IGSF11 is required for pericentric heterochromatin dissociation during meiotic diplotene

Meiosis initiation and progression are regulated by both germ cells and gonadal somatic cells. However, little is known about what genes or proteins connecting somatic and germ cells are required for this regulation. Our results show that deficiency for adhesion molecule IGSF11, which is expressed in both Sertoli cells and germ cells, leads to male infertility in mice. Combining a new meiotic fluorescent reporter system with testicular cell transplantation, we demonstrated that IGSF11 is required in both somatic cells and spermatogenic cells for primary spermatocyte development. In the absence of IGSF11, spermatocytes proceed through pachytene, but the pericentric heterochromatin of nonhomologous chromosomes remains inappropriately clustered from late pachytene onward, resulting in undissolved interchromosomal interactions. Hi-C analysis reveals elevated levels of interchromosomal interactions occurring mostly at the chromosome ends. Collectively, our data elucidates that IGSF11 in somatic cells and germ cells is required for pericentric heterochromatin dissociation during diplotene in mouse primary spermatocytes.


Author summary
For sexually reproducing species, the number of chromosomes in a mature germ cell is half that of a typical somatic cell, and its chromosome sequence is not identical to that of parental cell, these changes result from a highly specialized cell division process named meiosis. In contrast to mitosis, germ cells undergo many meiotic-specific regulatory processes during prophase I of meiosis. In mammals, the development of male and female meiotic germ cells relies on completely different microenvironment provided by sexually specialized gonadal somatic cells, but what gene is required for germ cells and gonadal somatic cells to mediate meiosis progression is largely unclear. Here, we construct a Introduction Germ cells carry the task of reproduction and creating diversity for sexually reproducing species. In mammals, although both male and female germ cells undergo meiosis with the support of gonadal somatic cells, their surrounding microenvironments are quite different. In mouse ovary, female germ cells are enclosed by pregranulosa cells and enter meiosis during fetal development. In contrast, postnatal male germ cells undergo meiosis in the space between Sertoli cells within the seminiferous tubules. Although somatic cells have been shown to be required for meiotic entry and progression in mice [1][2][3], little is known about what proteins connecting gonadal somatic cells and germ cells are required for this process. Igsf11, which belongs to the CTX family of immunoglobulin like cell adhesion molecule (IgCAMs, or immunoglobulin superfamily, IgSF), was identified in 2002 and found to be preferentially expressed in testis and brain [4]. This family of proteins are single-pass type I transmembrane glycoprotein and contain a membrane-distal V-type and a C2-type domain in their extracellular region [5]. IGSF11 functions as a cell adhesion molecule [6] and participates in regulation of synaptic plasticity [7]. Disruption of Igsf11 or conditional knockout of Igsf11 in Sertoli cells results in male infertility in mouse [8]. However, it is not clear if IGSF11 is required in germ cells or for progression of meiosis.
Mouse centromeres can be divided into centric and pericentric heterochromatin (PCH) domains [9]. Mouse PCH is mainly composed of major satellite repeats and enriched in repressive histone marks, such as H3K9me3 (trimethylation of histone H3 on lysine 9). During interphase and prophase I, PCH domains from different chromosomes are organized to form several chromocenters which can be cytologically visualized as bright DAPI staining chromosomal domains. During meiosis, chromosomes exhibit meiosis-specific chromosomal structures and behaviors at telomere [10], chromosome arm [11,12] and centromere [13]. In contrast to telomeres and chromosome arm regions, our understandings and characterizations of pericentric heterochromatin during meiosis is still limited.
In this study, we generated a fluorescent reporter for monitoring meiotic entry and prophase I progression. Using this system, we examined the potential role of an adhesion molecule, IGSF11, during meiosis. After analyzing the impact of Igsf11 deletion on meiosis progression, we uncover an unexpected role of IGSF11 in regulating nonhomologous pericentric heterochromatin dissociation during male meiosis.

