Systematic identification of Y-chromosome gene functions in mouse spermatogenesis

Abstract

The mammalian Y chromosome is essential for male fertility, but which Y-genes regulate spermatogenesis is unresolved. We addressed this by generating thirteen Y-deletant mouse models. In Eif2s3y, Uty, and Zfy2 deletants, spermatogenesis was impaired. We discovered that Uty regulates spermatogonial proliferation, revealed a role for Zfy2 in promoting meiotic sex chromosome pairing, and uncovered unexpected effects of Y-genes on the somatic testis transcriptome. In the remaining single Y-gene deletants, spermatogenesis appeared unperturbed, but testis transcription was still altered. Multi-gene deletions, including a human-infertility AZFa model, exhibited phenotypes absent in single Y-deletants. Thus, Y-genes may regulate spermatogenesis even if they show no phenotypes when deleted individually. This study advances our knowledge of Y-evolution and infertility and provides a resource to dissect Y-gene functions in other tissues.

The mammalian sex chromosomes evolved from a pair of autosomes, with the Y chromosome degenerating and ultimately losing around 92% of its ancestral gene content (1–3). The remaining non-recombining region of the mouse Y chromosome contains sixteen gene families, which encode proteins with predicted regulatory roles including transcription activation, ubiquitylation, chromatin modification, RNA stability, and translation (1, 2, 4, 5) (Fig. 1A). Four (Rbmy, Sly, Srsy, and Ssty1/2) are ampliconic, having been amplified into multiple copies, occupying the Y-long arm and centromeric end of the Y-short arm, and are implicated in spermiogenesis and sex ratio control (6–11). Of the remaining twelve, Uba1y, Kdm5d, Eif2s3y, Uty, Ddx3y, Usp9y, Sry, Zfy1, Zfy2, and Prssly were present on the ancestral XY pair, whereas Teyorf1 and the duplicated H2al2y (H2al2b and H2al2c) were more recently acquired in the mouse lineage (4, 12). The majority (twelve out of sixteen) of mouse Y-gene families exhibit testis-biased expression. This, together with the fact that large genomic deletions on the Y chromosome are associated with fertility defects in mice and men, highlights the importance of the Y chromosome in spermatogenesis (5). However, our understanding of the specific Y genes necessary for spermatogenesis and their precise roles remains incomplete. The functions of only a few genes have been identified. Sry is necessary and sufficient for testis determination (13). Eif2s3y is necessary for spermatogonial proliferation, but its mechanism of action remains unknown (14, 15). Zfy genes have been shown to be important in meiosis and sperm morphogenesis, but their exact functions are not fully understood (16–22). Other Y genes have been targeted with no overt fertility phenotypes (12, 23–27), but confirmation that the allele was a null was not always established, or deeper reproductive phenotyping was not performed (16, 23–26). For some genes, knockouts (KO) have never been made. In this study, we established a pipeline to systematically generate null Y-gene deletions and performed extensive fertility phenotyping, thereby determining which genes are involved in spermatogenesis and how they function.

Fig. 1. Generation and screening of thirteen Y-deletant mouse models for abnormal gamete production.

(A) Structure and gene content of the mouse Y chromosome. (B) Experimental strategy to generate and study Y-gene deletant mouse models. (C) Mean numbers of male and female pups born per litter from control and Y-deletant matings. n = total number of pups born. Error bars = 95% confidence interval. (D) Mean testis weight in controls and Y-deletants. n = number of males. Error bars = standard deviation. (E) Median seminiferous tubule area in control and Y-deletants. n = number of biological replicate males. At least 40 tubules were counted per replicate. Error bars = standard deviation. (F) Mean sperm count in controls and Y-deletants. Sperm was collected from the cauda epididymis. n = number of males. Error bars = standard deviation. (G) PAS-stained testis sections of control, Eif2s3y-KO, Uty-KO, AZFa-KO Zfy2-KO, and Zfy1&2-DKO. Tubules are circled by dotted lines. Asterisks indicate tubules with severe germ cell depletion. Black inset: control metaphase, red insets: dark cytoplasmic signal is visible, indicative of apoptosis. Scale bars = 20 μm. Statistical analysis by two-sided Chi-Square test for C, Mann Whitney test for D and F and two tailed t-test for E. Significant p values (<0.05) are shown.

Generating mouse models with Y-gene deletions

We targeted Y genes in XY mouse embryonic stem cells (mESCs), which were then used to produce Y-KO animals. For each Y gene, the entire coding region was excised using CRISPR-Cas9 with two flanking sgRNAs, thereby generating null alleles (Fig. 1B, Fig. S1A-C). The exception was the duplicated H2al2y, for which we used one sgRNA to generate indels in both copies. We used this approach for H2al2y because the two identical copies H2al2b and H2al2c flank Sry, meaning that a whole-gene deletion strategy could inadvertently remove Sry and cause sex reversal (Fig. 1A, Fig. S1D). For Zfy1 and Zfy2, we created individual deletions and a combined Zfy1&2-double knockout (DKO) deletion, allowing us to examine functional overlap between these paralogues. We also generated an animal model of the AZoospermia Factor a (AZFa) deletion, encompassing Usp9y, Uty, and Ddx3y, which in men causes the most severe form of infertility characterized by a complete absence of germ cells (28). We used this model to understand the etiology of AZFa-related infertility, and to examine functional divergence of the Y chromosome in mice and men. The multi-gene KOs also allowed us to interrogate combinatorial effects of Y-gene loss on spermatogenesis.

Once fully validated by low-pass whole genome sequencing (Fig. S1E, F) and MiSeq analysis of on- and off-target mutations (Fig. S1G, H), deletant mESC lines were used to make animals. This is usually achieved by blastocyst mESC injection, and breeding of resulting chimeras to attain germline transmission. Such an approach was not suitable for mutations that could compromise fertility. We therefore used tetraploid aggregation to generate founders that were fully derived from the mESCs, which means they were fully mutant. We improved the typically low birth and offspring survival rates associated with this technology (29–32) (see Methods for details) and generated founders from all our Y-KO mESC lines. We conclude that Eif2s3y, Zfy1, Zfy2, Uty, Uba1y, Usp9y, Ddx3y, Kdm5d, H2al2y, Teyorf1, and Prssly are all dispensable for embryonic survival.

Spermatogenic defects in a subset of Y-gene KOs

To determine which Y genes are necessary for reproduction, Y-KO males were mated with wild type females. Consistent with previous studies, Eif2s3y-KO and Zfy1&2-DKO founders were infertile, with severe spermatogenic defects (17, 33) (Fig.1C-G). The remaining eleven Y-deletant lines all produced offspring (Fig.1C). Thus, three of the eleven deleted Y gene families are necessary but not sufficient for mice to father offspring (13–15, 19): Sry (necessary for testis determination), Eif2s3y (necessary for spermatogonial proliferation), and Zfy (at least one copy needed to produce functional sperm).

