Structural basis of meiotic chromosome synaptic elongation through hierarchical fibrous assembly of SYCE2-TEX12

The synaptonemal complex (SC) is a supramolecular protein assembly that mediates synapsis between homologous chromosomes during meiosis. SC elongation along the chromosome length (up to 24 μm) depends on its midline α-fibrous component SYCE2-TEX12. Here, we report X-ray crystal structures of human SYCE2-TEX12 as an individual building-block and upon assembly within a fibrous lattice. We combine these structures with mutagenesis, biophysics and electron microscopy to reveal the hierarchical mechanism of SYCE2-TEX12 fibre assembly. SYCE2-TEX12’s building-blocks are 2:2 coiled-coils which dimerise into 4:4 hetero-oligomers and interact end-to-end and laterally to form 10-nm fibres, which intertwine within 40-nm bundled micrometre-long fibres that define the SC’s midline structure. This assembly mechanism bears striking resemblance with intermediate filament proteins vimentin, lamin and keratin. Thus, SYCE2-TEX12 exhibits behaviour typical of cytoskeletal proteins to provide an α-fibrous SC backbone that structurally underpins synaptic elongation along meiotic chromosomes.


Introduction
Meiosis, the specialised form of cell division in gametogenesis, involves an extraordinary chromosome choreography that generates haploid oocytes and spermatozoa 1 . The hallmark of meiosis is synapsis, in which aligned homologous chromosomes become tightly bound along their axial length whilst undergoing recombination and crossover formation 1 . This is achieved by the synaptonemal complex (SC), a supramolecular protein assembly that acts as a molecular 'zipper' of meiotic chromosomes 1,2 . SC integrity is essential for meiosis 3 , and mutations that disrupt SC structure have been identified in clinical cases of infertility 4,5 . Thus, the mechanism of SC assembly is fundamentally important for understanding human fertility and the molecular causes of infertility, miscarriage and aneuploidy.
In advance of synapsis, homologous chromosome pairs are established and aligned by recombination intermediates generated from 200-400 induced double-strand breaks per cell 1 . Concomitant with recombination, chromosomes adopt a 'lamp-brush' structure, in which chromatin is looped around linear axes 6,7 . Thus, the SC assembles between aligned linear arrays of homologous chromatin and converts discrete recombination-mediated links into a single continuous synapsis along the axial length. The SC's assembled threedimensional structure facilitates the resolution of recombination intermediates, forming typically one crossover per arm 3,8 . In addition to providing genetic diversity, crossovers provide the sole physical connections between homologues following SC disassembly, thereby enabling faithful homologous chromosome segregation at the end of meiosis I 1,2 .
The SC was discovered in 1956 through its iconic tripartite appearance on electron micrographs, initially in crayfish spermatocytes 9 , and subsequently throughout mammals, insects, plants and yeast 10,11 . The SC's tripartite electron-dense fibres correspond to the two chromosome axes (lateral elements), separated by 100-nm, with a midline central element ( Fig. 1a) 10,11 . Central and lateral elements have widths of approximately 20-40 nm and 50 nm, respectively, are continuous along the chromosome length (4-24 μm in humans), and are held together by interdigitated transverse filaments that act as the teeth of the SC 'zipper' [10][11][12][13][14] . Electron microscopy and super-resolution fluorescence microscopy have shown that the mammalian SC has a depth of up to 100 nm 13,15,16 . Assuming a square cross-section of 100-nm sides, with 20-80% solvent content, we can estimate a molecular weight of 1.6-6.4 GDa per μm of SC length. Thus, entire SCs may be between 6.4-154 GDa, placing them amongst the largest protein structures within a cell. associations but no tripartite SC 22,24,26 . In contrast, SYCE2 and TEX12 knockouts demonstrate short tripartite structures, containing SYCP1, SYCE1, SYCE3 and SIX6OS1, which fail to extend along the chromosome axis 27,28 . Thus, it has been proposed that SYCE1, SYCE3 and SIX6OS1 are initiation factors that provide short-range stabilisation of nascent SYCP1 synapsis, whereas SYCE2 and TEX12 are elongation factors that underpin long-range SC growth by providing longitudinal structural support 22,[27][28][29] . This dichotomy is underpinned by biochemical findings that SYCE1 directly interacts with SYCE3 and SIX6OS 14,30 , whilst SYCE2 and TEX12 exist in a seemingly constitutive complex 31 .
In the last decade, the molecular underpinnings of the SC have started to come into focus through analysis of SC proteins by structural biology 4,17,[30][31][32][33][34][35][36][37] . An emerging theme is that SC proteins, which are almost universally coiled-coils, exist as defined building-blocks that self-assemble into higher-order structures to provide the SC's distinct architectural elements. SYCP1 is a tetramer that assembles into a lattice-like array to fulfil the SC's fundamental role of tethering chromosome axes 17 . SYCP3 is a tetramer that assembles into paracrystalline fibres, which have been observed for recombinant proteins in vitro 32,37 , upon heterologous cellular expression 38 , and within the native SC 39 . SYCP3 fibres separate DNA-binding sites by 23-nm repeating units, facilitating the compaction of chromatin loops along the meiotic chromosome axis 32 . Similar structures and assemblies are achieved by an SYCP2-SYCP3 complex 36 . Central element proteins SYCE2 and TEX12 form a core 4:4 complex that assembles into fibres with polymorphic appearance, of 40-nm width, and up to 5-μm in length (Fig. 1a) 31 . The appearance and dimensions of SYCE2-TEX12 fibres match those of the native SC central element, suggesting that fibrous assembly along the SC axis underlies the structural role of SYCE2-TEX12 in synaptic elongation.
Here, we combine X-ray crystal structures of SYCE2-TEX12 at two critical stages of assembly with solution and electron microscopy studies to reveal a hierarchical mechanism for SYCE2-TEX12 α-fibre assembly, which intriguingly resembles intermediate filament assembly. The building blocks of SYCE2-TEX12 assembly are 2:2 complexes, which dimerise into 4:4 hetero-octamers that undergo end-to-end and staggered lateral interactions to form 10-nm fibres. These thread-like structures become intertwined within 40-nm fibres that extend to several micrometres in length and resemble the native SC central element. Thus, we define the molecular mechanism whereby SYCE2-TEX12 forms an α-fibrous axis that provides long-range structural support for synapsis as the 'backbone' of the SC.

