The chromosome axis controls meiotic events through a hierarchical assembly of HORMA domain proteins

The structure of HIM-3 reveals a conserved C-terminal “closure motif”

To establish the molecular basis for meiotic HORMA domain protein function, we sought to determine these proteins’ 3D structures. The four C. elegans meiotic HORMA domain proteins share a similar domain structure, with an N-terminal HORMA domain and a variable-length C-terminal tail predicted to lack intrinsic structure, ranging from ~50 residues in HIM-3 to ~100 residues in HTP-1 and HTP-2, and over 500 residues in HTP-3.

We first crystallized and determined the 3D structure of full-length HIM-3 to a resolution of 1.75 Å (Table 1). Residues 19-248 (of 291) of HIM-3 adopts a HORMA domain fold, consisting of a seven-stranded β-sheet backed on one side by three long α-helices (Figure 1A). This structure is extremely similar to those of Mad2 and Rev7 in their “closed” conformations (Figure S1A). The most significant structural difference between HIM-3 and previously characterized HORMA domain proteins is an extended loop between β5 and αC (residues 127-158 in HIM-3), which drapes over the β-sheet and forms an additional short α-helix not observed in Mad2 or Rev7. This loop is intimately associated with the C-terminal β-strands β8′ and β8″ through both ionic and Van der Waals interactions, potentially stabilizing HIM-3 in the closed state (see below).

In the closed states of Mad2 and Rev7, a peptide ligand forms a β-strand that packs against strand β6 of the HORMA domain and is held in place by a “safety belt” encompassing β-strands β7, β8′, and β8″ (Hara et al., 2010; Luo et al., 2002; Sironi et al., 2002). In our structure of HIM-3, experimental electron density maps clearly revealed a peptide forming a β-strand apposed to β6. The high resolution of our structure enabled us to unambiguously identify this peptide as residues 278-286 of HIM-3 (Residues ~249-277 are disordered in our structure; Figure 1, S3A). Because HIM-3 adopts a closed conformation around this C-terminal motif, we designate it the HIM-3 “closure motif”.

Sequence alignments of HIM-3 orthologs in related nematodes revealed that the HIM-3 closure motif is highly conserved, particularly a Pro-Tyr-Gly motif that is closely associated with the “safety belt” of the HORMA domain in our structure (Figure 1B,C, S1B). We also identified a similar conserved motif near the C-termini of HTP-1 and HTP-2 and their orthologs, which contains a nearly invariant Pro-Tyr-Ser motif (Figure 1C, S1C). The long C-terminal tail of HTP-3 has a repetitive sequence in which six motifs bearing signatures of the HIM-3 closure motif can be discerned (Figure 1C). Four of these motifs (#2-5) are nearly identical to one another, and flanking these four motifs are two variable motifs (#1 and 6), with motif #6 located at the extreme C-terminus of HTP-3. Sequence alignments of HTP-3 orthologs from related nematodes reveal only limited sequence homology outside the HORMA domain, but we were able to identify potential closure motifs in the C-termini of HTP-3 orthologs from C. briggsae, C. remanei, C. brenneri, C. japonica, and C. sp. 5 (Figure S1D).

C. elegans meiotic HORMA domain proteins form intermolecular complexes

The identification of several putative “closure motifs” within the tail of HTP-3, together with the knowledge that HTP-3 is required for association of the other meiotic HORMA domain proteins with the axis, suggested that axis assembly might be mediated by HORMA domain:closure motif interactions. To test this idea, and to identify other components associated with the C. elegans meiotic chromosome axis, we purified HTP-3-containing protein complexes from a worm strain expressing a functional htp-3-gfp fusion (ieSi6). Mass spectrometry analysis of soluble HTP-3 complexes indicated that the protein physically associates with HIM-3, HTP-1, and HTP-2 (Figure 2A). A parallel analysis of chromosome-associated HTP-3 complexes additionally detected multiple cohesin subunits, including the meiosis-specific kleisin subunits REC-8, COH-3, and COH-4 (Figure 2A). These results indicate that C. elegans axial elements are mainly composed of cohesin complexes and the four meiotic HORMA domain proteins, consistent with prior genetic and cytological studies (Couteau and Zetka, 2005; Goodyer et al., 2008; Martinez-Perez and Villeneuve, 2005; Zetka et al., 1999).