Ability to reproduce does not necessarily mean that spermatogenesis is normal. We therefore developed a phenotyping pipeline to exhaustively characterize the reproductive fitness of individual Y-deletants. We assayed litter sizes, offspring sex ratios, testis weights and histology, sperm counts, sperm head morphology, sperm motility, and in vitro fertilization (IVF) success rates. Notably, all these parameters remained unaffected for deletion of Zfy1, Uba1y, Usp9y, Ddx3y, Kdm5d, H2al2y, Teyorf1, and Prssly, showing that these genes are dispensable for mouse spermatogenesis when deleted individually (Fig. 1C-F, Fig. 2A-D, Fig. S2A-I). Litter sizes remained stable as the Y-deletants were progressively backcrossed to C57BL/6J, suggesting that strain background did not overtly influence fertility (Fig. S3A-I).

Fig. 2. Screening for reduced gamete quality in thirteen Y-deletant mouse models.

(A) Median sperm head area for control and Y-deletants. n = number of males. Error bars = standard deviation. On average, 260 sperm were analysed per male. (B) Mean percentage of motile sperm in the cauda epididymis of controls and Y-deletants. n = number of males. Error bars = standard error of the mean. (C) Mean sperm curvilinear velocity (VCL) in controls and Y-deletant. n = number of males. Error bars = standard error of the mean. (D) Mean fertilization rate of sperm from control and Y-deletant males in IVF assays. n = number of males. Error bars = standard deviation. Statistical analysis by Mann Whitney test, with significant p values (<0.05) shown. (E) DAPI-stained sperm heads from Zfy2-KO, Zfy1&2-DKO, and AZFa-KO showing a range of abnormalities, from least (top) to most severe (bottom), compared to a control sperm head with stereotypical shape. Scale bars = 2.5 μm. The average sperm profile of the three deletants is compared to that of the control. n = number of sperm heads used to build the average outline. (F) Sperm from control, Zfy2-KO, and AZFa-KO tracked over one second using computer assisted sperm analysis (CASA). Red trajectory represents a rapid progressive sperm and blue trajectories slow, non-progressive sperm. Scale bars = 50 μm.

In contrast, Eif2s3y-KO, Uty-KO, AZFa-KO, Zfy2-KO, and Zfy1&2-DKO males all exhibited defective germ cell production, manifested as reduced testis weight, seminiferous tubule area, and sperm counts (Fig.1D-F). We found Y genes to be involved in all the major steps of spermatogenesis: mitosis, meiosis, and spermiogenesis. Eif2s3y-KO males showed severe germ cell depletion, consistent with the role of this gene in spermatogonial proliferation (14) (Fig.1G). Interestingly, Uty-KO and AZFa-KO males also exhibited germ cell-depleted tubules, revealing a role for Uty in spermatogonia (Fig.1G, Fig.3C, Fig. S4A; and below).

Fig. 3. Defects in the establishment and differentiation of the spermatogonial pool in Uty-deleted mice.

(A) Number of undifferentiated spermatogonia per tubule in control and Uty-KO P1, P10, and adults, quantified in testis immunostaining for LIN28A and cKIT. n = number of tubules counted across three biological replicates. p values calculated by Mann Whitney test. Scale bars = 2.5 μm. (B) P10 testis sections immunostained for LIN28A. Dashed lines encircle tubules with no germ cells. Scale bars = 100 μm. (C) Adult PAS-stained testis sections, with Uty-KO exhibiting different missing generations compared to control. Epithelial stages are shown in roman numerals. Spermatocytes, round and elongated spermatids are expected at all these stages. Scale bars = 20 μm. (D) Percentage of tubules exhibiting stage-specific missing generations. n = number of tubules counted across four biological replicates. (E) Quantification of percentage of tubules with differentiating In-B spermatogonia using P1, P10, and adult testis sections immunostained with LIN28A and cKIT. Scale bars = 2.5 μm. Tubules lacking progenitor cells (with no LIN28A+ cells) were excluded. n = number of tubules counted across three biological replicates. All error bars = standard deviation. Statistical analysis by Mann Whitney test for A and D and two tailed t-test for E. Significant p values (<0.05) are shown.

Zfy2-KO and Zfy1&2-DKO males exhibited phenotypes at two stages of spermatogenesis. The first occurred during meiosis, with some metaphase cells undergoing apoptosis (Fig. 1G; and below). The second occurred during spermiogenesis, with abnormal sperm morphology and motility affecting fertilization (Fig. 2A-F, Fig. S2B-I, Fig. S4B-D). These defects were more severe in the Zfy1&2-DKO males, illustrating partially redundant roles of Zfy1 and Zfy2 in developing sperm (Fig. 1C, Fig. 2, Fig. S4B-D).

Notably, AZFa KO males also exhibited abnormal sperm morphology and motility, ultimately causing decreased IVF success (Fig. 2A-F). Sperm defects were not observed in Uty, Usp9y, or Ddx3y single mutants (Fig. S2A-I), indicating that multi-Y gene deletions can create phenotypes that are not observed in single gene deletants and unmask potential synergistic roles of Y genes. We conclude that genes exhibiting no observable effects when deleted individually may still impact spermatogenesis. Furthermore, the observation that, unlike men, AZFa-KO mice can produce sperm, provides experimental evidence that the mouse and human Y chromosomes are divergent in their spermatogenic functions.

Uty regulates spermatogonial pool establishment and differentiation

Adult Uty-KO males exhibited some tubules with few or no germ cells, suggesting a defect at the spermatogonial stage (Fig. 1G). This phenotype could arise through failure to establish, differentiate, or renew the spermatogonial stem cell (SSC) pool. There was no difference in testis weights, seminiferous tubule area, or sperm counts between adult (13-15 weeks) and aged (43 weeks) Uty-KO males, suggesting that SSC self-renewal was unaffected (Fig. S5A-C). We therefore focused on SSC establishment and spermatogonial differentiation. To examine SSC establishment, testis sections were co-stained for LIN28A (a marker of undifferentiated and A1-A4 spermatogonia) and cKIT (a marker of differentiating A1 to B spermatogonia) at postnatal day (P)1, P10, and adult (13-15 weeks old) (34, 35) (Fig. S5D). The number of undifferentiated spermatogonia (LIN28A+; cKIT-) was reduced in Uty-KO males at all ages (Fig. 3A). Accordingly, the number of differentiating A1-A4 and In-B spermatogonia (cKIT+; LIN28A-) was also lower (Fig. S5E-F). In some tubules, this phenotype was particularly severe and manifested as a complete absence of spermatogonial cells (Fig. 3B, Fig. S5G). The finding that spermatogonia were reduced as early as P1 demonstrates that Uty is required for the proper establishment of the spermatogonial pool.

