Proper timing of a quiescence period in precursor prospermatogonia is required for stem cell pool establishment in the male germline

Tissue-specific stem cells that support continual cycling of cell lineages through both self-renewal and production of differentiating progenitors are derived from precursors formed during embryonic, perinatal and neonatal development. These precursors progress through a diverse path of developmental milestones while gaining the capacity to support cell lineages long into adulthood. During development, precursors colonize different microenvironments, undergo epigenetic programming and dynamically alter proliferative states (Borgel et al., 2010; Saitou and Yamaji, 2012). A hallmark example of the latter is the perinatal mitotic arrest period of prospermatogonial precursors that occurs prior to formation of the undifferentiated spermatogonial population in which stem cell activity for the spermatogenic lineage is contained (Oatley and Brinster, 2008).

The subset of undifferentiated spermatogonia that possess stem cell activity is often referred to as spermatogonial stem cells (SSCs) and supports continuous spermatogenesis in adulthood (de Rooij and Russell, 2000; Oatley and Brinster, 2008). During development, SSCs originate from primordial germ cells (PGCs) that are induced from the proximal epiblast via WNT and bone morphogenetic protein signaling at embryonic day (E) 6.25 in mice (Lawson et al., 1999; Rivera-Perez and Magnuson, 2005). This signaling induction leads to the upregulation of Blimp1 (also known as Prdm1), Prdm14 and Tfap2c, which encode transcription factors that facilitate PGC specification, migration, and epigenetic reprogramming (Kurimoto et al., 2008; Magnusdottir et al., 2013; Yamaji et al., 2008). In subsequent developmental days, PGCs proliferate and migrate to the genital ridge (Hamer and de Rooij, 2018; Richardson and Lehmann, 2010). At E12.5 in mice, gonadal PGCs undergo sex determination in response to cues from the developing soma (Lin and Capel, 2015). If XX, germ cells initiate oogenesis and arrest in meiosis by E14.5 and remain in this state until puberty. By contrast, XY PGCs transition to form prospermatogonia, which continue to proliferate for several days before entering mitotic arrest around E16.5 (Saitou and Yamaji, 2012; Vergouwen et al., 1991). In the mouse, prospermatogonia progressively re-enter the cell cycle from postnatal day (P) 1.5-3.5 (Law et al., 2019). During this neonatal window, prospermatogonia transition to either from the foundational SSC pool that supports lifelong spermatogenesis or enter a differentiation pathway directly and contribute to a unique first round of spermatogenesis (Bellve et al., 1977; Law et al., 2019). Collectively, a variety of developmental milestones and dynamic cell-type transitions precede formation of the SSC population.

The fetal mitotic arrest and postnatal cell cycle re-entry periods of prospermatogonia are well-described developmental events, but the functional significance and key regulators of these processes have not been elucidated. During this period that precedes genesis of the spermatogenic lineage, prospermatogonia are known to re-establish DNA methylation patterns (Li and Sasaki, 2011), dynamically regulate histone modifications (Ly et al., 2015), migrate from the center of seminiferous cords to the basement membrane (Nagano et al., 2000), and preprogram postnatal SSC or differentiating fates (Law et al., 2019). At present, whether these processes and establishment of the spermatogenic lineage are tied to the mitotic arrest period or are developmentally independent is undefined.

Here, we used mouse models for conditional ablation of the master cell cycle regulator retinoblastoma transcriptional corepressor 1 (RB1) in vivo as a model system to explore the purpose of the prospermatogonial mitotic arrest period. Previous studies using ex vivo organ culture of testes from Rb1 null mouse fetuses indicate that PGC and prospermatogonial development are grossly normal but there is a delay in onset of the mitotic arrest (Spiller et al., 2010). As the previous studies did not create live animals with germ cell Rb1 deficiency, ramifications of the delayed onset of mitotic arrest on establishment of the spermatogenic lineage and fertility in postnatal life remain undefined. The current study extends the foundational observations of Spiller et al. (2010) to show that conditional inactivation of Rb1 in germ cells starting at the PGC stage leads to impaired perinatal mitotic arrest of prospermatogonia and subsequent ablation of the germline in vivo, thereby causing sterility in postnatal life. In addition, outcomes of single-cell RNA-sequencing (scRNA-seq) analysis revealed that a metabolic shift occurs in prospermatogonia during the perinatal quiescence period that is disrupted in cells that fail to make this transition. Also, inhibition of meiotic entry is impaired in prospermatogonia that do not initiate mitotic arrest. Collectively, these findings suggest that the purpose of the mitotic arrest period in prospermatogonia is to set programming for the undifferentiated stem cell fate in postnatal spermatogonia.

