Mouse EWSR1 is crucial for spermatid post-meiotic transcription and spermiogenesis

Reproductive failure affects 10-15% of couples worldwide, up to 55% of which is due to spermatogenic failure (Visser and Repping, 2010). Spermatogenesis is a tightly regulated process during which germ cells originating from spermatogonial stem cells progress through several developmental stages and ultimately produce highly differentiated spermatozoa. During this development, spermatogonial stem cells enter the proliferation cycle as spermatogonia, which in turn differentiate into spermatocytes that undergo meiosis and produce haploid round spermatids. The round spermatids in mammals then go through complex biochemical and morphological changes, and further develop into elongated spermatids and eventually into mature sperm cells. This dramatic post-meiotic differentiation process, termed spermiogenesis, includes acrosome and flagellum formation, nuclear condensation and cytoplasmic exclusion (Hess and Renato de Franca, 2008). A crucial feature of spermiogenesis is the repression of transcription before spermatid elongation. Thus, mRNAs required for post-meiotic spermatid differentiation are synthesized in spermatocytes and round spermatids, but then become translationally repressed and stored. As spermiogenesis progresses, those mRNAs are gradually released and spermatid differentiation-relevant proteins are synthesized (Steger, 1999). Regulation of the transcriptional repression and translational inactivation during spermatid development involves multiple factors and pathways, including non-coding RNAs (de Mateo and Sassone-Corsi, 2014; Meikar et al., 2011), replacement of histones by protamines (Barral et al., 2017; Rathke et al., 2014), RNA alternative splicing (Margolin et al., 2014; Schmid et al., 2013), transcription-factor and RNA-binding protein activation (Idler and Yan, 2012; Paronetto and Sette, 2010), and functions of RNA processing granules (de Mateo and Sassone-Corsi, 2014; Meikar et al., 2011). However, the mechanisms of action and the functions of these factors and pathways are not fully characterized.

Ewing sarcoma RNA-binding protein 1 (EWSR1) is a member of the TET family (TLS/FUS, EWS and TAF-II68) RNA- and DNA-binding proteins (Tan and Manley, 2009). In many somatic cell types, these proteins play important functions in transcription, RNA processing and tumorigenesis. The C-terminal regions of TET proteins contain several RNA-binding motifs, including three arginine-glycine-glycine (RGG) boxes, one RNA recognition motif (RRM) and one zinc-finger domain (Schwartz et al., 2015). The N-terminal domains of these proteins are rich in glutamine, serine and tyrosine; are capable of activating transcription; and are frequently involved in tumor-derived fusions with other genes (Delattre et al., 1992; Schwartz et al., 2015). In addition, EWSR1 interacts with multiple RNA-binding proteins and plays a role in various nuclear processes. For example, it is involved in transcriptional regulation via binding to the preinitiation complex TFIID and with subunits of RNA polymerase II (RNA POLII) (Bertolotti et al., 1998; Petermann et al., 1998; Yang et al., 2000); it influences microRNA (miRNA) activity by regulating expression of the miRNA processing protein Drosha (Kim et al., 2014); it interacts with several RNA alternative splicing factors, such as TASR1/2 (Yang et al., 2000), U1C (Knoop and Baker, 2000) and YB1 (Chansky et al., 2001); and it regulates DNA damage-induced RNA splicing (Paronetto et al., 2011).

We have reported that, during spermatogenesis, EWSR1 is one of the key regulators of meiotic recombination, by interacting with recombination hotspot determinator PRDM9 and meiotic-specific cohesin REC8 in leptotene and zygotene spermatocytes (Parvanov et al., 2017; Tian et al., 2021). Furthermore, although neither we nor others detected a function of EWSR1 in regulating gene expression or RNA binding during these prophase I stages, some of our findings indicated that EWSR1 may be involved in post-meiotic differentiation. Specifically, we showed that EWSR1 expression increased dramatically in pachytene and diplotene spermatocytes, and round spermatids in mouse testes (Parvanov et al., 2017). This prompted us to elucidate the function of EWSR1 in spermiogenesis.

