Meiotic chromosome synapsis depends on multivalent SYCE1-SIX6OS1 interactions that are disrupted in cases of human infertility

Sánchez-Sáez et al., Sci. Adv. 2020; 6 : eabb1660 2 September 2020 SCIENCE ADVANCES | RESEARCH ARTICLE 1 of 14 BIOCHEMISTRY Meiotic chromosome synapsis depends on multivalent SYCE1-SIX6OS1 interactions that are disrupted in cases of human infertility Fernando Sánchez-Sáez1*, Laura Gómez-H1*, Orla M. Dunne2, Cristina Gallego-Páramo2, Natalia Felipe-Medina1, Manuel Sánchez-Martín3, Elena Llano4, Alberto M. Pendas1†, Owen R. Davies2† Meiotic reductional division depends on the synaptonemal complex (SC), a supramolecular protein assembly that mediates homologous chromosomes synapsis and promotes crossover formation. The mammalian SC has eight structural components, including SYCE1, the only central element protein with known causative mutations in human infertility. We combine mouse genetics, cellular, and biochemical studies to reveal that SYCE1 undergoes multivalent interactions with SC component SIX6OS1. The N terminus of SIX6OS1 binds and disrupts SYCE1’s core dimeric structure to form a 1:1 complex, while their downstream sequences provide a distinct second interface. These interfaces are separately disrupted by SYCE1 mutations associated with nonobstructive azoospermia and premature ovarian failure (POF), respectively. Mice harboring SYCE1’s POF mutation and a targeted deletion within SIX6OS1’s N terminus are infertile with failure of chromosome synapsis. We conclude that both SYCE1-SIX6OS1 binding interfaces are essential for SC assembly, thus explaining how SYCE1’s reported clinical mutations give rise to human infertility. INTRODUCTION Meiotic cell division is defined by a unique and highly dynamic program of events that result in homologous chromosome synapsis, crossover (CO) formation, and subsequent homolog segregation into haploid germ cells (1–3). Homologous chromosome pairs are established through interhomolog recombination searches from up to 400 induced double-strand breaks (DSBs) per cell (4). Once established, local recombination-mediated alignments are converted into the single continuous synapsis of aligned homologous chromosomes through the zipper-like assembly of the synaptonemal complex (SC) (5). The SC’s supramolecular protein structure mediates continuous 100-nm tethering between homologous chromosome axes and provides the necessary three-dimensional framework for crossover formation (2). Following SC disassembly, crossovers provide the sole physical links between homologs at metaphase I, so are essential for ensuring correct homolog segregation in addition to providing genetic diversity (2). The SC has an iconic and highly conserved tripartite structure that has been observed across meiotically reproducing eukaryotes (6). This consists of lateral elements (LEs) that coat the two homologous chromosome axes and a midline central element (CE), with a series of transverse filaments that bind together these longitudinal electron-dense structures (Fig.1A) (7). The protein components of the mammalian SC have been identified as transverse filaments protein SYCP1 (Synaptonemal complex protein 1) (8), CE proteins SYCE1, SYCE2, and SYCE3 (Synaptonemal complex central element proteins 1 to 3), SIX6OS1, and TEX12 (Testis-expressed protein 12) (9–12), and LE proteins SYCP2 and SYCP3 (13,14). All transverse filament and CE components are essential for SC assembly, and their individual disruption leads to infertility owing to meiotic arrest with failure of DSB repair (10,11,15–18). In contrast, disruption of LE components produces a sexual dimorphism of male infertility and female subfertility (19,20), with SYCP3 deficiency in females promoting germ cell aneuploidy and embryonic death (21). In recent years, a variety of cellular imaging, biochemical and structural biology approaches have begun to uncover the molecular structures, interactions, and mechanisms responsible for mammalian SC assembly. SYCP1 self-assembles into a supramolecular lattice that provides the underlying 100-nm synapsis between chromosome axes (22,23), while SYCP3 assembles into regularly repeating filaments that support chromosomal looping (24,25). The five CE proteins provide essential structural supports for the SYCP1 lattice that enable its continuous and cooperative extension along the entire chromosome length. In this capacity, CE proteins have been categorized as synaptic initiation factors (SYCE3, SYCE1, and SIX6OS1) and elongation factors (SYCE2 and TEX12), of which their disruption leads to complete loss of tripartite SC structure and failure of extension of short SC-like stretches, respectively (10,11,16–18). Of synaptic initiation factors, SYCE3 forms dimers that undergo potentially limitless self-assembly (26,27), SYCE1 forms antiparallel dimeric assemblies (28), and SIX6OS1 is an SYCE1-interacting protein of unknown structure (11). These likely act as short-range structural supports between SYCP1 molecules, possibly in transverse, longitudinal, and vertical orientations to stabilize a local three-dimensional SYCP1 lattice (22). In contrast, SYCE2 and TEX12 exist as a seemingly constitutive complex that undergoes selfassembly into fibers of many micrometers in length (29), which likely provide the long-range structural supports that stabilize continuous growth of the SYCP1 lattice along the entire chromosome axis (22). 1Molecular Mechanisms Program, Centro de Investigación del Cáncer and Instituto de Biología Molecular y Celular del Cáncer (CSIC-Universidad de Salamanca), Salamanca, Spain. 2Biosciences Institute, Faculty of Medical Sciences, Newcastle University, Framlington Place, Newcastle upon Tyne NE2 4HH, UK. 3Departamento de Medicina, Universidad de Salamanca, Salamanca, Spain. 4Departamento de Fisiología y Farmacología, Universidad de Salamanca, Salamanca, Spain. *These authors contributed equally to this work. †Corresponding author. Email: [email protected] (O.R.D.); [email protected] (A.M.P.) Copyright © 2020 The Authors, some rights reserved; exclusive licensee American Association for the Advancement of Science. No claim to original U.S. Government Works. Distributed under a Creative Commons Attribution License 4.0 (CC BY). on January 12, 2021http://advances.sciencemag.org/Downloaded from Sánchez-Sáez et al., Sci. Adv. 2020; 6 : eabb1660 2 September 2020 SCIENCE ADVANCES | RESEARCH ARTICLE 2 of 14 Owing to the essential roles of meiotic recombination, synapsis, and chromosome dynamics in mammalian meiosis (15,30–34), their defects are associated with human infertility, recurrent miscarriage, and aneuploidies (35,36). As genetic causes of infertility, they typically fall within the category of idiopathic cases, having no readily diagnosable and clinically resolvable cause. Within the 10 to 15% of couples who suffer from infertility, approximately 25% are idiopathic and of likely genetic origin, comprising 50 to 80% of cases of nonobstructive azoospermia (NOA) and premature ovarian failure (POF) (36,37). While individual infertility mutations are inherently unlikely to become widespread in a population, they can be found within families, especially when consanguineous (38), and provide crucial insights into their common targets and the molecular mechanisms that they disrupt. Within the SC, familial infertility mutations have been identified for SYCP3 and SYCE1 (36). All identified SYCP3 mutations are autosomal dominant and alter or delete its structural core’s C terminus that mediates filamentous assembly, so likely sequester wildtype (WT) molecules into inactive complexes (24,36). In contrast, the three identified SYCE1 mutations are autosomal recessive and were found in two familial cases of NOA and one of POF (36). The two NOA cases are splice-site mutations, c.197-2A>G and c.3752A>G, which are predicted to result in a truncated product of amino acids 1 to 65 and an internal deletion of amino acids 126 to 155, respectively (39,40). These remove or delete part of human SYCE1’s structural core that is encoded by amino acids 25 to 179, so can be explained by disruption of its dimeric structure (Fig.1B) (28,36). The POF mutation c.613C>T generates a premature stop codon (p.Gln241*) to give a truncated product of amino acids 1 to 240, relative to the canonical 351–amino acid isoform (Fig.1B) (41). However, as this truncation lies outside SYCE1’s structural core, the molecular mechanism that is disrupted, and thereby responsible for infertility, remains unknown. Here, we combine mouse genetics and cellular and biochemical studies to reveal a multivalent interaction mode between SYCE1 and SIX6OS1 that is disrupted by infertility-associated mutations of SYCE1. We find that the SIX6OS1N terminus binds and disrupts the core dimeric structure of SYCE1 (amino acids 25 to 179) to form a 1:1 complex as the first interface, and its downstream sequence binds to SYCE1 amino acids 177 to 305 as the second interface. C C C C C C C C C C N N SYCE1 BA C Molecular weight MW (kDa) 80 0.5 dRI 1.0 Elution volume (ml) 12 13 0 20 60 0 120 14 15 100 40 11 SYCE1coreSYCE1POFSYCE1 86-kDa dimer 48-kDa dimer 39-kDa 36-kDa dimer (degraded) MBPSIX6OS1 N SYCE1 core SYCE1 POF SYCE1SIX6OS1 N SYCE1 POF - SIX6OS1 N 15 10 35 25 55 40 100 70 Mw (kDa) * * MBPSIX6OS1 N SIX6OS1 N SYCE1 SYCE1 core SYCE1 POF 5871 SIX6OS1 