DOT1L promotes spermatid differentiation by regulating expression of genes required for histone-to-protamine replacement

Mammalian spermatogenesis is a complex developmental process during which male germ cells undergo a series of tightly regulated molecular events to produce mature gametes capable of fertilizing the oocyte. These events include establishment of a spermatogonial stem cell pool, mitotic proliferation, meiosis and morphogenesis of haploid germ cells to produce spermatozoa. The last phase of this process, called spermiogenesis, involves complex biochemical and nuclear morphological transformations that eventually result in acrosome formation, sperm tail biogenesis, sperm DNA condensation and cytoplasm exclusion (Leblond and Clermont, 1952; Oakberg, 1956; Hermo et al., 2010). Although many of the molecular effectors of spermiogenesis have been identified, the gene regulatory factors responsible for their coordination are not well defined.

Spermiogenesis converts the spherical products of meiosis, called round spermatids, into elongated spermatids with the unique nuclear shape typical of mature sperm. This morphological change requires condensation of the spermatid nucleus and is accomplished by enormous chromatin remodeling. During this process, nucleosomes first incorporate histone variants, including TH2B and H2A.L.2, which promote loading of transition proteins TNP1 and TNP2 onto chromatin; the transition proteins promote recruitment and processing of protamines PRM1 and PRM2, and histones and transition proteins are eventually evicted, leaving protamines as the primary packaging protein in a condensed sperm genome (Meistrich et al., 2003; Lewis et al., 2003; Barral et al., 2017; Govin et al., 2007; Merges et al., 2022). This histone-to-protamine transition drastically reduces the sperm nuclear volume, and results in a compact and hydrodynamic nucleus that is resistant to DNA damaging agents, protecting the heritable genome during fertilization (Brewer et al., 1999; Oliva, 2006). Defects in the histone-to-protamine transition produce sperm with abnormal condensation, altered morphology and reduced motility, ultimately leading to failures in fertilization (Cho et al., 2001; Li et al., 2014; Wang et al., 2019; Okada et al., 2007; Shirley et al., 2004; Zhao et al., 2004a,b). How this crucial developmental event is regulated is therefore a fundamental question in reproduction and development.

Several factors important for regulating the histone-to-protamine transition are known. An essential prerequisite for histone removal in elongating spermatids is hyperacetylation of histone H4 (Hazzouri et al., 2000; Morinière et al., 2009; Meistrich et al., 1992; Qian et al., 2013). Acetylation occurs at multiple positions on the H4 protein and helps to loosen sperm chromatin by destabilizing its higher order structure. H4 hyperacetylation requires the histone acetyltransferases GCN5 and EPC/TIP60, recruitment of p300 by the nuclear protein NUT, and activity of the chromatin remodeler CHD5 along with other remodeling complexes (Morinière et al., 2009; Dong et al., 2017; Shiota et al., 2018; Qian et al., 2013; Gaucher et al., 2010; Luense et al., 2019; Li et al., 2014; Zhuang et al., 2014). A key reader of acetylH4 is bromodomain testis-specific protein (BRDT), which binds to acetylated lysines 5 and 8 of H4 (H4K5ac and H4K8ac) through its bromodomain and facilitates acetylH4 eviction and subsequent protamine recruitment (Dhar et al., 2012; Pivot-Pajot et al., 2003; Morinière et al., 2009; Sasaki et al., 2009; Gaucher et al., 2012). Mice specifically lacking the bromodomain of BRDT exhibit defects in histone removal and sperm head elongation due to altered acetylH4 binding (Gaucher et al., 2012; Goudarzi et al., 2016; Shang et al., 2007; Berkovits and Wolgemuth, 2011).

An underexplored candidate regulator of the histone-to-protamine transition is the histone methyltransferase DOT1L (disruptor of telomeric silencing 1-like, also called KMT4). DOT1L catalyzes mono-, di- and tri-methylation of histone H3 at lysine 79 (H3K79me1/2/3), and these modifications are enriched in meiotic and postmeiotic stages of spermatogenesis, peaking immediately before the histone-to-protamine transition (Ontoso et al., 2013; Dottermusch-Heidel et al., 2014; Feng et al., 2002; Lacoste et al., 2002; van Leeuwen et al., 2002; Ng et al., 2002). In humans, DOT1L is more strongly expressed in testis than in any somatic tissue (Mele et al., 2015; Thul et al., 2017; Uhlén et al., 2015), implying that it has an important role in spermatogenesis. In cell lines, H3K79me and DOT1L promote transcription by binding to RNA polymerase II and its associated protein complexes (Guenther et al., 2008; Steger et al., 2008; Jonkers et al., 2014; Kim et al., 2012; Mohan et al., 2010; Veloso et al., 2014; Wood et al., 2018). H3K79me3 has also been implicated in heterochromatin regulation and was recently shown to be essential for driving major satellite expression, heterochromatin maintenance and chromosomal integrity in mouse embryonic stem cells (Jones et al., 2008; Malla et al., 2021preprint; Ooga et al., 2013). In mice, DOT1L is important for maintenance of spermatogonial stem cell pools (Lin et al., 2022). However, the reported enrichment of H3K79 methylation in late meiosis raises the possibility that DOT1L could also be regulating meiotic and postmeiotic germ cell development.

Here, we investigate the role of DOT1L in spermiogenesis and demonstrate that DOT1L has an important role in histone-to-protamine replacement and in chromatin condensation during sperm differentiation. Specifically, DOT1L positively regulates expression of transition protein and protamine genes, as well as the acetylH4 chaperone gene Brdt. Loss of DOT1L in spermatogenic cells leads to defects in the histone-to-protamine transition and is incompatible with fertility, helping to explain why males lacking germline DOT1L are sterile, despite incomplete penetrance of the previously reported spermatogonial stem cell phenotype. Together, our results indicate that DOT1L is a crucial mediator of transcription in spermatids and an indispensable regulator of haploid germ cell differentiation.

DOT1L is broadly expressed in a variety of tissues, but is especially abundant in testis (Mele et al., 2015; Papatheodorou et al., 2020; Fonseca et al., 2014; Thul et al., 2017; Uhlén et al., 2015). To identify the precise testicular cell types that express DOT1L, we performed immunofluorescence in wild-type testis cross-sections after validating DOT1L antibody specificity in Dot1l knockout testis (see below for conditional knockout strategy and validation). We detected DOT1L expression in most testicular cells, including both germ and somatic cells, with the exception of late post-meiotic cells (elongated spermatids) (Fig. 1A). DOT1L was most highly expressed in nuclei of late pachytene spermatocytes, secondary spermatocytes and round spermatids (Fig. 1B), suggesting a possible requirement in meiotic and postmeiotic stages of spermatogenesis.