Generation of mVenus-P2A-HA-Sycp3 reporter mice
To examine the potential functional requirement of Igsf11 during meiosis, we generated a meiotic reporter system which allows us to examine the progression of meiotic prophase I and isolate meiotic cells from mouse testis.
As shown in Fig 1A, the reporter gene cassette consists of three components: a monomer fluorescent protein mVenus with higher fluorescent intensity than commonly used EGFP [14,15], a P2A linker which allows efficient in vivo self-cleavage of fusion proteins after translation [16], and a copy of HA epitope tag (See S1 Appendix for transgene DNA sequence). The reporter cassette was inserted right after the start codon of Sycp3 using the CRISPR/Cas9 "double nicking" strategy [17] by targeting at -168b and -130b upstream Sycp3 translation start site (S1A Fig). Out of 23 targeted live offspring, 1 contained a correct knockin allele without unwanted sequence variation from +2020b to -1741b around the target site, confirmed by DNA sequencing of the amplified fragment ( Fig 1A). The transgenic mice were named mVenus-P2A-HA-Sycp3 (VPHS) transgenic mice hereafter.
Heterozygote (Hetero) and homozygote (Homo) VPHS mice were fertile in both sexes, and generated similar litter sizes to those of wild type (WT) mice. Under the excitation light of a dissecting microscope, green fluorescence was detectable in both homozygote and heterozygote VPHS testis but not WT testis or surrounding tissues (Fig 1B), and brighter fluorescent signal was observed in seminiferous tubules of homozygotes compared to heterozygotes. Histological analysis revealed that spermatogenesis was normal in adult VPHS transgenic testis, since elongated sperms were observed in both seminiferous tubules and epididymides of homozygote or heterozygote VPHS mice, the same as wild type control (S1B Fig). However, some degenerated cells occasionally appeared in the homozygote tubules, and fewer sperms as well as more cell debris were found in the homozygote epididymis than heterozygote or wild type littermates (S1B Fig arrowheads), suggesting an overall reduced spermatogenic activity in homozygote VPHS testis, which may explain the smaller testis size for VPHS homozygote ( Fig  1B).
Subsequently, the expression of Sycp3 between different VPHS genotypes were evaluated by quantitative PCR with adult testis. Total Sycp3 transcript levels were comparable between different genotypes (S1C Fig). As expected, we found that the expression of wild type Sycp3 (WTS Sycp3) transcript decreased by half in the heterozygotes compared to wild type testes, and the expression of the knockin Sycp3 (KIS Sycp3) transcript in the heterozygotes was nearly half that of the homozygotes (S1D Fig). These results suggest that there was no obvious difference in either transcription or transcript stability between wild type Sycp3 and transgenic Sycp3 alleles in VPHS transgenic mice.
Total proteins were also collected from adult testis with different VPHS genotypes to verify SYCP3 expression by Western blot analysis. As expected, HA-SYCP3 and mVenus were only detected in the heterozygote and homozygote VPHS samples, and their expression in homozygotes is higher than that in heterozygotes ( Fig 1C). No additional larger bands were observed from blotting results using mVenus or HA antibodies (S1E and S1F Fig), suggesting the high cleavage efficiency of P2A linker under physiological conditions. Mouse Sycp3 gene has only one transcription isoform but 2 translation isoforms (predicted to be 27.2kD and 29.3kD) [18], and the VPHS transgene was integrated right after the upstream translation start site of Sycp3. If both translation start site were used in the transgenic Sycp3 transcript, there will be a 30.5kD HA tagged SYCP3 and a 27.2kD SYCP3 isoform. Interestingly, when SYCP3 polyclonal antibody was used to blot the protein extract, the 27.2kD shorter translational isoform that is the prominent band in wild type samples is absent from homozygous samples (Figs 1C and S1G). Since the total expression level of SYCP3 was not significantly different among the three VPHS genotypes (S1H Fig), we concluded that VPHS transgene insertion did not affect translational level of Sycp3 transcript, but may have altered its translation initiation site selection.
To confirm the localization of transgenic HA-SYCP3 within meiotic germ cells, frozen sections were prepared for immunostainings using homozygote gonads (Fig 1D). HA signal was hardly found in 17.5 dpc fetal testis that has not initiated meiosis. In the 17.5 dpc fetal ovary and the 17dpp postnatal testis, HA-SYCP3 signal showed linear patterns in the pachytene stage meiocytes that were also expressing the cytoplasmic germ cell marker MVH. Furthermore, in 24 dpp testis, no HA-SYCP3 was detected in round spermatids, which were identified by distinct nucleus morphology and a single MVH-positive chromatoid body right next to the nucleus (Fig 1D arrowheads). By comparing 17 dpp tubules of different VPHS genotypes, HA-SYCP3 staining was found to increase from heterozygote to homozygote, and colocalized well with SYCP3 staining (S1I Fig). These data suggest that transgene insertion did not influence the meiotic specific expression pattern, the localization to chromosome axes and the synaptonemal complex (SC), or the degradation after meiotic prophase of HA-SYCP3.
Mouse oocytes develop to leptotene, zygotene, pachytene and diplotene at 13.5dpc, 15.5dpc, 17.5dpc and 19.5dpc, respectively, whereas spermatocytes develop to corresponding stages at 9dpp, 15dpp, 17dpp and 19dpp, respectively. To confirm if the meiotic progress could be indicated by the fluorescence of the VPHS reporter, male and female gonads were analyzed by FACS (Fig 1E). Since the fluorescent intensity was substantially stronger in homozygote than that in heterozygote (S2A Fig), we only used homozygote VPHS mice in this analysis. In fetal ovaries, the mVenus fluorescence was hardly detected at 12.5 dpc, gradually increased from 13.5 dpc to 17.5 dpc when oocytes develop from leptotene to pachytene, and decreased at 19.5 dpc when diplotene oocytes appear (Figs 1E and S2B). In the postnatal testis, the mVenus fluorescence appeared at 9 dpp when spermatocytes enter meiosis, and increased from 15 dpp to 17 dpp. Because mVenus fluorescence decrease within spermatocyte after diplotene (20 dpp), these cells overlap with earlier stage spermatocytes in the FITC channel (Figs 1E and S2B). Furthermore, our meiotic spread staining results also confirmed the correlation between meiotic progression and mVenus fluorescent intensity in the spermatocytes (S2C and S2D  Fig). Unexpectedly, mVenus fluorescent is also detected in the fetal testis from 15.5 dpc to 19.5 dpc (S2E Fig). Indeed, SYCP3 is expressed in 13.5 dpc fetal testis, but not at later fetal stages [19], and our Q-PCR result revealed that the expression pattern of Sycp3 transcript within fetal gonads is not affected by VPHS transgene (S2F Fig), so the expression pattern of mVenus fluorescence may reflect a delayed degradation for mVenus protein within the dormant male germ cells.