SYCE2-TEX12 assembly is mediated by TEX12's C-terminal tip
A common feature of self-assembly by SC proteins SYCP1-3 is the presence of short motifs at their α-helical ends that direct higher-order assembly 17,32,36 . In each case, disruption of these motifs blocks assembly and restricts proteins to obligate 'building block' structures. Could a similar phenomenon apply to SYCE2-TEX12 fibres? To test this, we analysed the oligomeric state and assembly of short deletions of the SYCE2-TEX12 core (amino-acids 57-165 and 49-123; Fig. 1b Europe PMC Funders Author Manuscripts forms higher-order structures (Fig. 1c,d), we identified that deletion of TEX12's C-terminal tip (ΔCtip, amino-acids 49-113; Fig. 1b) restricted SYCE2-TEX12 to a 2:2 complex and blocked higher-order assembly (Fig. 1c,d). Circular dichroism (CD) determined that wildtype and ΔCtip complexes have comparable α-helical contents (93% and 91%) and melting temperatures (71°C and 67°C) (Extended Data Fig. 1a,b). Thus, the ΔCtip 2:2 complex may constitute a substructure of the wild-type core 4:4 complex that represents its minimum building block for assembly.
Full-length SYCE2-TEX12 has an increased propensity for assembling in solution, with SEC-MALS revealing its predominant formation of higher-order species, in addition to clearly discernible 4:4 and 2:2 complexes (Fig. 1c,e). The presence of a wild-type 2:2 complex supports its putative role as an obligate building block. In common with the core complex, deletion of TEX12's C-terminal tip restricted full-length SYCE2-TEX12 to a 2:2 complex and blocked higher-order assembly (Fig. 1c,e). The α-helical structure and stability of the wild-type protein was retained in the ΔCtip complex (Extended Data Fig.  1a,c). Thus, in a manner reminiscent of SYCP1-3 17,32,36 , deletion of a short motif at TEX12's C-terminal tip blocks higher-order assembly and stabilises SYCE2-TEX12 in its building-block 2:2 complex.
We next visualised SYCE2-TEX12 assembly by electron microscopy (Fig. 1f,g). This confirmed that core and full-length complexes form morphologically similar fibres, which are typically positively stained, vary in thickness up to approximately 40 nm and extend up to several micrometres in length (Fig. 1f, Fig. 3a,b). This suggests that flanking sequences provide additional stability to core assemblies. In agreement with SEC-MALS, TEX12 ΔCtip blocked fibre formation of core and full-length complexes (Fig. 1f,g and Extended Data Fig. 3b,d). We conclude that TEX12's C-terminal tip is essential for assembly of SYCE2-TEX12 fibres, and its deletion stabilises the building-block 2:2 structure, providing a critical means for elucidating the fibre assembly mechanism.
Crystal structure of SYCE2-TEX12 in a 2:2 complex The enhanced stability of the SYCE2-TEX12 ΔCtip core complex facilitated the growth of protein crystals, which diffracted anisotropically to 2.42-3.36 Å resolution and enabled structure solution by molecular replacement of ideal helical fragments using ARCIMBOLDO40 (Table 1 and Extended Data Fig. 4a). This revealed a 2:2 helical assembly in which an anti-parallel SYCE2 dimer spans the 14-nm molecular length and positions TEX12 chains in a staggered configuration (Fig. 2a-c). TEX12 chains are arranged with their N-termini towards the midline and their C-termini at either end of the molecule, immediately adjacent to SYCE2's N-and C-termini ( Fig. 2a- the C-terminal ends of SYCE2's core were not visible in electron density. Thus, deletion of TEX12's C-terminal tip corresponded with disorder of an analogous C-terminal sequence of the SYCE2 core (herein referred to as S2C; amino acids 155-165), such that the molecular ends of the 2:2 complex are formed of closely associated truncated C-termini of SYCE2 and TEX12 (Fig. 2a,c).
Does the 2:2 crystal structure represent the solution state of SYCE2-TEX12 ΔCtip? Sizeexclusion chromatography small-angle X-ray scattering (SEC-SAXS) of the ΔCtip core 2:2 complex revealed a real-space P(r) interatomic distance distribution and Guinier analysis indicative of a rod-like molecule with 19-nm length and 1.2-nm cross-sectional radius (Fig.  2g,h, Extended Data Fig. 4b,c and Supplementary Table 1). These dimensions match the length of the crystal structure with missing S2C ends modelled as ideal helices, and its observed 2-nm width ( Fig. 2c and Extended Data Fig. 4e). Further, the SAXS scattering curve was closely fitted by the crystal structure with modelled S2C helices and flexible N-termini (χ 2 = 1.99; Fig. 2g and Extended Data Fig. 4e). We also analysed a ΔS2C/ΔCtip complex in which the C-termini of SYCE2 and TEX12 were deleted. SEC-MALS confirmed that the ΔS2C/ΔCtip core complex is predominantly 2:2 in solution (Extended Data Fig. 4d). SEC-SAXS determined a rod-like molecule with 15-nm length and 1.2-nm cross-sectional radius, with an ab initio envelope that matches the dimensions of the 2:2 crystal structure ( Fig. 2g-i, Supplementary Table 1 and Extended Data Fig. 4c). Further, the SAXS scattering curve was closely fitted by the crystal structure upon ideal helical modelling of the few missing C-terminal amino-acids (χ 2 = 1.23; Fig. 2g). We conclude that the 2:2 crystal structure corresponds to the solution state of the ΔCtip core complex and hence represents the obligate 2:2 structure of SYCE2-TEX12 that acts as its building-block for assembly.