To achieve this goal, we created an Ewsr1 pachytene stage-specific knockout mouse model, allowing us to separate its functions during late meiotic prophase I and spermatozoa maturation from those in early meiosis that we have already described (Tian et al., 2021). We now show that loss of EWSR1 at this stage does not affect proper meiosis completion, but that developing germ cells exhibit defects in spermatid elongation and chromocenter formation. In Ewsr1 conditional knockout testes, expression of many genes regulating spermatid differentiation is impaired, suggesting that EWSR1 plays an important role in regulation of spermiogenesis-related mRNA synthesis, thereby enabling spermatid differentiation into mature sperm.

In a previous study, we showed that, in juvenile mouse testes, EWSR1 is expressed highly in both spermatogonia and Sertoli cells, and at lower levels in early spermatocytes (Parvanov et al., 2017). In the present study, we show that, in adult mouse testes, EWSR1 expression is high in spermatogonia and then decreases at the pre-leptotene stage, when STRA8, a spermatocyte development-regulating protein, is expressed (Fig. 1A, far right panel, stage VII-VIII). The EWSR1 signal remains relatively low in leptotene/zygotene spermatocytes, and then dramatically increases at pachynema and remains high in diplotene spermatocytes and round spermatids, eventually disappearing from elongated spermatids (stage X) (Parvanov et al., 2017; Tian et al., 2021). The EWSR1 signal is absent in heterochromatin regions, which are labeled by H3K9me3 (histone-3 trimethylated at lysine 9), such as the sex body at pachynema, as previously shown (Parvanov et al., 2017; Tian et al., 2021), or the chromocenter in round spermatids, as we show here (Fig. 1B). From this overall expression pattern, we speculated that EWSR1 may play a role in post-meiotic differentiation.

To determine the function of EWSR1 in late meiotic and post-meiotic stages, we used a conditional knockout approach to delete Ewsr1 from spermatocytes using Hspa2 promoter-driven CRE (Ewsr1^{loxp/Δ;Hspa2-Cre}; hereafter referred to as Ewsr1 CKO). Hspa2-Cre mice express CRE from zygonema onward (Inselman et al., 2010). We did not detect any adverse health effects due to potential Hspa2-Cre expression in other tissues. In the CKO testes, EWSR1 protein levels were lower than in testes of heterozygous controls (Ewsr1^{loxp/+;Hspa2-Cre}; hereafter referred to as Ewsr1 het) (Fig. 2A). Immunostaining of CKO seminiferous tubules showed that the EWSR1 signal in zygotene spermatocytes in stage XI seminiferous tubules was comparable in strength with that in the heterozygous control (Fig. 2B, left-hand panels, white arrows), but completely disappeared at pachynema and in round spermatids (Fig. 2B, right-hand panels, arrowheads). The EWSR1 signal in Sertoli cells was not affected (Fig. 2B, right-hand panels, arrows). These results indicate that EWSR1 expression is indeed abolished from the pachytene stage onward in Ewsr1 CKO testes.

Loss of Ewsr1 from pachytene spermatocytes led to a slight but significant reduction of testis size compared with heterozygous control testes in adult mice (Fig. 2C, P<0.05). Ewsr1 CKO mice were sterile. No offspring were produced by Ewsr1 CKO male mice when mated to C57BL/6J females, while control matings of Ewsr1 het male mice produced normal numbers of viable pups (7.0±3.0 pups per litter, Fig. 2D). Histological examination showed that Ewsr1-deficient seminiferous tubules lost spermatids, whereas the morphology of Sertoli cells, spermatogonia and spermatocytes was preserved (Fig. 2E). In addition, stage VII-VIII spermatid arrest in Ewsr1-deficient seminiferous tubules was confirmed by co-staining of EWSR1 and STRA8 on CKO testicular cross-sections (Fig. S1A). Furthermore, multinucleated cells with pyknotic nuclei were observed in CKO seminiferous tubules (Fig. 2E, arrows), suggesting both germ-cell apoptosis (Nantel et al., 1996) and seminiferous-tubule atrophy in Ewsr1-deficient testes.