To examine whether differentiation was also compromised, we examined testis histology and the proportion of undifferentiated versus differentiated spermatogonia in Uty-KO males. Of note, some tubules lacked entire generations of germ cells during stages of the seminiferous epithelium cycle where their presence would typically be expected. Because this defect was not cell-type specific and affected spermatocytes, round, and elongating spermatids at different stages of the seminiferous cycle, it was most likely caused by an early differentiation failure in spermatogonia (Fig. 3C, D). Supporting this hypothesis, we observed a reduction in the proportion of tubules containing late-stage differentiating spermatogonia (cKIT+; LIN28A-) and in the ratios of In-B to A1-A4 spermatogonia (Fig. 3E, Fig. S5H). We conclude that Uty has two functions in early spermatogenesis, first regulating spermatogonial stem cell establishment, and later spermatogonial differentiation.

Zfy2 regulates X-Y chromosome pairing at meiosis

In Zfy2-KO and Zfy1&2-DKO males, we observed apoptosis in meiotic metaphases (Fig. 1G), which we quantified by cPARP immunostaining (Fig. 4A). Metaphase apoptosis is commonly caused by misaligned chromosomes, which trigger the spindle assembly checkpoint (SAC) (36). Indeed, misaligned chromosomes were clearly apparent in DAPI-stained Zfy2-KO and Zfy1&2-DKO metaphase cells (Fig. 4A-C). Since sex chromosomes are more prone to misalignment than autosomes (36), we used X and Y chromosome painting to determine whether they are misaligned. In 99.51% of cases, the misaligned chromosome was the X or the Y, confirming a sex chromosome-specific effect (Fig. 4C). Despite this X-Y misalignment, there was no increase in sex chromosome aneuploid spermatids in Zfy2-deficient males (Fig. S6A). This finding indicated that, contrary to a previous report using transgenesis in a mouse model lacking a Y chromosome (20), the SAC functions efficiently without Zfy2.

Fig. 4. Defects in meiotic sex chromosome pairing in the absence of Zfy2.

(A) cPARP immunostaining in testis sections to quantify metaphase apoptosis in control and Zfy deletants. Arrows indicate apoptotic metaphase cells. Arrowheads point to misaligned chromosomes. Insets show example metaphase cells. Scale bars = 10 μm. n = number of metaphases counted across three-four biological replicates. Error bars = standard deviation. (B) Percentage of metaphase cells with aligned and misaligned chromosomes in control and Zfy deletants. n = number of metaphases counted across three biological replicates. Error bars = standard error of the mean. (C) Representative aligned and misaligned metaphase plates in testis sections painted with X (green) and Y (magenta) whole chromosome paints. Arrowheads point to misaligned chromosomes. Scale bars = 2 μm. (D) Pachytene spermatocytes immunostained for SYCP3 (green), CREST (magenta), and γH2AX (grey) to quantify X-Y asynapsis in control and Zfy deletants. Insets show sex chromosomes, with synapsed and asynapsed X-Y in control and Zfy2-KO, respectively. Scale bars = 20 μm. n = number of pachytene cells counted across three biological replicates. Error bars = standard error of the mean. Statistical analysis by unpaired two-tailed t test, with significant p values (<0.05) shown.

We suspected that the misalignment phenotype at metaphase resulted from a failure to establish synapsis and recombination between the X and Y chromosomes earlier, at pachynema. To test this hypothesis, we quantified synapsis using antibodies to the axial elements (SYCP3), centromeres (CREST), and the X-Y pair (γH2AX) (Fig. 4D). Zfy2-KO and Zfy1&2-DKO males exhibited a higher frequency of X-Y asynapsis than controls, while autosomal synapsis was unaffected (Fig. 4D; Fig. S6B). The incidence of X-Y asynapsis was similar between Zfy2-KO and Zfy1&2-DKO males, and Zfy1-KOs were unaffected, indicating a specific role for Zfy2 and not Zfy1 in promoting X-Y pairing. Asynapsed X-Y pairs failed to form crossovers, as demonstrated by MLH3 staining (Fig. S6C). ZFY2 thus promotes X-Y synapsis and reciprocal recombination.

The impact of Y genes on the testis transcriptome

To interrogate how Y genes impact the testis transcriptome, we performed bulk RNAseq on adult testes from our thirteen mutants. Consistent with their more dramatic spermatogenic defects, Eif2s3y-KO, Zfy1&2-DKO, and Zfy2-KOs showed the greatest separation from other samples on a Principal Component Analysis (PCA; Fig. S7A, B), and exhibited the highest number of differentially expressed (DE) genes compared to controls (Fig. 5A). Gene set enrichment analysis (GSEA) revealed that gene ontology terms related to spermatogenesis were deregulated in these mutants (Fig. S7C, D; Supplementary Table 1 and 2). In mutants with no phenotypes, we observed a surprisingly wide range of effects on gene expression. While some mutants exhibited few or no DE genes (such as Teyorf1-KO), others exhibited hundreds of DE genes (such as Ddx3y-KO and Kdm5d-KO) (Fig. 5A, Supplementary Table 1 and 2). In the case of Ddx3y-KO, other genes associated with RNA helicase activity were also downregulated (Fig. S7E). We conclude that Y genes for which deletion creates no apparent spermatogenic phenotype can still regulate multiple downstream targets in the testis. Since most Y genes have X-linked homologues, and Zfy and Ddx3y also have autosomal homologues, we assessed their expression in mutants lacking a phenotype. We found that these homologues were not upregulated, suggesting they do not compensate for Y-gene loss at the transcriptional level (Fig. S7F-J).

Fig. 5. Transcriptional deregulation in Y-deletant testes.

