. elegans, homologous loci remain closely apposed in most of the distal region of the gonad (MacQueen et al., 2002). Together with analysis of synapsis (below), this indicated that desynapsis initiates shortly after completion of synapsis in P. pacificus, resulting in partial separation of homologs.
We noted that the average distances between pairs of homologous FISH signals in spo-11 and dmc-1 mutants also decreased markedly upon meiotic entry, although clearly less so than in wild type (Figure 2D). In contrast, rad-51 mutants showed distributions of probe distances similar to wild-type animals (Figure 2C, D). We considered the possibility that the proximity between FISH signals might reflect the clustering of all chromosomes within a subregion of the nucleus, which is apparent during the “transition zone” (leptotene/zygotene), rather than specific homologous interactions. If so, the extended transition zone morphology in spo-11 and dmc-1 might obscure a pairing defect that would be more apparent in the absence of clustering (Figure 2 – Supplement 2). To address this, we measured the distances between pairs of heterologous FISH signals in the premeiotic region (dispersed) versus the transition zone (clustered). We observed that heterologous FISH signals were also significantly closer to each other in the transition zone compared to premeiotic nuclei in both wild-type and mutant animals (Figure 2E). Furthermore, the distances between heterologous versus homologous pairs of FISH loci were not significantly different in spo-11 and dmc-1 mutants (p=0.1777 and p=0.6774, respectively, by Student’s t-test), but were clearly different in wild-type meiocytes (p<0.0001 by Student’s t-test) (Figure 2E). These data support the idea that clustering, rather than specific pairing, promote proximity between these loci during leptotene/zygotene in spo-11 and dmc-1 mutants. Although we cannot conclude that transient homologous pairing is absent in these mutants, it is evident that these early recombination factors are required for stable pairing and extended association of homologous loci throughout prophase. In contrast, rad-51 is dispensable for homolog pairing, as in C. elegans.
SPO-11 and DMC-1 are required for homologous synapsis
To further investigate meiotic progression in P. pacificus and to probe the role of early recombination factors in synapsis, we developed cytological markers for the chromosome axis, which normally assembles upon meiotic entry, and the synaptonemal complex (SC), which assembles between paired axes during early prophase. Identification of a candidate axial element component was straightforward, due to the presence of the easily recognized HORMA (Hop1, Rev7, Mad2) domain among members of this family of proteins (Aravind and Koonin, 1998; Vader and Musacchio, 2014). We identified a gene encoding a HORMA domain protein that is most closely related to Cel-him-3 by reciprocal BLAST analysis (Table S1). We refer to this gene as Ppa-hop-1, after the founding member of the meiotic HORMA proteins, S. cerevisiae Hop1. We raised a polyclonal antibody by genetic immunization using a fragment that encodes a peptide of 100 amino acids in length, including part of the HORMA domain, and found that this antibody indeed recognized chromosome axes from meiotic entry through late pachytene.
To enable cytological detection of synaptonemal complex (SC) assembly, we searched for homologs of SC proteins. This is notoriously challenging due to rapid divergence of these proteins, and their extensive regions of coiled-coil potential, which is associated with a strongly skewed amino acid composition. C. elegans expresses four known SC proteins, known as SYP-1, SYP-2, SYP-3, and SYP-4 (Colaiácovo et al., 2003; MacQueen et al., 2002; Smolikov et al., 2009, 2007). SYP-4 contains a distinctive C-terminal domain with several unusual motifs enriched in glycine and phenylalanine residues. We generated and sequenced an RNA library from isolated gonads, which facilitated identification of full-length Ppa-syp-4. We inserted an HA epitope tag at the C-terminus of the coding sequence via template-directed repair of a CRISPR/Cas9-induced break, and found that immunofluorescence with an epitope-specific antibody localized specifically between paired meiotic chromosomes, confirming SYP-4-HA as a marker for the SC (Figure 3A, B). The tagged protein supported normal meiosis, demonstrated by the low percentage of inviable embryos and males in the strain population (Figure 3 – Supplement 1).