3511 SYCE1 Core SYCE1 (1–351) SYCE1POF (1–240) 177–305 25–315 25–240 SYCE1core (25–179) POF c.613C>T 1–240 1–587 1–528 SIX6OS1N (1-67) 1–75 1–163 1–262 *Degradation NOA c.197-2A>G 1–65 D C C C C C C C C C C C C C C C C C C C C C C C C C Central element Lateral element SYCP1 SYCP1 Synaptonemal complex Lateral element Meiotic chromatin loops (maternal) Meiotic chromatin loops (paternal) 100 nm C C C C C C C C C C C C C C C C N YC P1 YC P1 Cen ele m N Cen N N SYC P nm SYCP N N N N N N N N N SYCE3 SYCE1-SIX6OS1 SYCE2-TEX12 -Helical region Fig. 1. SYCE1POF retains its core dimeric structure. (A) Schematic of the SC demonstrating its tripartite structure of two chromosome-bound LEs and a midline CE. Synapsis is achieved through N-terminal head-to-head assembly of SYCP1 molecules, which are bound via their C termini to meiotic chromosomes. SYCP1 head-to-head assembly is structurally supported within the CE by SYCE3 (red), an SYCE1-SIX6OS1 complex (yellow), and SYCE2-TEX12 fibrous assemblies (green). (B) Human SYCE1 (top) and SIX6OS1 (bottom) sequence schematics indicating the location and consequence of infertility-associated mutations of SYCE1 and 10–21 internal deletion of SIX6OS1, alongside the principal constructs used in this study. (C) SDS–polyacrylamide gel electrophoresis (SDS-PAGE) analysis of the purified recombinant proteins used in this study. The dominant degradation product of SYCE1POF is indicated by an asterisk; its identity was confirmed by the observed cleavage of degraded MBPand His-SYCE1POF fusion proteins upon treatment with TEV protease (fig. S1, A and B), consistent with it representing C-terminal degradation down to SYCE1’s structural core. Mw, weight-average molecular weight. (D) SEC-MALS analysis. SYCE1core (yellow), SYCE1POF (green), and full-length SYCE1 (violet) are dimeric species of 36, 48 (39 kDa for the degradation product), and 86 kDa, respectively (theoretical dimers: 37, 55, and 80 kDa). dRI, differential refractive index. Data for SYCE1core and full-length SYCE1 are reproduced from (28). on January 12, 2021http://advances.sciencemag.org/Downloaded from Sánchez-Sáez et al., Sci. Adv. 2020; 6 : eabb1660 2 September 2020 SCIENCE ADVANCES | RESEARCH ARTICLE 3 of 14 SYCE1’s infertility-associated mutations c.375-2A>G (NOA) and c.613C>T (POF) specifically disrupt the first and second interfaces, respectively. Mice harboring the SYCE1 POF mutation and a targeted deletion within SIX6OS1 (which disrupts the first interface) are infertile, with failure of SC assembly. We conclude that both SYCE1-SIX6OS1 binding interfaces are essential for SC assembly and meiotic division, thus explaining how human infertility results from the differential targeting of binding interfaces by SYCE1’s reported clinical mutations. RESULTS SYCE1 POF mutation c.613C>T retains its core dimeric structure The SYCE1 POF mutation c.613C>T encodes a premature stop codon (p.Gln241*) that is predicted to generate a truncated protein product of amino acids 1 to 240, relative to SYCE1’s canonical 351– amino acid isoform (Fig.1B) (41). We previously demonstrated that an N-terminal structural core encoded by amino acids 25 to 179 (SYCE1core) forms an -helical antiparallel coiled-coil structure that mediates head-of-head dimerization of SYCE1 (28). As this core region is retained (Fig.1B), we predicted that SYCE1’s antiparallel dimeric structure would be maintained within the 1to 240-amino acid truncated product of the POF mutation (SYCE1pof). To test this, we purified recombinant SYCE1pof, generating purified material that contained approximately equal quantities of the full protein and a degradation product of apparent size consistent with degradation to the C-terminal boundary of its structural core (Fig.1C and fig. S1, A and B). Circular dichroism (CD) spectroscopy confirmed that SYCE1pof contains a proportion of -helical structure consistent with retention of the 25–179 core structure (fig. S1C), and SYCE1pof and SYCE1core demonstrated identical melting temperatures (Tm) of 39°C (fig. S1D). Furthermore, analysis by size exclusion chromatography multiangle light scattering (SEC-MALS) confirmed that the full and degraded proteins are homodimers of 48and 39 kDa, respectively (Fig.1D). We conclude that SYCE1pof retains the dimeric structure imposed by its core 25–179 region, so its SC and meiotic defects must result from additional structural or functional roles of its deleted C terminus. The SYCE1 POF mutation leads to failure of SC assembly and infertility in mice Having established its retention of core dimeric structure, we next sought to determine the structural and functional consequence of the SYCE1 POF mutation