Igsf11 is expressed by both spermatogenic cells and Sertoli cell within mouse seminiferous tubules
We generated a Igsf11 knockout mouse by CRISPR/Cas9-mediated gene targeting in mouse embryos. The knockout allele lacks 19 base pairs in exon 2 of Igsf11 (Figs 2A and S3B), leading to a frameshift mutation and premature translation termination. Western analysis of Igsf11 -/testis extract confirmed no expression of IGSF11( Fig 2B). Igsf11 +/male, Igsf11 +/female and Igsf11 -/female mice were fertile, but Igsf11 -/male mice were infertile. Testes of the Igsf11 -/mice were significantly smaller than testes of Igsf11 +/+ and Igsf11 +/- (Fig 2C and 2D). Although spermatocytes exist in Igsf11 -/tubules, no elongated spermatids were found in both tubules and epididymis, leaving only round-shaped cell debris in the caput epididymis ( Fig 2E).
Next, the expression pattern of Igsf11 was analyzed by quantitative PCR. During fetal development, elevation of Igsf11 transcript was detected in testis at 17.5 dpc, but not in littermate ovary (S3C Fig). While in postnatal testis, Igsf11 was expressed at a low level from 1-19 dpp, the expression increased significantly at 21-23 dpp, and became even higher at 25 dpp ( Fig  2F). When the expression of Igsf11 was compared between Sertoli cells and spermatogenic cells (Figs 2G, S3D and S4C), we found a low but detectable level of Igsf11 in Sertoli cells, which exhibited a slight upregulation after 25 dpp, and the expression level was always lower than that in adult testis or non-Sertoli seminiferous tubule cells. Igsf11 expression level was low in C-Kit + differentiated spermatogonia, increased from primary spermatocyte to secondary spermatocyte, and reached the highest level in spermatid. A recent single-cell transcriptome study also showed an increasing expression pattern of Igsf11 in spermatogenic cells from the pachytene stage onwards (S3E Fig) [20].
Next, immunostaining was performed to localize IGSF11 within squashed seminiferous tubules ( Fig 2H). The specificity of our IGSF11 antibody was confirmed by Western blot and immunostaining with 293FT cells that overexpressed HA-IGSF11 (S3F and S3G Fig). First, we observed that Sertoli cell derived IGSF11 colocalizes with ZO-1, a tight-junction marker of blood-testis barrier (BTB) in the seminiferous tubules ( Fig 2H, panels a-d). Besides its localization to cellular interfaces, IGSF11 also appeared as punctate aggregates (putative secretory vesicles) in cytoplasm, which was not detected in Igsf11 -/testicular cells (Fig 2H, panels s-v). Coimmunostaining of IGSF11 with Sertoli cell marker GATA4 (Fig 2H, panels e-h) or primary spermatocyte marker SYCP3 (Fig 2H, panels i-n) further confirmed IGSF11 expressed in the cytoplasm of both cell types. Consistent with our transcription analysis, IGSF11 expression was much higher in spermatids than in spermatocytes ( Fig 2H, panels o-r). The expression of IGSF11 in primary spermatocytes was also confirmed by collecting VPHS+ primary spermatocytes (4C) cells for Western analysis (Fig 2I). In summary, mRNA and protein expression analysis both confirm that Igsf11 is expressed in both Sertoli cells and multiple stages of spermatogenic cells.

Adhesion molecule IGSF11 from Sertoli cells and spermatogenic cells are both required for the completion of meiosis
Cells that express IGSF11 at high level tend to form aggregates in vitro, indicating that IGSF11 functions as a cell adhesion molecule [6]. When ectopically expressed in human 293FT cells, IGSF11 was specifically enriched on the contact surface of two IGSF11-positive cells (S4A Fig,  arrowhead), but was not enriched at the interface between expressing and non-expressing cells (S4A Fig, triangular arrowhead).
Since homophilic adhesion molecule IGSF11 is expressed in both Sertoli cells and germ cells, we next examined the effect of IGSF11 deletion in Sertoli cells or germ cells on meiotic progression by two sets of testicular cell transplantation experiments, as illustrated in Figs 3A and S4B and S1 Table. After surgical transplantation of testicular cells into the seminiferous tubules of the germ-cell depleted recipient mice, donor SSCs will colonize in the recipient testis and restore spermatogenesis. Thus, the requirement of IGSF11 in somatic cells or germ cells can be verified by the combination of donors with different Igsf11 genotypes and recipients with different Igsf11 genotypes, respectively. In order to evaluate the development status of the donor cells, we combined an assay utilizing the VHPS reporter system to distinguish meiotic  As a control, we first applied these assays to the VPHS transgenic Igsf11 +/or Igsf11 -/testis. The mVenus fluorescent signal was much weaker in VPSH; Igsf11 -/testis than that in VPHS; Igsf11 +/testis ( Fig 3B, panels a-d). Accordingly, V&H assay revealed that in the VPHS; Igsf11 +/testis, mVenus+ cells contained 4C, 2C and 1C cells (Fig 3C, panels a and b). In contrast, the mVenus+ cells in VPSH; Igsf11 -/testis only contained 4C cells, suggesting that the germ cells had entered meiosis and expressed HA-SYCP3, but meiosis I had not been completed (Fig 3C, panels c and d).
In the first set of experiments, we tested whether absence of IGSF11 in Sertoli cells affects meiotic progression by transplantation of VPHS; Igsf11 +/+ testicular cells into the germ-celldepleted Igsf11 +/or Igsf11 -/recipient mice ( Fig 3A, panels b and c). Four months after transplantation, when recipient testes were recovered, we found that the Igsf11 +/control recipient testis showed relatively strong mVenus fluorescent in the seminiferous tubules (