RB1 is well known as a repressor of cell cycle progression in many cell types and its actions are inhibited by post-translational phosphorylation (Burke et al., 2012; Dick and Rubin, 2013). To explore the relationship of RB1 abundance as an active and inactive form and proliferation status of prospermatogonia, we used co-immunofluorescence staining for phosphorylated RB1 at serine residues 807/811 (designated p-RB1), a marker of active proliferation (Ki67) and the germ cell-specific marker TRA98 on cross-sections of testes from fetal and neonatal wild-type mice (Fig. 1). Co-staining revealed that RB1 is present in prospermatogonia and somatic cells throughout development, consistent with previous reports (Spiller et al., 2010; Western et al., 2008), but p-RB1 is detectable in prospermatogonia during the period of proliferation from E12.5 to E13.5 and becomes low to undetectable at E14.5-P1.5, which is coincident with the period of mitotic arrest. Thereafter, staining for p-RB1 increased by P6.5 when establishment of spermatogonial subsets, including the foundational SSC pool, occurs (Law et al., 2019). Taken together, these observations indicate that RB1 is active as a cell cycle repressor coincident with the period of mitotic arrest of prospermatogonia during late fetal and early neonatal development.

Previous studies showed that conditional inactivation of Rb1 in prospermatogonia after the initiation of quiescence led to loss of the germline following one or two rounds of spermatogenesis in postnatal life, implying breakdown in the SSC pool (Hu et al., 2013; Yang et al., 2013). However, the ramifications of impaired cell cycle regulation caused by Rb1 inactivation in prospermatogonia prior to initiation of quiescence during fetal development was not explored. To address this gap in knowledge, we used the Cre-Lox approach by backcrossing mice harboring Rb1 floxed alleles and a Blimp1-Cre transgene to generate a model of conditional Rb1 inactivation at the PGC stage, which is prior to prospermatogonial formation and the onset of quiescence, designated hereafter as Rb1-cKO^Blimp1 (Fig. 2A). The Blimp1-Cre transgene is known to initiate expression starting at E6.5 and the efficiency of recombining floxed genes has been measured at >75% (Li et al., 2015; Ohinata et al., 2005). Rb1-cKO^Blimp1 mice were born at the expected frequency and postnatal development was grossly normal with a lifespan that was similar to that of wild-type mice. As early as P14.5, testis weight of Rb1-cKO^Blimp1 males was significantly (P≤0.05) reduced to 68.8% of that of control mice with one or two functionally intact Rb1 alleles. (Fig. 2B), and these animals were sterile at all ages examined from P45 to P90 (Fig. 2C), demonstrating that, unlike males with conditional inactivation of Rb1 in prospermatogonia after initiation of quiescence, even a first round of spermatogenesis does not occur. In accordance, assessment of testis cross-sections at P35, when the first round of spermatogenesis normally completes, and P60-180, when steady-state spermatogenesis has normally commenced, confirmed the cause of sterility as germline ablation in Rb1-cKO^Blimp1 mice (Fig. 2D). In addition, body weights for Rb1-cKO^Blimp1 mice were not different (P≥0.05) to controls (Fig. S1A) and copulatory plugs were observed at all ages examined, confirming that mating had occurred. Furthermore, Rb1-cKO^Blimp1 females were fertile well into adulthood despite significantly (P≤0.05) smaller litter sizes compared with controls (Fig. S1B), similar to the phenotype of female mice with germline conditional inactivation of Rb1 after sex determination, which may be accounted for by dysregulation of follicle growth (Yang et al., 2013, 2015), and/or uterine deficiencies (Goolam et al., 2020). Collectively, these findings suggest that proper cell cycle regulation during the prospermatogonial stage of development to ensure correct timing of mitotic arrest is crucial for establishment of the SSC pool and laying of the foundation for continuity of the spermatogenic lineage in postnatal life.

In considering the phenotype of mice with conditional inactivation of Rb1 either prior to or after onset of prospermatogonial quiescence, we were intrigued that disruption in the normal timing of mitotic arrest (i.e. Rb1-cKO^Blimp1 males) led to germline ablation throughout all of postnatal life, but inactivation after initiation of quiescence (generated by backcrossing Rb1-floxed and Ddx4-Cre transgenic mice) did not disrupt the first round of spermatogenesis (Hu et al., 2013; Yang et al., 2013). These findings suggested that lack of RB1 per se does not impair prospermatogonial survival, but disruption of proper initiation of mitotic arrest is crucial. In global Rb1-null mice, germline development appears phenotypically normal up to E12.5 with the number of PGCs and prospermatogonia being unaltered, at which point embryonic death occurs (Spiller et al., 2010). We aimed to validate and extend these findings in vivo using Rb1-cKO^Blimp1 males. In confirmation of previous results, both germ cell number and proliferation were found to be no different (P≥0.05) between control and Rb1-cKO^Blimp1 mice at E13.5 (Fig. S2A,B). However, based on positive immunostaining for the marker Ki67, 85.9% of prospermatogonia were found to be proliferating in Rb1-cKO^Blimp1 mice at E14.5, which was significantly greater (P≤0.01) compared with 59.1% of prospermatogonia in controls (Fig. 3A,B). This difference was even more striking at E16.5 when 65.2% of Rb1-cKO^Blimp1 prospermatogonia were Ki67⁺ compared with 1.9% in control mice (Fig. 3A,B). By E18.5, the percentage of proliferating prospermatogonia in Rb1-cKO^Blimp1 mice dropped to 4.9% but was still significantly greater (P≤0.05) compared with control mice (Fig. 3A,B). These findings demonstrate that lack of Rb1 altered the normal timing of mitotic arrest during fetal prospermatogonial development.