on the SC and meiotic division invivo. We thus generated mice harboring mutations of Syce1 alleles to introduce stop codons at amino acid position 243, equivalent to the human p.Gln241* mutation (figs. S2 and S3). While heterozygotes (designated Syce1POF/WT) were fertile, both male and female homozygotes (designated Syce1POF/POF) were infertile, replicating the autosomal recessive pattern of the POF mutation in humans (41). In male mutant mice, we observed reduced testis size (63% smaller, n=3 mice at 2 months of age; fig. S4A) and a zygotene-like arrest similar to that observed in the SYCE1 knockout (16). There was defective SC assembly, with reduced staining for SYCP1 (Fig.2A) and SYCE3 (Fig.2B) and no staining for SYCE1 (Fig.2C), SIX6OS1 (Fig.2D), and SYCE2-TEX12 (fig. S4, B and C). Analysis of SYCE1 expression in the testis of Syce1POF/POF mice confirmed the presence of Syce1 transcript and a protein product of the correct molecular weight, albeit at reduced levels in comparison with WT (fig. S4, D and E, and table S1A). The Syce1POF open reading frame achieved WT levels of protein expression in a heterologous 293T cellular system (fig. S4F). We next studied the kinetics of DSB repair. Meiotic DSBs are generated by the nuclease SPO11 and are then resected to form single-stranded DNA ends that invade into the homologous chromosome by the recombinases RAD51 (DNA repair protein RAD51 homolog 1) and DMC1 (Meiotic recombination protein DMC1/LIM15 homolog) (42). DSBs are labeled by the presence of phosphorylated H2AX (-H2AX) (43). The distribution of -H2AX in mutant spermatocytes was similar to that found in WT cells at early prophase I but show increased staining at zygotene-like arrest (Fig.2E). The distributions of RAD51 and DMC1 were detected on aligned LEs (Fig.2,FandG) but in absence of mismatch repair protein MLH1 (DNA mismatch repair protein Mlh1) (marker of crossingovers) (Fig.2H). Together, these data indicate generation of DSBs but with failure of their repair and CO formation in Syce1POF/POF. In female mutant mice, we observed no follicles in adult ovaries (fig. S5A), and embryonic oocytes demonstrated zygotene arrest with mostly unaligned chromosome axes, recapitulating the human POF syndrome. Analysis of the SC revealed similar defects, with reduction in SYCP1 and SYCE3 (Fig.3,AandB) staining (though to a lesser extent than males), and absence of SYCE1, SIX6OS1 (Fig.3,CandD), and SYCE2-TEX12 (fig. S5, B and C). The distribution of -H2AX, RAD51, and DMC1 labeling in zygotene-like mutant oocytes was also increased and lacked MLH1 foci (Fig.3,EtoH). Thus, the SYCE1 POF mutation leads to male and female infertility with phenotypes of failed DSB repair, synapsis, and lastly SC assembly, similar to those previously observed upon disruption of structural components of the SC CE (10,11,16–18). SYCE1POF retains SIX6OS1 binding but lacks SYCE3 binding in heterologous systems As the Syce1POF/POF mouse strain indicated a clear structural defect in the SC, we wondered whether the POF mutation may disrupt the known interaction between SYCE1 and fellow SC CE components SIX6OS1 and SYCE3 (11). The expression of SYCE1 and SIX6OS1in COS7 cells produced cytoplasmic signals that became colocalized in foci upon coexpression (95% cells; Fig.4A and fig. S6), in keeping with our previous findings (11). SYCE1pof formed similar or slightly reduced numbers of foci that equally colocalized with SIX6OS1, indicating a retention of SIX6OS1 binding (89% cells; Fig.4A). We further demonstrated a similar coimmunoprecipitation of SIX6OS1 by WT SYCE1 and SYCE1pof upon coexpression in human embryonic kidney (HEK) 293 cells (Fig.4B). Thus, the SYCE1-SIX6OS1 interaction is retained in the SYCE1 POF mutation. Could other disrupted functions contribute to the effect of the POF mutation? The only other known SYCE1 interactor is SYCE3, which undergoes lowaffinity binding, as determined by its dissociation during purification (fig. S7, A and B). In contrast with the WT protein, the expression of SYCE1pof (cytoplasmic foci) in COS7 cells failed to recruit SYCE3 (preferentially nuclear) to their cytoplasmic foci (colocalization between SYCE3 and SYCE1 was observed for 95% of cells expressing WT SYCE1 and 21% of cells expressing SYCE1pof; Fig.4C and fig. S6). Similarly, SYCE1pof failed to coimmunoprecipitate SYCE3 upon coexpression in HEK293 cells (Fig.4D). Thus, while the SYCE1-SIX6OS1 complex is retained, the low-affinity SYCE1-SYCE3 complex is largely abolished in the SYCE1 POF mutation. on January 12, 2021http://advances.sciencemag.org/Downloaded from Sánchez-Sáez et al., Sci. Adv. 2020; 6 : eabb1660 2 September 2020 SCIENCE ADVANCES | RESEARCH ARTICLE 4 of 14 AB DC E G F H Fig. 2. Syce1POF/POF spermatocytes are not able to synapse and DSBs are deficiently repaired. (A) Double immunolabeling of WT pachytene and Syce1POF/POF zygotene-like spermatocytes with SYCP3 (red) and SYCP1 (green). In Syce1POF/POF spermatocytes, AEs fail to synapse and show a weak staining of SYCP1 along the axial elements (AEs). a.u., arbitrary units. (B to D) Double immunolabeling of spermatocyte spreads with SYCP3 (red) and the CE proteins (green). Syce1POF/POF zygotene-like spermatocytes showed a highly reduced signal of SYCE3 (B) and the absence of (C) SYCE1 and (D) SIX6OS1 from the AEs. (E) Double immunolabeling of -H2AX (green) and SYCP3 (red) in spermatocyte spreads from WT and Syce1POF/POF mice. -H2AX staining was persistent in Syce1POF/POF zygotene-like spermatocytes, but was restricted to the sex body in WT pachytene cells. (F and G) Double immunofluorescence of (F) RAD51 or (G) DMC1 (green) and SYCP3 (red). Syce1POF/POF zygotene-like spermatocytes showed increased numbers of foci of RAD51 and DMC1 along the AEs in comparison with WT, indicating unrepaired DSBs. (H) Double immunolabeling of MLH1 (green) and SYCP3 (red) showing the absence of COs (MLH1) in arrested Syce1POF/POF spermatocytes. Fluorescence intensity levels (A, B, and E) and number of foci (F and G) from WT and zygotene-like arrested spermatocytes are quantified in the right-hand plots. Welch’s t test analysis: ***P < 0.0001. Scale bars, 10 m. on January 12, 2021http://advances.sciencemag.org/Downloaded from Sánchez-Sáez et al., Sci. Adv. 2020; 6 : eabb1660 2 September 2020 SCIENCE ADVANCES | RESEARCH ARTICLE 5 of 14 AB DC E G F H Fig. 3. Syce1POF/POF oocytes fail to synapse and do not properly repair DSBs. (A) Double immunolabeling of oocyte spreads from WT and Syce1POF/POF mice with SYCP3 (red) and SYCP1 (green). Syce1POF/POF oocytes became arrested in a zygotene-like stage where AEs remain unsynapsed and unaligned, with reduced levels of SYCP1. (B to D) Double immunolabeling of oocyte spreads with SYCP3 (red) and the CE proteins (green). Syce1POF/POF zygotene-like oocytes showed reduced SYCE3 signal (B) and a complete absence of (C) SYCE1 and (D) SIX6OS1 from the AEs. IP, immunoprecipitation. (E) Double immunostaining of spread preparations of WT pachytene and Syce1POF/POF zygotene-like oocytes with -H2AX (green) and SYCP3 (red). In Syce1POF/POF oocytes, the levels of -H2AX increased and were more restricted to AEs in comparison with WT pachytene cells. (F to G) Double immunolabeling of (F) RAD51 or (G) DMC1 (green) and SYCP3 (red), showing higher numbers of foci in AEs from mutant oocytes. (H) Labeling of MLH1 (green) and SYCP3 (red). MLH1 foci are absent from the AEs of Syce1POF/POF oocytes. Fluorescence intensity levels (A, B, and E) and number of foci (F and G) from WT and Syce1POF/POF zygotene-like oocytes are quantified in the right-hand plots. Welch’s t test analysis: ***P < 0.0001. Scale bars, 10 m. on January 12, 2021http://advances.sciencemag.org/Downloaded from Sánchez-Sáez et al., Sci. Adv. 2020; 6 : eabb1660 2 September 2020 SCIENCE ADVANCES | RESEARCH ARTICLE 6 of 14 Fig. 4. SYCE1POF retains SIX6OS1 binding but fails to retain the SYCE3-interaction in heterologous systems. (A) Mouse SIX6OS1 colocalized with mouse SYCE1 and SYCE1POF in a cytoplasmatic punctate pattern upon coexpression in COS7 cells; the percentage of cells exhibiting colocalization is shown in the right-hand plot (n = 100 cells). DAPI, 4′,6-diamidino-2-phenylindole. (B) HEK293T cells were cotransfected with the indicated expression vectors. Protein complexes were immunoprecipitated with anti-Flag or anti–enhanced green fluorescent protein (EGFP) antibodies, or mouse immunoglobulin G (IgG) as a negative control, and were analyzed by immunoblotting with the indicated antibody. GFP-mSIX6OS1 coimmunoprecipitated with Flag-mSYCE1 and Flag-mSYCE1POF, suggesting that the POF mutation of SYCE1 alone is insufficient to block the interaction. (C) COS7 cells were transfected with mouse Syce3 in combination with mouse Syce1 or Syce1pof as indicated. SYCE1 colocalized with SYCE3 in its own cytoplasmatic punctate pattern, and colocalization was substantially diminished for SYCE1POF (n = 100 cells). (D) Immunoprecipitation of protein complexes from HEK293T-cotransfected cells with an anti-Myc or anti-EGFP antibody or mouse IgG. SYCE1 coimmunoprecipitated with SYCE3, and the interaction was disrupted for SYCE1 POF, suggesting that the C-terminal region of SYCE1 is required for its interaction with SYCE3. The untransfected lanes in (B) and (D) show the absence of all the proteins in total protein extracts from untransfected 