HOP-1 was detected in the nucleoplasm in the premeiotic region of the germline and formed linear structures along chromosomes upon meiotic entry. SYP-4 was detected along chromosome segments shortly thereafter, and fully colocalized with HOP-1 tracks during pachytene. Notably, the region of the germline containing nuclei with fully aligned stretches of SYP-4 and HOP-1 was very short compared to C. elegans, in which SC disassembly occurs close to the bend of the gonadal arm, shortly prior to diakinesis. In contrast, SC disassembly initiated much earlier in P. pacificus; the major fraction of prophase nuclei is best described as “pachy-diplotene,” since chromosomes remain only partially synapsed. Six short stretches of SYP-4 were apparent in these nuclei, and persisted over an extended region (Figure 3A, B). This asymmetrical pattern of the SC is highly reminiscent of a more transient stage in C. elegans, in which the SC remains associated with the “short arm” of each homolog pair and subsequently contributes to the step-wise loss of cohesion during the first and second meiotic divisions (Lui and Colaiácovo, 2012).
HOP-1 localized normally to chromosome axes in spo-11 and dmc-1 mutants, but extensive SC assembly failed. Instead, small, dispersed puncta of SYP-4 were observed along chromosome axes, with occasional longer tracks (Figure 3C). In contrast to spo-11 and dmc-1 mutants, rad-51 mutants displayed robust synapsis with a distribution of stages similar to that seen in wild-type hermaphrodites (Figure 3C). These observations indicate that spo-11 and dmc-1 play crucial roles during homologous synapsis in P. pacificus, while rad-51 is dispensable for this process. This further suggests that synapsis initiation may be tightly coupled to homolog pairing and that SPO-11 and DMC-1, but not RAD-51, play central roles in this process.
DMC-1 and RAD-51 localize sequentially during distinct stages of meiotic prophase
To investigate the functions of and interplay between DMC-1 and RAD-51 in P. pacificus, we inserted a V5 epitope tag at the C-terminus of the Ppa-dmc-1 coding sequence via CRISPR/Cas9 and raised a polyclonal antibody that recognizes Ppa-RAD-51 (see Materials and Methods). DMC-1::V5 supported normal meiosis, as evidenced by a normal brood size, high embryonic viability and low percentage of males (Figure 3 – Supplement 1). Surprisingly, the two proteins showed distinct and nonoverlapping patterns of localization. DMC-1 localized broadly on chromatin in transition zone nuclei and disappeared immediately upon completion of synapsis. RAD-51 displayed a much more restricted, punctate distribution along chromosomes, and was only detected in nuclei in which DMC-1 no longer coated the chromatin (Figure 4A, B). Occasional nuclei at the border between the transition zone and pachytene region exhibited both DMC-1 and RAD-51, although DMC-1 was very faint in these nuclei and did not overlap with RAD-51 (Figure 4C). Additionally, DMC-1 remained strongly associated with chromosomes in some late nuclei that retained clustered DAPI morphology, presumably either “straggler” cells that were delayed in complete synapsis or crossover designation, or were undergoing apoptosis (Figure 4D).
We also tested the interdependence of DMC-1 and RAD-51 recombinases for their localization. In S. cerevisiae and A. thaliana, Dmc1 functions as an essential catalyst for interhomolog joint molecule formation during meiotic DSB repair, while Rad51 acts as an accessory protein for proper Dmc1 nucleofilament formation (Cloud et al., 2012; Da Ines et al., 2013). We did not detect RAD-51 along chromosomes in the transition zone, where DMC-1 was abundant on chromatin, and we found that DMC-1::V5 was normal in rad-51 mutants, indicating that RAD-51 does not play an essential role in the recruitment of DMC-1 (Figure 4 – Supplement 1A). Conversely, RAD-51 was also observed in some nuclei in dmc-1 mutants, specifically in late prophase nuclei proximal to the extended transition zone (Figure 4 – Supplement 1B). RAD-51 foci were both more abundant and larger in dmc-1 mutants compared to wild-type pachytene nuclei, perhaps due to persistent unrepaired DSBs that accumulate in the absence of DMC-1 protein. Alternatively, the bright foci of RAD-51 observed in late prophase nuclei could be indicative of an apoptotic response to unrepaired breaks and/or extensive asynapsis.