Localizations of Sertoli cells and spermatocytes in seminiferous tubules are not affected in the absence of Igsf11
Expression and requirement of adhesion molecule IGSF11 suggest that cellular interaction between Sertoli cells and germ cells may be important for meiosis completion, so we examined whether the absence of IGSF11 may impede cellular localization of Sertoli cells and germ cells in seminiferous tubules. Since Igsf11 -/testis contains mVenus+ cells arrested before meiosis I as 4C cells (Fig 3C, panels c and d), we examined cellular localization of Sertoli cells and germ cells in seminiferous tubules of 18 dpp testis before the first wave of meiosis I completion occurs and in adult testis in which some spermatocytes have completed meiosis I.
Immunostaining was performed for γH2AX, a DSB marker, and GATA4, a Sertoli cell transcriptional factor (Fig 4A, see schematic diagrams for staining results in Fig 4B, panels a-d). Within 18 dpp Igsf11 +/seminiferous tubules, early-leptotene (ep) spermatocytes were still close to the basal lamina, located between Sertoli cells and basal lamina, or surrounded by Sertoli cell nuclei on both sides (Fig 4A, panel a). Leptotene (le) spermatocytes migrate from the basal compartment to the adluminal compartment, whereas zygotene (zg) and pachytene (pc) spermatocytes are in the middle region of the seminiferous tubules. In 18 dpp Igsf11 -/seminiferous tubules, Sertoli cells and various stage spermatocytes localized similarly to the Igsf11 +/cells ( Fig 4A, panel b). Furthermore, the localization pattern of Sertoli cells and germ cells was also not apparently different between adult Igsf11 +/and Igsf11 -/testes (Fig 4A, panel c and d).
Multiple layers of leptotene spermatocytes appeared in some adult Igsf11 -/seminiferous tubules, which is similar to that observed in 18dpp testis, but zygotene and pachytene spermatocytes also appeared in the same seminiferous tubules, suggesting that meiosis proceeds through pachytene. Hence, the localization pattern of germ cells at different stages in seminiferous tubules and their positions relative to Sertoli cells are not significantly affected in Igsf11 -/testis.
The BTB is a highly specialized barrier formed by Sertoli cells, which divides the intratubular space into a basal compartment and an adluminal compartment. And leptotene spermatocytes migrate from the basal compartment into the adluminal compartment during the process of BTB reconstruction [22]. We next analyzed whether the migration of primary spermatocytes through the BTB is affected by the loss of IGSF11. For this purpose, the testis sections were stained for γH2AX and CX43, a gap junction protein that localizes to the BTB [23] (Fig 4C, see schematic diagrams for staining results in Fig 4B, panels e-h). In 18 dpp Igsf11 +/testis, early-leptotene spermatocytes were located between the BTB and basal lamina, leptotene spermatocytes were enclosed by two layers of BTB, while pachytene spermatocytes were located in the center surrounded by BTB (Fig 4C, panel a). Similar localizations of the spermatocytes and BTB were observed in 18 dpp Igsf11 -/seminiferous tubules (Fig 4C, panel b).
In adult Igsf11 +/seminiferous tubules, many sperm cells occupied the adluminal compartment, confining BTB to the region close to the basal lamina (Fig 4C, panel c). Although the adult Igsf11 -/seminiferous tubules showed a less confined BTB, presumably because the missing post-meiotic cells no longer occupied space in the adluminal compartment, the distribution pattern of migratory spermatocytes was nonetheless similar to wild type in 18 dpp testes (two layers), and the positions of spermatocytes at different stages relative to the BTB were not affected in adult Igsf11 -/seminiferous tubules (Fig 4C, panel d).
Moreover These results suggest that there is no significant disruption of BTB structure in the absence of IGSF11.