To investigate how the four HORMA domain proteins interact with one another, we used a bacterial polycistronic expression system to co-express various combinations of the four proteins. When expressed individually, both HIM-3 and HTP-3 were soluble and monomeric in solution, as judged by size exclusion chromatography and multi-angle light scattering (SEC-MALS) (Figure 2B, C, S2). HTP-1 formed soluble aggregates when expressed on its own (not shown), but when HTP-1 was co-expressed with HIM-3 we detected soluble complexes with a stoichiometry of 1 HIM-3:1 HTP-1 (Figure 2B, C). Co-expression of HIM-3, HTP-1, and HTP-2 yielded a mixture of 1 HIM-3:1 HTP-1 and 1 HIM-3:1 HTP-2 complexes (not shown).

When HTP-3 was co-expressed with different combinations of HIM-3, HTP-1, and HTP-2, the resulting complexes were heterogeneous. Analysis of the composition and average molecular weights of these complexes indicated that multiple copies of the smaller HORMA domain proteins can associate simultaneously with a single HTP-3 (Figure 2, S2). Importantly, we found that HTP-1 and HIM-3 could each associate independently with HTP-3 (Figure 2C). Together, these results reveal that HTP-3 is able to associate simultaneously with multiple copies of HIM-3, HTP-1, and HTP-2.

Specific HORMA domain-closure motif interactions define a hierarchical assembly

Our reconstitution experiments showed that HIM-3, HTP-1, and HTP-2 can each bind directly to HTP-3, and that HTP-1 and HTP-2 can also independently bind HIM-3. By co-expressing truncated proteins, we further found that the HIM-3:HTP-1 interaction requires the C-terminal tail of HIM-3 (co-expression of HTP-1 with HIM-3 (1-245) did not result in a complex despite expression of both proteins; not shown), and also required the HORMA domain of HTP-1, but not its C-terminal tail. These results strongly suggest that the HIM-3-HTP-1 interaction is mediated by HORMA domain-closure motif binding in trans (Figure 2B; compare measured stoichiometry of HTP-1:HIM-3 with HTP-1 (1-253):HIM-3). Using a fluorescence polarization (FP) peptide-binding assay, we measured the affinity and specificity of HORMA domain:closure motif interactions. We found that HIM-3 binds strongly to a peptide that includes the HTP-3 motif #4 sequence (K_d 0.3 μM; Figure 3A), and is therefore likely to also bind motifs #2, #3, and #5. However, HIM-3 did not bind a randomized version of HTP-3 motif #4, nor did it bind the divergent HTP-3 motifs #1 or #6. As full-length HIM-3 was used for this binding assay, our results show that HTP-3 motif #4 can effectively compete for binding to the HORMA domain of HIM-3 in the presence of the closure motif within HIM-3’s tail. By contrast, full-length HIM-3 did not effectively bind in trans to the motif found within its own tail. To further characterize the selectivity of the HIM-3 HORMA domain, we co-expressed untagged HIM-3 HORMA domain (residues 1-245) in E. coli together with tagged closure motif peptides and measured HIM-3 association using a pulldown assay. In this assay, the HIM-3 HORMA domain bound strongly to HTP-3 motif #4, but again did not detectably bind its own closure motif (not shown). Together with results from the FP peptide-binding assay described above, this indicates that binding of the HORMA domain of HIM-3 to the closure motif within its own tail is a low-affinity interaction that depends on the high effective concentration imparted by covalent association between the HORMA domain and tail.

We next measured the binding of HTP-1 and HTP-2 to various closure motif peptides (Figure 3B,C). As noted above, both HTP-1 and HTP-2 form soluble aggregates when expressed in isolation, and the resulting purified proteins did not specifically bind any peptides in the FP assay (not shown). However, co-expression of His₆-tagged HTP-1 or HTP-2 with untagged HTP-3 resulted in a mixed population of HTP-3:HTP-1/2 complexes and free monomeric HTP-1/2, suggesting that HTP-3 may act as a chaperone for these proteins when they are co-expressed in E. coli. When these monomers were purified away from their respective HTP-3 complexes, both HTP-1 and HTP-2 bound strongly to the HIM-3 closure motif and to HTP-3 motifs #1 and #6, but did not associate detectably with HTP-3 motif #4 (Figure 3B,C). The two proteins showed similar specificities, with only subtle differences in binding affinities (Figure 3D).