(A) Quantification of differentially expressed (DE) genes in bulk RNAseq of Y-KO testes compared to controls. A threshold of Log2 fold change ≥0.5 was used. Asterisks indicate Y-genes not annotated in the mouse genome assembly which thus won’t appear as downregulated. (B-C), Uniform Manifold Approximation and Projection (UMAP) visualization of testis single nuclei RNA sequencing (snRNA-seq) datasets. (B) shows integrated P10 UMAP from control, Eif2s3y, and Uty KO samples (see Fig. S8E for individual UMAPs) and (C) displays integrated adult samples from control, Zfy1, Zfy2, Zfy1&2, and AZFa KOs (see Fig. S8F for individual UMAPs). Undifferentiated and differentiating (Undif and Dif) spermatogonia (SG); leptotene (Lep); zygotene (Zyg); pachytene (Pach); diplotene (Dip); metaphase (M) and secondary (2ry) spermatocytes (SC); round and elongating spermatids (rSD and eSD); foetal Leydig; adult Leydig; Sertoli; peritubular (PT); endothelial (EC) and macrophage (MP) cells are shown. Asterisks show unknown clusters with no clear signatures. (D) Quantification of DE genes for each cell type in P10 Eif2s3y and Uty KOs and adult Zfy1, Zfy2, Zfy1&2, and AZFa KOs. Cell types with less than 10 nuclei in the mutants are marked as na. (E) Expression of the differentiation markers Dmrt1, Uchl1 and the progenitor marker Gfra1 in Eif2s3y-KO and control Undif SG1. (F) Expression of the differentiation markers Dmrt1 and Dazl in Uty-KO, AZFa-KO, and control Dif SG. (G) Expression of members of the steroid synthesis pathway in Eif2s3y-KO, Uty-KO, and control Leydig cells. (H) Expression of the recombinases Dmc1 and Mnd1 in Zfy1, Zfy2, Zfy1&2 KOs and control zygotene cells. Statistical analysis by Kolmogorov Smirnov test, with significant p values (<0.05) shown. For box plots, center line is the median; box limits, 25th and 75th percentile; whiskers, minimum to maximum; points, outliers.

To gain deeper molecular insights into each spermatogenic defect, we performed single nuclei RNA sequencing on Eif2s3y-KO, Uty-KO, AZFa-KO, Zfy2-KO, and Zfy1&2-DKO males. P10 testes from Eif2s3y and Uty-KOs were used to enrich for spermatogonia, since this was where the phenotypes manifested, and adult testes were used for Zfy1, Zfy2, Zfy1&2, and AZFa-KOs to capture all cell types (Fig. 5B, C). Using published marker genes (37, 38), we identified all the expected somatic testis and germ cell populations (Fig. 5B-C, Fig. S8A-F).

By quantifying DE genes across clusters, we identified transcriptomic deregulation in unexpected cell types (Fig. 5D, Supplementary Tables 3 to 8). Interestingly, in Eif2s3y-KO, the earliest undifferentiated spermatogonial cluster had the most DE genes (Fig. 5D). The transcriptional deregulation therefore begins much earlier than the known proliferation defect, which manifests in differentiating spermatogonia (14, 33, 39). Moreover, Eif2s3y-KOs, Uty-KOs, and AZFa-KOs had extensive deregulation in Leydig and Sertoli cells, revealing a surprising impact on the somatic testis transcriptome (Fig. 5D). Notably, despite Zfy2 being predominantly expressed in spermatids (18), the highest number of DE genes in the Zfy2-KO was found at zygonema (Fig. 5D). Strikingly, in AZFa-KOs, the number of DE genes increased throughout pachynema, even though no visible defects were observed at this stage (Fig. 5D). In AZFa-KO, we also observed changes in the expression levels of the corresponding X-linked and autosomal homologues in some cell types, which could compensate for the loss of the three Y-chromosome genes (Fig. S9A-J).

Using gene set enrichment analysis, we identified multiple pathways affected by Y-gene loss that may contribute to the spermatogonial defects in Eif2s3y, Uty, and AZFa deletants, the meiotic defects in Zfy2 and Zfy1&2 mutants, and the spermatid defects in Zfy2, Zfy1&2, and AZFa mutants (Supplementary Tables 9 to 14). To find candidate genes contributing to each phenotype, we also examined DE genes along with the expression of known spermatogenic regulators. In Eif2s3y-KOs, genes involved in spermatogonial differentiation, including Dmrt1, Uchl1, and Stra8 were downregulated, while markers of the undifferentiated state, including Gfra1 and Etv5 were upregulated (Fig. 5E, Fig. S10A, B). Eif2s3y could therefore act to suppress the program that maintains an undifferentiated state (40), or it could promote activation of the differentiation program. In Uty-KOs, known regulators of spermatogonial development were downregulated, including Dmrt1 and Dazl (Fig. 5F). Strikingly, the steroidogenesis pathway, including genes such as Igf1r, Star, Cyp11a1, and Cyp17a1, was downregulated in Leydig cells of Eif2s3y, Uty, and AZFa deletants (Fig. 5G, Fig. S10C-E). Because production of steroid hormones like testosterone is crucial for germ cell homeostasis (41), it is possible that the observed germ cell defects are, at least partially, caused by dysfunctional somatic cells, which could be investigated in future endocrine studies (41).

We also found that multiple factors regulating meiotic synapsis and recombination, including Dmc1, Mnd1, and Rad51ap2 were downregulated in Zfy2-deficient testes, implicating these factors in the X-Y pairing phenotype (Fig. 5H, Fig. S10F). Comparing DE genes in all our Zfy deletants allowed us to identify putative stage-specific shared and unique targets of ZFY1 and ZFY2 transcription factors (Fig. S10G, H, Supplementary Table 15). Zfy1&2-DKOs shared a higher number of DE genes with Zfy2 than with Zfy1 mutants, further indicating that Zfy2 is the main contributor to the Zfy1&2-DKO phenotypes. Surprisingly however, most DE genes in the Zfy1&2-DKOs were not shared with either Zfy1 or Zfy2 mutants. This indicated that the effects of deleting both Zfy genes are not merely additive, but also combinatorial.

Discussion

Here, we generated a deletion series of the non-ampliconic Y genes and used an exhaustive reproductive phenotyping pipeline to provide a comprehensive overview of Y-gene functions in spermatogenesis. We also generated a transcriptomic atlas of Y-mutant testes, revealing how Y genes regulate testis gene expression and providing a resource to the research community. We demonstrated that more Y genes are required for normal murine spermatogenesis than previously known, and they function in more spermatogenic processes than previously thought.

As shown by others (1, 2, 42), our comparative analyses across species and within the human population revealed that all the broadly conserved Y-chromosome genes we studied are subject to purifying selection, regardless of whether their deletion causes spermatogenic defects (Fig. S11A, B). While some Y genes are paramount for normal spermatogenesis, others have more subtle effects, which are revealed when deleted in combinations. This provides insights into human infertility. Most known cases of human Y chromosome-related infertility are caused by either AZFa, AZFb, or AZFc deletions (28). Since these deletions remove multiple Y genes, identifying which is causative remains a challenge in clinical genetics (43). Our work raises the possibility that the etiology of some infertility cases may result from the cumulative loss of several Y genes, each of which only having a minimal impact on spermatogenesis. Moreover, Y-mutations may sensitize individuals to environmental insults and autosomal mutations. Subtle Y-gene functions may be shared with their X or autosomal homologues, and presence of these homologues may be sufficient to prevent severe defects in some Y-deletants. Variation in the extent to which homologues can compensate for Y-mutations could explain difference in phenotypes between mice and humans, as seen when AZFa is deleted.