Igsf11 deficient mouse primary spermatocytes arrest at the diplotene stage
During meiotic prophase I, programmed DSB formation and subsequent DSB repair process facilitate homolog chromosome pairing, synapsis and recombination, which is accompanied by the dynamic assembly and de-assembling of the synaptonemal complex [12]. The largely unsynapsed X and Y chromosomes are transcriptionally inactivated to form distinct chromosomal domain called XY body in spermatocytes [24]. Since Igsf11 deficient spermatocytes may be arrested during meiosis prophase I, we co-immunostained SYCP3 with REC8, a meiosis specific component of cohesin complexes, or SYCP1, a structural component of the central element of the SC, to check whether Igsf11 deficient spermatocytes undergo normal pairing and synapsis. Both SYCP3 and REC8 and colocalize to lateral element of the SC throughout meiotic prophase I. SYCP1 and SYCP3 start to colocalize at the homologous chromosomes undergoing synapsis since zygotene, and becoming fully colocalize at pachytene. During diplotene, homologous chromosomes start to dissociate and SYCP1 is gradually degraded from the lateral elements [25]. Results showed that both REC8 and SYCP1 colocalize normally with SYCP3 before pachytene in the mutant spermatocytes (Figs S6A and 5A, indicating that IGSF11 is not required for homologous chromosome pairing and synapsis in mouse spermatocytes. Interestingly, Igsf11 deficient spermatocytes also exhibited an apparently normal early/ mid-diplotene SC conformation (Fig 5B) before their SC became fragmented (Figs S6A, panel h and 5A, panel h). DSB formation in the mutant spermatocytes was also confirmed to be similar to the control heterozygous spermatocytes before diplotene by co-staining for γH2AX and SYCP3 (Fig 5C). The staining results and analysis of over 2,000 mutant cells (Fig 5D and S2  Table) indicate that Igsf11 -/spermatocytes cannot develop beyond diplotene, but the appearance of earlier stage meiotic spreads was similar to the controls. These results suggested that meiotic DSB formation, DSB processing and repair, and sex body formation in Igsf11 -/spermatocytes are all carried out normally. Furthermore, meiotic entry and progression of the mutant spermatocytes were also not delayed before their developmental arrest in diplotene, as indicated by quantitative, age-specific analysis of VPHS reporter expression (S6B Fig).
Besides synapsis and DSB formation [26], germ cells with defects in a variety of different developmental pathways, including retrotransposon silencing [27,28], and meiotic sex chromosome inactivation (MSCI) [29] are also blocked by the pachytene checkpoint, which is very close to the stage that Igsf11 mutant spermatocytes were arrested. We therefore analyzed whether Igsf11 deficient spermatocytes were defective in these aspects. The expression levels of several retrotransposons in purified Igsf11 -/spermatocytes were comparable to that of the controls at both leptotene and pachytene stages (S6C Fig). We performed mRNA-seq using purified leptotene and pachytene stage spermatocytes, and found no differentially expressed gene (DEG) between leptotene Igsf11 -/and Igsf11 +/spermatocytes, especially for those genes that have reported function in meiosis (S6E Fig and S3 Table). 8 downregulated and 58 upregulated DEG are identified in Igsf11 -/pachytene spermatocytes. Gene ontology of 58 up-regulated genes in pachytene Igsf11 -/spermatocytes only revealed an enrichment of genes involved in aminoacyl-tRNA and amino-acid biosynthesis pathways (S6F Fig). Our mRNA-seq data also showed that from leptotene to pachytene, the expression patterns of the sex chromosomelinked genes (Cdc25c, Hprt, Mecp2, Pgk1, Rbmy1a1 and Ube1y1) and compensatory genes on autosomes (Cetn1 and Pgk2), are similar between Igsf11 +/and Igsf11 -/spermatocytes (S6G Fig), indicating a normal MSCI in the absence of Igsf11. Incorporation of histone H1t begins at mid-pachytene and increases apparently after late-pachytene in spermatocytes [30]. The incorporation of H1t started at mid-pachytene in Igsf11 -/spermatocytes as observed in the control group. Furthermore, H1t level in the arrested Igsf11 -/spermatocytes was comparable to that in diplotene stage Igsf11 +/spermatocytes (Figs 5E and S6H), suggesting that Igsf11 -/spermatocytes did develop beyond pachytene checkpoint. Interestingly, when apoptosis was analyzed by TUNEL staining, spermatocytes with various TUNEL signal intensity were observed in Igsf11 -/epididymis, but not in the adluminal compartment of the Igsf11 -/seminiferous tubules where the arrested spermatocytes were first appeared (Fig 5F and 5G), showing that apoptosis was induced in the Igsf11 -/spermatocytes after their release into the epididymis.
In summary, Igsf11 deficient spermatocytes did not show detectable defects in synapsis, DSB formation and repair, sex body formation, retrotransposon silencing, meiotic sex chromosome inactivation, or genome-wide transcription profiles in known meiotic related genes. The meiotic entry and progression of Igsf11 -/spermatocyte was not delayed, nor were spermatocytes arrested by the pachytene checkpoint. However, Igsf11 deficient mouse spermatocytes cannot develop beyond diplotene stage.  Table  Igsf11 deficiency compromised dissociation of nonhomologous pericentric heterochromatin in late-pachytene mouse spermatocytes Chromocenters are clusters of PCH from different chromosomes, and are characterized by the enrichment of H3K9me3 [9]. Compared to controls, unusually bright and large DAPI staining of chromosomes was often detected within the arrested Igsf11 -/diplotene nuclei (Figs 5A-5C and 5G and S6A and S6H), which may be putative chromocenters. Co-staining of SYCP3 and H3K9me3 revealed that PCH clusters into a few compact spherical chromocenters in early leptotene nuclei (S7A Fig) and these PCH gradually dissociate from each other from leptotene to zygotene. Complete assembly of the synaptonemal complex occurs simultaneously with the clustering of homologous and nonhomologous PCH into the same chromocenter during midpachytene. Despite similar assembly and dissociation patterns of PCH from leptotene to midpachytene in control and mutant spermatocytes, PCH patterns became significantly different starting from late-pachytene (Fig 6A-6C). Nonhomologous PCH of the control nuclei dissociated from each other since late-pachytene whereas the homologous PCH remained associated from mid-pachytene until late diplotene. In contrast, nonhomologous PCH of the mutant stayed associated to each other from mid-pachytene until late diplotene, forming 3 to 6 large chromocenters within the arrested mutant nuclei (Fig 6A-6C). Noticeably, decomposition of the SC was observed from mid-to-late diplotene in these mutant spermatocytes, which happened after PCH dissociation defect occurred (Fig 6A). In line with the staining result, proportion of spermatocytes with defect in both the SC and PCH rose and overtook proportion of spermatocytes with defect only in the PCH from 20 dpp to 26 dpp in Igsf11 mutant (S7B Fig). To further confirm that the putative chromocenters in the arrested Igsf11 -/spermatocytes result from clustering of PCH, we co-stained the meiotic spread samples with SYCP3 and TERF1 antibodies. TERF1 binds to telomeric DNA and forms a focus on each end of a chromosome. Since all of the mouse chromosomes are acrocentric [31], one TERF1 focus (from proximal telomere) is close to the PCH for each chromosome. While the majority of the chromocenters were small and colocalized with TERF1 foci of homologous chromosomes in the control diplotene nuclei, the chromocenters in the arrested mutant nuclei were surrounded by multiple TERF1 foci which linked to the partially degenerated SC (Fig 6D, panel c, d and Fig  6E). Further co-staining of SYCP3 and anti-centromere protein antibody (ACA) revealed that all the proximal telomeres clustered to several big chromocenters in Igsf11 mutant spermatocytes except the PCH from sex chromosome (Fig 6D, panel b arrowhead). In conclusion, absence of Igsf11 compromised nonhomologous pericentric heterochromatin dissociation in late-pachytene mouse spermatocytes, and the PCH clustering happened before SC degeneration in the mutant nuclei.

Igsf11 deficiency leads to elevated interchromosomal interactions in mouse spermatocytes
As indicated by the immunostaining results that chromosomal interactions might be altered in the Igsf11 deficient spermatocytes (Fig 6A and 6D), we further examined the arrested mutant spermatocytes by Hi-C. Toward this aim, diplotene stage spermatocytes were enriched from testes of 23 dpp Igsf11 +/+ and Igsf11 -/mice with the help of the VPHS reporter. Although the for additional information about each biological repeat. n = 3 animals/genotype. Values and error bars are mean and SD. � , P < 0.05; �� , P < 0.01 by Student's t test. (E) Statistics of relative H1t intensity in spermatocytes of different Igsf11 genotypes. n, number of spermatocytes. (F) Apoptotic analysis using testis or epididymis sections. (G) Apoptosis analysis using meiotic spread specimens from Igsf11 mutant testis. A very small number of TUNEL-positive cells, which represent a basal level of apoptosis events happened during normal spermatogenesis, were used as internal positive controls. While all the Igsf11 mutant spermatocytes that exhibited fragmented SC complex were TUNEL-negative.  (S7C and S7D Fig). The distribution pattern of compartment A and B on each chromosome (showing chromosome 9 as example) was also similar between the wild type and mutant chromatins (Fig 7A), indicating that overall chromatin distribution is not affected in the absence of IGSF11. Furthermore, the P(s) curve analysis and intrachromosomal chromatin interaction heatmap revealed only minor differences between Igsf11 +/+ and Igsf11 -/spermatocytes ( Fig 7B  and 7C), suggesting that intrachromosomal chromatin interaction is also not affected in the absence of IGSF11.
However, we observed elevated interchromosomal interactions, especially at chromosome ends, in the Igsf11 -/spermatocytes compared to the control (Fig 7D). Although the regions with elevated interchromosomal interactions locate mainly at chromosome ends, these interactions were not limited to interactions between proximal telomere (Fig 7D, blue box), indicating that both ends of chromosomes have higher interchromosomal interactions. Therefore, we concluded that Igsf11 deficiency leads to elevated interchromosomal interactions during meiotic diplotene.