Full-length HTP-3 did not detectably associate with either the HIM-3 or HTP-2 closure motif peptides in our peptide-binding assay (not shown). In co-expression/pulldown tests, the HTP-3 HORMA domain was also unable to interact with any of the identified closure motifs in its own C-terminus (not shown); we were unable to confirm this result through peptide binding assays due to the difficulty in purifying the truncated HTP-3 construct. Overall, these results are consistent with the idea that HTP-3 binds an as-yet-unidentified partner to mediate its localization to the meiotic chromosome axis.

Finally, none of the four proteins associated with the putative closure motif at the C-terminus of HTP-2. Notably, this motif and that of HTP-1 contain a serine in place of the highly conserved glycine in other closure motifs; our in vitro and in vivo data (see below) suggest that larger amino acid side chains at this position might interfere with binding. It is possible that the highly-conserved C-terminal motifs in HTP-1 and HTP-2 act as low-affinity closure motifs that facilitate HTP-1/2 folding or stability, but evidence supporting this idea is currently lacking. Taken together, our peptide binding assays are consistent with the knowledge that HIM-3, HTP-1, and HTP-2 require HTP-3 to associate with the meiotic chromosome axis, and suggest that that they are recruited to the axis by binding of their HORMA domains to specific closure motifs within HTP-3 and/or HIM-3 (Figure 3E).

Specific closure motifs within HTP-3 recruit HIM-3 to mediate homologous synapsis

To gain insight into the functional importance of the interaction between HIM-3 and HTP-3, we determined the structure of a chimeric protein in which the closure motif of HIM-3 was replaced by HTP-3 motif #4. This structure shows the HIM-3 HORMA domain in essentially the same conformation as in the native HIM-3 structure (Cα r.m.s.d. of 0.62 Å over 228 residues), and shows strong electron density for residues 649-662 of HTP-3 (Figure 4A, B, S3B) within the binding pocket. The interaction of the HIM-3 HORMA domain with HTP-3 motif #4 is much more extensive than with its own closure motif, burying 1620 Ų of total surface area versus 1150 Ų for the intramolecular binding interaction, consistent with the higher affinity measured in our peptide-binding and pull-down assays. HTP-3 residues 650-654 adopt a β-strand conformation and form a continuous β-sheet with HIM-3 β6 (within the HORMA domain) and β7 (in the “safety belt”). The tyrosine residue conserved in all closure motifs (Y651 in HTP-3 motif #4) is positioned close to the HIM-3 safety belt, and makes a hydrogen bond with asparagine 139 of HIM-3. The following residue, glycine 652, is buried against the HIM-3 HORMA domain. Arginine 653, conserved in HTP-3 motifs #2-5, makes a salt bridge with HIM-3 glutamate 115. This arginine residue is conserved in the majority of closure motifs we identified in Caenorhabditis HTP-3 orthologs (Figure S1D), and glutamate 115 is highly conserved in Caenorhabditis HIM-3 orthologs; this interaction may therefore represent a specific feature of HTP-3:HIM-3 complexes. The C-terminus of the closure motif (HTP-3 residues 657-662) forms an amphipathic α-helix that packs against a hydrophobic cavity in the HIM-3 HORMA domain. The hydrophobic residues in this helix are conserved within HTP-3 motifs #2-5, but in motifs #1 and #6 the corresponding side chains are charged or polar. Thus, several distinct features of HTP-3 closure motifs #2-5 contribute to HIM-3’s preference for these sequences.

In all of our HORMA domain:closure motif structures, a conserved glycine near the N-terminus of the closure motif is packed tightly against a hydrophobic surface in the HORMA domain (Figure 4B,C). We found that mutation of this glycine to lysine in HTP-3 motif #4 (G652K) abolished binding to HIM-3 in two independent in vitro assays (Figure 4D, S4A). We therefore created HTP-3^4GK, in which the conserved glycine residues in the four central HTP-3 closure motifs (#2-5) were all mutated to lysine. As expected, this 4GK mutation strongly abrogated binding of HIM-3 to HTP-3 when the proteins were co-expressed in E. coli, but did not affect binding of HTP-1 to HTP-3 (Figure 4E). This result agrees with our peptide binding data, and validates the idea that HTP-3 motifs #2-5 are specific binding sites for HIM-3.