Some Y genes are also expressed outside of the testis (Fig. 1A) and potential roles in non-reproductive organs may contribute to their evolutionary survival. A rapidly increasing volume of research supports this, linking Y chromosome loss to ageing and a wide range of conditions including cancer and heart disease (44–49). Our Y-deletant mouse models present a valuable resource with which to unravel these phenotypes at the single gene level.

Materials and Methods

Mouse colony maintenance

Mouse lines (Mus musculus) were maintained under UK Home Office Regulations, UK Animals (Scientific Procedures) Act 1986, and according to ethical guidelines at the Francis Crick Institute. Permission for animal experiments was granted by The Crick Biological Research Facility Strategic Oversight Committee (BRF-SOC) incorporating Animal Welfare and Ethical Review Body (AWERB) (Project Licence P8ECF28D9). Mice were kept as previously described (50). Generated Y-deletant F0s were backcrossed to C57BL/6J females. For Eif2s3y-KO and Zfy1&2-DKO analyses, only F0 animals were used, as they could not reproduce. F0 animals were 50% C57BL/6J and 50% CBA (see below). For the other genotypes including controls, F1 animals were used (75% C57BL/6J and 25% CBA), except for IVF assays and litter size/sex ratio analyses, for which later (more backcrossed) generations were also used. Adult males were processed at ages between 13-15 weeks.

mESC derivation and maintenance

We wanted the mESC line used for targeting to satisfy three requirements: 1) it should carry a reporter for easy quantification of ESC contribution in founder, 2) the Y chromosome should originate from a C57BL/6J background, and 3) it should possess high developmental potential. First, a B6D2F1-Tg(CAG/Su9-DsRed2, Acr3-EGFP RBGS002Osb) (51) female was crossed to a C57BL/6J male and the line backcrossed to C57BL/6J for six generations. A resulting male carrying the transgenes was crossed to a CBA female, and the resulting blastocysts used for mESC derivation as previously described (50). The resulting mESCs were therefore CBAB6F1, with a C57BL/6J Y chromosome and carrying a mitochondrial DsRed2 and acrosomal GFP reporters. All mESC lines were maintained in 2i/LIF conditions on laminin-coated plates as previously described (52, 53).

Preparation of the CRISPR-Cas9 components

sgRNAs were designed using the online tool CRISPRdirect (54). sgRNAs with predicted low off-target activity were chosen (Supplementary Table 16). Additionally, for each selected sgRNA, potential off-target sites were identified using the Cas-OFFinder tool (55) (Supplementary Table 17). sgRNA oligonucleotides (synthesized by Eurofins Genomics, Germany) were annealed, and then ligated into the targeting plasmid px333-puro at the BbsI or BsaI (NEB, USA) restriction enzyme sites. The px333-puro vector was engineered by cloning a puromycin resistance cassette from px459v2 (gift from Feng Zhang, Addgene #62988) into the px333 vector (gift from Andrea Ventura, Addgene #64073) using standard directional cloning with FseI and NotI restriction enzymes (NEB, USA) (Fig. S1B).

Generating the Y-deletant mESCs

150,000 cells were seeded onto a laminin-coated well of a 6-well plate. On the following day, mESCs were transfected. First, 2.5 μg of px-333-puro plasmid (encoding sgRNAs) was added to 100 μl of Opti-MEM (ThermoFisher Scientific, USA), followed by 10 μl of FuGENE HD (Promega, USA) reagent. The solution was then mixed, incubated at room temperature for 10 minutes and then pipetted onto the cells. Two days post transfection, cell medium was changed to 2i+LIF containing puromycin (1.8 μg/ml). Cells were kept in selection medium for 48 h, then put back into 2i+LIF for 4 days of recovery (Fig. S1B). Individual surviving colonies were picked manually into the 96-well format and expanded. DNA extracted from individual clones was used for PCR screening. PCR reactions were carried out using the Q5 High-Fidelity DNA polymerase kit (New England Biolabs (NEB), USA) in a total volume of 25 μl following manufacturer’s instructions. Primers (Supplementary Table 16) were designed using Primer3 (http://bioinfo.ut.ee/primer3/), synthesised by Eurofins Genomics (Germany), and used at the final concentration of 10 μM. Agarose gel electrophoresis was used to confirm correct PCR amplification. Clones with a detectable deletion band and no exonic and boundary bands were amplified, frozen down and submitted for MiSeq and low pass WGS (Fig. S1A, D, Supplementary Table 18).

Hyperovulation for oocyte and embryo collection

Hyperovulation was used instead of superovulation to retrieve a higher number of oocytes or embryos, as previously published (56). Female mice were administered 3.75 IU pregnant mare serum gonadotropin (PMSG) + 0.1 ml inhibin antiserum (CARD HyperOva, KYD-010-EX-X5, 2BScientific, UK). 48h later, mice were administered 7.5 IU human chorionic gonadotropin (hCG).

Tetraploid aggregations

The F0 mice were produced using tetraploid CD1 host embryos, which do not contribute to the embryo proper. The method allows the generation of F0 mice entirely derived from the Y-KO ES cells used during aggregation. All culture media were made up in house according to the EMMA cryopreservation methods [https://www.infrafrontier.eu/emma/cryopreservation-protocols/#Consumables]. 2-cell embryos from plugged hyperovulated CD-1 mice were harvested and transferred into a culture dish with potassium supplemented simplex optimised medium (KSOM) micro drops. They were then transferred to Flushing and Holding Media (FHM), then to 0.3 M mannitol. The embryos were then aligned into the fusion chamber containing mannitol, in between CUY5001P1 fusion electrode (Nepa Gene, Japan). The embryos were then pulsed using the Nepa ECFG21 (Nepa Gene, Japan) with a 10-volt 30 second AC pulse, followed by a 150-volt 30 μs DC pulse, with one 0.1 s interval and 10% decay and positive polarity. Embryos were then washed through three drops of FHM, a dish of KSOM, and cultured in a dish. Embryos that had successfully fused one hour post pulse were cultured until day 2. Acid Tyrode’s solution (Sigma, T1788) was used to remove the zona pellucida, and the 4-cell stage embryos were then washed in 3 drops of FHM. Denuded embryos were washed in KSOM and individually placed in aggregation wells. Clumps containing 8-15 mESCs in KSOM were then added to the aggregation wells containing half of the individual embryos. The remaining half were individually deposited into the aggregation wells, “sandwiching” the mESCs in between embryos. Aggregated embryos were cultured overnight. On day 3, aggregated embryos at the blastocyst stage were washed and transferred into E2.5 pseudopregnant CD-1 females via uterine transfer. Overall, we achieved an average birth rate of 15.47% and average survival rate to adulthood of 85.39%.