As DOT1L is a histone H3 lysine 79 (H3K79) methyltransferase, we next assessed the dynamics of the DOT1L-deposited modifications H3K79me2 and H3K79me3 in meiotic and postmeiotic male germ cells using immunofluorescence in chromosome spreads (Fig. 1C-E). Consistent with previous results in other cell types (Ontoso et al., 2013; Malla et al., 2021 preprint; Ooga et al., 2008), H3K79me2 was present throughout the euchromatin but excluded from DAPI-dense heterochromatin in pachytene spermatocytes and round spermatids, while H3K79me3 was strongly enriched in heterochromatin but mostly depleted from euchromatin (Fig. 1C,D). In elongating spermatids, H3K79me2 was broadly present throughout the chromatin, while H3K79me3 was present in restricted DAPI-dense domains that were reminiscent of chromocenters, which are the DAPI-dense heterochromatic structures composed primarily of pericentromeric repeats (Fig. 1E). These results suggest that DOT1L is active during meiotic and postmeiotic stages of spermatogenesis, and imply that it may be important for either meiotic division or postmeiotic spermatid differentiation.

To evaluate the function of DOT1L in spermatogenesis, we generated a germ cell conditional knockout for Dot1l (Dot1l cKO) by crossing mice carrying a conditional Dot1l allele (Dot1l^f/f) (Bernt et al., 2011) with mice carrying a Ddx4-Cre transgene (Gallardo et al., 2007; Hu et al., 2013) (Fig. 2A). Ddx4-Cre is expressed in prenatal germ cells beginning by embryonic day 15.5 and mediates recombination of target alleles by birth (Gallardo et al., 2007; Hu et al., 2013). DOT1L protein expression was completely abolished in Dot1l cKO adult testis (Fig. 2B). Dot1l cKO males produced no offspring after mating with wild-type females for 6 months (Fig. 2C), indicating that DOT1L is required for male fertility. Testes of young adult 3-month-old Dot1l cKO mice were reduced in size compared with control mice but did not display any gross morphological defects (Fig. 2D). Epididymides of Dot1l cKO mice were mostly devoid of mature sperm and instead contained abnormally condensed and round cells (Fig. 2E). DAPI staining of germ cells isolated from the cauda epididymis of Dot1l cKO and control mice confirmed that the small number of sperm-like cells in the epididymides exhibited abnormal elongation and condensation in the Dot1l cKO, compared with normal sperm head morphology in control animals (Fig. 2F). The abnormally elongated sperm in Dot1l cKO epididymis contained fragmented DNA detected by TUNEL staining, suggesting that they have undergone apoptosis (Fig. 2G). These findings demonstrate that loss of DOT1L in spermatogenic cells produces defective sperm with decondensed chromatin, depletes the number of normal sperm in the epididymis and ultimately causes infertility.

We next performed histological analysis of Hematoxylin and Eosin stained testes of control and Dot1l cKO mice in order to more precisely dissect the role of DOT1L in sperm formation. We observed that Dot1l cKO testis exhibited two types of seminiferous tubules. About 25% of tubules displayed a Sertoli cell-only phenotype devoid of any germ cells, indicative of defective spermatogonial stem cell proliferation, as recently reported (Lin et al., 2022). However, most tubules possessed both meiotic and postmeiotic germ cells, indicating successful completion of meiosis (Fig. 3A,B). Immunostaining for DDX4 (also known as MVH), a marker for meiotic and postmeiotic germ cells, revealed that 65% of tubules in our Dot1l cKO contained MVH positive germ cells (Fig. 3C,D). At tubule stages X and I, which are distinguished by presence of late pachytene spermatocytes and elongating spermatids (stage X), or by early pachytene spermatocytes, round spermatids and early condensing spermatids (stage I), there were no apparent differences in the composition and arrangement of testicular cells between Dot1l cKO and control (Fig. S1). However, Dot1l cKO stage VII and VIII tubules, which should contain spermatids in the final stages of maturation (stage VII) or sperm just released into the lumen (stage VIII), were deficient in fully formed elongated spermatids, although early stage elongating spermatids could be seen (Fig. 3A, Fig. S1). Some tubules that should contain elongating or condensing spermatids were missing these cells altogether, and tubules that did contain elongating spermatids displayed both normally and abnormally condensed spermatids (Fig. 3A, Fig. S1). These results indicate that the previously reported defect in spermatogonial stem cell proliferation does not fully explain the completely penetrant male infertility we observe following loss of DOT1L in male germ cells. Instead, our data suggest that infertility is also due to abnormal spermiogenic differentiation of round to elongated spermatids.

To test whether there is any delay in the formation of round spermatids, we collected testis from juvenile control and Dot1l cKO mice at 25 days post partum (dpp), when spermatogenesis has progressed to the round spermatid stage but elongating spermatids are not yet present, and performed histological and flow cytometric analysis. Consistent with our results from adult testes, we did not observe any histological differences in composition and arrangement of germ cells in tubules containing postmeiotic cells (Fig. 3E). Similarly, flow cytometry analysis based on DNA content in isolated testicular cells showed similar fractions of cells corresponding to haploid round spermatids (1C), diploid spermatogonia and somatic cells (2C), and tetraploid primary spermatocytes (4C) (Fig. 3F,G), indicating that the defect occurs after the round spermatid stage.

The known function of DOT1L as a chromatin regulator in combination with our data indicating a role for DOT1L in spermiogenesis suggest that DOT1L might participate in the chromatin condensation that is a central aspect of spermiogenesis (Brewer et al., 1999; Pogany et al., 1981; Zhao et al., 2004b; Ribas-Maynou et al., 2022). Mouse spermiogenesis is divided into 16 steps of spermatid differentiation (1-16) encompassing the major cellular changes of sperm differentiation, including acrosome formation, mitochondrial biogenesis, sperm tail formation and sperm head elongation and condensation (Russell et al., 1990). Sperm head condensation specifically begins at step 9 (corresponding to tubule stage IX) and is complete by step 13 (tubule stage I). To determine the steps of spermiogenesis during which DOT1L is most active, we performed immunostaining for H3K79me2 in wild-type adult testis sections. We analyzed tubules corresponding to stage VIII that have just released mature sperm into the tubule lumen (spermiation), stages IX-XI containing elongating spermatids that are undergoing the histone-to-protamine transition, and stage XII tubules that contain condensing spermatids. We observed intense staining of H3K79me2 in steps 9-11 (tubule stage IX-XI) elongating spermatids, implying high activity of DOT1L at the stages during which sperm-specific chromatin remodeling and condensation take place (Fig. 4A-D). As expected, H3K79me2 was completely absent in step 12 (tubule stage XII) condensing spermatids of control mice because histone removal is complete at this stage (Fig. 4E). H3K79me2 was absent in stage XI tubules of Dot1l cKO testes, confirming the specificity of the H3K79me2 antibody (Fig. 4F). These results support a role for DOT1L in regulating sperm head condensation.