To determine the consequences of disrupting HIM-3:HTP-3 binding in vivo, we engineered an htp-3-gfp transgene with the 4GK substitutions, which was used to generate transgenic animals by Mos1-mediated single copy insertion (MosSCI) (Frøkjaer-Jensen et al., 2008) at a defined genomic locus (ttTi5605 on Chr II). Worms expressing either wild-type or 4GK mutant HTP-3-GFP were crossed to the null allele htp-3(tm3655). The wild-type htp-3-gfp transgene supported normal chromosome axis and SC assembly dynamics (Figure 4F-H) and fully rescued the viability of embryos produced by homozygous animals (100% of embryos from htp-3(tm3655) I; htp-3^WT II were viable, compared to 13% from htp-3(tm3655) I) (Figure S4C, D). HTP-3^4GK-GFP showed slightly lower protein expression compared to the wild-type tagged protein (Figure S4B), but localized properly to the chromosome axes in meiotic nuclei (Figure 4F). However, animals expressing HTP-3^4GK-GFP produced mostly dead eggs (only 3.1% of embryos produced by self-fertilizing htp-3(tm3655) I; htp-3^4GK II survived to adulthood), and 27.5% of the few survivors were males (Figure 4F, S4C, D), indicating that mutation of these central four motifs leads to severe defects in meiotic chromosome segregation. Consistent with our in vitro data, HIM-3 failed to load onto the chromosomes in the HTP-3^4GK mutant, while HTP-1 was still detected along the axes (Figure 4F). As in him-3 mutants, chromosomes failed to pair and synapse in the HTP-3^4GK mutant (Figure 4G, H). These results demonstrate that HTP-3 closure motifs #2-5 play a critical role in chromosome axis assembly and homolog synapsis through their recruitment of HIM-3 to the chromosomes. The localization of HTP-1 to the chromosomes in the HTP-3^4GK mutant further confirms that HTP-1/HTP-2 recruitment can be mediated by direct binding to HTP-3, presumably through motifs #1 and #6, in the absence of HIM-3.

Redundant mechanisms recruit HTP-1 and HTP-2 to the meiotic chromosome axis

The HIM-3 HORMA domain binds specifically to HTP-3 closure motifs #2-5. In contrast, HTP-1 and HTP-2 preferentially bind to HTP-3 motifs #1 and #6, and bind the HIM-3 closure motif with similarly high affinity (Figure 3B-D). To determine the basis for their specificity, we co-expressed and purified the HORMA domains of both HTP-1 and HTP-2 (residues 1-253 of 352) with different closure motifs. HTP-2 readily co-crystallized with the HIM-3 motif, HTP-3 motif #1, and HTP-3 motif #6, and we determined structures for each of these complexes (Figure 5D-G, Table 1). When initial crystallization screens for HTP-1:closure motif complexes failed, we examined the crystal packing interactions of HTP-2 complexes. Based on this analysis, we designed a point mutant of HTP-1 (proline 84 to leucine, P84L) to mimic a specific crystal packing interaction observed in HTP-2 structures. This residue resides on a surface-exposed α-helix (αB) and is poorly conserved among HTP-1/HTP-2 orthologs, so this mutation is unlikely to have a significant effect on HTP-1’s structure or binding. Complexes containing the mutant HTP-1(1-253)^P84L crystallized readily in conditions similar to those for HTP-2(1-253) complexes, and we determined its structure bound to both the HIM-3 closure motif and HTP-3 closure motif #1 (Figure 5A-C).