In spo-11 mutants, which are expected to lack any meiotic DSBs, DMC-1 failed to localize to chromosomes, and instead formed nuclear aggregates (Figure 4 – Supplement 1C). It was unclear whether these DMC-1 aggregates were bound to chromatin. Since DMC-1 does not appear to bind specifically at recombination intermediates in wild-type nuclei, this suggests that the absence of DMC-1 reflects a spo-11–dependent regulatory mechanism, likely through activation of a DNA damage signaling pathway, rather than an absence of potential binding sites. Consistent with cytological evidence that RAD-51 localizes specifically with recombination intermediates, the protein was absent from chromosomes in spo-11 mutants during meiotic prophase (Figure 4 – Supplement 1E). Detection of RAD-51 foci in the premeiotic region of spo-11 mutants provides a positive control for immunofluorescence (Figure 4 – Supplement 1D). Thus, our data show that DMC-1 and RAD-51 bind to chromatin at different stages of meiotic prophase and are not interdependent, although both require DSBs for their localization.
RAD-51 is not required for crossover formation or completion of meiosis
To assess the roles of DMC-1 and RAD-51 in crossover formation, we quantified the number of DAPI-staining bodies at diakinesis in dmc-1 and rad-51 mutants. Wild-type oocytes at this stage contained ∼6 DAPI-staining bodies (average= 5.6), as expected, while in spo-11 mutants, ∼12 DAPI-staining bodies were present (average= 11.5), consistent with a complete failure of crossover formation in the absence of DSBs (Figure 5A,B). Interestingly, in dmc-1 mutant germlines we frequently failed to detect oocytes at diakinesis, indicative of a defect in meiotic progression and the likely activation of a checkpoint in response to unrepaired DSBs. In cases when we did observe nuclei at diakinesis, they contained an average of 11.6 DAPI-staining bodies, reflecting a complete absence of crossovers as in spo-11 mutants (Figure 5C, D).
Somewhat surprisingly, disruption of rad-51 resulted in homozygous hermaphrodites that were viable and fertile, although animals produced smaller broods and their embryos showed greatly reduced viability (Figure 5E). Homozygous rad-51 mutant gonads also displayed diakinesis nuclei more frequently than dmc-1 mutants, although they were absent in 2 out of 20 gonads scored, indicating that loss of DMC-1 function impairs meiotic progression more severely than loss of RAD-51 (Figure 5C, D). Consistent with this observation, while rad-51 mutants had a lower average brood size compared to wild-type animals, dmc-1 mutants had even smaller broods, ranging from zero to 35 embryos laid per mutant homozygote (Figure 5E). In striking contrast to Cel-rad-51 mutants, which display fragmented chromatin aggregates at diakinesis (Martin et al., 2005; Rinaldo et al., 2002), Ppa-rad-51 mutants displayed an average of 6 DAPI-staining bodies, similar to wild-type (Figure 5B). Together with the relatively high viability of progeny of rad-51 homozygous mutants, this indicates that RAD-51 does not play an essential role in crossover formation in P. pacificus.
Conserved crossover factor COSA-1 marks Class I COs in P. pacificus
To further analyze crossover formation in P. pacificus, we identified the gene encoding the metazoan meiotic cyclin-related protein COSA-1/Cntd1 and inserted a 3xFLAG epitope tag at the C-terminus of the coding sequence via CRISPR/Cas9. The strain expressing COSA-1::3xFLAG yielded progeny with high embryonic viability and low percentage of males, demonstrating that the tagged protein functions sufficiently to support normal meiosis (Figure 3 – Supplement 1). Immunostaining with anti-FLAG antibodies revealed discrete foci along the SC, beginning as early as zygotene, which decreased in number and became more intense during the brief pachytene region (Figure 6A, B). Most pachytene nuclei displayed 6 COSA-1 foci, each of which was associated with an individual SC between each pair of homologous chromosomes (Figure 6C and Supplementary Video 1). Previous genetic mapping experiments revealed that homolog pairs often experience multiple crossovers in P. pacificus (Srinivasan et al., 2002). To investigate whether these extra crossovers might arise specifically during spermatogenesis, we analyzed younger J4 hermaphrodites whose germlines had not yet undergone the switch from spermatogenesis to oogenesis, which occurs in early adulthood, and found that pachytene nuclei undergoing spermatogenesis also displayed ∼6 COSA-1 foci (Figure 6-Supplemental Figure 1). This suggests that P. pacificus, like C. elegans, exhibits complete chromosome-wide Class I CO interference and, combined with previous genetic data, implies that additional crossovers likely arise through the Class II CO pathway, which does not require COSA-1.