IVF assays

On day one, sperm was harvested by nicking the apex of one cauda epididymis and transferring the sperm into a 90 μl drop of sperm capacitation media (Toyoda, Yokoyama and Hosi medium supplemented with methyl-beta-cyclodextrin (TYH+MBCD)), which was made in house according to the EMMA cryopreservation method [https://www.infrafrontier.eu/emma/cryopreservation-protocols/#Consumables]. The sperm solution was incubated for 30 minutes at 37°C. In the meantime, oocytes from B6CBAF1 hyperovulated females were collected and incubated into 200 μl drops of fertilization medium (0.25 mM glutathione in Human Tubal Fluid medium (HTF)) for maximum 30 minutes. 3-5 μl of sperm was added into the fertilization medium drops containing oocytes. 3 to 4 hours later, oocytes were washed in 4 drops of HTF, and the presumptive zygotes cultured overnight. On day two, the percentage of 2-cell embryos was assessed.

Testis histology

Mouse testes were dissected, weighed, and fixed in Bouin’s solution (Sigma, UK) overnight at room temperature. The next day, they were washed in distilled water and stored in 70% EtOH at room temperature. Fixed testes were embedded in wax, sectioned transversely at 4 μm, and stained with Periodic Acid-Schiff (PAS). Sections were imaged using an Olympus VS120 Slide Scanner with a 40X objective. Analysis of testis histology was blinded.

Immunostaining of testis sections

Testes were fixed in 4% paraformaldehyde (PFA) overnight at 4°C, washed in 70% EtOH, embedded in paraffin, and sectioned at 4 μM on slides. Slides were incubated at 60°C for 10 minutes, then washed in Histo-Clear (National Diagnostics, USA) twice for 5 minutes. Samples were then rehydrated through ethanol washes (2x 100% ethanol 1 minute, 2x 95% ethanol 1 minute, once in 75% ethanol 1 minute, and once in distilled water 1 minute). Antigen retrieval was performed by boiling the slides in 0.01 M Sodium Citrate for 10 minutes and then letting the solution cool for 15 minutes at room temperature. Sections were blocked in 5% fetal bovine serum (FBS) in PBS with 0.01% Triton at room temperature for one hour. For apoptosis quantification, slides were incubated with rabbit anti-cleaved poly-ADP-ribose polymerases (cPARP), (1:200, ab32064, Abcam) in a humidified chamber at 37°C for 2 hours. Samples were incubated in the secondary AlexaFluor 647 anti-rabbit antibody (1:200, ThermoFisher) at room temperature for 1 hour. For spermatogonia quantification, slides were incubated with LIN28A (1:200, ab63740, Abcam) and goat cKIT (1:100, AF1356, R&D systems) at 4°C overnight. Samples were incubated in the secondary AlexaFluor 647 anti-goat and AlexaFluor 594 anti-rabbit antibodies (1:250, ThermoFisher) at room temperature for 1 hour. Washes were performed with 0.01% PBS-Triton. Sections were counterstained with the DNA-binding fluorescent stain DAPI (D9564, Sigma-Aldrich).

Immunostaining of testis nuclear spreads

Nuclear spreads were prepared from frozen samples, except for MLH3 quantification, where fresh tissue was used. For frozen samples, a piece of testis was thawed in a drop of PBS and chopped into a suspension using 2 scalpels. One drop of the testis suspension was dropped onto a pre-boiled slide from 15-20 cm height. One drop of 0.05% Triton X-100 (dissolved in distilled water) was added to each drop and left for 10 minutes. 8 drops of 2% formaldehyde, 0.02% Sodium dodecyl sulfate (SDS) in PBS were then added and slides left incubating at room temperature in a humidified chamber for one hour. Slides were then quickly dipped in distilled water 6 times and air-dried for 5 minutes. Slides were either used for immunostaining straight away or stored at -80°C.

Samples from fresh tissues were prepared by adapting a previously published protocol (57). A small piece of testis was incubated in 2.5 mL dissociation buffer (0.25% Trypsin, 20 μg/μl DNAse I in PBS) at 37°C for 15 min shaking at 250 rpm. The tissue was further dissociated by pipetting for ~2 min using a wide-mouthed Pasteur pipette, and the reaction quenched by adding 150 μl of FBS. Cells were moved through a 70 μm cell strainer and pelleted by spinning at 1000 g for 5 min. The pellet was resuspended in 15 mL of PBS with 20 μg/ml DNAse I then further centrifuged at 1000 g for 5 min. This was repeated for a total of 3 washes and cells finally resuspended in PBS (~200 μl of PBS per 100 mg of tissue). 10 μl of suspension was added into a tube containing 90 μl of 75 mM sucrose in water. 45 μl of the dissociated cells was added to a slide, inside a square drawn with a lipophilic marker containing 100 μl fixing solution (1% PFA, 0.15% triton, in PBS). Slides were incubated in a humidified chamber overnight. Slides were dried, rinsed in dH₂O, washed 1min in PBS 0.4% Kodak PhotoFlo 200, air-dried, and stored at -80°C.

For immunostaining, slides were incubated in PBT (PBS, 0.15% BSA, 0.1%Tween-20) for one hour at room temperature. The primary antibodies diluted in PBT were then added to the slides and incubated overnight at 37°C in a humidified chamber. The following antibodies were used: guinea pig SYCP3 (in house) at 1:100, human CREST (gift from W. Earnshaw) at 1:100, mouse γH2AX (05-636; RRID: AB_309864, Millipore) at 1:250, rabbit MLH3 (gift from Paola Cohen) at 1:500. The slides were washed in PBS for 5 minutes 3 times. The secondary antibodies (Alexa Fluor 488, Alexa Fluor 594 and Alexa Fluor 647) diluted in PBS (1:250) were then added to the slides and incubated for 1 hour at 37°C. Slides were washed for 5 minutes 3 times and mounted in Vectashield plus DAPI (4’,6-diamidino-2-phenylindole) (Vector Laboratories, USA). Stained testis spreads were imaged using an Olympus delta vision or Zeiss Observer microscope using 40X or 63X objectives.

Chromosome painting

Testis nuclear spreads or testis sections (see previous two sections) were washed twice in PBS for 5 minutes and dehydrated through ethanol series, 70% ethanol twice, 90% ethanol twice, 2 minutes each, and 100% ethanol once for 5 minutes. Slides were then washed three times in 2xSSC for 5 minutes and denatured in 2xSSC at 80°C for 5 minutes. Slides were quenched in ice-cold 70% ethanol and a second dehydration step performed as described above. The X (XMP X green, D-1420-050-FI) and Y (XMP Y orange, D-141421-050-OR) chromosome paints (Metasystem, Germany) were added to the slides in equal volumes (10-20 μl total), a coverslip added, and the slides incubated, first at 75°C for 5 minutes and then at 37°C overnight in a humidified chamber. Slides were washed in 2xSSC at room temperature, then in 0.4xSSC at 72°C for 2 minutes, then in 2xSSC with 0.05% Tween-20 at room temperature for 30 seconds. Slides were rinsed in distilled water and stained with DAPI for 10 minutes.