HTP-1 and HTP-2 share 82% sequence identity, and their HORMA domains have nearly identical crystal structures, with an overall Cα r.m.s.d of less than 0.5 Å (Figure 5A, D). Both are also highly similar to HIM-3 (~1.0 Å r.m.s.d between 180 aligned Cα atoms), with most of the differences in flexible loop regions. The residues that form the closure motif binding pockets within HTP-1 and HTP-2 are identical, and their modes of interaction with closure motifs are also virtually indistinguishable (e.g. compare Figure 5B and E). The highly conserved Tyr-Gly pair within the closure motifs adopts an identical conformation in all HTP-1/2 structures, tightly engaged by strand β7 of the associated HORMA domain’s “safety belt”. The major difference when compared to HIM-3 is the conformation of the C-terminal portion of the bound closure motifs: while a HIM-3-bound closure motif adopts a short amphipathic α-helix (Figure 4B), all three motifs that bind HTP-1 and HTP-2 adopt a β-strand conformation at both ends of the motif, with a small bulge in the middle (Figure 5). This bulge is one residue longer in HTP-3 motif #1 than in the other two motifs, but otherwise the three motifs are bound by the HTP-1 HORMA domain in a nearly identical manner (compare Figure 5E with 5F and 5G; alignment in Figure 5H). Each of these three motifs also possesses an arginine N-terminal to the conserved Tyr-Gly pair that packs into the same negatively charged pocket on the HORMA domain, despite variation in its position within each closure motif . Other interactions in the HTP-1 and HTP-2 complexes are mostly backbone-backbone hydrogen bonds characteristic of β-sheets, providing little sequence specificity. This is reminiscent of Mad2, whose known binding peptides in Cdc20, Mad1, and Shugoshin bear only weak sequence similarity (Luo et al., 2002; Orth et al., 2011).

We next generated disruptive glycine-to-lysine mutations in HTP-3 motifs #1 and #6 (motif #1, G490K; motif #6, G728K; both, 2GK), and tested for binding by HIM-3 or HTP-1 when co-expressed in E. coli. None of these mutations affected HIM-3 binding (Figure 5I), while HTP-1 binding was clearly affected: Mutation of HTP-3 motif #1 (G490K) alone significantly reduced HTP-1 binding. Unexpectedly, mutation of motif #6 (G728K) had no detectable effect on HTP-1 binding, yet did appear to synergize with mutation of motif #1, since the HTP-3^2GK mutant did not detectibly bind to HTP-1 (Figure 5I). This suggests that HTP-1 interacts more strongly with HTP-3 motif #1 than #6, consistent with results from our peptide binding assays (Figure 3B, C). Together with our peptide binding and structural data, these results confirm that HTP-3 motifs #1 and #6 are specific binding sites for HTP-1 and HTP-2.

We next introduced these same mutations (HTP-3 motif #1 (G490K), motif #6 (G728K), and the double mutant (2GK)) into worms by MosSCI and crossed each transgene to the htp-3(tm3655) null allele to test the roles of these motifs in meiotic chromosome axis assembly and synapsis. Surprisingly, all three mutant transgenes supported high levels of embryonic viability (G490K, 96.1%; G728K, 100%; 2GK, 98.6%). Consistent with this, in each of these mutants, HTP-1/2 and HIM-3 were detected at the meiotic chromosome axis (Figure 5J, S5A), and robust homologous synapsis was observed, albeit with a modest delay in animals expressing HTP-3^G490K or HTP-3^2GK (Figure 5K, S5B, C). Because the double mutant (HTP-3^2GK) protein failed to bind HTP-1 in vitro (Figure 5I), we reasoned that axis localization in this mutant is likely mediated through binding of HTP-1/2 to HIM-3, which in turn is recruited by HTP-3. Consistent with this, when HTP-3^2GK was crossed into the him-3(gk149) null background, HTP-1/2 was drastically reduced on meiotic chromosomes, and chromosome synapsis was eliminated (Figure 5J, K). Similar results were observed when all six closure motifs within HTP-3 were mutated (HTP-3^6GK), which eliminated recruitment of HIM-3, HTP-1, and HTP-2 (Figure 5J, K). Notably, we found that animals expressing a truncated HIM-3 protein lacking the closure motif (residues 1-245) were proficient to recruit HTP-1/2 to the chromosome axis and showed no meiotic nondisjunction (Figure S5A). Taken together, these results indicate that HTP-1/2 can be recruited to the axis by binding to HTP-3 motifs #1 and #6, or by binding to the closure motif within HIM-3, and that either mode of HTP-1/2 recruitment is sufficient to support homologous synapsis and meiotic chromosome segregation.