Intriguingly, SC disassembly appeared to be regulated by the position of the lone Class I CO between each chromosome pair, as in C. elegans. By mid-pachy-diplotene, 6 short stretches of SYP-4::HA were observed, each associated with a single COSA-1::3xFLAG focus near one end (Figure 6B). During late pachy-diplotene, COSA-1::3xFLAG foci were no longer visible, although short stretches of SYP-4::HA could still be observed. A major difference from C. elegans that occurs in P. pacificus during SC disassembly is the visible splaying of chromosome axes upon removal of the central element component(s) (Figure 6D). HOP-1 is retained on the axes upon disassembly of the SC though the signal is more difficult to visualize as two separate axial structures. At this stage, short stretches of SYP-4::HA colocalize with corresponding bright stretches of HOP-1 (Figure 3B and Figure 6D). Bivalents at diakinesis also show a cruciform structure similar to that seen in C. elegans, indicative that only the Class I crossovers give rise to chiasmata that persist until diakinesis.
We next looked at the pattern of COSA-1::3xFLAG in various mutant backgrounds. As expected, dmc-1 mutants showed a complete absence of COSA-1 foci throughout prophase, while 6 foci were readily observed during pachytene in rad-51 mutants (Figure 7A). These observations were consistent with the number of DAPI-staining bodies we observed in diakinesis (Figure 5B). Surprisingly, a few bright COSA-1::3xFLAG foci were present throughout prophase in spo-11 mutants (Figure 7A). However, since ∼12 DAPI-staining bodies were observed during diakinesis, we conclude that these COSA-1 foci do not mark designated COs. A similar phenomenon has been reported in C. elegans spo-11 mutants (Nadarajan et al., 2017; Pattabiraman et al., 2017), suggesting that COSA-1 can coalesce at sites lacking bona fide recombination intermediates.
DISCUSSION
Distinct roles of Dmc1 and Rad51
Rad51 and Dmc1 show very similar activities in vitro, but these proteins clearly play non-redundant functions in meiosis. This is at least in part due to the activity of Dmc-1 specific cofactors, as well as differential regulation of the timing and activities of the two paralogs. Our analysis of the two RecA homologs in P. pacificus, RAD-51 and DMC-1, defines their distinct contributions during meiosis.
The pattern of extensive DMC-1 loading in transition zone nuclei is not likely to reflect an exorbitant number of DSBs or recombination intermediates present in the genome. Instead we favor the idea that DMC-1 or its cofactors is specifically regulated at this stage to promote its nonspecific association with double-stranded DNA or chromatin, and that this regulation depends on spo-11 activity. Purified yeast and human Dmc1 protein are capable of binding both ssDNA and dsDNA in vitro, although there is a strong preference for ssDNA binding (Hong et al., 2001; Li et al., 1997; Masson et al., 1999).
The sequential localization of DMC-1 and RAD-51 first suggested that they function independently, and this is supported by our analysis of loss-of-function mutations. By contrast to budding yeast and A. thaliana, Ppa-RAD-51 is dispensable for the activity of DMC-1 in pairing, synapsis, and crossover formation. Instead, RAD-51 appears to play a supporting role in DSB repair during pachytene, processing excess DSBs that remain after crossover designation has occured. In C. elegans, which expresses only RAD-51, a similar switch between two modes of double-strand break repair is observed during meiotic prophase (Hayashi et al., 2007). Association of Ce-RAD-51 with repair intermediates is differentially regulated from the onset of meiosis until a mid-pachytene transition that coincides with crossover designation; at this time, competence to convert DSBs to interhomolog COs is also lost. Recent work revealed an analogous switch from a “meiotic” repair to a “somatic”-like repair pathway during the transition from mid- to late pachytene in mouse spermatocytes (Enguita-Marruedo et al., 2019). In P. pacificus, it is thus likely that DMC-1 and RAD-51 have highly specialized functions: formation of interhomolog COs by DMC-1 in the transition zone and a more general mode of double-strand break repair by RAD-51 during pachytene. Furthermore, our observation that nuclei in rad-51 mutants display cruciform bivalents and lack fragmented chromatin at diakinesis suggests that excess DSBs can be repaired through an alternate pathway which does not depend on RAD-51 activity, such as non-homologous end joining, or that DMC-1 can compensate for RAD-51 but not the other way around. The nature of RAD-51-dependent DSB repair and how the activities of RAD-51 and DMC-1 are regulated in early meiotic prophase will be a focus of future investigation in this species.
