Sexual dimorphism in the width of the mouse synaptonemal complex

The synaptonemal complex (SC) is a meiosis-specific protein complex conserved across species (reviewed in von Wettstein et al., 1984; Zickler and Kleckner, 1998). The SC functions as a scaffold for meiotic recombination, facilitating the exchange of genetic information and creating physical links (chiasmata) between homologous chromosomes that are required for correct chromosome segregation.

The SC has a characteristic tripartite ladder-like structure where transverse filaments (TFs) connect a central element (CE) to the lateral elements (LEs) of each homologue. In mammals, the length of the SC ranges between 220 and 360 nm, with bigger chromosomes having longer SCs (Lynn et al., 2002), whereas its width is relatively constant at ∼200 nm (Moses, 1956; Westergaard and von Wettstein, 1972; Schucker et al., 2015). Several SC proteins have been identified in mammals: LE proteins SYCP2 and SYCP3, the TF protein SYCP1, and CE proteins SYCE1, SYCE2, SYCE3 and TEX12, and SIX6OS1 (Costa et al., 2005; Hamer et al., 2006; Heyting et al., 1989; Lammers et al., 1994; Meuwissen et al., 1992a; Offenberg et al., 1998; Schramm et al., 2011; Gómez et al., 2016).

Given the relatively small dimensions of the SC, this structure has been analysed mostly by electron microscopy. However, in recent years, advances in light microscopy have enabled the imaging of specimens with spatial resolutions below the diffraction limit of visible light (∼250 nm). One such technique is structured illumination microscopy (SIM). SIM works by recording wide-field images while illuminating the sample with a grid pattern of light. The patterned illumination ‘interferes’ with the spatial distribution of the sample, setting up beat-frequencies (Moiré fringes), which carry high-resolution information. After computational reconstruction of several images with the grid pattern imaged in different orientations, an image of the object with extended resolution is computationally calculated (Gustafsson et al., 2008). Another super-resolution technique shown to render the SC tripartite structure is stochastic optical reconstruction microscopy (STORM) (Schucker et al., 2015). This method relies on the detection of individual molecules based on their photoswitchable properties. Each activated molecule, in turn, leads to the emission of sufficient photons to enable precise nanoscale localization before it becomes deactivated (Betzig et al., 2006; Hess et al., 2006; Rust et al., 2006).

In mammals, studies of the SC structure have been predominantly done in male germ cells, as these are more abundant and easier to retrieve than female germ cells. However, sexual dimorphism in SC morphology and how the SC is generated, are known to occur during mammalian meiosis. For instance, male SCs are known to be shorter and to have fewer recombination sites than those of females (Kleckner et al., 2003; Bojko, 1983; Gruhn et al., 2013).

We set out to analyse whether the SC displays any other forms of sexual dimorphism aided by several super-resolution techniques and electron microscopy. Importantly, we found that SC width in oocytes is narrower than in spermatocytes at the pachytene stage of meiosis I. We show that this difference is not a result of altered LE organization, instead, the central region (CR) and the CE were found to have significant morphological differences between the sexes. Thus, our data suggest that in mice, the sexual dimorphism in SC width may be due to non-identical organization of the CR components.

In pachytene cells, a distance of ∼100 nm separates the fully paired LEs of the SC (Fig. 1A). Thus, in order to analyse the organization of male and female SCs, we employed super-resolution structured illumination microscopy (SIM), a technique that allows for a twofold increase in lateral resolution when compared with conventional light microscopy (∼250 nm). Wild-type spermatocytes and oocytes at the pachytene stage of meiosis I were immunostained with an antibody against the LE protein SYCP3. In male and female pachytene cells, despite the two LEs of each SC being readily detected as two parallel SYCP3-labelled tracks, the SCs displayed an apparent difference in the width of the CR, judging by the space detected between the two LEs (Fig. 1B, insets and Fig. 1C, straightened chromosomes).

We further confirmed the difference in SC width using stochastic optical reconstruction microscopy (STORM), a super-resolution imaging method for single-molecule localization detection with higher resolving power, and a well suited technique for quantitative evaluation of SC topology (Schucker et al., 2015) (Fig. 1D,E, straightened chromosomes). Using this method, we quantified the width between the centres of each SYCP3-labelled LE (inter-LE distance) in areas of the SC in frontal view. The inter-LE distance was found to be on average 210.0±3.7 nm (mean±s.e.m.) in male SCs (Fig. 1F). Remarkably, the inter-LE distance of female SCs was significantly narrower, 143.0±2.8 nm (Fig. 1F). In addition, plotting all inter-LE widths clearly showed a narrower bimodal distribution of female SYCP3 compared with male SYCP3 (Fig. 1G), further demonstrating the difference in inter-LE distance. These data demonstrate that SC width varies between the sexes, and suggests a different organization of LEs or CR components. In fact, sexual dimorphism in CR width has been described in the silk worm Bombyx mori (von Wettstein et al., 1984). In this insect, the CR of female SCs is 70–80 nm wide, whereas in spermatocytes, the CR has a standard width of 100–120 nm (Rasmussen and Holm, 1979).