To demonstrate the effect of loss of DOT1L on sperm head morphology, we then performed transillumination-assisted microdissection of tubules containing step 13-16 elongated spermatids and stained with DAPI to determine their nuclear shape. Compared with control spermatids, Dot1l cKO spermatids exhibited large, poorly condensed nuclei with shapes that were distinct from the typical comma-shaped spermatid nuclei of control mice (Fig. 4G). Although it is possible that the developmental changes occurring in Dot1l cKO testes could alter the cell populations recovered from transillumination-based microdissection, the shapes observed did not correspond to normal spermatid morphology at any stage (Oakberg, 1956), indicating a defect in sperm head morphogenesis. Mis-shapen sperm heads could arise either due to problems with nuclear condensation or due to defective DNA damage signaling, which is required for histone eviction immediately before chromatin condensation (Leduc et al., 2008; Meyer et al., 2017). Because DOT1L has been implicated in DNA damage repair (Wood et al., 2018; Kari et al., 2019), we tested the levels of DNA damage in Dot1l cKO spermatids by staining for the DNA damage marker γH2AX. We observed no difference in γH2AX staining between Dot1l and control spermatids, making it unlikely that defects in DNA damage induction explain the head shape defect we observe in Dot1l cKO spermatids (Fig. 4H). Further supporting this conclusion, 8-OHdG, a marker of oxidative damage, was occasionally but rarely detected in Dot1l cKO testicular germ cells (Fig. 4I), but was elevated in germ cells isolated from the caput epididymis (Fig. 4J), consistent with elevated DNA damage occurring after the onset of nuclear condensation defects.

One of the central events of mammalian spermiogenesis is the histone-to-protamine transition. During this drastic remodeling event, the histone variant H2A.L.2 is incorporated into chromatin and enables loading of transition proteins (TNP1 and TNP2), which in turn drive recruitment and processing of protamine proteins (PRM1 and PRM2) (Barral et al., 2017; Meistrich et al., 2003; Lewis et al., 2003; Govin et al., 2007; Merges et al., 2022; Hazzouri et al., 2000). Eventually, histones and transition proteins are evicted, resulting in protaminated, hyper-compact sperm chromatin (Meistrich et al., 2003; Lewis et al., 2003; Qian et al., 2013). As Dot1l cKO spermatids appeared to exhibit chromatin condensation defects (Fig. 4G), we evaluated the possibility that the histone-to-protamine transition was disrupted in the absence of DOT1L. We assessed the presence of histone H3 in Dot1l cKO compared with control seminiferous tubules using immunofluorescence. H3 was observed in stage X elongating spermatids of both control and Dot1l cKO mice, as expected (Fig. 5A). However, although histone H3 was no longer detected in elongated spermatids of control mice at stage II, after the replacement of histones by protamines should be complete, H3 persisted in stage II spermatids of Dot1l cKO mice (Fig. 5B,C, arrow), pointing to incomplete histone eviction in the Dot1l cKO spermatids.

We next evaluated whether protamine loading was also defective by performing immunostaining for PRM1 in squash preparations of control and Dot1l cKO elongated spermatids. Compared with control spermatids that exhibited strong PRM1 signal and were normally condensed into a comma shape, Dot1l cKO spermatids isolated from testes displayed a variety of abnormalities in PRM1 recruitment (Fig. 5D). The majority (>50%) of Dot1l cKO spermatids showed reduced PRM1 staining intensity, whereas another 30% showed high intensity but abnormal localization of PRM1 in the cytoplasm. In the remaining 20% of Dot1l cKO spermatids, PRM1 recruitment appeared similar to control (Fig. 5D, bottom row). Among sperm-like germ cells isolated from the Dot1l cKO epididymis, the majority showed reduced PRM1 intensity, although the range of PRM1 phenotypes observed in testicular spermatids was also evident (Fig. 5E). Together, these data show that loss of DOT1L impairs histone eviction and protamine recruitment in spermatids.

We then asked how DOT1L acts to promote the histone-to-protamine transition and nuclear condensation. As transition proteins are crucial for guiding recruitment and processing of protamines on chromatin (Barral et al., 2017), we next asked whether transition protein expression was disrupted in the absence of DOT1L. Immunofluorescence staining in testis sections revealed that levels of the transition protein TNP1 were drastically reduced in Dot1l cKO testes (Fig. 5F), suggesting that loss of DOT1L impairs expression of both transition proteins and protamines. The major upstream event triggering transition protein and protamine loading is hyperacetylation of histone H4 in stage IX-XI elongating spermatids, with subsequent removal of all acetylated H4 as the spermatids progress to stage XII (Hazzouri et al., 2000; Meistrich et al., 1992; Morinière et al., 2009; Sasaki et al., 2009; Qian et al., 2013). Therefore, we next tested whether DOT1L is required for H4 hyperacetylation. We focused on spermatids undergoing this transition, which occurs at steps 9-12 (tubule stages IX-XII), and performed dual immunostaining of acetylated histone H4 (pan acetylH4) and protamine 2 (PRM2) in control and Dot1l cKO testis cross-sections. Similar to our observations for PRM1 (Fig. 5G), we observed reduced staining of PRM2 in most Dot1l cKO spermatids at all stages with a fraction of spermatids expressing abnormally high but mislocalized staining. As expected, in wild-type testis acetylH4 was detectable only in step 9-11 (tubule stage IX-XI) elongating spermatids, and PRM2 was highly expressed in spermatids beyond steps 12-16 (tubule stage XII) (Fig. 5G, left panel). In Dot1l cKO spermatids, acetylH4 was detected at tubule stages X and XI, similar to control, but abnormally retained in stages III-V, when it is absent in wild type (Fig. 5G, right panel). These results indicate that acetylH4 induction is not affected in Dot1l cKO mice, but its removal is impaired. Histone retention can lead to DNA fragmentation, and we found that TUNEL staining was more frequently positive in Dot1l cKO elongated spermatids compared with control (Fig. 5H), suggesting that histone retention in Dot1l cKO sperm might be triggering apoptosis. Together, our results demonstrate that Dot1l cKO mice exhibit incomplete histone replacement and abnormal acetylH4 retention in spermatids during chromatin remodeling, which is correlated with elevated apoptosis.

At the molecular level, DOT1L functions as a transcriptional activator by interacting with RNA polymerase II (Pol II) at promoters and promoting transcriptional elongation (Jonkers et al., 2014; Kim et al., 2012; Mohan et al., 2010; Veloso et al., 2014; Wood et al., 2018; Steger et al., 2008). To understand whether its transcriptional activation function contributes to its role in promoting the histone-to-protamine transition, we performed single cell RNA-sequencing (scRNA-seq) on Dot1l cKO and heterozygous control testes. After quality filtering, removal of doublets and dead cells, and harmonization of the control and cKO datasets (see Materials and Methods, Fig. S2A-D), we obtained a final dataset comprising 9823 control and 11,172 Dot1l cKO cells. Identities were assigned to an initial set of 20 harmonized clusters based on marker genes (Fig. S2E) and followed a continuous trajectory spanning spermatogenic development from spermatogonia to elongating spermatids, along with four populations of testicular somatic cells, similar to published scRNA-seq datasets from mouse testes (Lukassen et al., 2018; Green et al., 2018; Hermann et al., 2018). Four clusters for which high confidence identities could not be assigned were excluded from downstream analysis, leaving a final set of 16 clusters (Fig. 6A, Table S1).