Crossover regulation in P. pacificus
Genetic mapping in P. pacificus revealed lengths of ∼100-250 centimorgans for each chromosome, corresponding to 2-5 interhomolog crossovers per meiosis (Srinivasan et al., 2003, 2002). By contrast, the genetic map of each chromosome in C. elegans is 50 cM, reflecting a single crossover per pair (Hillers et al., 2017). Surprisingly, our analysis of Ppa-COSA-1 localization revealed only a single COSA-1 focus per chromosome by pachytene, very similar to what is seen in C. elegans (Yokoo et al., 2012). This suggests that either multiple Class I COs occur but only one retains COSA-1, or that all but a single CO forms by an alternate, COSA-1-independent pathway. In many eukaryotic systems, two CO pathways co-exist (Gray and Cohen, 2016). Class I COs show spatial interference and depend on factors including MSH-4/5, ZHP-3/Rnf212, and COSA-1/Cntd1. On the other hand, Class II COs do not exhibit interference and undergo an alternate resolution pathway that requires the structure-specific endonuclease complex Mus81-Eme1/Mms4. In some species, notably in A. thaliana and S. cerevisiae, Class II COs can occur in the absence of Class I COs, but this may not be the case in all organisms. Future work will examine whether all COs in P. pacificus arise only after the single, obligate Class I CO has been designated, and whether the designation of Class I COs is necessary for the formation and resolution of Class II COs. We further note that there appears to be only one chiasma formed between each homolog pair, which is likely created by the lone COSA-1-associated Class I CO. Our work reveals the potential of P. pacificus to address some long-standing questions about mechanisms and regulation of crossover recombination that are not accessible in the meiotic model C. elegans due to the absence of Class II COs during normal meiosis.
Comparative analysis of meiosis reveals major variations within the nematode lineage
In addition to establishing key aspects of meiosis in P. pacificus, this work also illuminates the evolutionary history of meiosis in C. elegans. A body of prior work has revealed that recombination-independent homologous synapsis in C. elegans relies on pairing centers, specialized chromosome regions that interact with nuclear envelope and drive chromosome movement during early prophase. These meiosis-specific dynamics are typically mediated by telomeres, but have shifted to a unique region on each chromosome in C. elegans. Pairing centers also act as the sites of synapsis initiation (Rog and Dernburg, 2013). By contrast, in other organisms, telomere-led chromosome movement is thought to promote homologous interactions, but stabilization of pairing and initiation of synapsis occur at early recombination intermediates, which depend on Spo11 and Dmc1 activity. Pairing center activity depends on and is largely defined by the recruitment of a family of zinc finger proteins that bind to DNA sequence motifs in these regions (Phillips et al., 2009). These proteins, known as ZIM-1, ZIM-2, and ZIM-3, and HIM-8 in C. elegans, also act as scaffolds to recruit a cascade of kinase activities required for pairing and synapsis (Harper et al., 2011; Kim et al., 2015; Labella et al., 2011). Surprisingly, most of the sequenced genomes of nematodes in Clade V include both homologs of the HIM-8/ZIM family and orthologs of Dmc1, Hop2, and Mnd1 (Figure 2 – Supplement 1). The Pristionchus genus is unusual within this Clade in that it lacks apparent homologs of the pairing center proteins, while Caenorhabditids are among the few genera that have lost Dmc1/Mnd1/Hop2, which were independently lost along the branch leading to Oscheius tiuplae and Auanema rhodensis. This suggests that the dominant role of pairing centers in homologous synapsis is likely to be recently derived, perhaps restricted to Caenorhabditids. Future analysis of species in which Dmc1 and pairing center proteins co-exist may illuminate how synapsis initiation activity relocalized from sites of Dmc1-mediated strand exchange to pairing centers, and thus how pairing centers acquired their central meiotic roles in C. elegans.