Sperm morphology analyses

To extract spermatozoa, three incisions were made on the cauda epididymis, which was then dropped into a tube containing 200 μl of pre-warmed TYH+MBCD and incubated in a 37°C 5% CO₂ incubator for 8 minutes. Using tweezers, the cauda was then removed from the tube and additional 200-400 μl of warm TYH+MBCD was added. After gently mixing by pipetting with a wide-bore tip, 50 μl of sperm solution was transferred drop-by-drop to a new tube containing an equal volume of 4% PFA (final concentration of 2% PFA). After 5 minutes of incubation at room temperature, spermatozoa were centrifuged at 4°C for 3 minutes at 300 g. The sperm pellet was resuspended in a staining solution (0.1% polyvinyl alcohol (PVA), Hoechst (1:1000, ThermoFisher, 62249), Phalloidin (1:250, ThermoFisher, 12381), Triton (0.1%, Sigma, 93443) in PBS). The sperm solution was then pipetted onto SUPERFROST slides (Fisher Scientific, UK) (15 μl per slide) and kept sealed at 4°C. With a Nikon LTTL 3 microscope, micromanager, and a Python script, a strategy to automatically map and image individual sperm heads was designed. First, slides containing fixed spermatozoa were scanned at the 20X magnification on the DAPI channel and positions of individual sperm heads were recorded. Switching to a Nikon Plan Apo 100x/1.4 objective, the saved locations (sperm heads) were then automatically imaged with a z-stack. Z-stacks were then analyzed using an ImageJ script to identify the best focused image in the stack and to normalize the exposure settings across files. Sperm head images were imported into the Nuclear_Morphology software for morphology analysis (58). Analysis settings were left to default except for “flattening threshold” which was changed to “120”. After detection, identified sperm heads were manually curated to exclude any poorly outlined sperm heads from the analysis.

Sperm motility analyses

After isolation (see sperm morphology analysis), 3 μl of sperm solution was loaded into a Leja 20 μm 4 chamber slide (IMV Technologies, France). The slide was kept on a portable warming stage (HS-PT-USB, Microptic, Spain) set to 37°C. Once on the slide, the sperm solution was let to settle for 2 minutes before analysis. Altogether, sperm motility analysis was performed using the SCA Motility and concentration TOX edition (SCA-TOX-01, Microptic, Spain). Analysis started 15 minutes after dissection of the cauda epididymis. Set parameters were used (area min = 20, area max = 200, drifting = 0, progressive STR> 70, connectivity = 25, VAP points = 9, VCL/VAP = VCL, heads points restriction min and max = 35 and 2000) and at least 20 video fields captured throughout the slide to obtain data for at least 300 spermatozoa. After capture, each field was inspected manually to delete any sperm-tracking errors or duplicated sperm tracks.

Sperm counts

Cauda epididymis were dissected and shredded in pre-warmed TYH+MBCD medium (usually 1.1 ml) to extract all the sperm. Samples were incubated in a 37°C 5% CO₂ incubator for up to 30 minutes. 3 μl of sperm solution was loaded into a Leja 20 μm 4-chamber slide and the total cauda sperm count was estimated using the SCA software.

MiSeq

MiSeq was used to characterize on-target deletion outcome and to screen for potential off-target editing. PCR amplicons were purified using SPRI bead clean, and library prepared according to the Illumina MiSeq library prep manufacturer’s instructions (Nextera Index Kit V2). Samples first underwent PCR indexing and libraries were then normalized and pooled after fluorometric quantification using a dsDNA dye. Library were sequenced using the Illumina MiSeq platform with a PE 250 bp run configuration on a Nano flowcell. An average of 2000-5000 reads were obtained per sample. After sequencing, the Fastq sequence files were collapsed using the FastX Toolkit (v0.0.13) [https://github.com/agordon/fastx_toolkit]. To determine the whole-locus deletion outcome, the collapsed MiSeq reads were aligned to the reference mouse genome (Mm10) using blastn (Fig. S1C, G). To examine potential off-target editing, the collapsed MiSeq reads were aligned using Burrows-Wheeler Alignment tool (BWA, v0.7.170) (59) and analysed using the R package CrispRVariants (60) (Fig. S1h). This was used to calculate the proportion of wild type reads (perfect match to Mm10) around the off-target site compared to the sum of all other reads containing single nucleotide variants ± 20 bp away from the sgRNA binding site.

Low pass whole genome sequencing

Low pass WGS was used to confirm whole-locus deletion and to ensure that the rest of the Y chromosome, as well as the X homologues were intact. It also allowed to characterize the karyotype of each cell line. After genomic DNA purification, samples were prepared with the Illumina DNA Prep with Enrichment kit following the Nextera flex protocol. Coverage of 0.1X-0.3X was obtained on average. Resulting FastQ reads were aligned to the reference genome using BWA (Fig. S1E). The R package QDNAseq (61) was used for karyotype analysis (Fig. S1F).