The distribution of cells across clusters was largely similar between control and Dot1l cKO datasets (Fig. 6B). The exception was the most terminal population of elongating spermatids (ES-3), which were substantially reduced in number in the Dot1l cKO, consistent with a model where DOT1L is required for spermiogenesis (Fig. 6B). We assessed differential gene expression between control and Dot1l cKO cell populations within each cluster, accounting for the reduced numbers of elongating spermatids in the Dot1l cKO. We defined a large number of gene expression changes induced by the loss of DOT1L: a total of 2204 unique genes were misregulated in the Dot1l cKO across all clusters, with the highest number of differentially expressed genes (DEGs) occurring in postmeiotic stages (Fig. 6C, Table S2). To address the possibility that the apparent differences in gene expression were secondary to differences in cluster assignments between control and Dot1l cKO datasets, we assessed the probabilities that cluster assignments in the Dot1l cKO matched those defined in the control dataset and confirmed that clusters were equivalently assigned between the datasets (Fig. S3A). Expression of Dot1l itself was significantly downregulated across multiple clusters, further validating the knockout (Fig. 6D, Fig. S3B). Contrary to expectations based on the known role of DOT1L as a transcriptional activator, there was a mild bias toward upregulation among DEGs (Fig. 6E). Upregulation of this set of genes may be an indirect effect of DOT1L loss. We plotted the relative expression of each DEG across spermatogenesis in wild-type cells and found that while DEG expression patterns varied, a large fraction were most highly expressed at postmeiotic stages (Fig. 6F). Functional enrichment analysis of the sets of upregulated or downregulated genes across all stages revealed enrichment for spermatogenesis functions among genes of both classes, as well as metabolic processes and cell cycle regulation among upregulated genes, and sperm motility and aerobic respiration among downregulated genes (Fig. 6G). Finally, we examined the specific genes that were most strongly downregulated in Dot1l cKO cells. These encompassed multiple genes required for spermiogenesis, including Tnp1, Tnp2, Prm1 and Prm2 (Fig. 6H), but not H2al2a, which encodes the histone variant H2A.L.2 (Fig. S3C). We also observed a more modest downregulation of Brdt in late round spermatids and the bromodomain homolog Brd8 in early meiotic cells (Fig. 6I). Unexpectedly, we also found misregulation of transcripts corresponding to epigenetic repressors, suggesting that some of the effects of DOT1L on chromatin regulation may be indirect and mediated by misexpression of other chromatin regulators, such as Erh, Kdm5b and Suv39h2 (Fig. S3D). Together, these results indicate that although loss of DOT1L does not significantly influence the distribution of spermatogenic cell populations, it causes substantial transcriptional misregulation of genes involved in nuclear condensation during spermiogenesis, with the strongest effect concentrated in postmeiotic cells.

The results of our scRNA-seq analysis indicate that although all spermatogenic cell types are present in Dot1l cKO testes, there are significant changes in gene expression within each cell type. Although many of these changes are likely to be indirect effects, the extensive transcriptional defects and known molecular function of DOT1L as a transcriptional activator strongly suggest that DOT1L directly regulates the spermatogenic transcriptome. We selected some known markers of spermatogenesis that are required at meiotic or postmeiotic spermatid differentiation stages (Sycp3, Tnp1, Tnp2, Prm1, Prm2, Brdt, Brd2 and Brd4) and validated the observed changes in expression by RT-qPCR. Consistent with the scRNA-seq data, this analysis showed significantly reduced levels for the protamine and transition protein genes Prm1, Prm2 and Tnp1 (Fig. 7A), and a trend toward downregulation for the other markers. The stronger effect on transition protein and protamine genes is consistent with the scRNA-seq data and supports a specific role for DOT1L in regulating postmeiotic chromatin remodeling and sperm head condensation. We also tested the protein levels of PRM2 and BRDT, both of which are essential for sperm-specific chromatin compaction, by western blotting and found that levels of both proteins were reduced in Dot1l cKO testis compared with control (Fig. 7B).

The finding that BRDT is reduced in the absence of DOT1L is intriguing because BRDT is a sperm-specific histone reader that binds to acetylH4, has bromodomain-dependent functions in post-meiotic spermatid development and is required for appropriate spermatid condensation (Gaucher et al., 2012). DOT1L may therefore regulate spermiogenesis partly through direct regulation of transition protein and protamine genes, and partly through indirect regulation of these genes via regulation of Brdt. To more deeply understand how DOT1L regulates Brdt expression during spermiogenesis, we performed chromatin immunoprecipitation followed by qPCR (ChIP-qPCR) for the DOT1L-dependent histone modification H3K79me2 in control and Dot1l cKO round spermatids isolated by flow cytometry. H3K79me2 was strongly enriched at the Brdt locus in control spermatids and this enrichment was dependent on DOT1L (Fig. 7C). We then performed ChIP-qPCR for elongating RNA Pol II (phosphorylated at serine 2, RNA Pol II S2-P) in flow-sorted round spermatids. DOT1L interacts with RNA Pol II at transcriptional start sites to mediate gene transcription (Kim et al., 2012; Steger et al., 2008), and we reasoned that DOT1L could regulate Brdt expression by recruiting RNA Pol II and/or promoting transcriptional elongation. We found that Pol II S2-P enrichment at the Brdt promoter was also DOT1L dependent, as a 50% reduction in RNA Pol II S2-P at Brdt was observed in Dot1l cKO compared with the control spermatids (Fig. 7D). These data implicate DOT1L as a direct transcriptional regulator of Brdt, a gene that is crucial for sperm-specific chromatin remodeling.

In addition to its role as a transcriptional activator at single-copy genes, DOT1L has also been shown to promote major satellite transcription in mouse embryonic stem cells (Malla et al., 2021 preprint). Major satellite transcripts increase in abundance in condensing spermatids and have been suggested to be important for the histone-to-protamine transition (Hoghoughi et al., 2020). Major satellite transcription is required for structural integrity of chromocenters, which are DAPI-dense heterochromatic regions composed mostly of major satellite repeats that are thought to facilitate compaction of the genome in elongating spermatids (Saksouk et al., 2015; Probst et al., 2010; Velazquez Camacho et al., 2017; Meyer-Ficca et al., 1998; Haaf and Ward, 1995; Prakash Yadav et al., 2023; Yadav et al., 2020). Therefore, we considered the possibility that, in addition to regulating expression of genes important for the histone-to-protamine transition, DOT1L also promotes this transition by enhancing expression of major satellite transcripts. We evaluated transcript levels corresponding to the mouse major satellite sequence and found a substantial decrease in major satellite transcripts in Dot1l cKO testis (Fig. 7E). Correspondingly, we observed a significant disruption of chromocenter morphology in round spermatids of Dot1l cKO mice (Fig. 7F,G). Changes in major satellite expression and chromocenter integrity may combine with reduced production of transition proteins and protamines to contribute to defective chromatin condensation in Dot1l cKO spermatids. Interestingly, BRDT has itself been implicated in regulation of chromocenter formation in testis, although this effect was variable and dependent on genetic background (Berkovits and Wolgemuth, 2011). DOT1L could thus regulate chromocenter structure both through direct regulation of major satellite transcription and indirectly through regulation of BRDT levels. Together, these results establish DOT1L as an essential chromatin regulator that drives transcriptional programs needed to bring about the complex nuclear remodeling required for sperm formation and function.