Discussion
By profiling the expression of Igsf11 within seminiferous tubules and conducting testicular transplantation experiments, we showed that presence of the homophilic adhesion molecule IGSF11 is required in both Sertoli cells and spermatogenic cells for meiotic completion by mouse primary spermatocytes. Although the BTB structure is not significantly affected and mouse primary spermatocytes develop beyond the pachytene checkpoint in the absence of IGSF11, the process by which nonhomologous PCH clusters dissociate is arrested from latepachytene onwards. IGSF11 deficiency ultimately leads to male infertility.
Although several CTX family members are expressed in mouse seminiferous tubules [32][33][34][35], IGSF11 is the only member that is essential for meiotic diplotene progression and primary spermatocyte development. Since IGSF11 is expressed in both Sertoli cells and primary spermatocytes, it has several possible nonexclusive routes of action, including mediating interactions between Sertoli cells and between primary spermatocytes independently, or mediating interactions between Sertoli cells and spermatocytes. Although it is more likely that meiotic completion by mouse spermatocytes depends on IGSF11-mediated adhesion between Sertoli cells and spermatogenic cells, our current experimental evidence has not ruled out other possible routes. Future molecular and cellular experiments are needed to decipher the molecular and cellular mechanisms of IGSF11 during spermatogenesis.
Spermatocytes that have defects in meiosis-related developmental processes such as DSB formation and DNA repair are mainly eliminated through apoptosis during the pachytene stage [36,37]. In the absence of IGSF11, mouse primary spermatocytes were instead arrested at diplotene without apparent defects at earlier stages. The novel spermatocyte elimination stage that we observe in Igsf11 mutants suggests the existence of a previously unknown surveillance H3K9me3 as markers. The upper right corner of each image presents a partial magnification of SYCP3 staining results to distinguish the developmental stages of each spermatocyte. (B, C) Statistics of chromocenter number in mid-pachytene (B) and diplotene (C) spermatocytes. Values and error bars are mean and SD. �� , P < 0.01 by Student's t test. n, number of spermatocytes. (D) Organization dynamic of telomeres and centromere within meiotic prophase I spermatocytes of the control and mutant mice. ACA is the abbreviation of anti-centromere protein antibody. Triangular arrowhead indicated PCH cluster of sex chromosomes. Note that SYCP3 signal was more preserved on sex chromosomes when it became fragmented from autosomes. (E) Schematic view of mouse acrocentric chromosome depicting the staining results shown in Fig 6D. https://doi.org/10.1371/journal.pgen.1009778.g006 mechanism and a novel checkpoint that works after the pachytene checkpoint. This putative checkpoint may be responsible for surveying defects during or after pachytene such as failure of nonhomologous PCH clusters to dissociate. Since IGSF11 is a typical membrane protein that may serve as a starting point of the potential signal transduction pathway, downstream factor(s) of IGSF11 in cytoplasm may be needed in spermatocytes to mediate this putative checkpoint.
The organization of PCH in cells varies significantly in different cellular states. Within mouse spermatocytes, PCH undergoes dynamic assembly and dissociation during meiotic prophase I, and more importantly, the dynamics of PCH is highly correlated with that of the SC (Figs 6A and S7A, and [38,39]), suggesting that the dynamics of PCH is also regulated and coordinated by the meiotic program. A few proteins have been reported to be involved in the regulation of PCH in mouse spermatocytes, including SUV39H1/SUV39H2 [40], YY1 [41], H1Pγ [42], and DICER [43], but the functions of none of these proteins is restricted only to the PCH region. It is therefore important to note that the phenotypes observed in these mutants may also be the result of a deficiency of protein functions in other chromosomal or cellular regions. Interestingly, in the absence of IGSF11, elevated clustering in the PCH region occurs prior to SC degeneration at the chromosome arm, indicating that IGSF11 is more likely to restrict its regulation on PCH but not SC and other chromosomal events during meiosis prophase I.
Meiotic prophase I is one of the most complicated developmental stage during germ cell development, yet we still lack reported system that allows us to dissect each stage of prophase in male and female meiocytes. Although several germ cell reporter mice have been generated (Summarized in S4 Table), none of these reporters could precisely indicate meiotic entry and prophase I progression. In this study, we generated a SYCP3 reporter system for monitoring meiotic entry and progression of meiotic prophase I. Combining the VPHS reporter with Hoechst33342 staining analysis, the V&H assay enable precise analysis of the developing meiotic cells, even when the percentage of isolated germ cells is low (S4E Fig). Therefore, we believe the VPHS reporter established in this study will provide a reliable tool to study meiotic prophase in greater details, such as chromosomal dynamics in both male and female meiocytes.

Ethics statement
All animal experiments and procedures are approved by Animal Ethics Committee of Tsinghua University (approval no. 17-JJK1).

Mice and welfare statement
Igsf11 -/mice were crossed to VPHS transgenic reporter mice for V&H assay. See S5 Table for detailed information about genotyping of VPHS and Igsf11 -/mice. Amh-Cre mice [44] were mated to mT/mG mice, a double-fluorescent Cre reporter line [45], for purification of Sertoli cells by FACS (S3D Fig). All mice were maintained in C57BL/6 background and housed under controlled lighting conditions (12L:12D).