During spermiogenesis, spermatid elongation begins in step 8 spermatids, in which microtubules are assembled to form a semi-circular (skirt-like) band structure, termed a manchette, around the caudal pole of elongated spermatid nuclei (Fig. S1B, left panel). β-Tubulin, one of the proteins participating in manchette assembly (Fujii et al., 2017; Lehti and Sironen, 2016), can be used as a marker of spermatid elongation. In CKO tubules, a skirt-like manchette structure was not observed (Fig. S1B, right panel), further indicating that no elongated spermatids were present in CKO testes. In additional, no spermatozoa were observed in the Ewsr1 CKO epididymis, whereas they were plentiful in the het control (Fig. 2E, lower panels). Together, these results suggest that the absence of spermatozoa in in testes lacking EWRS1 expression from the pachytene stage onward is likely the result of a block of spermatid differentiation.

During meiosis, EWSR1 plays a role in meiotic recombination initiation and determination by interacting with PRDM9 and pREC8, a chromosome axis protein, and stabilizing the PRDM9-hotspot DNA/chromosomal axis complex at the leptotene and zygotene stages (Tian et al., 2021). Knocking out Ewsr1 in mice, either systemically or conditionally at the onset of meiosis, leads to meiotic prophase I arrest with increased chromosomal asynapsis (Li et al., 2007; Tian et al., 2021). To investigate whether loss of EWSR1 from pachytene cells also affects meiotic progression, we first examined DNA double-strand break (DSB) formation and repair using phosphorylated H2AX (γH2AX, a marker for unrepaired DNA lesions) staining on Ewsr1 CKO and het spermatocyte chromosome spreads. CKO spermatocytes showed no increase in the level of unrepaired DSBs or asynapsis (98.9% fully synapsed pachynema) relative to het controls (97.8% fully synapsed pachynema) (Fig. 3A and Fig. S2), while about 30% of systemic Ewsr1 knockout spermatocytes showed chromosome asynapsis (Li et al., 2007). Next, we used ATR and HORMAD1 staining as markers for unsynapsed chromosomes, to examine the effect of EWSR1 loss on sex body formation and chromosomal synapsis. In both control and CKO pachytene spermatocytes, round-spread ATR signal covers only the sex chromosomes (Fig. 3B), and HORMAD1 is retained only on the sex chromosomes (Fig. 3C). Together, these data indicate that chromosomal synapsis and DSB formation are not impaired in Ewsr1 CKO testes.

Systemic knockout of Ewsr1 leads to remarkably reduced numbers of crossovers and increased numbers of bivalents without crossovers (Li et al., 2007; Tian et al., 2021). However, loss of EWSR1 from the pachytene stage onward showed no impact on crossover number (24.0±2.0 MLH1 foci per pachytene cell, n=70, from two individual mice; Fig. 3D,E) relative to het controls (23.7±2.0 MLH1 foci per pachytene cell, n=63, from two individual mice). Chiasma formation in the diplotene stage in Ewsr1 CKO testes was also normal (Fig. S2). Stra8-iCre-mediated Ewsr1 CKO spermatocytes showed both insufficient numbers of crossovers and shorter synaptonemal complex (SC) length at the pachytene stage (Tian et al., 2021). However, in Hspa2-Cre-mediated Ewsr1 CKO spermatocytes, SC length (184.1±20.5 µm, n=64, from two individuals; Fig. 3F) was similar to that in heterozygous controls (179.3±26.6 µm for total SC length, n=65, from two individuals), consistent with the results for crossover numbers in Hspa2-Cre-mediated Ewsr1 CKO testes. Together, these results suggest that loss of EWSR1 from the pachytene stage onward does not affect crossover determination.

Overall, these results suggest that loss of Ewsr1 from the pachytene stage onward does not cause any meiotic defects. It demonstrates that during spermatid differentiation and spermiogenesis, EWSR1 has a post-meiotic function that is independent of its function in meiotic recombination.