Here, we demonstrate that the histone modifier DOT1L is essential for histone removal and sperm head condensation during spermiogenesis, and that its loss in spermatids leads to defective chromatin remodeling, which ultimately results in abnormal sperm head morphogenesis and infertility. We find that enrichment of DOT1L-dependent H3K79 methylation coincides with the timing of spermatid elongation and condensation during spermatogenesis, and show that DOT1L serves as an important driver of transcriptional programs that are required for postmeiotic maturation of round spermatids to elongated spermatids. We find that DOT1L deposits H3K79me2 at Brdt and may help to promote transcriptional elongation by RNA Pol II, ultimately enhancing expression at this locus and other genes that are essential for sperm DNA condensation. Furthermore, we show that abnormal elongating spermatids in Dot1l cKO testes are likely eliminated by apoptosis. Our data suggest that the male sterility phenotype observed in Dot1l germline conditional knockouts is due in part to defects in postmeiotic spermatid differentiation in addition to the loss of premeiotic spermatogonial stem cells that has been reported previously (Lin et al., 2022). Our findings are consistent with those of Lin et al. (2023) showing a requirement for DOT1L beyond the spermatogonial stem cell stage, aberrant retention of histones in epididymal sperm of Dot1l cKO males, and co-dependence between DOT1L and the transcriptional elongation co-factor MLLT10 (AF10) during spermatogenesis.

We found that the expression of DOT1L peaks during meiotic to postmeiotic transition and that a large fraction of transcriptional changes in expression occur after meiosis (Fig. 6C), supporting the involvement of DOT1L in postmeiotic germ cell development and spermatid differentiation. Specifically, we found that genes required for normal spermatid elongation and chromatin condensation, including Tnp1, Tnp2, Prm1 and Prm2 (Cho et al., 2001; Shirley et al., 2004) are significantly downregulated in Dot1l cKO spermatids (Figs 6H and 7A). We also detected a robust downregulation of Brdt transcript by qPCR and downregulation of BRDT protein by western blot, as well as a trend toward downregulation in spermatids in our scRNA-seq dataset. DOT1L is known to interact with the C-terminal domain of RNA Pol II and to recruit it to active chromatin (Mohan et al., 2010; Kim et al., 2012). Together with altered deposition of H3K79me2 and differential enrichment of elongating RNA Pol II at spermiogenic loci, our findings suggest that a requirement for DOT1L-mediated transcriptional activation of spermiogenesis genes explains the defects in sperm head packaging and chromatin condensation observed in the Dot1l cKO. DOT1L could be acting via the DotCom complex to promote sperm-specific transcriptional activation, as has been reported for somatic cells (Mohan et al., 2010), consistent with the genetic and physical interaction with MLLT10 in testes reported by Lin et al. (2023), and possibly with participation of some unknown spermatogenesis-specific complex components.

The central finding of this study is that DOT1L loss causes defective histone removal that results in abnormal chromatin condensation in elongating spermatids, as well as partial disruption of chromocenter formation in round spermatids. Several consequences of DOT1L loss could mediate these effects. First, expression of protamines and transition proteins, which are essential for normal histone removal and chromatin condensation, were reduced in Dot1l cKO spermatids (Merges et al., 2022; Cho et al., 2001; Zhao et al., 2004a,b). Second, we find that loss of DOT1L impairs transcription of the gene encoding BRDT (Figs 6I, 7B-D), a testis-specific bromodomain protein known to bind acetylated H4 and facilitate histone eviction during spermiogenesis (Morinière et al., 2009; Gaucher et al., 2012; Goudarzi et al., 2016; Shiota et al., 2018), implying that reduced BRDT expression may mediate some of the phenotypes observed in Dot1l cKO spermatids. Finally, reduced expression of major satellite repeat transcripts in Dot1l cKO spermatids may contribute to chromocenter destabilization and associated protamine replacement defects. Major satellite transcription has been shown to stabilize chromocenter structures in spermatids and other cell types, facilitating histone-to-protamine replacement; our group recently reported that DOT1L selectively promotes major satellite expression to stabilize heterochromatin in mouse embryonic stem cells (Malla et al., 2021preprint; Saksouk et al., 2015; Probst et al., 2010; Velazquez Camacho et al., 2017; Meyer-Ficca et al., 1998; Haaf and Ward, 1995; Yadav et al., 2020; Prakash Yadav et al., 2023). Interestingly, spermatids homozygous for mutant BRDT lacking the first bromodomain (Brdt^ΔBD1/ΔBD1) also exhibit chromocenter disruption, although this effect is variable and dependent on genetic background (Berkovits and Wolgemuth, 2011), implying that DOT1L may regulate chromocenter stability at multiple levels.

Our scRNA-seq data reveal that there are few substantial differences in the populations of germ cells present in the testis of the Dot1l cKO, a result supported by our flow cytometry data, although scRNA-seq analysis did confirm the reduction in elongating spermatids detected in our histological and immunofluorescence data (Figs 3 and 6). We note that it is possible that changes in cell differentiation induced by loss of DOT1L could affect cell type assignment in the scRNA-seq data and bias cell type-specific differential gene expression calls, although our analysis confirms high confidence in cluster assignments across genotypes. DOT1L was recently shown to be required for maintenance of spermatogonial stem cells (SSCs) (Lin et al., 2022). Using the same conditional knockout model, we observed that the SSC defect appeared incompletely penetrant, with about 25% of seminiferous tubules exhibiting complete loss of germ cells (Sertoli cell only phenotype) at 3-6 months of age (Fig. 3). This result suggests that loss of DOT1L does not affect all SSCs equally. Current models support the co-existence of multiple subpopulations of neonatal spermatogonia defined by different cell surface markers, where SSCs expressing high levels of Id4 constitute a true stem cell population and are a subset of undifferentiated spermatogonia identified by PLZF/ZBTB16 and CDH1, while a separate pool of initial differentiating spermatogonia expresses STRA8 and cKIT (Niedenberger et al., 2015; Law and Oatley, 2020). It is possible that loss of DOT1L specifically depletes only some of these populations, resulting in the incompletely penetrant spermatogonial phenotype we observed. Our scRNA-seq analysis did not reveal loss of a specific spermatogonial population in Dot1l cKO testis, but the small size of spermatogonial subpopulations may have precluded their detection in our whole-testis dataset. We confirmed complete loss of male fertility upon DOT1L deletion, in agreement with the previous report (Lin et al., 2022). The presence of abundant meiotic and postmeiotic cells in our sterile adult Dot1l cKO males indicates that postmeiotic defects also contribute substantially to the sterility phenotype.