Crystal structure of SYCE2-TEX12 in a 4:4 assembly
We obtained an alternative crystal form of SYCE2-TEX12 ΔCtip core, which diffracted anisotropically to 3.33-6.00 Å resolution, and was solved by molecular replacement using a three-helical bundle of the 2:2 structure as a search model (Table 1 and Extended Data Fig. 5a). This revealed a 4:4 structure in which two 2:2 complexes interact laterally, through tessellation of S-shaped SYCE2 dimers, to create a molecule of the same length (14 nm) but twice as wide (4 nm) as individual 2:2 complexes (Fig. 3a,b). The 2:2 crystal structure is largely retained by the two interacting 2:2 complexes (r.m.s. deviation = 1.91; Extended Data Fig. 5b), which seamlessly generate an SYCE2 tetrameric core with staggered TEX12 chains arranged on opposing surfaces (Fig. 3a). In common with the 2:2 structure, S2C sequences were missing from electron density, meaning that pairs of truncated SYCE2 and TEX12 C-termini are juxtaposed at both ends of the 4:4 molecule (Fig. 3c). The observed tessellation of 2:2 complexes provides an enticing mechanism for assembly of a 4:4 complex from structurally intact obligate 2:2 molecules, in agreement with our biochemical findings.

TEX12's Ctip stabilises the SYCE2-TEX12 4:4 complex
The 4:4 crystal structure lacks TEX12's C-terminal tip, which is required for 4:4 assembly in solution, suggesting that this conformation was supported by the crystal lattice. We reasoned that the tessellating interface likely drives 4:4 assembly of the wild-type protein and is stabilised in solution by additional TEX12 Ctip interactions. To test this, we introduced glutamate mutations of amino-acids H89 and Y115, which mediate the tessellating interaction but have no structural role within constituent 2:2 complexes (Fig.  3d). Within the core complex, the H89E Y115E mutation partially blocked 4:4 assembly, restricting 60% of material to a 2:2 complex, despite the presence of Ctip sequences (Fig.  3e). SEC-SAXS determined that the wild-type 4:4 complex has a 19-nm length and 1.9-nm cross-sectional radius (Fig. 4a,b and Extended Data Fig. 6b-d), matching the observed geometry of the 4:4 crystal structure in which the length of the 2:2 complex is retained and its width is doubled from 2 nm to 4 nm (Fig. 3a-c). Further, multi-phase SAXS ab initio modelling of the wild-type 4:4 structure placed ΔCtip 2:2 ab initio envelopes in parallel, consistent with their lateral tessellating interaction, with missing mass -corresponding to S2C and Ctip sequences -placed at either end of the molecule (χ 2 = 1.69; Fig. 4c). Thus, the lateral tessellating interaction of 2:2 complexes within the 4:4 crystal structure describes the underlying architecture of the wild-type 4:4 assembly.
How does TEX12's Ctip stabilise the wild-type 4:4 complex? The location of pairs of truncated TEX12 and SYCE2 C-termini at either end of the 4:4 complex suggests that Ctip and S2C sequences could interact in four-helical structures that clamp together tessellated 2:2 molecules. Accordingly, deletion of SYCE2's S2C blocked 4:4 assembly and restricted the SYCE2-TEX12 core to a 2:2 complex (Extended Data Fig. 6a), confirming that S2C and Ctip have equivalent roles in assembly. The predicted heptads of S2C-Ctip sequences are in-phase with the crystallographically-observed heptads of their upstream sequences, so we modelled an extended S2C-Ctip parallel heterodimeric coiled-coil (Extended Data Fig. 7a). This model docked seamlessly and in-phase with overlapping sequences within the 4:4 crystal structure (Extended Data Fig. 7b), producing a model in which S2C-Ctip coiled-coils emanate from both ends of constituent 2:2 complexes (Extended Data Fig. 7c). Upon energy minimisation and geometry idealisation, juxtaposed pairs of S2C-Ctip dimers interacted together (Extended Data Fig. 7c), forming S2C-Ctip 2:2 parallel bundles that clamp together 2:2 complexes at either end of the tessellating 4:4 interface (Fig. 4d). The resultant 'clamped' 4:4 structure closely fitted the SAXS scattering curve of the wild-type 4:4 core complex upon flexible modelling of missing N-termini (χ 2 = 2.47; Fig. 4a). Thus, formation of S2C-Ctip C-terminal bundles provides a mechanism for clamping together laterally associated 2:2 complexes within the architecture of the wild-type 4:4 complex.