Bulk RNAseq

About 20 mg of frozen testis pieces were placed in Precellys Homogenizer tubes 2.8 mm ceramic reinforced (Precellys 50722019, Bertin Technologies, France) on dry ice. 500 μl of Trizol reagent (Sigma-Aldrich, USA) was added to the tubes and the samples placed in a Bertin Precellys 24-Dual High-throughput Tissue Homogenizer (Bertin Technologies, France). The machine was then run with the factory programme 5: 6,500 rpm for 20 seconds, automatic 25 seconds of pause, and a repeat of 6500 rpm for 20 seconds. The lysate was then centrifuged for 5 minutes at 12,000 g at 4°C and the clear supernatant transferred to a MaXtract High Density phase lock tube (Qiagen, UK). After 5 minutes of incubation at room temperature, 100 μl of chloroform was added and tubes shaken vigorously for 2-3 minutes. Tubes were centrifuged for 15 minutes at 12,000 g at 4°C. The upper aqueous phase was transferred to a new tube and 250 μl of isopropanol added and mixed by pipetting. After 10 minutes of incubation at room temperature samples were centrifuged for 10 minutes at 12,000 g at 4°C. The supernatant was discarded, the RNA pellet washed with 1 ml of 75% ethanol, and the samples centrifuged for 5 minutes at 7,500 g at 4°C. Finally, the pellet was air-dried for 5-10 minutes and resuspended in 40 μl of nuclease-free water. Samples were checked using a TapeStation (Agilent Technologies, USA). Libraries were prepared using the NEBNext Ultra II Directional PolyA mRNA kits according to manufacturer instructions. Raw RNA-seq reads were processed using the RNA-seq nf-core pipeline (v3.14), star_rsem was used to generate raw reads counts. The read counts were processed in R using the DESeq2 (v1.34) package. Very lowly expressed genes were filtered out by applying a rowSums filter of >=5 to the raw counts table. Principal component analysis (PCA) plots were generated using the top 500 most variable genes, applying the vst() function and limma::removeBatchEffect() to remove batch effects for PCA visualization. Raw counts were batch corrected and normalized using the DESeq() function, specifying “~batch + Sample_Genotype” in the design formula. Log2 fold change (log2FC) and adjusted p-values between KO and WT were calculated using the results() function in DESeq2, specifying the “BH” (Benjamini-Hochberg p-value adjustment). Log2FCs were shrunk using the lfcShrink(type=“apeglm”) function. Differentially expressed genes (DEGs) were defined as genes which had an adjusted p-value < 0.05 and log2FC > 0.5 (Supplementary Table 1). Enrichment of genomic functions and cellular processes was done using the gseGO() function, as part of the R package, clusterProfiler (v4.2.2) (Supplementary Table 2).

Single nuclei RNAseq

For Eif2s3y and Uty deletants, P10 testes were used because, unlike adult testes, they are enriched in spermatogonia, the cell type affected in these mutants. Moreover, for Eif2s3y, we did not want to capture potential secondary effects of the total spermatogonia arrest. For Zfy2, Zfy1-KO, Zfy1&2-DKO, and AZFa-KO, which have all germ cell types, we sequenced nuclei from adult testes.

Testis nuclei isolation was adapted from a previously published protocol (37). On ice, around 10 mg of frozen testis was homogenized in 1 ml of lysis buffer (250 mM sucrose, 25 mM KCl, 5 mM MgCl₂, 10 mM HEPES pH 8, 1% bovine serum albumin (BSA), 0.1% IGEPAL and freshly added 1 μM DTT, 0.4 U μl−1 RNase Inhibitor (New England BioLabs), 0.2 U μl−1 SUPERasIn (ThermoFischer Scientific)) and incubated on ice for 4 minutes. The lysate was filtered through a 30 μm filter and then centrifuged at 500 g for 5 minutes at 4°C. The supernatant was removed, and the pellet resuspended in 1 ml of wash buffer 1 (250 mM sucrose, 25 mM KCl, 5 mM MgCl₂, 10 mM HEPES pH 8, 1% BSA, and freshly added 1 μM DTT, 0.4 U μl−1 RNase Inhibitor, 0.2 U μl−1 SUPERasIn). This was centrifuged at 500 g for 5 minutes at 4°C and washing step repeated. Next the pellet was resuspended in 670 μl of freshly diluted dithio-bis (succinimidyl propionate) (DSP) (1 mg/ml in PBS) and incubated at room temperature for 30 minutes. Fixation was quenched by adding 14.07 μl Tris-HCL, pH 8. Nuclei were centrifuged at 500 g for 5 minutes at 4°C and resuspended in 1 ml of wash buffer 2 (250 mM sucrose, 25 mM KCl, 5 mM MgCl₂, 10 mM Tris-HCL, pH 8, 1% BSA, and freshly added 1 μM DTT, 0.4 U μl−1 RNase Inhibitor, 0.2 U μl−1 SUPERasIn). The samples were centrifuged one more time at 500 g for 4 minutes at 4°C and the pellet resuspended in 0.04% BSA in PBS.

Replicates from the P10 and adult 10x nuclei datasets were aligned to the mm39 genome using STARsolo (v2.7.9), specifying the parameters –soloMultiMappers EM –outFilterScoreMin 30 -- soloFeatures Gene GeneFull --soloType CB_UMI_Simple --soloFeatures Gene GeneFull –soloUMIlen 12 --soloCBmatchWLtype 1MM_multi_Nbase_pseudocounts --soloUMIdedup 1MM_CR. Empty droplets were separated from nuclei using the proportion of reads mapping to intronic reads (Supplementary Table 19). Replicates were then processed in the Seurat R package (v4.3.0). Cells were retained if they had <5% reads aligning to mitochondrial and rRNA genes, if the total number of molecules within a cell was between 1000 and 50000 and the number of genes detected per cell was between 500 and 10000. The data were normalized using NormalizeData, then FindVariableFeatures. The dataset was scaled using ScaleData. Principle Component Analysis was applied using RunPCA followed by FindNeighbours and FindClusters. Uniform Manifold Approximation and Projection was estimated for the integrated dataset using RunUMAP. Doublets were identified using the DoubletFinder R package (v2.0.3) (Supplementary Table 19). Replicates from each KO were integrated together, then the integrated KOs from their respective P10 and adult datasets were integrated with each other using SelectIntegrationFeatures, FindIntegrationAnchors, and IntegrateData, generating one integrated dataset for the P10 10x data, and another for the adult 10x data. Each dataset was processed using ScaleData, then RunPCA and RunUMAP. Clusters were identified using a 0.5 resolution for both the adult and P10 datasets. In the adult dataset, ambient RNA clusters were identified using the FindAllMarkers function, and subsequently removed. Clusters were annotated using published marker genes(37, 38). Differentially expressed genes were identified between wildtype control and KOs on a cluster-by-cluster basis using FindMarkers (Supplementary Tables 3 to 8). Enrichment of genomic functions and cellular processes was done using the gseGO() function, as part of the R package, clusterProfiler (v4.2.2) (Supplementary Tables 9 to 14).

dN/dS and LOEUF scores

The ratios of nonsynonymous to synonymous substitutions in protein-coding genes (dN/dS) were calculated for deeply conserved mouse Y genes. This was done to compare mouse, rat, chimp and human sequences to each other. dN and dS were identified using biomaRt R package (v2.56.1). The function useEnsembl was used to connect to the mouse, human, chimp and rat Ensembl databases, specifying archived Ensembl version 99. dN and dS scores were extracted using the getBM() function. The dN/dS ratio was then calculated by dN over dS. Loss-of-function observed/expected upper bound fraction (LOEUF) scores were used to assess Y-genes’ intolerance to loss-of-function variation in humans. The scores were directly taken from the gnomAD v2.1 browser.

Data and materials availability

Bulk RNAseq and single nuclei RNAseq data have been deposited at GEO (accession number GSE274100 and GSE274095 respectively). Supplementary Tables have been deposited on Dryad under (DOI): doi:10.5061/dryad.0p2ngf295.