In summary, our results establish DOT1L as an essential regulator of sperm-specific gene transcription and highlight its importance in histone removal and chromatin condensation during spermatid differentiation. Together, our data reveal that a major reason for male sterility observed upon loss of DOT1L is defective histone-to-protamine replacement in elongating spermatids. Interestingly, H3K79me2 is one of a handful of histone modifications reported to be retained at a higher frequency in defective compared with normal human sperm (La Spina et al., 2014). Along with the previously reported spermatogonial defect, our findings emphasize that DOT1L has multiple crucial functions during spermatogenesis. Our results advance mechanistic understanding of sperm morphogenesis and suggest new directions for future studies of infertility and contraception.

These studies were approved by the Yale University Institutional Animal Care and Use Committee under protocol 2020-20169. All mice (Mus musculus) were maintained and euthanized according to the principles and procedures described in the National Institutes of Health guide for the care and use of laboratory animals. All mice were on a C57BL/6J genetic background. Dot1l cKO male mice were generated by crossing Dot1l^flox/flox (Bernt et al., 2011) females with male mice harboring one copy of Ddx4-Cre transgene (B6-Ddx^{4tm1.1(cre/mOrange)Dcp}), which is first expressed in germ cells at embryonic day 15.5 (Gallardo et al., 2007; Hu et al., 2013). Mouse genotyping was performed as defined previously (Bernt et al., 2011). Males between 2 and 5 months of age were used except where otherwise specified. Primers used for genotyping are listed in Table S3.

Testes and epididymides were dissected from control and Dot1l cKO male mice at matched ages. After isolation, tissues were briefly washed with phosphate-buffered saline (PBS) and fixed in Hartman's fixative (Millipore-Sigma; H0290-500ML) for 48 h at room temperature. To remove the last traces of fixative, samples were washed in 70% ethanol, dehydrated and embedded in paraffin wax. Sections (4 µm) were prepared on glass slides, cleared in xylene and dehydrated in a graded series of ethanol. The sections were then stained with Hematoxylin and Eosin. Images were acquired with a bright-field light microscope (Zeiss Axio Lab A1 Bright Field Microscope, 20×0.5 NA or 40×0.75 NA objectives). Stages of the seminiferous epithelia were identified according to established methods (Russell et al., 1990; Ahmed and de Rooij, 2009).

Paraffin wax-embedded testis sections were deparaffinized in xylene, dehydrated in a graded series of ethanol, boiled in 10 mM sodium citrate buffer (pH 6) in a microwave oven for 20 min to retrieve the antigen, washed in PBS, permeabilized in 0.5% Triton X-100 for 10 min, blocked in blocking buffer (5% BSA+0.1% Triton X-100) for 1 h at room temperature and then incubated overnight at 4°C with primary antibodies (Table S4) diluted in blocking buffer. The slides were washed with PBS and stained with fluorophore-conjugated secondary antibodies, including AlexaFluor 488-conjugated or AlexaFluor 568-conjugated goat anti-rabbit-IgG (Molecular Probes) and AlexaFluor 568-conjugated goat anti-mouse-IgG (Molecular Probes). See Table S4 for a complete listing of antibodies and dilutions. All secondary antibodies were used at 1:500 dilution in blocking buffer and incubated at room temperature for 1 h. Finally, slides were mounted in antifade mounting medium containing DAPI (H-1200, Vector Labs). Images were acquired with an LSM 880 (LSM acquisition software) or LSM 980 airyscan confocal microscope (Zeiss) equipped with 405, 488 or 555/561 nm lasers, and fitted with a 63×1.4 NA objective. Images were processed with Zen acquisition software.

TUNEL staining was performed using an APO-BrdU TUNEL assay kit (Sigma-Aldrich, A35125) according to the manufacturer's instructions. Briefly, testis sections were deparaffinized as stated above and then labelled with BrdU in the presence of terminal deoxynucleotidyl transferase (TdT) enzyme for 2 h at 37°C. After two washes, the sections were incubated with AlexaFluor 488-conjugated anti-BrdU antibody supplied with the kit for 1 h at room temperature. The slides were mounted antifade mounting medium containing DAPI and imaged as described above.

Testes from control and Dot1l cKO male mice were dissected and briefly washed in chilled 1× PBS. Tunica albuginea were removed, and seminiferous tubules were minced and collected in RIPA buffer [50 mM Tris-HCl (pH 7.4), 150 mM NaCl, 1% Triton X-100, 0.5% sodium deoxycholate, 0.1% SDS, 1 mM EDTA, 10 mM NaF and 1 mM PMSF] containing freshly prepared cOmplete EDTA-free protease inhibitor cocktail (Roche, 11873580001). Tissues were subjected to homogenization and sonication in a Bioruptor Pico (Diagenode). Cell extracts were centrifuged at 20,000× g for 15 min at 4°C and supernatants were immediately mixed with 4× SDS sample buffer to a final dilution of 1×. The samples were heated at 95°C for 5 min, resolved on a Mini-PROTEAN TGX gel (Bio-Rad, 456-8093) by SDS-PAGE for 2 h at 80 V and transferred onto PVDF membranes (GE Healthcare) in 20 mM Tris-HCl (pH 8.0), 150 mM glycine and 20% methanol. After transfer, membranes were blocked with 5% skimmed (non-fat) milk for 1 h at room temperature, followed by incubation with the primary antibody overnight at 4°C. After washing three times with TBST buffer [20 mM Tris-HCl (pH 7.6), 150 mM NaCl and 0.1% Tween 20), membranes were incubated at room temperature for 1 h with HRP-conjugated secondary antibodies as follows: goat anti-rabbit IgG conjugated to HRP (Jackson ImmunoResearch, 111-035-003, 1:20,000) for DOT1L and BRDT, and goat anti-mouse IgG conjugated to HRP (Jackson ImmunoResearch, 111-035-003, 1:20,000) for PRM2 and alpha- tubulin. See Table S4 for a complete list of antibodies and dilutions. After washing the membrane three times with TBST, specific protein bands were detected using a SuperSignal West Pico PLUS Chemiluminescent Substrate (Thermo Fisher Scientific, 34577). Chemiluminescence was detected using the FluorChem E (Protein Simple) documentation system. Densitometry analysis of bands was performed by using ImageJ (Fiji) software. The band intensity of the protein of interest was normalized to the loading control.