Nuclear topology is reorganized during spermiogenesis (Brinkley et al., 1986), when the centromeric heterochromatin domains form one or two chromocenters in round and elongated spermatids (Berkovits and Wolgemuth, 2013; Hoyer-Fender et al., 2000; Namekawa et al., 2006). To investigate whether EWSR1 functions in chromocenter formation, we carried out immunostaining with antibodies against H3K9me3 to label chromocenters and against CREST to label centromeres. In Ewsr1 het round spermatids, chromocenter formation was normal; centromeres coalesced and colocalized with the H3K9me3 signal to form a single focus (Fig. 4A, top panels), and only ∼8.4% of het round spermatids showed more than two chromocenters per cell (n=1097 spermatids, Fig. 4B). However, we observed chromocenter fragmentation in Ewsr1 CKO round spermatids, most of which contained three or more H3K9me3/CREST foci (Fig. 4A, lower panels). The proportion of round spermatids with fragmented chromocenters was significantly increased (∼84.5%) in Ewsr1 CKO spermatids. (n=1031 spermatids; Fig. 4B, P<0.001) To further confirm these results, we performed the same staining on Ewsr1 CKO and het haploid round spermatids isolated by flow cytometry (Fig. S3), and observed the same chromocenter fragmentation phenotype in Ewsr1 CKO spermatids (Fig. S4). These data indicate that absence of EWSR1 from the pachytene stage onward leads to abnormal chromatin condensation during spermatid differentiation.

To investigate whether spermiogenesis defects in the Ewsr1 CKO mice are a consequence of altered gene expression, we performed RNA-seq in flow cytometry-sorted diploid cells, which contain primarily pachytene and diplotene spermatocytes (4c DNA content, hereafter referred to as P/D spermatocytes), and haploid cells, which represent round spermatids (1c DNA content, hereafter referred to as RS) (Figs S3 and S5A).

In P/D cells, 558 genes were differentially expressed between Ewsr1 CKO and het control mice (P<0.05, Wilcoxon rank sum test). Of these, 408 were upregulated and 150 were downregulated in CKO mice (Fig. 5A, green dots, upregulated; yellow dots, downregulated). In RS, 937 genes were differentially expressed between Ewsr1 CKO and het RS. Of these, 794 were upregulated and 143 were downregulated in CKO RS (Fig. 5A). Gene ontology analysis indicated enrichment of meiotic cell cycle-related genes (GO:0051321) in the group of genes upregulated in the CKO RS (FDR=1.14×10⁻²).

The upregulation of meiosis-related genes in CKO RS suggested that the meiotic cell cycle was altered in these cells. To test whether this could be the case, we first compared the P/D het and RS het RNA-seq datasets, and identified 2577 genes that showed higher expression in P/D cells than in RS(p_adj <0.05, Wilcoxon rank sum test; Fig. S5B left, columns 1 and 3). We termed these genes P/D-enriched genes. Conversely, 1476 genes showed significantly higher expression in RS compared with P/D cells (p_adj <0.05, Wilcoxon rank sum test; Fig. S5B right, columns 1 and 3). We termed these genes RS-enriched genes.

Next, we compared the expression of P/D-enriched genes between Ewsr1 CKO P/D cells and het P/D cells, and between CKO RS and het RS; and expression of RS-enriched genes between Ewsr1 CKO P/D cells and het P/D cells, and between Ewsr1 CKO RS and het RS. We found that expression of P/D-enriched genes was not significantly affected in CKO P/D cells (p_adj=0.042, Wilcoxon rank sum test; Fig. 5B left, red dots; Fig. 5C left, columns 1 and 2). However, these genes retained higher expression levels in CKO RS than in het controls (p_adj=2.2×10⁻¹², Wilcoxon rank sum test; Fig. 5B right, red dots; Fig. 5C left, columns 3 and 4). We tested these overall results for expression of P/D-enriched genes, using qRT-PCR, and identified four meiosis-specific genes that showed no differential expression between P/D CKO cells and P/D het cells, and that also showed higher expression levels in Ewsr1 CKO RS than in het RS (Fig. 5D). Of these four genes, Cntd1 and Mlh1 are crucial for normal meiotic crossover formation (Holloway et al., 2014; Santucci-Darmanin et al., 2000); Mov10l1, a gene involved in piRNA biogenesis, is predominantly expressed in pachytene spermatocytes, and its mRNA normally disappears in post-meiotic cells (Frost et al., 2010; Vourekas et al., 2015); and Sycp3 codes for a component of the synaptonemal complex, and is expressed only in spermatocytes during spermatogenesis.