In order to pinpoint the origin of the inter-LE difference, we analysed male and female SCs with electron microscopy. In pachytene cells, the two LEs are observed as dark, electron-dense structures, defining the limits of the SC, whereas inwards, the CR is composed mostly of poorly stained TFs perpendicular to a densely stained CE (Fig. 2A,B). Taking advantage of these different properties in electron density, we could confirm the reduced inter-LE distance in female SCs (138.0±2.3 nm) compared with male SCs (150.5±1.2 nm) in internal SC regions (Fig. 2B, graph). In addition, we analysed the inter-LE width in telomeric SC regions, i.e. associated with the nuclear membrane (Fig. S1). The female inter-LE distance was also narrower than in male SCs (Fig. S1, top graph), while the mean inter-LE distance did not differ between membrane-associated and internal SC regions within each sex (Fig. S1, bottom graph). Furthermore, while the LEs of female and male SCs were identical in width (Fig. 2C), the female CE (22.1±0.6 nm) was significantly narrower than that of males (24.6±0.5 nm) (Fig. 2D). These data, together with the results discussed above, exclude a different organization of the LEs (i.e. LE width) as the cause of the narrower inter-LE in females, instead suggesting that the difference in width arises from a different organization of the CR or TFs.

The major contribution to the SC width comes from the TF protein SYCP1; mutations altering the length of the central α-helical domain of SYCP1 influence the width of SC-like polycomplexes formed in a heterologous system by overexpressing SYCP1 protein (Ollinger et al., 2005; Tung and Roeder, 1998; Page et al., 2008). We analysed the possibility of the existence of two differently sized SYCP1 isoforms for each sex by comparing the size of SYCP1 isolated from male and female germ cells, using E17.5 embryonic ovaries and adult testes. We did not detect a difference in the number and/or size of the bands corresponding to the detection of SYCP1 by immunoblotting analysis (Fig. S2A), suggesting that a different isoform of SYCP1 is not expressed in female pachytene cells.

Next, we asked whether SYCP1 filaments could be organised differently in females. TFs are built up of SYCP1 homodimers and it has been shown that SYCP1 C-termini localize within the LEs, whereas SYCP1 N-termini interact in the CE (Hernández-Hernández et al., 2016; Liu et al., 1996; Meuwissen et al., 1992b; Schmekel et al., 1996). Given that the mass of the SYCP1 protein was found to be the same both in males and females (Fig. S2A), a different integration site for SYCP1 C-termini (e.g. the C-terminus protruding deeper into the LEs in females than in males) could explain the inter-LE difference detected in females. To test this hypothesis, SYCP3 and the C-terminal region of SYCP1 were detected and imaged with SIM, in male and female pachytene cells (Fig. 3A and Fig. S2B). Two SYCP1 C-termini tracks could occasionally be resolved in male SCs, but not in female SCs where a single SYCP1 track was always detected (Fig. 3A,B, insets and graph, respectively), thus precluding a quantitative analysis with this method. Using STORM microscopy, we could readily detect the two C-termini tracks, apparently close to the inner edge of the LEs (Fig. 3C insets and Fig. S2C), in male SCs. Remarkably, in female SCs, despite the increase in resolution with this method, the two C-termini tracks could not be resolved in the majority of chromosomes (Fig. 3C, insets). The combined intensity profiles of SC frontal views clearly showed that in males, SYCP1 C-termini were present at the inner edge of the LEs, as previously reported (Schucker et al., 2015) and spanning ∼130 nm (Fig. 3D, top graph, green arrows). In contrast, in female SCs only a single peak of SYCP1 was detected, i.e. C-termini are closer to each other (Fig. 3D, middle graph).

To further analyse the difference in the C-terminal peaks, the female SYCP1 distribution was fitted using the width parameters from the male SYCP1 analysis (see Materials and Methods). This model deduced that the two female SYCP1 peaks are separated by ∼50 nm (Fig. 3D, bottom graph, green arrows and dashed lines), which is less than twice of that found in males. Importantly, the peak to peak distance between SYCP1 C-terminus and SYCP3 was 58 nm in male SCs (Fig. 3D, top graph, red arrows), whereas in female SCs this distance now modelled to be ∼13 nm shorter (45 nm, Fig. 3D, bottom graph, red arrows). Considering that SYCP1 protein size was found to be the same (Fig. S2A) and that the LEs were found to have the same width in both sexes (Fig. 2C), a shorter distance measured between the LE and SYCP1 C-terminal suggests that in females, the C-terminus is integrated deeper into the LE. In agreement, the double immunostaining of SYCP3 and SYCP1 in females appeared to colocalize to a large extent (Fig. 3C).