C. elegans is a popular and powerful model system for molecular studies of meiosis. Pristionchus pacificus shares many of the same experimental advantages, although naturally has fewer experimental tools, since it has been developed far more recently. Genome engineering using CRISPR/Cas9 is somewhat more challenging, and has so far been refractory to large insertions, such as fluorescent proteins. However, this barrier will likely be overcome through advances in editing efficiency. Perhaps a greater obstacle is the absence of balancer chromosomes, which are invaluable for maintaining mutations that reduce viability or fertility in C. elegans. Because balancers are unavailable for P. pacificus, most of the mutations described in this work have been maintained in unbalanced heterozygotes through frequent, labor-intensive genotyping assays. Moreover, the analysis presented here suggests that it may not be possible to construct crossover-suppressing balancer chromosomes, given that recombination is essential for homolog pairing and synapsis. By contrast, C. elegans uses a derived, recombination-independent mechanism to robustly pair and synapse homologous chromosomes, which makes it possible to propagate large-scale chromosome rearrangements that suppress recombination over large genomic regions while maintaining regular segregation of chromosomes during meiosis.
The advent of broadly applicable techniques for genome editing has enabled rapid progress towards developing P. pacificus for molecular studies, along with many other experimental models. Future exploration of meiosis in P. pacificus, and perhaps in other nematode models, will further expand our understanding of core mechanisms and plasticity of sexual reproduction.
MATERIALS AND METHODS
P. pacificus strains and maintenance
Animals were cultured on NGM media with E. coli OP50 at 20°C under the same standard conditions as C. elegans (Brenner, 1974). The wild-type strain is a derivative of PS312 designated as “97,” which was provided by Ralf Sommer. This isolate was found to be more amenable to genome editing by CRISPR-Cas9 than the parental strain. Mutant alleles were maintained in a heterozygous state. Every few generations and before each immunofluorescence or FISH experiment, single adult hermaphrodites were picked to new plates and allowed to lay embryos for two days, after which the genotype of the parent was determined by PCR genotyping. Progeny from heterozygous mothers, one-fourth of which are homozygous for the meiotic mutation, were analyzed using the assays described here; heterozygous and wild-type siblings were frequently used as controls, in addition to analysis of unedited wild-type animals.
CRISPR/Cas9 genome editing
To modify the P. pacificus genome, we adapted our preferred CRISPR/Cas9 protocol from C. elegans to P. pacificus. Equimolar quantities of Alt-R®CRISPR-Cas9 crRNA and tracrRNA molecules (Integrated DNA Technologies, Coralville, IA) were hybridized using a thermocycler (95°C for 5 minutes, then 25°C for 5 minutes). 4µl of 100µM hybridized tracrRNA/crRNA was combined with 4µl of 40µM S. pyogenes Cas9-NLS purified protein (QB3 Macrolab, UC Berkeley, Berkeley, CA) and incubated at room temperature for 5 minutes. 2µl of 100µM stock of an Ultramer® DNA oligo (IDT) repair template containing 50-60 bp homology arms and the desired epitope or mutation sequence was added to the mixture, for a total volume of 10µl, and injected into the gonads of 24 hour-post J4 adult hermaphrodites. Following a 2-4 hour recovery period, injected animals were allowed to lay embryos at 20°C for 16-20hr. Four days later, a fraction of the F1 population (typically 150-200 progeny from 6-8 injected P0s) was screened for the presence of the mutation or epitope tag sequence by PCR genotyping, and candidate alleles were verified by Sanger sequencing. A complete list of crRNA, repair template, and genotyping primer sequences used to generate alleles in this study is provided as Table S1.
TALEN constructs were generated using a protocol adapted from Zhang et al. (2011) and designed using the TAL Effector Nucleotide Targeter 2.0 website (https://tale-nt.cac.cornell.edu/).
Viability and fertility
To quantify embryonic viability, brood size, and male progeny of wild-type and rad-51 mutants, J4 hermaphrodites were picked onto individual plates and transferred every 24 hours over 72 hours total. Embryos were counted each day, after transferring the adult hermaphrodite to a new plate, and kept at 20°C. Three to four days later, adults were counted on each plate. To analyze spo-11 and dmc-1 mutants, 24 individual J4 hermaphrodites were picked from progeny of a verified heterozygous mutant hermaphrodite. Quantification was performed as in wild type, but after 72 hours, the adult hermaphrodite was lysed and genotyped for the presence of the mutation. Thus, although 24 animals total were quantified from a mixed population of spo-11/+ or dmc-1/+ animals, data from 5 homozygous spo-11 and 7 homozygous dmc-1 mutant animals are reported in the data table, Figure 5E.