In contrast to these results for PD-enriched genes, most of the RS-enriched genes were downregulated in Ewsr1 CKO RS compared with het RS (p_adj=0.006, Wilcoxon rank sum test; Fig. 5B right, blue dots; Fig. 5C right, columns 3 and 4). The expression of these genes in CKO P/D cells showed no major differences in comparison with expression in het P/D controls (p_adj=0.16, Wilcoxon rank sum test; Fig. 5B left, blue dots; Fig. 5C right, columns 1 and 2). Genes downregulated in CKO RS compared with het RS controls included genes required for normal spermatid elongation, DNA packaging and chromatin condensation, including genes coding for protamines (Prm1 and Prm2) (Cho et al., 2001), and transition nuclear proteins (Tnp1 and Tnp2) (Shirley et al., 2004) (Fig. 5E). This misregulation could explain the cell-cycle arrest and DNA-packaging and -condensation defects that we observed in CKO spermatids.

Because we observed disruption of chromocenter formation, we also analyzed the expression of genes known to have functions related to RS chromocenter formation: H1fnt (H1f7), Brdt, Hmgb2 and Tbpl1. H1fnt codes for a histone H1-like, haploid germ cell-specific nuclear protein that is required for DNA condensation and chromocenter formation in spermatids (Martianov et al., 2005; Tanaka et al., 2005); Brdt codes for a testis-specific double bromodomain protein that is required for chromocenter organization, depending on the genetic background (Berkovits and Wolgemuth, 2011); Hmgb2 codes for a protein that is crucial for chromocenter and acrosome formation in spermatids (Catena et al., 2006; Ronfani et al., 2001); and Tbpl1 is required for chromocenter formation and maintenance (Catena et al., 2006).

We found that H1fnt expression is significantly reduced in CKO RS compared with het RS controls, but is not changed in P/D cells in comparison with het P/D controls (Fig. 5F). This reduction in P/D-cell gene expression was confirmed by qRT-PCR with 21 dpp CKO and het control mouse testis extracts (Fig. S5C). The expression levels of the other known RS chromocenter formation-related genes we examined were not affected (Fig. 5F and Fig. S5C).

Taken together, these data suggest that the transition of a meiotic to a post-meiotic gene-expression program is impaired in Ewsr1 CKO testes, and that this, in turn, leads to RS arrest and a block in spermiogenesis. In addition, the expression of some chromocenter formation-related genes is affected, which could explain the DNA condensation defects, which in turn prevent spermatid elongation.

Many of the genes with functions related to spermatid elongation and DNA condensation, including Tnp2 and Prm2, are transcribed in spermatocytes and RS and stored in an RNA-protein complex termed the chromatoid body (Meikar et al., 2014; Peruquetti, 2015; Yokota, 2008). In chromatoid body-disrupted spermatids, in which the proper association between RNA-binding proteins and RNA cannot be retained, the mRNA content is altered by RNA degradation (Fanourgakis et al., 2016; Nagamori et al., 2011). To investigate whether the dysregulated mRNA levels we observed in Ewsr1 CKO RS is due to disruption of the chromatoid body, we tested chromatin-body integrity by staining for MVH, an RNA helicase enriched in this structure. In both Ewsr1 CKO and het controls, we observed a conspicuous perinuclear spot of MVH staining in RS, which suggested that the integrity of chromatoid bodies in the CKO was not affected (Fig. 6A). In addition, there was no significant difference between CKO and het control RS in expression of chromatoid body-related genes, including Piwil1, Piwil2, Tdrd5, Tdrd6, Nsun and Mvh (Fig. 6B), indicating that the major components of the chromatoid body are not affected by the absence of EWSR1. Our results strongly suggest that loss of EWSR1 in late pachytene spermatocytes leads to transcriptional changes but does not affect mRNA storage and processing.