Controlling the hierarchical assembly of SYCE2-TEX12
Our model of the clamped 4:4 assembly predicts that TEX12 amino-acids L110, F114, I117 and L121 contribute to the C-terminal bundle's hydrophobic core, whilst F102, F109 and V116 are solvent-exposed (Fig. 5a,b). Thus, the first group is likely essential for clamped 4:4 assembly, whereas the second group is likely dispensable. Accordingly, hydrophobic core mutation L110E F114E I117E L121E (LFIL) restricted core and full-length SYCE2-TEX12 to 2:2 complexes and blocked fibrous assembly, thereby mimicking ΔCtip ( Fig.  5c-f). We similarly observed a restricted 2:2 complex upon analogous mutation of S2C hydrophobic core amino-acids V149, V153, V156 and L160 ( Fig. 5a and Extended Data Fig. 6a). In contrast, surface mutation F102A F109A V116A (FFV) retained the 4:4 core complex, and wild-type pattern of full-length assembly in solution, but blocked higher-order assembly ( Fig. 5c-f and Extended Data Fig. 3b,d). Importantly, SEC-SAXS confirmed that LFIL and FFV mutants adopt native 2:2 and 4:4 core structures, respectively (Extended Data Fig. 6b-g). Hence, whilst the core complex relies solely on FFV amino-acids for assembly, the full complex undergoes residual assembly through its unstructured termini, with FFV residues required for full fibre formation. Thus, we have identified separation-of-function mutations that block SYCE2-TEX12 in its building-block 2:2 and intermediate assembly structures. The mimicry of ΔCtip by LFIL mutation validates our clamped 4:4 assembly model, whilst blockade of higher-order assembly by FFV mutation implicates these surfaceexposed residues in SYCE2-TEX12 fibre assembly.