Guide RNA design and in vitro transcription
Guide RNAs were designed by an online software [46], and inserted into pX335 plasmid (addgene ID: 42335). T7 promoter was added upstream the expression cassette of Cas9 D10A nickase and guide RNA, respectively, through polymerase chain reaction (PCR) from the pX335 plasmid. The PCR products were purified by DNA Clean & Concentrator (ZYMO, D4033) and used as in vitro transcription template. Capped Cas9 mRNA and uncapped sgRNA were synthesized by an in vitro transcription kit (New England Biolabs, E2040) according to the manufacturer's protocol. After removal of the template by DNase I (New England Biolabs, m0303), RNAs were cleaned up by RNeasy Mini kit (Qiagen, 74104) and dissolved in RNase-free water.

Generation of transgenic mice
To confirm the target efficiency of our system, 20ng/μL in vitro-transcripted Cas9 D10A nickase mRNA and 10ng/μL guide RNAs were microinjected into the pronuclear mouse embryos with Eppendorf TransferMan NK2 microinjection system, and DNA fragment flanking the target site was amplified when embryos developed to blastocysts. Five different indels that caused by Cas9 induced double strands breaks were detected from twenty sequenced fragment clones (S1A Fig).
In order to generate the transgenic mice, pronuclear stage C57BL/6 mouse embryos were injected with targeting mix (2ng/μL donor DNA, 20ng/μL Cas9 mRNA, and 10ng/μL of each sgRNA), and cultured overnight in KSOM medium. The next day, 2-cell stage embryos were surgically transferred into oviduct of pseudopregnant mice.

Living imaging
Testes from different genotypes were examined under a fluorescent dissecting microscope (Leica, M165Fc) to detect mVenus signal. Then, seminiferous tubules were placed under an inverted fluorescent microscope (Nikon, EclipseTi) for imaging.

Quantitative real-time polymerase chain reaction
One whole testis (8-week-old) was first stabilized in RNAlater (Qiagen, 76104), and homogenated in 1.5mL the buffer RLT. Then, 350μL lysate was taken and progressed to RNA purification using RNeasy Mini kit (Qiagen, 74104). 1ug of each total RNA was used for revers transcription by PrimeScript RT reagent Kit with gDNA Eraser (TaKaRa RR047A). Primers specific for each gene were designed by Primer Blast and primers for retrotransposons were described previously [47]. See S6 Table for primer sequences. Quantitative RT-PCR was performed with iTaq Universal SYBR Green supermix (Bio-Rad, 172-5122).

Western blot
Mice testes from different genotypes were lysed in RIPA buffer (CoWin Biosciences, CW2333). Protein lysates were separated by 8%-12% SDS-PAGE gel electrophoresis, electrotransferred onto a nitrocellulose membrane, and processed according to standard procedures. Antibodies were diluted in TBST containing 5% non-fat milk (BD Biosciences, 232100). First antibodies were incubated overnight at 4˚C, and secondary antibodies were incubated at room temperature for 1 hour. There times washing by TBST were performed after each incubation. See S7 Table for antibody information. After adding HRP substrate (Millipore, WBKLS0100), the chemiluminescent signals were detected by an imaging system (Bio-Rad, ChemiDoc XRS + ).

Hematoxylin-eosin staining
Testes and epididymides from 8-week-old mice with different genotypes were fixed in Bouin's solution overnight at 4˚C. Then, samples were imbedded in paraffin, sectioned at 6μm, and subjected to hematoxylin-eosin staining.
Fresh-frozen samples. Freshly collected mouse gonads were imbedded in OCT, frozen in liquid nitrogen, and sectioned at 6μm. After fixation in 2% chilled PB-PFA (pH7.4) at 4˚C for 10min and PBS washing, antigen retrieval was carried out with PBS containing 1% SDS [48].
Meiotic spread samples [49]. After released from tunica albuginea, seminiferous tubules were rinsing in PBS to remove adherent extratubular tissue and then immersed in hypotonic extraction buffer (30mM Tris, 50mM sucrose, 17mM citric acid, 5mM EDTA, 2.5mM DTT, 1mM PMSF, pH 8.2) for 30min. Subsequently, each of one-inch-long seminiferous tubules were torn into pieces with tweezers in 40μL 0.1M sucrose and the result cell suspension was dispersed in 1% PFA containing 0.15% TritonX-100 (pH 9.2) on slides. The slides were kept in a humidified box for one night, air dry in a fume hoods, and washed in 0.04% Photo-Flo (Kodak) for 4min.
Seminiferous tubule Squash samples [50]. Seminiferous tubules were obtained through above stated procedure, and every 20mm long tubules were fixed in 100μL fix-lysis solution (0.1M PB, 0.8% PFA, 0.1% TritonX-100, pH9) on slide for 10min at room temperature. After removal of the excess solution with filter paper, a coverslip was placed on the slide, and pressure was applied on the coverslip by palm to disperse cells from the tubules. Slide was frozen in liquid nitrogen for 15 seconds immediately. Followed by removing coverslips with a lancet, slides were washed in PBS for three times.
TUNEL assay. TUNEL labeling was performed before antibody incubation procedure with a one step TUNEL assay kit (Beyotime, C1089). Specimens were incubated in detection buffer at room temperature for one hour, followed by three times washing in PBS.
Staining procedure. Specimens were first blocked in TBS buffer (with 10% normal goat serum, 3% bovine serum albumin) at room temperature for 60min, and incubated with primary antibodies overnight at 4˚C. Secondary antibodies incubation was performed at room temperature for 2 hours. 6μg/mL Hoechst33342 (Beyotime, C1022) was used to counterstain cell nuclear. All stains were diluted in blocking buffer, and washing was conducted in TBST between each incubation. Finally, sections were mounted by antifade reagent (Life Technologies, 9071S) and coverslips. Immunostainings were visualized and captured by confocal microscopy (Nikon, A1 RSiMP) or super-resolution structure illumination microscopy (Nikon, SIM-S). Images were further analyzed by NIS-Elements software (Nikon). Information about antibodies were listed in S7 Table. Testicular cell transplantation Testicular cell transplantation was performed according to previously published protocol with slight modifications [51,52]. Briefly, 6 week-age mice were intraperitoneally injected with 40mg/kg Busulfan (Sigma, B2635) to deplete the endogenous germ cells. One month after Busulfan treatment, the mice were used as recipient. For donor cells, 3 dpp-10 dpp seminiferous tubules were dissociated into single cells by trypsin-EDTA, filtered by 70μm cell strainer, and resuspended in ice-cold PBS supplementary with 0.04% Trypan blue at concentration of 1×10 8 cells/mL. Cells were then loaded into the injection instrument, a glass capillary with 40μm sharpened tip connected to a 1mL syringe by teflon capillary. Immediately after loading, the tip of the glass capillary was inserted through the bundle of efferent ducts and piercing into the rete testis of anesthetized recipient (S4B Fig). By applying gently pressure to the syringe, about 10μL donor cells were filled into the seminiferous tubules of each recipient testis. 4 months after transplantation, testes were collected from the recipient mice for subsequent analysis. At the time of recovery, all the non-SSCs donor cells have been excreted out from the recipient testis as indicated by trypan blue.