In our previous work, using CKO mice lacking Ewsr1 from the onset of meiosis, we showed that EWSR1 plays an essential role in meiotic recombination initiation (Tian et al., 2021). In the present study, using CKO mice lacking Ewsr1 from the pachytene stage onward, we demonstrate that Ewsr1 also acts as a regulator of spermiogenesis. Specifically, we show that EWSR1 is excluded from heterochromatin regions in pachytene/diplotene spermatocytes and round spermatids. Unlike the phenotype in CKO mice lacking Ewsr1 from the onset of meiosis, deletion of Ewsr1 from pachytene stage on does not affect any meiotic events, including chromosomal synapsis, DSB formation and crossover number. However, these CKO mice are sterile and lack elongated spermatids and mature sperm.

To identify the specific molecular mechanisms underlying the spermiogenesis defects caused by a lack of EWSR1, we performed RNA-seq with CKO pachytene spermatocytes and round spermatids, and, interestingly, determined that gene expression in CKO pachynema is not substantially altered, but that meiotic cell cycle-related genes are misregulated in round spermatids. Specifically, we showed that although high mRNA levels of the meiosis-specific proteins CNTD1, MLH1, L MOV10L1 and SYCP3 are retained in Ewsr1 CKO RS, genes required for normal spermatid elongation and chromatin condensation, such Prm1 and Prm2, genes that encode protamines (Cho et al., 2001), and Tnp1 and Tnp2, which encode transition nuclear proteins (Shirley et al., 2004), are downregulated in CKO round spermatids. This misregulation most likely leads to DNA-packaging and chromatin-condensation defects in CKO spermatids.

In addition to meiotic cell cycle gene expression dysregulation, Ewsr1 CKO round spermatids show severe chromocenter fragmentation. H1FNT (also known as HIST1T2 and H1T2) is a histone H1 variant specifically expressed in round and elongated spermatids. H1FNT is not required for acrosome formation per se (Martianov et al., 2005; Tanaka et al., 2005), but it associates with chromatin and is necessary for the polarity of nuclear chromatin organization and DNA condensation (Catena et al., 2006; Martianov et al., 2005; Tanaka et al., 2005). Therefore, reduction of H1fnt expression may affect the chromatin packaging required for spermatid elongation. The bipolar localization of H1FNT is regulated by TBPL1 and HMGB2 (Catena et al., 2006). Tbpl1 and Hmgb2 knockout mice both show chromocenter fragmentation defects, as well as disorganization of heterochromatin and dyslocalization of H1FNT (Catena et al., 2006; Martianov et al., 2002; Ronfani et al., 2001). BRDT is also known to be associated with spermiogenesis and chromocenter formation. The deletion of the first bromodomain of Brdt affects acetylated histone removal in elongating spermatids (Gaucher et al., 2012). This mutant also causes chromocenter fragmentation, depending on the genetic background (Berkovits and Wolgemuth, 2011), although this effect is dependent on the genetic background. However, expression of Tbpl1, Hmgb2 and Brdt remains unaffected in pachytene spermatocytes or RS in the absence of EWSR1.

There are three possible explanations of how EWSR1 could be involved in regulating gene expression during spermiogenesis. First, EWSR1 may be involved in mRNA processing. Unlike in any other tissues, genes that are required for the final steps of spermatogenesis are transcribed in late meiotic cells and round spermatids, and their mRNAs are repressed and stored in a protein-RNA granule termed the chromatoid body (Braun, 1998). However, EWSR1 involvement in chromatoid body function is unlikely as the chromatoid-body integrity is not affected in Ewsr1 CKO spermatids.

Second, EWSR1 may transcriptionally control the mRNA expression of the RS-upregulated genes. It is unlikely that EWSR1 regulates the altered gene expression directly, because our attempts to detect EWSR1 binding to DNA by EWSR1 ChIP-seq and affinity pull-down assays did not indicate that EWSR1 binds to DNA in germ cells on its own (data not shown). However, EWSR1 is reported to bind to and modulate the functions of other RNA-binding proteins and transcriptional factors (Bertolotti et al., 1998; Petermann et al., 1998; Yang et al., 2000). In this regard, it remains possible that EWSR1 mediates gene transcription as a component of complexes involving those RNA-binding proteins and transcriptional factors.