Testes from 25 day post partum (dpp) mice were dissected, the tunica albuginea was removed and the seminiferous tubules were minced in Ca²⁺- and Mg²⁺-free PBS (Gibco). Cells were dispersed by gentle aspiration, filtered using a 40 µm nylon filter, and washed in PBS by centrifuging at 800× g for 5 min. The cells were re-suspended in PBS, fixed in 70% chilled ethanol, and stored at 4°C for 24 h or at −20°C for up to 1 week until further analysis. 1×10⁶-2×10⁶ ethanol-fixed testicular cells were washed with PBS and treated with 0.25% pepsin solution for 10 min at 37°C. Finally, cells were stained with propidium iodide (PI) staining solution (25 µg/ml PI, 40 mg/ml RNase and 0.03% Nonidet P-40 in PBS) at room temperature for 20 min. The PI-stained cells were analyzed on an Bio-Rad S3e cell sorter (Bio-Rad; excitation 488 nm; emission 585/40 nm).

Decapsulated testes were incubated in DMEM containing 0.75 mg/ml collagenase type IV (Gibco, 17104-019) for 10 min at 37°C with occasional inversion. An equal volume of DMEM was added to the suspension of dissociated tubules before centrifuging at 400 g for 5 min at 4°C. After removing the supernatant, the cell pellet was washed in DMEM and resuspended in 0.05% trypsin-EDTA (Gibco, 25200-056) for 10 min incubation at room temperature. An equal volume of ice-cold DMEM+10% FBS was then added to the cell suspension before washing the pellet twice with media. The cell pellet was resuspended in DPBS (Sigma-Aldrich, D8537) containing 0.04% bovine serum albumin (Sigma-Aldrich, A9647) and filtered through a 100 µm filter (Falcon, 22363549) then a 40 µm filter (Falcon, 352340) to generate a single cell suspension.

Testicular cell suspensions were prepared as previously reported (Yeh et al., 2021). Decapsulated testes were briefly washed with ice-cold Dulbecco's phosphate-buffered saline (DPBS, Gibco). Seminiferous tubules were gently dispersed and untangled with forceps and transferred to fresh petri dish. This was repeated three times to remove interstitial cells as much as possible. Untangled seminiferous tubules were incubated in DMEM containing 1.5 mg/ml hyaluronidase (Millipore Sigma, H3506), 0.25 mg/ml DNase I (Stem Cell Technologies, 07900), collagenase type IV (2 mg/ml) and FBS (2%) in a 15 ml tube at 37°C for 20 min. Tubules were gently pipetted until no visible chunks remained. An equal volume of ice-cold DMEM+10%FBS was then added to the cell suspension. Cells were centrifuged at 300 g for 5 min at room temperature and supernatant was discarded. The pellet was resuspended in 10 ml of FACS buffer (DPBS with 2% FBS) and cells were filtered through a 40 μm filter (Falcon, 352340) to generate a single cell suspension.

Testicular cells in suspension were crosslinked with 1% formaldehyde at room temperature for 10 min. Formaldehyde was quenched with 0.1375 M glycine at room temperature for 10 min. Fixed cells were spun down at 300 g for 5 min at 4°C. The pellet was washed twice with FACS buffer. Total testicular cells were stained with PI staining solution as described above and round spermatids (1C cell population) were isolated using a Bio-Rad S3e cell sorter. 1×10⁶ round spermatids were washed with ice-cold DPBS and pelleted. The pellet was resuspended in 50 μl chromatin immunoprecipitation (ChIP) lysis buffer (1% SDS, 10 mM EDTA and 50 mM Tri-HCl at pH 8.1) and frozen at −80°C. To prepare antibody-bound Dynabeads (Thermo Fisher Scientific, 10007D), 10 μl Dynabeads were washed twice with 150 μl of block solution (0.5% BSA in PBS) and resuspended in 30 μl block solution. 1 μg antibody per million cells was added to each aliquot of beads and incubated for 8 h rotating at 4°C. Frozen cells were thawed on ice and ChIP dilution buffer (0.01% SDS, 1.1% Triton X-100, 1.2 mM EDTA, 167 mM NaCl and 16.7 mM Tris-HCl at pH 8.1) was added to reach a total volume of 150 μl. Cells were sonicated at 4°C for 20 cycles (30 s on/off) using a Bioruptor bath sonicator (Diagenode). Aliquots of the same samples were pooled and spun down at 12,000 g for 5 min. The supernatant was moved to a new 1.5 ml Eppendorf tube and brought up to 1 ml with 750 μl ChIP dilution buffer and 100 μl protease inhibitor cocktail (Roche, 11836153001). 50 μl of each sample was set aside as input, and then an aliquot of antibody-bound Dynabeads was added to the lysate and incubated overnight at 4°C with rotation. After overnight incubation, beads were washed twice with low-salt immune complex wash buffer (0.1% SDS, 1% Triton X-100, 2 mM EDTA, 150 mM NaCl and 20 mM Tris-HCL at pH 8.1), twice with LiCl wash buffer (0.25 M LiCl, 1% NP40, 1% deoxycholate, 1 mM EDTA and 10 mM Tris-HCl at pH 8.1) and twice with TE (1 mM EDTA and 10 mM Tris-HCl at pH 8.0). Bound DNA was eluted twice with 125 μl elution buffer (0.2% SDS, 0.1 M NaHCO3 and 5 mM DTT in TE) at 65°C and crosslink reversal was performed by incubating at 65°C for 8-15 h. ChIP and input samples were incubated for 3 h at 37°C with 0.2 mg/ml RNAse A (Millipore, 70856-3) and for 4-10 h at 55°C with 0.1 mg/ml Proteinase K (NEB P8107S). Sample DNA was isolated using a Zymo ChIP DNA Clean & Concentrator kit (Zymo Research, D5201) according to the manufacturer's instructions. DNA was eluted with 20 μl Elution Buffer into fresh Eppendorf tubes, then re-eluted with the same eluate to enhance the yield.

Testes from control and Dot1l cKO male mice were dissected and briefly washed in chilled 1× PBS. After removing tunica albuginea, seminiferous tubules were minced and thoroughly homogenized in 1 ml of TRIzol reagent (Invitrogen, 15596026). After 5 min of incubation, the lysate was mixed with 200 μl of chloroform and centrifuged at 12,000 g for 15 min at 4°C. The aqueous phase was mixed with an equal volume of 100% ethanol and transferred to an RNeasy MinElute spin column (Qiagen, 74104) for purification of total RNA according to the manufacturer's instructions. Reverse transcription of 1 µg of total RNA was performed with random hexamer primers and iScript reverse transcriptase (Bio-Rad, 1708896) in a total volume of 20 µl according to the manufacturer's instructions. Reaction mixtures were incubated in a thermocycler at 25°C for 5 min then 42°C for 60 min before stopping the reaction at 85°C for 5 min.