HTP-3 promotes chromosome cohesion and DSB formation, independent of HIM-3 and HTP-1/HTP-2 recruitment to the meiotic chromosome axis

In addition to promoting homolog pairing and synapsis and double-strand break formation, HTP-3 is required to maintain sister chromatid cohesion until meiosis II (Severson et al., 2009). Our evidence that HTP-3 recruits the other HORMA domain proteins to the axis through motifs in its C-terminal tail enabled us to test whether these roles of HTP-3 require the other HORMA domain proteins. To determine how HTP-3 contributes to cohesion, we compared cohesin loading in htp-3 null mutants vs. animals expressing the engineered HTP-3^6GK variant, which shows normal axis association but fails to recruit HIM-3 and HTP-1/2. In wild-type animals, cohesin complexes containing the common SMC subunits SMC-1 and SMC-3, but with distinct kleisin subunits (REC-8 or COH-3/COH-4), localize along meiotic chromosome axes from meiotic entry through late pachytene (Pasierbek et al., 2001; Severson et al., 2009). While SMC-1, SMC-3 and COH-3/COH-4 were detected on chromosomes in two htp-3 null mutants (tm3655 and y428; data not shown for y428) (Figure 6B, S6A, B), REC-8 staining was greatly diminished (Figure 6A), as previously reported (Severson et al., 2009). By contrast, REC-8 was clearly retained on meiotic chromosomes in HTP-3^6GK mutants (Figure 6A). These results indicate that HTP-3 promotes sister chromatid cohesion through loading and/or maintenance of REC-8, but not COH-3/COH-4, and that the requirement for HTP-3 in REC-8 loading is independent of its role in recruiting HIM-3 or HTP-1/HTP-2 to the axis.

Previous work has indicated that HTP-3 is required for the formation of meiotic DSBs (Goodyer et al., 2008). Thus we examined DSB formation in HTP-3^6GK mutants by staining for RAD-51, a DNA strand exchange protein that marks recombination intermediates (Colaiácovo et al., 2003; Sung, 1994). While most germline nuclei in htp-3 null animals lacked RAD-51 foci, consistent with prior work, RAD-51 foci appeared and disappeared with normal kinetics in HTP-3^6GK animals despite the absence of pairing, synapsis, or crossover formation (Figure 6C, S6C). These results indicate that HTP-3’s role in DSB formation and/or processing, which has been proposed to involve a direct interaction with the Rad50/Mre11 complex (Goodyer et al., 2008), is also independent of its recruitment of HIM-3 and HTP-1/HTP-2 (Figure 6D). The timely disappearance of RAD-51 foci suggests that axial association of HTP-3 is insufficient to pose a barrier to intersister recombination, consistent with previous evidence that HIM-3 and HTP-1 are required for this barrier in C. elegans (Couteau and Zetka, 2005; Couteau et al., 2004).

Meiotic HORMA domain:closure motif interactions are a conserved feature of axis organization

In mouse spermatocytes and oocytes, two meiosis-specific HORMA domain proteins, HORMAD1 and HORMAD2, co-localize on the chromosome axis (Wojtasz et al., 2009). These proteins also physically interact both in vivo and in vitro (Wojtasz et al., 2012). Like the four C. elegans proteins analyzed here, these mammalian proteins contain a HORMA domain followed by a C-terminal tail predicted to lack inherent structure. The tails of mammalian HORMAD1 and HORMAD2 orthologs are highly conserved along their entire lengths, making it impossible to identify putative closure motifs from sequence alone. However, when we compared HORMAD1 to HORMAD2, we noticed two short regions of homology within their tails, including one at their extreme C-termini (Figure 7A), suggesting that this region might mediate protein-protein interactions.