Third, EWSR1 has been shown to be broadly involved in primary miRNA processing (He and Ding, 2020; Kim et al., 2015; Ouyang et al., 2017), raising the possibility that, in the absence of EWSR1, many downstream genes may be affected by aberrant post-transcriptional control resulting from dysregulation of miRNAs. With our current data, we cannot confirm or rule out either of the second or third possibility.

In summary, our results show that EWSR1 has two completely different functions in germ cell development – one in recombination regulation (Tian et al., 2021) and another in spermatid development (this article). These functions are related to the distinct expression pattern of EWSR1 – low expression in early meiosis, and dramatic increase at and after the pachytene stage. The latter function of EWSR1 is more reminiscent of those of the TET family proteins in somatic cells (Tan and Manley, 2009). Together, our data emphasize the nature of EWSR1 as a multifunctional protein. Further research characterizing the functions of its individual domains in different cell types would be necessary to elucidate the mechanisms of its multifunctionality.

The animal care rules used by The Jackson Laboratory are compatible with the regulations and standards of the US Department of Agriculture and the National Institutes of Health. The protocols used in this study were approved by the Animal Care and Use Committee of The Jackson Laboratory (Summary #04008). Euthanasia for this study was carried out by cervical dislocation.

Information on the sources of the strains used to generate the Ewsr1 CKO mice used in this study is provided in the section ‘Mouse strains used in this project’. The Ewsr1 CKO mice used in this study were produced by a two-step deletion scheme. C57BL/6N-Ewsr1^{<tm1c(EUCOMM)Wtsi>}/Tcp mice were mated to Tg(Sox2-cre)1Amc/J mice to generate one Ewsr1 allele deleted mice, and then the Ewsr1 hemizygous mice (Ewsr1^Δ/+) were mated to C57BL/6-Tg(Hspa2-cre)1Eddy/J to obtain Ewsr1^{Δ/+;Hspa-Cre} mice. The Ewsr1^{Δ/+;Hspa-Cre} mice were mated to homozygous Ewsr1 loxp mice to generate heterozygous control mice (Ewsr1^{loxp/+;Hspa-Cre}, Ewsr1 het) or conditional knockout mice (Ewsr1^{loxp/Δ;Hspa-Cre}, Ewsr1 CKO).

The mouse line C57BL/6N-Ewsr1^{<tm1c(EUCOMM)Wtsi>}/Tcp was made as part of the NorCOMM2 project at the Toronto Centre for Phenogenomics (Bradley et al., 2012) and obtained from the Canadian Mouse Mutant Repository. C57BL/6J (stock number 000664), Tg(Sox2-cre)1Amc/J (stock number 004783) and C57BL/6-Tg(Hspa2-cre)1Eddy/J (stock number 008870) mice were obtained from The Jackson Laboratory. All animal experiments were approved by the Animal Care and Use Committee of The Jackson Laboratory (Summary #04008).

For histological evaluation, testis or epididymis were dissected out, fixed with Bouin's solution and embedded in paraffin wax, and 5 μm sections were prepared. Sections were stained with Hematoxylin and Eosin or periodic acid-Schiff-diastase (PAS) using standard techniques.

For preparation of nuclear spreads from germ cells, the drying-down technique (Peters et al., 1997) was used, followed by immunolabeling with anti-SYCP3 (NB300-231, Novus Biologicals, 1:500), anti-γH2AX (ab26350, Abcam, 1:1000), anti-ATR (sc-1887, Santa Cruz, 1:500), anti-HORMAD1 (13917-1-AP, Protein Tech, 1:500), anti-MLH1 (550838, BD Pharmingen, 1:100) or anti-centromere (CREST) (15-234-0001, Antibodies Incorporated, 1:50) antibodies.