Immunofluorescence and FISH
To stage animals for each experiment, 30-40 J4s were picked from a PCR-verified heterozygous mother onto a fresh plate and allowed to develop for an additional 24 or 48 hours at 20°C. Young adult hermaphrodites were dissected on a clean coverslip in egg buffer containing 0.05% tetramisole and 0.1% Tween-20. Samples were fixed for 2 minutes in egg buffer with 1% formaldehyde and transferred to a 1.5-ml tube containing PBST. After 5 minutes, the PBST was replaced with ice-cold methanol and incubated at room temperature for an additional 5 minutes. Worms were washed twice with PBST, blocked with Roche blocking reagent, and stained with primary antibodies diluted in Roche blocking solution for 1.5-2 hours at room temperature. Samples were washed with PBST and incubated with secondary antibodies raised in donkey and conjugated with Alexa-488, Cy3 or Cy5 (Jackson ImmunoResearch Laboratories, West Grove, PA). Worms were then incubated with 1µg/ml DAPI in PBST, washed with PBST, and mounted in ProLong™ Gold antifade mounting solution (Invitrogen) before imaging.
For embryo staining, 20 plates of mixed-stage worms were harvested with water and treated with 1:8 solution of bleach:water for 5 minutes at room temperature. Embryos were collected by centrifugation and washed twice with PBS. To dissolve the vitelline membrane, a solution containing 2.6 ml of n-heptane, 2 ml of PBS, and 1% paraformaldehyde (final) was added to the embryo pellet for 5 minutes at room temperature with shaking. Treated embryos were collected by centrifugation, washed twice with 5 ml MeOH, three times with PBS, and incubated with Roche blocking reagent, primary, and secondary antibodies as described above. Embryos were mounted on agarose pads for imaging.
For FISH experiments, age-matched animals were dissected and fixed as for immunofluorescence experiments described above, except that the initial fixation was for 4 minutes in 2% formaldehyde. After incubation in ice-cold methanol, worms were washed with 2x SSCT twice and incubated in 50% 2x SSCT/50% formamide solution overnight at 37°C. The next day, the worms were transferred to a small PCR tube, excess solution was removed from the sample, and a 40 µl hybridization mix containing 250ng of each probe in hybridization buffer (3.5xSSC, 58% formamide, 12.75% dextran sulfate) was added.
The sample was immediately denatured in a thermocycler at 91°C for 2 minutes and incubated overnight at 37°C. On the last day, the worms were transferred to a 1.5-ml tube and washed with 2xSSCT. After 5 minutes, the solution was replaced with fresh 2xSSCT and mounted with ProLong™ Diamond Antifade Mountant with DAPI (Invitrogen).
FISH probes
Probes for a central locus on chromosome IV and the left end of chromosome X were designed based on two short tandem repeat motifs. Tandem Repeat Finder v4.09 (Benson, 1999) was used to identify tandem repeats in P. pacificus “El Paco” genome assembly (Rödelsperger et al., 2017) using default parameters and a maximum periodicity of 200 bp. The output was then filtered to identify repeats that spanned more than 8 kb. These were compared to the genome sequence using BLAST to identify the subset of sequences restricted to a single major locus per genome. A subset of these repeats was then tested for specific and robust hybridization with oligonucleotide probes. The Chromosome IV probe targets the 30-base repeated motif TCATTGAAATGATCACAATCATTGA, which spans 40.1 kb at a position 11.3 Mb from the left end of chromosome IV. The Chromosome X probe (GGTGGTCGACGGCTGCGTCG) targets a 30-base repeat motif that spans two very close regions of 29.3kb and 11.1kb on the left end of the X chromosome. Single-stranded oligonucleotides labeled at the 3’ end with 6-FAM or Cy3 dyes were purchased from IDT and used directly as FISH probes.
Antibodies
Antibodies against Ppa-RAD-51 were generated against a 6x His-tagged N-terminal fusion protein (aa 1-103) expressed and purified from bacteria. Four mice were immunized with the antigen. Serum from one animal, designated S148, was used without purification at 1:300 dilution (Pocono Rabbit Farm and Laboratory, Canadensis, PA). Antibodies against Ppa-HOP-1 were generated by genetic immunization against aa 177-276 (SDIX, Newark, DE) and used in the following experiments at 1:300 dilution. Additional antibodies were purchased from commercial sources and diluted as follows: mouse anti-FLAG (1:500, Sigma #F1804), mouse anti-V5 (1:500, Thermo Fisher #R960-25), rabbit anti-V5 (1:250, Millipore Sigma #V8137), and goat anti-HA (1:500, Novus Biologicals #NB600-362). Secondary antibodies raised in donkey and labeled with Alexa 488, Cy3, or Cy5 were used at 1:400 dilution (Jackson ImmunoResearch Laboratories).