SYCE2-TEX12 forms 2-nm and 4-nm fibres
How does TEX12's Ctip drive assembly of SYCE2-TEX12 fibres? The crystal lattice of the 4:4 structure is a fibrous array in which 4:4 molecules are arranged in end-to-end chains, with 1-nm gaps between juxtaposed C-termini, and a lateral stagger (Fig. 6a). Could this represent the fibrous structure of SYCE2-TEX12, with end-to-end chains corresponding to fibres of 4:4 molecules that are thickened by lateral staggering? We obtained protein crystals of various SYCE2-TEX12 core constructs, including ΔCtip, which displayed a fibrous pattern of X-ray diffraction that is typical of the k-m-e-f family (keratin, myosin, epidermin, fibrinogen) of fibrous α-proteins ( Fig. 6b) 41,42 . This consists of a 5.1 Å meridional arc arising from the coiled-coil repeat, and 12 Å equatorial reflections that correspond to inter-helical distances of three-/four-helical bundles (contrasting with 9.8 Å for dimeric coiled-coils) 41,42 . This is consistent with the end-to-end arrangement of molecules within the 4:4 crystal lattice. Further, upon rotation about the meridional axis, equatorial reflections resolved into orthogonal 15-nm and 5-nm regularly-spaced reflections, closely matching the repeating units of the 4:4 crystal lattice (Fig. 6a,b). Thus, the fibrous diffraction pattern is consistent with the fibrous array of the 4:4 crystal lattice, demonstrating that this molecular arrangement is commonly exhibited by SYCE2-TEX12 constructs.
If the 4:4 crystal lattice represents the true fibrous assembly, then it should be possible to rationalise S2C-Ctip's structural role within this fibrous array. We modelled S2C and Ctip as α-helical extensions of SYCE2 and TEX12 chains within 4:4 complexes in situ. The modelled S2C-Ctip helices of juxtaposed 2:2 ends immediately tessellated and interacted favourably within anti-parallel bundles, with two stacked S2C-Ctip bundles mediating endto-end association of 4:4 molecules ( Fig. 6c and Extended Data Fig. 8a). Importantly, LFIL and S2C heptad residues are in the hydrophobic core, whilst FFV residues are surfaceexposed, in agreement with the clamped 4:4 model and their mutant phenotypes (Fig. 6c).
Our molecular model suggests that constituent SYCE2-TEX12 fibres could be 2-nm or 4-nm wide, formed from 2:2 or 4:4 building-blocks, and stabilised by individual S2C-Ctip bundles or the tessellating 4:4 interface with stacked bundles (Fig. 6c). Further, it predicts that the lengths of constituent fibres are defined by a 20-nm initial subunit with 15-nm increments for additional subunits (Fig. 6c).
We tested the fibre assembly model through SAXS analysis across the size-exclusion chromatography elution profile of SYCE2-TEX12 core (Fig. 6d). There was a stepwise increase in molecular length from 19 nm (4:4) to 35 nm and 65 nm at elution points preceding the 4:4 complex, with a concomitant reduction in cross-sectional radius from 1.9  Table 1). We confirmed these findings by visualising the smallest SYCE2-TEX12 core assemblies within negativelystained regions of electron micrographs (Fig. 6f). Automated image analysis determined populations with mean fibre widths of 2.3 ± 0.4 nm and 3.6 ± 1.0 nm, corresponding to 2:2 (2-nm width) and 4:4 (4-nm width) fibres, respectively ( Fig. 6f-h). Further, the observed lengths of 2-nm and 4-nm fibres conformed to the predicted discrete lengths of end-to-end assemblies (Fig. 6f). Thus, SAXS and EM data support our molecular model of SYCE2-TEX12 assembly into 2-nm and 4-nm fibres.

Hierarchical assembly of SYCE2-TEX12 fibres
How do 2-nm and 4-nm fibres assemble into 40-nm fibres? We visualised full-length SYCE2-TEX12 within negatively-stained regions of electron micrographs (Fig. 7a). This revealed the presence of 2-nm (2.1 ± 0.8 nm) and 4-nm (4.1 ± 1.6 nm) fibres, corresponding to the 2:2 and 4:4 fibres observed for the core complex, but of considerably greater length ( Fig. 7a-c). Importantly, the full-length complex also formed 10-nm fibres (9.6 ± 2.4 nm), which frequently intertwined within bundles of up to 40 nm that accumulated stain ( Fig. 7a-c). The 10-nm fibres likely represent laterally-associated 4:4 fibres, interacting through Ctip/FFV residues and flanking termini, and seemingly represent an upper limit for the width of individual 'smooth' fibres ( Fig. 7d,e). The 40-nm bundled fibres demonstrated no overt helical or linear structural regularity, but instead represent an irregular packing of 10-nm fibres, consistent with gliding surface interactions. The resultant fibres have the same dimensions as the positively-stained SYCE2-TEX12 fibres that dominate electron micrographs (Fig. 1f,g), suggesting that they could represent fortuitously stained representative structures, or could be a subpopulation of loosely-packed fibres. Further, whilst we lack information on the internal structure of the SC central element, its overall dimensions and polymorphic appearance match those of the 40-nm fibres (Fig. 1a), consistent with SYCE2-TEX12 constituting the major structural component of the central element.
In summary, SYCE2-TEX12 undergoes fibrous assembly through a hierarchical mechanism in which building-block 2:2 molecules laterally tessellate into 4:4 molecules, and assemble via end-to-end S2C-Ctip bundles into 2-nm and 4-nm fibres (Fig. 7d,e). These structures thicken into 10-nm fibres, which are interwoven within 40-nm bundled fibres (Fig. 7d,e). Hence, we define the molecular mechanism whereby SYCE2-TEX12 self-assembles into rope-like fibres that structurally underpin longitudinal growth of the SC to enable synaptic elongation along the axial length of meiotic chromosomes.