FACS was essentially carried out according to Lam et al. (2019). Adult testes from Ewsr1 het and CKO mice were placed in 10 ml EKRB [1×KRB with 100 nM CaCl₂, 200 mmM L-Glutamine-Penicillin-Streptomycin solution, 1× MEM essential amino acids, 1× MEM non-essential amino acids, 0.05 mg/ml lactate sodium, 0.72 mg/ml sodium pyruvate and 20 mM HEPES buffer (pH 7.25)] with 0.5 mg/ml collagenase IV and 0.5 µg/ml DnaseI, and incubated in a shaker at 32°C for 20 min at 120 rpm. After incubation, seminiferous tubules were pipetted 20 times with a 10 ml pipette for, and spun down at 800 rpm (300 g) for 10 min. The supernatant was then removed and tubules were washed with 10 ml EKRB with 0.5 mg/ml collagenase IV and 0.5 µg/ml DnaseI, and incubated in a shaker at 32°C for 20 min at 120 rpm. After pipetting again 20 times, tubules were filtered with a 40 μm mesh, spun down at 1000 rpm (400 g) for 10 min, and then washed three times with 20 ml EKRB . The cells were resuspended in 1 ml EKRB containing 1% FBS, filtered with a 40 μm mesh and diluted to 6000 cells/μl with 1% FBS/EKRB. The cells were then stained with 5 µl/ml Hoechst 33342 (ThermoFisher, H1399) and 0.5 µl/ml DnaseI in the dark at 32°C. Before loading the cell for sorting, 2 µl/ml propidium iodide was added to samples. Then, 0.5 million 4c pachynema and 1 million 2c round spermatids were sorted into EKRB with 1% FBS, after which the RNA was isolated by a standard method and RNA-seq libraries are prepared with a KAPA RNA Hyper Prep Kit (KAPABIOSYSTEM, KK8540).

Libraries were sequenced on an Illumina HiSeq 2500, with 75 bp paired-end reads. Fastq files for paired-end sequenced DMC1 samples were trimmed using Trimmomatic (v0.32) (Bolger et al., 2014). The trimmed fastq files were aligned to mouse genome mm10 using Bowtie (2.2.0) (Langmead and Salzberg, 2012). The software package RSEM (1.2.12) (Li and Dewey, 2011) was then used for gene annotation, calculation of fragments per kilobase million (FPKM) and normalization. For each cell group (Ewrs1 het P/D and RS samples, Ewsr1 CKO P/D and RS samples), the correlations among the replicates are high (r2=0.99 for Ewsr1 het P/D and RS samples, and Ewsr1 CKO RS samples; r²=0.96 for Ewsr1 CKO P/D samples). Differentially expressed genes were analyzed using Gene Ontology Enrichment Analysis (http://www.geneontology.org/page/go-enrichment-analysis).

Total RNA from 21 dpp Ewsr1 het and CKO mouse testes was isolated with a mirVana miRNA isolation kit (AM1560, Applied Biosystems). qPCR was performed with Quantifast SYBR Green PCR kit (204054, Qiagen). β-Actin mRNA levels were used for normalization. Primer sequences for qPCR were: β-actin forward, TGGCTCCTAGCACCATGAA; β-actin reverse, CTCAGTAACAGTCCGCCTAGAAGCA; Brdt forward, GGTCCAAATTACAAAGGGTGTGAAG; Brdt reverse, GTCGGAGGAGATTCGCTACTTG; Tbpl1 forward, TCTGGCAAGCTGGTCTCTGT; Tbpl1 reverse, CCTCCAGTCACCCTTCACAT; Hmgb2 forward, GGCCAGCCTGTCTACAGAG; Hmgb2 reverse, GTTGGGTAGGCTGCTATCAAGA; H1fnt forward, CCAGTAAAACAGACCATCCAGA; and H1fnt reverse, CTGGCATTAAGGAGACTCAGTG.

We thank Tim Billings for technical help and Mary Ann Handel for critical reading of the manuscript. This article is dedicated to the memory of Ken Paigen, an excellent colleague, mentor and visionary.

The peer review history is available online at https://journals.biologists.com/dev/article-lookup/doi/10.1242/dev.199414