293T cells. Scale bars, 20 m. on January 12, 2021http://advances.sciencemag.org/Downloaded from Sánchez-Sáez et al., Sci. Adv. 2020; 6 : eabb1660 2 September 2020 SCIENCE ADVANCES | RESEARCH ARTICLE 7 of 14 SYCE1core undergoes conformational change to form a 1:1 complex with SIX6OS1 What is the molecular basis of SIX6OS1 binding by SYCE1? As this is retained in SYCE1pof, we reasoned that SIX6OS1 binding must be mediated by SYCE1’s structural core. We screened SYCE1core against a library of SIX6OS1 constructs through bacterial coexpression and identified a robust interaction with amino acids 1 to 67 of SIX6OS1, herein referred to as SIX6OS1N (Figs.1B and 5A). We were able to purify the SYCE1core-SIX6OS1N complex by reciprocal affinity chromatography, ion exchange, and size exclusion chromatography (Fig.5B) and found it to be stable under all experimental conditions tested. We were further able to purify similar complexes for SYCE1pof (with the same degradation product as upon isolated expression) and full-length SYCE1 (Fig.1C and fig. S1B), confirming that SIX6OS1 binding is retained by all constructs containing the 25–179 core. CD analysis revealed similar -helical content for SYCE1-SIX6OS1N complexes as for their isolated SYCE1 proteins (fig. S1C). CD thermal denaturation revealed slightly increased cooperativity of unfolding and melting temperatures for SYCE1SIX6OS1N complexes relative to their isolated SYCE1 proteins (increasing from 39° to 43°C, 39° to 41°C, and 38° to 40°C for SYCE1core, SYCE1pof, and full length, respectively; Fig.5C and fig. S1D). SEC-MALS analysis revealed that all three SYCE1-SIX6OS1N complexes are 1:1, with molecular weights of 27, 37, and 46 kDa, respectively (Fig.5D and fig. S7C). Thus, the SYCE1core undergoes conformation change from an antiparallel homodimer to a 1:1 complex upon binding to SIX6OS1N (Fig.5E). We analyzed the conformation of the SYCE1core-SIX6OS1N complex by size exclusion chromatography small-angle x-ray scattering (SEC-SAXS; fig. S7, D and E). The SAXS real-space pair-distance P(r) distribution (the distribution of interatomic distances within a protein structure) demonstrates positive skew, indicating that SYCE1core-SIX6OS1N retains the rod-like structure of SYCE1core, but with a reduction in its molecular length from 186 to 138 Å (Fig.5F). Furthermore, its cross-sectional radius is slightly increased from 9 to 11 Å (fig. S7F), suggesting an increase from a twoto four-helical coiled coil. These geometric changes are consistent with the SYCE1core-SIX6OS1N 1:1 complex forming a shorter but wider coiled coil than the isolated SYCE1core dimer, as indicated by their SAXS ab initio models (Fig.5G). Furthermore, the SAXS P(r) distribution of SYCE1pof indicates a similar elongated structure but with an increased tail to a maximum dimension of 180 Å (Fig.5F), consistent with it containing the same SYCE1core-SIX6OS1N structure with an extended and potentially unstructured C terminus to amino acid 240. We conclude that SYCE1core mediates a direct interaction with SIX6OS1N that imposes a conformational change to a 1:1 complex that adopts a shorter and wider coiled-coil conformation than the isolated SYCE1core antiparallel homodimer. SYCE1POF disrupts a second SYCE1-SIX6OS1 binding interface Does the SYCE1core-SIX6OS1N complex represent the sole means by which SYCE1 interacts with SIX6OS1? We were unable to obtain soluble biochemical complexes containing SIX6OS1 sequences beyond its N terminus and so used yeast two-hybrid (Y2H) to test SYCE1 binding by full-length SIX6OS1. Having confirmed direct binding of SYCE1core to full-length SIX6OS1, we used C-terminal truncation to dissect its minimal binding site to amino acids 1 to 75, in keeping with our biochemical findings, and identified an additional interaction between SYCE1 177–305 and full-length SIX6OS1 (Fig.6A). To establish whether SYCE1core and 177–305 bind to the same or distinct sites within SIX6OS1, we established an internal deletion of SIX6OS1 amino acids 10 to 21 (10–21) that blocks formation of the SYCE1core-SIX6OS1N biochemical complex (Fig.5A). SIX6OS1 1–22 did not interact with any SYCE1 construct (Fig.6A), indicating that amino acids 10 to 21 are necessary but not sufficient for SYCE1core binding. While 10–21 completely abrogated the Y2H interaction of full-length SIX6OS1 with SYCE1core (25–179), it retained a robust interaction with SYCE1 177–305, suggesting distinct SIX6OS1-binding sites (Fig.6A). Furthermore, 10–21 blocked the ability of SIX6OS1 1–262 to interact with SYCE1core and SYCE1pof (amino acids 25 to 240) while retaining its binding to full-length and 25–315 SYCE1 (Fig.6A). Thus, SYCE1 