Discussion
The formation of supramolecular cellular structures through protein self-assembly is fundamentally important in numerous physiological systems, including cytoskeletal and chromosomal structure, and pathological processes such as amyloid formation [43][44][45] . Selfassembling systems can be defined as obligate building blocks that interact recursively to generate assembled structures. In meiotic chromosome synapsis, the supramolecular SC is formed by three major self-assembling systems: a zipper-like SYCP1 lattice, chromosomeassociated SYCP3 (and SYCP2-SYCP3) axes, and a midline SYCE2-TEX12 fibrous backbone 17,31,32,36 . Here, we report that SYCE2-TEX12 is formed of 2:2 hetero-tetrameric building blocks, which assemble into fibres through short motifs at their C-terminal αhelical ends. The deletion of TEX12's C-terminal tip blocked fibrous assembly and enabled X-ray crystallographic structure solution of its 2:2 building-block complex, revealing an unusual three-and four-helical coiled-coil structure. SYCP1-3 consist of tetrameric rod-like building blocks that self-assemble via N-and C-terminal α-helical end motifs 17,32,36 . Thus, an apparent theme in mammalian SC assembly is the formation of tetrameric coiled-coil building blocks, which self-assemble via α-helical end motifs that may be disrupted to stabilise their non-assembled building-block states in solution.
Full-length SYCE2-TEX12 shows a higher propensity for assembly than its structural core, and exists in a far wider range of solution species, including 2:2, 4:4 and megadalton complexes. Thus, SYCE2 and TEX12 unstructured termini likely provide additional longitudinal stability to cores assemblies formed through lateral and end-to-end interfaces.
In their absence, a compensatory increase in the number of end-to-end interfaces within the cross-section, through fibre thickening, may be necessary for stability, explaining our EM finding that core fibres are slightly thicker than full-length fibres. Further, unstructured termini could stabilise 2:2 complexes by acting in 'cis' through termini interacting with their own core, explaining the formation of wild-type 2:2 complexes by full-length but not core SYCE2-TEX12. Thus, the core complex may represent a restricted form of SYCE2-TEX12 in which the absence of unstructured termini reduces the stability of both higher-order assemblies and building-block 2:2 complexes, resulting in its stabilisation within a partiallyassembled clamped 4:4 structure.
How does the SYCE2-TEX12 structure relate to our previous findings that isolated SYCE2 and TEX12 form tetramers and dimers, respectively 31 ? The SYCE2 tetramer interface within the 4:4 structure is disrupted by TEX12 ΔCtip so is unlikely to support an isolated tetramer, and the linear arrangement of TEX12 chains is incompatible with isolated TEX12's compact dimer structure 46 . Thus, these likely represent alternative conformations. Isolated SYCE2 is unstable in solution and has no known cellular role, so the tetramer may be artefactual 31 . In contrast, isolated TEX12 is highly stable and has a defined cellular function in centrosomes, so likely represents a physiological conformation 31,46 . These findings reflect a wider phenomenon that coiled-coils often interact through conformational change in which homotypic interfaces are replaced by heterotypic interfaces, rather than through additive binding of existing oligomers. Thus, whilst we previously attributed the 4:4 complex to an SYCE2 tetramer binding to two TEX12 dimers 31 , it is now apparent that these species undergo conformational change into a structure that consists of interfaces distinct from those of its isolated protein constituents.
The varied cellular functions of IF proteins can be attributed to their high tensile strength coupled with torsional and flexional freedom, which allows them to define axes, connect distant structures, form interaction hubs, and establish architectural meshworks that define cellular milieu [47][48][49][50][51][52][53][54][55][56][57]59 . What does this tell us about SYCE2-TEX12 function within the SC? Firstly, linear SYCE2-TEX12 fibres may establish a single continuous SC backbone that structurally underpins SC structure over micrometre distances, preventing disjunct and overlapping synapses, and providing a structural means for rapid signal transduction along the SC. This explains the presence of only short discontinuous stretches of nascent synapsis upon disruption of SYCE2 or TEX12 27,28 . Secondly, SYCE2-TEX12 fibres likely possess significant longitudinal strength, which could constrain the length of compacted synapsed meiotic chromosomes, whilst permitting substantial bending and torsional freedom during chromosomal dynamics and DNA exchanges of meiotic prophase and recombination 1 . Such flexion and torsional motions could be enhanced by gliding interactions of 10-nm fibres within bundled 40-nm fibres, in which a single fixed-length axis could be highly adaptive and responsive to external forces directed away from the longitudinal axis. Finally, SYCE2-TEX12 fibres could act as interaction hubs for formation of a meshwork assembly that defines SC architecture. How do SYCE2-TEX12 fibres assemble with other SC components? The most prominent and conserved surface-exposed amino acids of the SYCE2-TEX12 structure are SYCE2 H89 and Y115, which mediate 4:4 assembly, and aromatic residues at the end-to-end assembly interface (including SYCE2 58-LYF-60 and TEX12 FFV). Thus, predicted interacting sites correspond to regions of the structural core that perform critical roles in fibrous selfassembly. Could interactions with other SC components directly regulate SYCE2-TEX12 assembly? Such interactions could trigger SYCE2-TEX12 fibre formation, or impose a specific SYCE2-TEX12 structure (such as by dictating the formation of only 2-nm rather than 4-nm fibres by binding to the SYCE2 tetramer interface) at key axial locations, such as recombination sites. This could mediate signalling, by propagating certain SYCE2-TEX12 fibrous conformations along the SC axis that impose function by regulating the local access of recombination factors. Similarly, phosphorylation events that are implicated in SC disruption are located at the N-terminal end of TEX12's α-helical core 60 , and thereby in close proximity of SYCE2-TEX12 end-to-end assembly interfaces. Could post-translational modifications precipitate SC disassembly by directly disrupting SYCE2-TEX12 fibres? Further, SYCE2 and TEX12 have long unstructured termini, which contribute to assembly and may also contain protein binding sites. In this case, they may emanate from the fibrous surface -in a 'bottle-brush' conformation -to mediate cooperative interactions of SYCE2-TEX12 fibres with SC transverse filament and central element components. It thus remains a critical challenge to uncover how the SC's major self-assembling systems of SYCP1, SYCP3 (and SYCP2-SYCP3) and SYCE2-TEX12 are bound together to achieve the enigmatic structure and function of the fully assembled SC.