Although it is difficult to predict the implications of the SC structural dimorphism we observe, a crucial role for a non-identical assembled CR on meiotic recombination cannot be ruled out. In human and mouse females, SC length is longer than that of males, with the increase in length correlating with an increase in number of crossing-over events (COs). Yet, in human females 25% of all designated COs are absent, leading to a reduction of 33% in the overall CO density (Wang et al., 2017). This reduction is due to inefficient CO maturation (Wang et al., 2017). It is tempting to speculate in light of our results that a narrower SC in human females similar to that observed in mouse females, could generate a restrictive environment for the recombination machinery to process all designated COs, thus limiting and/or preventing their maturation.

In summary, we have observed a narrower CE and a deeper integration site for SYCP1 C-termini, in female SCs (Fig. 3E). Yet, these features do not fully account for the narrower inter-LE measured. One explanation for the narrower width in females could be the existence of post-translational modifications (PTMs) acting on SYCP1. Given that SYCP1 is the major determinant of SC width, PTMs leading to different folding of its coiled-coil region could render a more compact conformation. At present, we also cannot exclude a different assembly and/or interaction of SYCP1 N-termini with the remaining CE components within the CE, or that a longer piece of N-terminal SYCP1 be accommodated in the female CE, thus contributing to the narrower female inter-LE detected.

The animal experiments were approved by the Stockholm-North Animal Ethical Committee. Wild-type C56BL/6J mice were used in this study. Experimental animals were compared with controls from the same litter (when possible) or from other litters from the same matings. To detect pregnancy, females were caged with males and examined for vaginal plugs the following morning, and then on a daily basis. The day that a plug was found was marked as embryonic day (E) 0.5. For ovary sampling at embryonic stages, pregnant female mice were killed at days 17.5–18.5 post-partum. Mating males were also sacrificed and testes were used for further analysis.

Testicular and ovarian extracts were prepared as previously described (Fukuda et al., 2014). In brief, testes or ovaries were homogenized in a buffer containing 50 mM Tris-HCl (pH 7.5), 150 mM NaCl, 1 mM EDTA, 1% NP-40, 0.5% Na deoxycholate, 0.1% SDS, 1 mM phenylmethylsulfonyl fluoride (PMSF) and the complete protease inhibitors (Roche). Proteins were separated on 4–15% Mini-PROTEAN TGX gels (Bio-Rad) in Tris-glycine running buffer and were subsequently transferred onto PVDF membranes (Bio-Rad). Signals were detected with horseradish peroxidase-conjugated secondary antibodies and visualized by ECL Prime (GE Healthcare). The quantitative evaluation of the bands was carried out with Image Lab software (Bio-Rad).

Slides with embryonic oocyte and spermatocyte spreads were fixed in 0.8% and 1% paraformaldehyde (PFA), respectively, and as previously described (Kouznetsova et al., 2005; Agostinho et al., 2016), using a drying-down technique (Peters et al., 1997). The following primary antibodies were used at the indicated dilutions: guinea pig anti-SYCP1, 1:100 (C-terminal amino acids 978–993 of mouse SYCP1, Q62209 SwissProt; Kouznetsova et al., 2005) and mouse anti-SYCP3 (sc-74569) 1:500–1:200 from Santa Cruz Biotechnology. Secondary antibodies used for detection with SIM were: goat anti-guinea-pig Alexa Fluor 555 (A-21435) and goat anti-mouse Alexa Fluor 488 (A-11001) 1:800–1:600, from Life Technologies. Incubation of primary antibodies was done at room temperature overnight. Secondary antibodies used for detection with STORM were: goat anti-guinea-pig 647 (A-21450) and donkey anti-mouse 555 (A-31570), 1:400–1:200 from Life Technologies. Samples for SIM and STORM imaging were mounted in a medium consisting of 10% (v/v) Vectashield and 90% (v/v) glycerol in 50 mM Tris-HCl, pH 7.4 (Olivier et al., 2013).

SIM was performed on a Zeiss Elyra PS.1 system. Images were captured with an Andor iXon DU 855 camera and a Plan-Apochromate 100×/1.46 NA oil immersion objective. Excitation wavelengths used were 488 and 561 nm, and fluorescence emission was collected through appropriate dichroic mirrors and single colour band-pass filters (495–550 nm and 570–620 nm). An EMCCD gain of 10–20 and 5 grid rotations were applied with camera integration times of 150 to 250 ms, Calibration on 40 nm green fluorescent beads delivering a maximal focal precision of approximately 85 nm laterally and 250 nm axially (Liu et al., 2016).