To determine whether HORMAD1 and HORMAD2 might interact in a manner similar to the C. elegans HORMA domain proteins, we purified the HORMA domain (residues 2-235 of 394) from human HORMAD1, expressed in E. coli. Separately, we expressed and purified the C-terminal tails of HORMAD1 (residues 236-394) and HORMAD2 (residues 245-307), each fused to an N-terminal His₆-maltose binding protein (MBP) tag. In a Ni²⁺ pulldown assay, we found that both full-length tails strongly interacted with HORMAD1(2-235) (Figure 7B, C), suggesting that each tail contains a peptide that can act as a HORMAD1 closure motif. By expressing and purifying tail fragments, we mapped this binding activity to the conserved region at these proteins’ C-termini (residues 375-394 of HORMAD1 and 283-307 of HORMAD2). Thus, human HORMAD1 binds a putative closure motif present in the C-terminal tails of both HORMAD1 and HORMAD2 (Figure 7D), indicating that the HORMA domain-tail interactions we have identified in C. elegans are conserved in mammals. It is likely that some details of these interactions differ among lineages, given the variable number of meiotic HORMA proteins and our evidence that mammalian HORMAD1 can bind either its own C-terminus or that of HORMAD2, in contrast to the hierarchical interactions we have documented for the C. elegans protein family.

Mass Spectrometry Analysis

C. elegans expressing htp-3-gfp (ieSi6) were synchronized by bleaching and grown in liquid culture at 20°C until worms reached the young adult stage. Animals were harvested by sucrose flotation, frozen, and disrupted using a mixer mill. HTP-3 complexes (soluble or chromatin-associated) were purified using a GFP-binding protein (Rothbauer et al., 2008). Trypsin-digested samples were analyzed for protein identification by MudPIT (Washburn et al., 2001). To enrich for chromatin-associated proteins, worms were incubated in hypotonic buffer and then homogenized in a dounce homogenizer, then spun at low speed to separate nuclei from cell debris, further purified over a sucrose cushion, then treated with micrococcal nuclease to digest chromatin prior to immunoprecipitation as above.

In vitro reconstitution and peptide binding

ORFs were amplified from a cDNA library and cloned into E. coli expression vectors fused to His₆-, His₆-maltose binding protein (MBP), or Strep-tag II sequences (in vitro reconstitution). Expressed proteins were purified from bacterial lysates using affinity (Ni²⁺ or Strep), ion-exchange, and size-exclusion chromatography. For crystallization, proteins were concentrated and stored at 4°C, and for biochemical experiments, proteins were aliquoted and frozen at −80°C. Purified samples were characterized by size-exclusion chromatography (Superose 6, GE Life Sciences) and multi-angle light scattering (Wyatt Technologies mini-DAWN Treos) to determine molecular weight and complex stoichiometry. FITC-labeled peptides (BioMatik) at 50 nM were incubated with 12 nM-12.5 μM HORMA protein, fluorescence polarization was read using a TECAN Infinite M1000 PRO fluorescence plate reader, and binding data were analyzed with Graphpad Prism v. 6 using a single-site binding model. For Ni²⁺ pulldown assays, human HORMAD1 was purified as above, and HORMAD1/2 tail sequences were fused to His₆-MBP and purified. Proteins were mixed, then bound to Ni-NTA beads (Qiagen), washed, and bound proteins were visualized by SDS-PAGE.

Crystallization and Structure Solution

Wild-type HIM-3 was purified as above, crystals were grown and cryoprotected in sodium malonate, and the structure was determined using a 2.0 Å resolution single-wavelength anomalous (SAD) selenomethionine dataset. For crystallization of HIM-3 bound to HTP-3 closure motif #4, residues 277-289 of HIM-3 were replaced by residues 649-663 of HTP-3. For crystallization of HTP-2 complexes, HTP-2 residues 1-253 was co-expressed with a closure motif fused to a TEV protease-cleavable His₆-MBP tag, and complexes purified as above. For crystallization of HTP-1 complexes, HTP-1 residues 1-253, with proline 84 mutated to leucine, was co-expressed with closure motifs as for HTP-2. All HTP-1/HTP-2:closure motif complexes were crystallized and cryoprotected in polyethylene glycol, and structures were solved by SAD or molecular replacement (Table 1).

C. elegans Strains and Immunofluorescence

Strains carrying wild-type, mutant htp-3-gfp, and a truncated him-3 (residues 1-245) transgenes were generated by MosSCI (Frøkjaer-Jensen et al., 2008). Homozygous insertions at the ttTi5605 locus (Chr II) were verified by PCR, and crossed into htp-3(tm3655) or him-3(gk149) to examine the loading of HORMA domain proteins, cohesin loading and synapsis. Immunofluorescence of dissected gonads was performed as previously described (Phillips et al., 2009).