For ChIP-qPCR, 1 μl each of eluted ChIP and input DNA was used for qPCR in 20 μl reaction containing 1 μM forward/reverse primer mix, 1× Power SYBR Green PCR Master Mix (Applied Biosystems, 4367659) and 5.6 μl nuclease-free water. Reactions for each target gene were performed in duplicate in a 96-well plate loaded into an Applied Biosystems QuantStudio 3 Real-Time PCR system under standard cycling conditions.

For reverse transcription qPCR (RT-qPCR), 1 µl of undiluted cDNA was used for a 20 μl reaction volume consisting of 4 µl of 10 µM forward plus reverse primer mix, 10 µl of Power SYBR Green PCR Master Mix (Applied Biosystems, 4367659) and 5 µl nuclease-free water. Relative fold change in transcript abundance was calculated using the delta-delta Ct method by normalizing target gene expression levels to either Actb or Wt1.

Reactions for each target gene were performed in duplicate in a 96-well plate loaded into an Applied Biosystems QuantStudio 3 Real-Time PCR System. Standard cycling conditions were used: for hold stage (×1), 50°C for 2 min, 95°C for 10 min; for PCR stage (×40), 95°C for 15 s, 60°C for 1 min. Melt curve stage conditions were: 95°C for 15 s, 60°C for 1 min and 95°C for 15 s. Primer sequences used for qPCR are listed in Table S3.

Isolation of elongated spermatids was performed as described previously (Kotaja et al., 2004). Briefly, 12-week-old male mice were euthanized, testes were removed and decapsulated, and seminiferous tubules transferred to a Petri dish containing PBS. When viewed under a transilluminating dissection microscope, seminiferous tubules produce different patterns of light absorption ,giving four different regions: weak spot (stages XII-I), strong spot (stages II-VI), dark zone (stages VII-VIII) and pale zone (stages IX-XI). The dark zone, containing elongated spermatids, was cut into 1 mm long pieces and placed in 50 µl of 100 mM sucrose. Elongated spermatids were obtained in suspension by gentle pipetting of the minced tissue for 1 min. Cells were dried on slides pre-dipped in 1% PFA containing 0.15% Triton X-100. The dried slides were briefly washed in PBS followed by cell fixation in 4% PFA for 10 min and immunofluorescence staining with anti-PRM1 antibody was performed as described above. Slides were mounted in antifade mounting medium containing DAPI (H-1200, Vector Labs). Images were acquired as described earlier.

Meiotic chromosome spreads were generated as described previously (Peters et al., 1997). Testes were dissected out from mice and collected in DMEM. After removing the tunica albuginea and extracellular material, the seminiferous tubules were briefly washed in cold PBS followed by incubation in hypotonic extraction buffer [30 mM Tris (pH 8.2), 50 mM sucrose, 17 mM trisodium citrate dihydrate, 5 mM EDTA, 0.5 mM dithiothreitol and 0.5 mM phenylmethylsulfonyl fluoride) for 1 h at room temperature. The tubules were then removed from the hypotonic buffer, transferred to a glass slide, and minced with tweezers and scissors to release the cells. 10 μl of cell suspension was diluted with 40 μl 100 mM sucrose and spread onto a glass slide pre-dipped in 1% paraformaldehyde (PFA) containing 0.15% Triton X-100. Slides were dried for 6 h in a humidified chamber before proceeding with immunofluorescence staining. After two washes in PBS, the dried slides were treated with 2% PFA and 0.15% Triton X-100 for 10 min at room temperature and then blocked in blocking buffer (5% BSA+0.1% Triton X-100) for 1 h at room temperature. Slides were then incubated in anti-SYCP3 and H3K79me2/3 antibodies overnight at 4°C. After staining with fluorophore-conjugated secondary antibodies, slides were mounted in antifade mounting medium containing DAPI (H-1200, Vector Labs). Images were acquired using an LSM 980 airyscan confocal microscope (Zeiss, Zen acquisition software) as described above.

Libraries were prepared with a Chromium Next GEM Single Cell 3′ v3.1 kit (10X Genomics, CG000204) according to the manufacturer's instructions and sequenced on an Illumina NovaSeq machine with 150 base pair paired-end reads. Raw count matrices were generated using CellRanger (10X Genomics). Matrices for the ‘WT’ and ‘KO’ samples were imported to R (R Core Team, 2015) using Seurat 4.1.0 (Hao et al., 2021) and combined into one Seurat object. Reads were filtered to exclude cells likely to be doublets or multiplets (nCount_RNA<100,000), dead or dying cells (nFeature_RNA<8000), or cells with high mitochondrial content (percent.mt<mean+standard deviation). This filtered subset was log normalized and scaled. A linear dimensional reduction was performed using principal component analysis (PCA), and 17 dimensions were determined to be appropriate using the ‘ElbowPlot()’ function. Clusters were identified at a resolution of 0.5, and the ‘RunUMAP()’ function was used to generate Uniform Manifold Approximation and Projection (UMAP) objects for further analysis. Clusters were associated with stages of spermatogenesis based on expression of key spermatogenic genes. Three of these initial clusters had a low number of detected molecules per cell, indicating dead or dying cells, and did not have any top genes associated with spermatogenesis. Therefore, these three clusters were removed, and the data were normalized, scaled and regressed again using the same parameters described above. The dataset was then integrated using Harmony (Korsunsky et al., 2019), and the ‘RunUMAP()’, ‘FindNeighbors()’ and ‘FindClusters()’ functions were used at a dimensionality of 17 and resolution of 0.5 to generate a final UMAP for further analysis. Cluster identities were again assigned using the same set of key spermatogenic genes applied above. To evaluate the fidelity of cluster assignment between control and Dot1l cKO genotypes, cluster identities defined in the control dataset were transferred to the Dot1l cKO using the ‘TransferData()’ function in Seurat and prediction scores were evaluated to confirm a high probability that developmentally equivalent cells were assigned to the same clusters. Differentially expressed genes were determined using the ‘FindMarkers()’ function. Graphs were produced in R using the ggplot2 package.

GO enrichments were evaluated using the GOStats package (Falcon and Gentleman, 2007) in R. P-values were adjusted both by conditioning out child categories and by subsequent correction for multiple testing using the Benjamini-Hochberg method.

Statistical analyses were performed as indicated in the figure legends. In general, unpaired two-tailed t-tests were used for comparison of continuous data, and Welch's correction was applied when the conditions being compared could not be assumed to have the same distribution. Fisher's Exact test was used for categorical data. The threshold for statistical significance was defined as P<0.05. No randomization or blinding was performed.

We thank Kathrin Bernt for the gift of Dot1l^fl mice. We appreciate assistance with genotyping from Y. Huang, K. Marshall and C. Kaya, and advice on scRNA-seq analysis from N. Dias. We thank the staff of the Yale Center for Genome Analysis for Illumina sequencing of scRNA-seq data.

The peer review history is available online at https://journals.biologists.com/dev/lookup/doi/10.1242/dev.201497.reviewer-comments.pdf