STORM was performed using the same Zeiss Elyra PS.1 microscope as for SIM. For single-molecule localization, de-activation and excitation 405, 561 and 642 nm high power lasers were used. The objective applied was the 100×/1.46 oil immersion lens and fluorescence was detected on a liquid cooled EMCCD camera (iXon DU 897 Andor Technology). Single-molecule fluorescence emission was collected through appropriate multi-colour dichroic mirrors and dual colour filters, collecting at 570–635 nm for detection of Alexa Fluor 555, and 655 nm long-pass for detection of Alexa Fluor 647. Alexa Fluor 555 de-activation and excitation was done with 100% laser power of 561 nm (maximum input of 57.8 mW at the back-focal plane of the objective). For STORM imaging of Alexa Fluor 647, back-pumping with 0.5% of 405 nm laser power (maximum input 10.1 mW) was used together with 100% laser power of 642 nm de-activation and excitation (maximum input power 25.4 mW). Generated focal illumination was achieved over a 25 µm² area (1/e² spatial irradiance distance) in the sample in inclined HILO-mode (Tokunaga et al., 2008). Single-molecule fluorescence detection with the EMCCD camera was acquired over the full field of view, with 100×100 nm pixel sizes, 25 ms integration time and 100 gain for Alexa Fluor 555 and Alexa Fluor 647, respectively. A total of 10,000 images were acquired for each cell imaged in an interleaved manner where 500 images of Alexa Fluor 555 were acquired followed by 500 images of Alexa Fluor 647 which was then repeated 20 times. All filters and mirrors used were such that only back-pumping and excitation/de-activation lasers were either switched on or off to alter which dye was imaged.

Raw SIM datasets were processed with the integrated ELYRA PS.1 system analysis software (Zen 2011 SP2 Black) with selection of automatic settings for SIM evaluations (i.e. theoretical PSF, noise filter setting, frequency weighting, baseline handling, etc.). After evaluation, SIM images were checked for possible artefacts (e.g. honeycomb patterns of intensity in the image) to confirm suitable evaluation settings were applied (Komis et al., 2015).

Single-molecule emission events in the STORM datasets were localized using the ImageJ plugin SMLocalizer (Kristoffer Bernhem and Hjalmar Brismar; SMLocalizer, a GPU accelerated ImageJ plugin for single molecule localization microscopy, https://doi.org/10.1093/bioinformatics/btx553). To avoid edge artefacts due to incomplete single-molecule switching, localization analysis was performed over the central 15×15 µm STORM imaged area. Resulting localization tables were filtered based on photon count (>100 photons; i.e. removal of background) and accuracy of fit (R²>0.8: i.e. removal of multi-emission events skewing detection profiles). For visualisation, 5×5 nm pixel plotted images were smoothed with a 10-nm-wide Gaussian. Rendered images were then used to calculate line profiles of the protein distributions across individual chromosomes, using the mean of a 10-pixel-wide line, orthogonal to the long axis of the chromosome. The peak to peak distances were measured at the widest points on each chromosome (for at least 3 points per chromosome), and the mean of all values obtained was used as the reported width for males and females.

To analyse the protein's topological distribution from the rendered STORM images further, the data was also fitted with a double Gauss function for male and female line profile calculations: f(x)=a₁*exp{−[(x−b₁)/c₁]^2}+a2*exp{−[(x−b₂)/c₂]^2}, where a₁(2), b₁(2), c₁(2) model the line profiles' heights, central spatial positions of peaks, and the widths of the peaks. Fitting of the SYCP3 data generated line profile distributions with R-square (goodness of fit) values of 0.9975 and 0.9908 for male and female, respectively (i.e. Fig. 3D top and middle). For SYCP1 the male distribution fitted to the double Gaussian function generated a goodness of fit with R-square=0.9651 (i.e. Fig. 3D top). Fitting the female SYCP1 distribution, assuming male peak widths, allowing amplitudes and centres to move freely, gave a goodness value of R-square=0.9909.

Small pieces of testis and fetal ovaries were fixed in 2.5% glutaraldehyde and 4% PFA in cacodylate buffer (pH 7.2). After post-fixation with 2% OsO₄, pre-embedding staining was performed with 0.5% uranyl acetate. Samples were dehydrated through graded ethanol solutions, embedded in epoxy resin durcupan ACM (Electron Microscopy Sciences) and polymerized for 48 h at 60°C. 70 nm ultra-thin sections were collected on Formvar/carbon-coated one-slot copper grids (Agar Scientific), contrasted with uranyl acetate and lead citrate before examination in a transmission electron microscope (Philips, CM120) at 100 kV.

The statistical analysis was performed using GraphPad Prism 6 software. P-values reported in figure legends are two-tailed probabilities calculated by Mann–Whitney two-sided non-parametric test. The non-parametric tests were used as the data distributions did not pass the D'Agostino and Pearson omnibus normality test with α=0.05%.

We are grateful to S. Valentiniene, J. G. Liu and D. Jans for technical support.