undergoes multivalent interactions with SIX6OS1, with the first binding interface mediated by SYCE1core and SIX6OS1N (1–67), and the second interface mediated by SYCE1 177–305 and downstream sequence within SIX6OS1 1–262. Furthermore, the first and second binding interfaces are specifically disrupted by SIX6OS1 deletion 10–21 and the SYCE1 POF mutation, respectively, and in both cases, an SYCE1-SIX6OS1 complex is retained through the unaffected alternative site (Fig.6B). SIX6OS1 10–21 retains SYCE1 binding in heterologous systems Our biochemical and Y2H analyses concluded that SIX6OS1 10–21 would disrupt the first SYCE1-SIX6OS1 binding interface while retaining complex formation through the second interface. In support of this, we found that SIX6OS1 10–21 retained its ability to form intense colocalized foci with SYCE1 upon coexpression in COS7 cells (98% of the cells; Fig.6C), similar to our previous observations for the SYCE1 POF mutation (Fig.4A). Similarly, SIX6OS1 10–21 retained its ability to coimmunoprecipitate SYCE1 upon coexpression in HEK293 cells (Fig.6D). Thus, localization and coimmunoprecipitation data from heterologous systems support our Y2H findings that the second SYCE1-SIX6OS1 binding interface is retained in SIX6OS1 10–21, mirroring the retention of only the second binding interface that is predicted for the 126–155 deletion of the SYCE1 c.375-2A>G NOA mutation (40). SIX6OS1 10–21 leads to failure of SC assembly and murine infertility Having established that the severe phenotype of the SYCE1 POF mutation likely results from the disruption of the second SYCE1SIX6OS1 binding interface and its interaction with SYCE3, we wondered whether a similar phenotype would result from the sole disruption of the first SYCE1-SIX6OS1 binding interface. To test this, we generated mice harboring mutations of Six6os1 alleles encoding internal in-frame deletions of amino acids 10 to 21 (equivalent numbering to the human protein) (fig. S8, A and B). While heterozygotes (designated Six6os110–21/WT) were fertile, both male and female homozygotes (designated Six6os110–21/10–21) were infertile, similar to the SYCE1 POF mutation. In males, we observed reduced testis size (Fig.7A) and a zygotene-like arrest similar to that observed in the Six6os1 and Syce1 knockouts (11,16). The mutant spermatocytes were defective in synapsis and SC assembly, with reduced staining for SC proteins SYCP1 (Fig.7B) and SYCE3 (Fig.7C) and no staining for SYCE2-TEX12 (Fig.7,FandG). In on January 12, 2021http://advances.sciencemag.org/Downloaded from Sánchez-Sáez et al., Sci. Adv. 2020; 6 : eabb1660 2 September 2020 SCIENCE ADVANCES | RESEARCH ARTICLE 8 of 14 MBPSIX6OS1N D C B A 15 10 35 25 55 40 100 70 Mw (kDa) Bacterial supernatant Amylose TEV cleavage HiTrap Q Ni-NTA MBPSIX6OS1 N HiTrap Q Gel filtration SIX6OS1 N HisSYCE1core SYCE1core + His-SYCE1core M W (kDa) 100 70 55 40 35 25 15 1–75 1–67 (N) Empty MBPSIX6OS1 MBPSIX6OS1 HisSYCE1core MBP 90˚ 186 Å SAXS ab initio models 138 Å G F 1 0.4 0.2 0.8 0.6 80 1000 120 140 0 P(r) Interatomic distance (Å) 160 SYCE1core Dmax = 186 Å Rc = 9 Å 6040 20 SYCE1POF-SIX6OS1N Dmax = 180 Å 180 200 SYCE1core-SIX6OS1N Dmax = 138 Å Rc = 11 Å SYCE1coreSIX6OS1N SYCE1core Temperature (°C) 60 20 40 0 80 100 % Unfolded ([θ]222,x-[θ]222,5)/[θ]222,95-[θ]222,5) 10 20 30 40 50 60 70 80 90 1000 SYCE1-SIX6OS1N Tm = 40°C SYCE1 Tm = 38°C Molecular weight MW (kDa) 0.5 dRI 1.0 Elution volume (ml) 12 13 0 20 60 0 80 14 15 40 SYCE1POFSIX6OS1N SYCE1coreSIX6OS1N SYCE1SIX6OS1N 46 kDa 1:1 37 kDa 1:1 29 kDa 1:1 1:1 1:1 1:1 1: 1:1 (degraded) d) d ) 27 kDa 1:1 SYCE1core antiparallel dimer SYCE1core-SIX6OS1N 1:1 complex E 57kDa monomer Fig. 5. SYCE1core undergoes conformational change to form a 1:1 complex with SIX6OS1n. (A) Amylose pulldown following coexpression of MBP-SIX6OS1 1–75, 1–67, 1–75 10–21, and free MBP with His-SYCE1core. (B) SDS-PAGE of the copurification of the SYCE1core-SIX6OS1n complex. Ni-NTA, Ni–nitrilotriacetic acid. (C) CD thermal denaturation recording the CD helical signature at 222 nm between 5° and 95°C, as % unfolded; estimated melting temperatures (Tm) are indicated. (D) SEC-MALS analysis. SYCE1core-SIX6OS1n (blue), SYCE1POF-SIX6OS1n (red) and full-length SYCE1-SIX6OS1n (black) are 1:1 complexes of 27, 37 (29 kDa for the degradation product complex), and 46 kDa, respectively (theoretical 1:1 to 27, 36, and 48 kDa), while MBP-SIX6OS1n (gray) is a 57-kDa monomer (theoretical, 53 kDa). SDS-PAGE of the SYCE1POF-SIX6OS1n sample is shown in Fig. 1C. (E) Schematic of the conformational change of the SYCE1core antiparallel dimer (yellow) into a 1:1 SYCE1core-SIX6OS1n complex (yellow-blue). (F and G) SEC-SAXS