Recombinant protein expression and purification
Sequences corresponding to regions of human SYCE2 (full-length,  67 . The structure was refined using PHENIX refine 66 , using isotropic atomic displacement parameters with one TLS group. The structure was refined against data to anisotropy-corrected data with resolution limits between 2.42 Å and 3.36 Å, to R and Rfree values of 0.2871 and 0.3296 respectively, with 100% of residues within the favoured regions of the Ramachandran plot (0 outliers), clashscore of 6.89 and overall MolProbity score of 1.3868. The R-factors are higher than is typical for isotropic diffraction at this resolution, but are consistent with the combined effects of anisotropic diffraction and data modulation owing to the orientation of coiled-coils in parallel densely-packed arrays in which repetitive structures give rise to systematically strong and weak reflections within regions of reciprocal space 69 . , with a 2:2 SYCE2-TEX12 heterotetramer in the asymmetric unit. Structure solution was achieved by molecular replacement using PHASER 65 , in which two copies of one end of the higher resolution SYCE2-TEX12 2:2 structure (PDB accession 6R17) were placed in the asymmetric unit. The structure was completed through manual building in COOT 67 , with iterative refinement using PHENIX refine 66 . The structure was refined against data to anisotropy-corrected data with resolution limits between 3.33 Å and 6.00 Å, to R and Rfree values of 0.3266 and 0.3672 respectively, with 100% of residues within the favoured regions of the Ramachandran plot (0 outliers), clashscore of 3.48 and overall MolProbity score of

Europe PMC Funders Author Manuscripts
Europe PMC Funders Author Manuscripts 1.1468. Similar to the 2:2 structure, the R-factors are higher than is typical for isotropic diffraction at this resolution but are consistent with the combined effects of anisotropic diffraction and data modulation in a fibrous coiled-coil lattice 69 .

Crystallisation and X-ray diffraction analysis of SYCE2-TEX12 core ΔCtip fibres
SYCE2-TEX12 core ΔCtip protein crystals were obtained through vapour diffusion in hanging drops, by mixing 1 μl of protein at 17 mg/ml in buffer containing 20% glycerol with 1 μl of crystallisation solution (0.05 M caesium chloride, 0.1 M MES pH 6.5, 30% v/v Jeffamine M-600) and equilibrating at 20°C. Crystals were cryo-cooled in liquid nitrogen in crystallisation solution supplemented with 25% glycerol. X-ray diffraction data were collected at 0.9763 Å, 100 K, as individual 0.50° frames of 0.500 s exposure on a Pilatus3 6M detector at beamline I03 of the Diamond Light Source synchrotron facility (Oxfordshire, UK). Diffraction images were analysed using Adxv.

Size-exclusion chromatography multi-angle light scattering (SEC-MALS)
The absolute molecular masses of SYCE2-TEX12 complexes were determined by sizeexclusion chromatography multi-angle light scattering (SEC-MALS). Protein samples at >1 mg/ml were loaded onto a Superdex™ 200 Increase 10/300 GL size exclusion chromatography column (GE Healthcare) in 20 mM Tris pH 8.0, 150 mM KCl, 2 mM DTT, at 0.5 ml/min using an ÄKTA™ Pure (GE Healthcare). The column outlet was fed into a DAWN® HELEOS™ II MALS detector (Wyatt Technology), followed by an Optilab® T-rEX™ differential refractometer (Wyatt Technology). Light scattering and differential refractive index data were collected and analysed using ASTRA® 6 software (Wyatt Technology). Molecular weights and estimated errors were calculated across eluted peaks by extrapolation from Zimm plots using a dn/dc value of 0.1850 ml/g. SEC-MALS data are presented as differential refractive index (dRI) profiles, with fitted molecular weights (M w s) plotted across elution peaks.

Circular dichroism (CD) spectroscopy
Far UV circular dichroism (CD) spectroscopy data were collected on a Jasco J-810 spectropolarimeter (Institute for Cell and Molecular Biosciences, Newcastle University). CD spectra were recorded in 10mM Na 2 HPO 4 /NaH 2 PO 4 pH 7.5, at protein concentrations between 0.1-0.5 mg/ml, using a 0.