Orthology analysis and phylogenetic inference
Accessions to all data used in orthology analysis are available in Table S2. We downloaded the predicted protein sequences of 65 nematode species and five outgroup taxa and filtered for the longest isoform of each gene. OrthoFinder (Emms and Kelly, 2015) was used to cluster all protein sequences into putative orthologous groups (OGs) using the default inflation value of 1.5. OGs containing loci which were present in at least 75% of species and which were, on average, single copy (mean < 1.3) were selected. We aligned each selected OG using MAFFT (Katoh and Standley, 2013) and generated a maximum likelihood tree along with 1000 ultrafast bootstraps (Hoang et al., 2018) using IQ-TREE (Nguyen et al., 2015), allowing the best-fitting substitution model to be selected automatically (Kalyaanamoorthy et al., 2017). Each tree was screened by PhyloTreePruner (Kocot et al., 2013); collapsing nodes with bootstrap support <90), and any OGs containing paralogues were discarded. If two representative sequences were present for any species (i.e., “in-paralogs”) after this paralog screening step, the longest of the two sequences was retained and the other discarded. We then realigned the remaining OGs using MAFFT and trimmed spuriously aligned regions using trimAl (Capella-Gutiérrez et al., 2009). The trimmed alignments were subsequently concatenated using catfasta2phyml (available at https://github.com/nylander/catfasta2phyml) to form a supermatrix. We inferred the nematode species tree using IQ-TREE with the general time reversible model (GTR) with gamma-distributed rate variation among sites. The resulting tree was visualized using iTOL (Letunic and Bork, 2016).
We identified the OGs which contain orthologs of DMC-1, MND-1, HOP-2 and RAD-51 using BLASTP to search using the orthologous protein sequences from Homo sapiens as queries. The OG containing the C. elegans proteins HIM-8, ZIM-1, ZIM-2, ZIM-3 was identified using the appropriate transcript IDs. Each OG was aligned using MAFFT and a gene tree was inferred using IQ-TREE, allowing the best-fitting substitution model to be selected automatically. Each gene tree was visually inspected using iTOL.
Total RNAseq
Total RNA was isolated from 20 whole worms or 30 dissected gonads from 48 h post J4 animals using TRIzol (Invitrogen). TruSeq Stranded Total RNA (Illumina) sequencing libraries were constructed following the manufacturer’s instructions. For both conditions, three independent libraries were constructed. Libraries were pooled and sequenced on a HiSeq4000 (150bp, PE, QB3 Vincent J. Coates Genomics Sequencing Laboratory). Reads were mapped to the “El Paco” genome assembly including annotated splice sites (Rödelsperger et al., 2017) using STAR. To correct misannotated splice sites, a transcriptome was then reconstructed de novo using StringTie. Transdecoder was used to generate potential open reading frames. To identify potential meiotic genes, we identified genes enriched in dissected gonads over whole worms using HTSeq and edgeR.
Supplementary Video 1. COSA-1/Cntd1 accumulates at a single site per chromosome pair. 3D volume rendering of a single nucleus from the pachytene region of a wild-type worm expressing COSA-1::3xFLAG (green) and SYP-4::HA (magenta). Each frame is a maximum intensity projection from a region of a deconvolved 3D image. Each stretch of SYP-4::HA is associated with a COSA- 1::3xFLAG focus.
Acknowledgements
This work was supported by funding from the Howard Hughes Medical Institute to AFD and a fellowship from the Helen Hay Whitney Foundation to JB. This work used the Vincent J. Coates Genomics Sequencing Laboratory at UC Berkeley, supported by NIH S10 OD018174 Instrumentation Grant. We are grateful to Ralf J. Sommer, Ray Hong, and other members of the P. pacificus research community for providing us with P. pacificus strains and abundant helpful advice. We thank members of the Dernburg lab for helpful discussions about this work and critical reading of the manuscript.