analysis. (F) SEC-SAXS P(r) interatomic distance distributions of SYCE1core-SIX6OS1n (blue), SYCE1POF-SIX6OS1n (red), and SYCE1core (yellow), revealing maximum dimensions (Dmax) of 138, 180, and 186 Å, respectively. Their cross-sectional radii (Rc) are indicated (fig. S7D). (G) SAXS ab initio models of SYCE1core-SIX6OS1n (blue) and SYCE1core (yellow); averaged models were generated from 20 independent DAMMIF runs. Data for SYCE1core and full-length SYCE1 are reproduced from (28). on January 12, 2021http://advances.sciencemag.org/Downloaded from Sánchez-Sáez et al., Sci. Adv. 2020; 6 : eabb1660 2 September 2020 SCIENCE ADVANCES | RESEARCH ARTICLE 9 of 14 contrast with their complete absence in the SYCE1 POF mutation, we observed some residual staining for SYCE1 (Fig.7D) and SIX6OS1 (Fig.7E) even though the levels of transcription of Six6os110–21 appeared to be increased in the mutant testis (fig. S9 and table S1B). We detected -H2AX (fig. S10A) and DMC1/RAD51 foci (fig. S10, B and C) on aligned axial elements but no MLH1 foci (fig. S10D), indicating the proper induction of DSBs with their failed repair and absence of COs. Thus, SIX6OS1 10–21 leads to infertility with a phenotype of failed DSB repair and SC assembly, similar to the SYCE1 POF mutation and those reported for disruption of structural components of the CE (10,11,16–18). Thus, we conclude that both first and second SYCE1-SIX6OS1 binding interfaces are essential for SC assembly and meiotic progression. Furthermore, these findings explain how the sole disruption of individual SYCE1-SIX6OS1 binding interfaces by SYCE1 NOA (c.375-2A>G) and POF (c.613C>T) mutations result in the reported familial cases of human infertility. DISCUSSION The structural and functional integrity of the SC is contingent on the structure and assembly of is constituent protein components. Here, we report that SC assembly depends on multivalent interactions between CE components SYCE1 and SIX6OS1 that are disrupted by infertility-associated mutations of SYCE1. The first binding interface is formed by the structural core of SYCE1 (SYCE1core; amino acids 25 to 179), which undergoes conformational change from an antiparallel homodimer to a 1:1 complex upon interaction with SIX6OS1’s N terminus (SIX6OS1N; amino acids 1 to 67). The second binding interface is formed by downstream sequence within SIX6OS1 1–262 interacting directly with SYCE1 177–305. Through the generation of mice harboring an internal deletion of SIX6OS1’s N terminus (10–21) and the SYCE1 POF mutation (murine p. Gln243*), which specifically block the first and second binding interfaces, respectively, we find that integrity of both SYCE1SIX6OS1 binding interfaces is essential for SC assembly and meiotic progression invivo. What is the structure of the SYCE1-SIX6OS1 complex? SEC-SAXS analysis revealed that the SYCE1core-SIX6OS1N 1:1 complex formed by the first binding interface has a length and cross-sectional radius of 138and 11 Å, in comparison with 186 and 9 Å for the SYCE1core dimer. We previously reported a model for SYCE1core in which amino acids 52 to 179 form an antiparallel dimeric coiled coil containing a midline “kink”, with  helices of amino acids 25 to 50 packing Fig. 6. SYCE1 undergoes multivalent interaction with SIX6OS1 in yeast, but SIX6OS1 10–21 retains SYCE1 binding in heterologous systems. (A) Y2H analysis of interactions between SYCE1 and SIX6OS1 in which positive reactions are indicated by the growth of blue colonies. These data are representative of three repeats. (B) Schematic of the SYCE1-SIX6OS1 interaction based on the Y2H data in (A), with the two binding sites highlighted in red and green. The SYCE1 POF mutation blocks the second binding interface between SYCE1 177–305 and SIX6OS1 downstream sequence within region 1–262, whereas the SIX6OS1 10–21 deletion blocks the first binding interface between SYCE1core (25–179) and SIX6OS1n (1–67). (C) COS7 cells were transfected with mouse Six6os1 10–21 alone or in combination with mouse Syce1. SIX6OS1 10–21 showed nuclear localization with some cytoplasmatic signal and colocalized in cytoplasmic foci with SYCE1; the percentage of cells exhibiting colocalization is shown. Scale bars, 20 m. (D) Coimmunoprecipitation of SIX6OS1 10–21 and Flag-SYCE1 from cotransfected HEK293T cells using anti-Myc or anti-EGFP antibodies, or mouse IgG as a negative control. SIX6OS1 10–21 coimmunoprecipitated SYCE1, indicating that the second SYCE1 binding interface is retained. The untransfected lanes confirm the absence of SIX6OS1 10–21 and SYCE1 in total protein extracts of untransfected 293T cells. on January 12, 2021http://advances.sciencemag.org/Downloaded from