Electron Microscopy
Electron microscopy (EM) was performed using FEI Philips CM100 and 120 kV Hitachi HT7800 transmission electron microscopes at the Electron Microscopy Research Services, Newcastle University. Protein samples at 0.005-3 mg/ml were applied to carbon-coated grids and negative staining was performed using 2% (weight/volume) uranyl acetate. Images were analysed in the Fiji distribution of ImageJ 73 , by applying background correction and an FFT Bandpass filter, and measuring the number, mean width and length of fibres using the Ridge detection algorithm. Data processing and statistical analysis were performed using Prism (GraphPad).

Size-exclusion chromatography small-angle X-ray scattering (SEC-SAXS)
SEC-SAXS experiments were performed at beamline B21 of the Diamond Light Source synchrotron facility (Oxfordshire, UK). Protein samples at concentrations >5 mg/ml were loaded onto a Superdex™ 200 Increase 10/300 GL size exclusion chromatography column (GE Healthcare) in 20 mM Tris pH 8.0, 150 mM KCl at 0.5 ml/min using an Agilent 1200 HPLC system. The column outlet was fed into the experimental cell, and SAXS data were recorded at 12.4 keV, detector distance 4.014 m, in 3.0 s frames. Data were subtracted and averaged, and analysed for Guinier region Rg and cross-sectional Rg (Rc) using ScÅtter 3.0 (http://www.bioisis.net), and P(r) distributions were fitted using PRIMUS 74 . Ab initio modelling was performed using DAMMIF 75 , in which 30 independent runs were performed in P1 or P22 symmetry and averaged. Multi-phase SAXS ab initio modelling was performed using MONSA 76 . Crystal structures and models were docked into DAMFILT and MONSA molecular envelopes using SUPCOMB 77 , and were fitted to experimental data using CRYSOL 78 and FoXS 79 , and flexible termini were modelled and fitted to experimental data using CORAL 80 .

Protein sequence and structure analysis
Multiple sequence alignments were generated using Jalview 83 , and molecular structure images were generated using the PyMOL Molecular Graphics System, Version 2.0.4 Schrödinger, LLC.    (top) that was subjected to iterative energy minimisation and geometry idealisation such that S2C-Ctip sequences of adjacent dimers combined into capping four-helical bundles at either end of the 4:4 molecule (bottom). Fig. 8. Modelling of the SYCE2-TEX12 core fibrous assembly. K1/K10-1B complex is a dimer that forms an anti-parallel hetero-tetramer (PDB accession 6EC0) 53 . Higher order assembly occurs through unit length filaments, into 10-nm fibres, which are directly comparable to SYCE2-TEX12 10-nm fibres, and are shown here for vimentin (middle). Scale bars, 100 nm. Reproduced from Helfand, et al. 54 . The 10-nm IF fibres can further assemble into thick bundles, which are directly comparable with SYCE2-TEX12 40-nm fibres. Please see Kayser, et al. 58 for electron micrographs of keratin's thick bundles.

Supplementary Material
Refer to Web version on PubMed Central for supplementary material. (a) Schematic of the synaptonemal complex (SC) mediating full synapsis between homologous chromosomes that are physically tethered at both ends to the nuclear envelope. The SC has a tripartite structure of an electron-dense central element lying midway between chromosome-bound lateral elements (electron micrograph reproduced from Kouznetsova,et al. 3        (a) Molecular model of the SYCE2-TEX12 core ΔCtip 'closed' 4:4 assembly. The Cterminal bundle is formed of interactions between two S2C-Ctip coiled-coils, each of which involves heptad interactions between S2C residues V149, V153, V156 and L160, and Ctip residues L110, F114, I117 and L121, whilst F102, F109 and V116 are solvent-exposed. (b) Multiple sequence alignment of the TEX12 C-terminus, highlighting the Ctip sequence and the presence of LFIL residues at heptad positions, and FFV residues at non-heptad positions.  Europe PMC Funders Author Manuscripts back through stacked S2C-Ctip anti-parallel four-helical bundles (left), of which each bundle contains a hydrophobic core of Ctip LFIL and S2C residues, with FFV residues remaining solvent-exposed (right). This assembly mechanism generates fibres of discrete lengths corresponding to the 20 nm initial subunit and integer multiples of the 15 nm repeating unit, with a width of 2 nm or 4 nm if it is formed of 2:2 or 4:4 complexes, respectively. (d-e) SEC-SAXS analysis of SYCE2-TEX12 core assemblies. (d) SAXS P(r) maximum dimensions (Dmax, red), cross-sectional radii (Rc, yellow) and corresponding MALS molecular weights (diamonds, blue) are shown across an elution peak; Dmax is not determined for higher-order species in which the Shannon limit is exceeded. (e) SAXS P(r) interatomic distance distributions of the SYCE2-TEX12 core 4:4 complex (blue) and its fibrous assemblies of lengths 35 nm (red) and 65 nm (yellow).  into 'closed' 4:4 structures that directly assemble into 4-nm fibres. The lateral association of 4-nm fibres generates 10-nm fibres, which become intertwined in bundled fibres of up to 40 nm width and many micrometres in length, and fulfil the known dimensions of the SC central element.