Flow cytometry
Fetal gonads were digested with TrypLE Express (Gibco, 12605010) to obtain single cell suspension. After removing the tunica albuginea, postnatal testes were first digested in collagenase IV (Gibco, 17104019) to remove the interstitial cells, followed by 3 times washing with PBS, seminiferous tubules were further digested with 0.25% trypsin-EDTA (Gibco, 25200056), collagenase IV and DNase I (Solarbio, D8071). After neutralization with fetal bovine serum, cells were washed with PBS and filtered with 70μm strainer. Single cells were finally resuspended in HBSS buffer containing 2% fetal bovine serum and 10mM HEPES. For Hoechst33342 staining assay, seminiferous tubules derived single cell suspensions were stained with 5μg/mL Hoechst33342 for 1 hour at 32˚C as previously described [53,54]. See S4C and S4D Fig for detailed control and gating strategy. Cells were analysis or sorted by Influx (BD Biosciences), and data were analyzed and displayed by FlowJo. Every FACS result was repeated for at least three times with three biological duplicates.

RNA-seq
Leptotene (9dpp) and pachytene (17dpp) spermatocytes were sorted from VPHS; Igsf11 -/or VPHS; Igsf11 +/mouse testis, respectively, using the VPHS reporter system. Gatings for FACS were shown in Fig 1E ( [55]. Briefly, 80ng of total RNA from each sample was used as input for reverse transcription with SuperScript II reverse transcriptase (Life Technology, 18064), followed by 8 cycle pre-amplification with a hot start PCR kit (KAPA Biosystems, KR0370). After purification with magnetic beads (Beckman Coulter, A63881), 5ng of each amplified cDNA was fragmented with a library preparation kit (Vazyme, TD502) and index (Vazyme, TD202) was added through 8 cycle enrichment PCR according to the manufacturer's protocol. Fragments around 350 bp were selected and all libraries were sequenced by Illumina HiSeqX10.
For RNA-seq data analysis, adaptors were removed with Cutadapt [56], gene expression levels were quantified with Kallisto [57], differentially expressed genes (criteria: P value<0.05 and fold change>2) were identified by DESeq2 [58], and gene ontology was performed with DAVID [59]. Sequencing data is available in the GEO database (GSE174752). The expression pattern of selected DEG genes were further verified by Q-PCR.

sisHi-C library generation and data processing
Diplotene stage spermatocytes were sorted by using the VPHS reporter system. Purity of the sorted cells were analysis by meiotic spread (S2C and S2D Fig). And sisHi-C libraries were produced as described [60]. Briefly, samples were cross-linked with 1% formaldehyde at room temperature (RT) for 10 min. Formaldehyde was quenched with glycine at RT for 10 min. The cells were lysed on ice for 50min and the chromatin was solubilized with 0.5% SDS. The nuclei were digested with MboI at 37˚C overnight. After fill-in with biotin-14-dCTP, the fragments were ligated at RT for 5.5h. These was followed by reversal of cross-linking, DNA purification, and sonication. Then the biotin-labeled DNA was pulled down with 10μl Dynabeads MyOne Streptavidin C1 (Life Technology). The fragments that included a ligation junction were subjected to Illumina library preparation, which including end repair, dATP tailing and adaptor ligation. 12-15 cycles of PCR amplification were performed with TransStart Fastpfu (Transgene), and the products were purified and size-selected with AMPure XP beads. Fragments ranging from 200 bp to 500 bp were selected. All libraries were sequenced on an Illumina HiSeq 1500 according to the manufacturer's instructions.
Paired end Hi-C sequencing reads were processed using HiC-Pro [61]. Briefly, paired reads were iteratively mapped to the mm9 reference genome via bowtie2 (v 2.3.5.1). The unmapped reads, multiple mapped reads and uninformative reads pairs from self-circle ligation, dangling ends, undigested DNA fragments, PCR artifacts were discarded, only the valid read pairs from two different restriction fragments were retained and we merged biological replicates valid pairs for following analysis. The Hi-C matrices were further converted to Hi-C contact matrices binned into 1Mb, 300kb, 100kb sizes, respectively, and normalized by an iterative correction method [61]. Then, Hi-C matrices were transferred into the hic format to facilitate visualization with juicebox [62]. The pairwise differential heatmap were calculated using HiC-Plotter under 300kb resolution of the log2 comparison matrix [63]. For P(s) curves calculation, the contact probability versus genomic distance (P(s)) curves were calculated with normalized interaction matrices in 100kb resolution as previously described in [60,64]. For A/B Compartment analysis, the compartmentalization strength was calculated at 100kb resolution using HiCExplorer [65]. The A/B compartment profile were identified using the eigenvector of the first principal component [66]. The eigenvector of the first principal component defines the A/ B compartment profile ('A': active/euchromatic compartments and 'B': inactive/heterochromatic compartments) primarily based on mRNA transcription. For Hi-C biological replicate correlation analysis, the Peason correlation between sisHi-C biological replicates was calculated for each pair of 100kb bins within a maximum distance of 50 bins (5Mb) as previously described [67]. Sequencing data is available in the GEO database (GSE174752).