At E14.5, the number of prospermatogonia in Rb1-cKO^Blimp1 mice was not different compared with the normal situation in control mice but, consistent with elevated proliferation, prospermatogonial number at E16.5 was significantly (P≤0.05) greater in Rb1-cKO^Blimp1 mice compared with controls (Fig. 3C). Thereafter, the number of prospermatogonia in both control and Rb1-cKO^Blimp1 mice declined by E18.5, consistent with normal development, but the reduction in Rb1-cKO^Blimp1 was significantly (P≤0.05) greater compared with controls (Fig. 3C). Examination of testis cross-sections revealed that the percentage of seminiferous cords devoid of germ cells increased through developmental time in Rb1-cKO^Blimp1 mice beginning with 11.2% at E18.5, jumping to 87.2% at birth (i.e. P1.5), and eventually leading to germline ablation (i.e. >95%) by P3.5-6.5 (Fig. 3D,E). In contrast, no cords lacked germ cells at P6.5 in control mice (Fig. 3D,E). Based on co-immunostaining for TRA98 and cleaved CASPASE-3, the level of germ cell apoptosis was not different (P≥0.05) between control and Rb1-cKO^Blimp1 mice at E14.5 or E15.5 (Fig. 3F,G, Fig. S2C). However, by E16.5 44.2% of prospermatogonia were apoptotic in Rb1-cKO^Blimp1 mice, which was significantly greater (P≤0.05) compared with the 6.9% of prospermatogonia undergoing apoptosis in control mice (Fig. 3F,G). Collectively, these findings indicate that failure of prospermatogonia to enter quiescence on a normal developmental timeline leads to programmed cell death.

Considering that RB1 has diverse cellular functions beyond regulation of cell cycle progression, we next aimed to explore whether additional mechanisms underpin the impaired survival of prospermatogonia in Rb1-cKO^Blimp1 males. To achieve this, we performed scRNA-seq on E14.5 testes, which was the developmental age point when interruption of the normal entry point to quiescence occurred and apoptosis was not different between Rb1-cKO and control prospermatogonia (Fig. 3A,B,F,G), thus allowing for assessment of the root causes of impaired programming due to Rb1 deficiency rather than ramifications that would manifest later in development. To focus the analysis on prospermatogonia, we generated a multi-transgenic mouse model in which simultaneous inactivation of Rb1 and constitutive activation of tdTomato expression occurs in all PGCs (Fig. S3A). Prospermatogonia from E14.5 Rb1-cKO^Blimp1;tdTomato⁺ (n=3) and control (Blimp1-Cre^Tg/+;Rb1^fl/+;tdTomato⁺, n=3) mice were isolated by fluorescence-activated cell sorting (FACS) and single-cell mRNA libraries generated with the 10x Genomics platform followed by Illumina sequencing analysis. Notably, Rb1 transcripts were undetectable in all libraries generated for Rb1-cKO^Blimp1 prospermatogonia (Fig. S3B), thus demonstrating complete penetrance of the Cre-Lox approach. In addition, consistent with findings of Spiller et al. (2010) that examined Rb1 null prospermatogonia, expression of the Rb1 family members p107 (Rbl1) and p130 (Rbl2) in Rb1-cKO prospermatogonia was not different compared with Rb1-sufficient control prospermatogonia (Fig. S3B).

A total of 3560 and 4025 cells were captured from control and Rb1-cKO^Blimp1;tdTomato⁺ males, respectively (Fig. S3B,C). Transcriptomes were sequenced at an average depth of 49,332 and 41,976 reads per cell representing an average sequencing saturation of 51% and 54% for control and Rb1-cKO^Blimp1;tdTomato⁺, respectively. Averages of 17,149 and 17,237 genes were detected across pooled triplicate libraries for control and Rb1-cKO^Blimp1;tdTomato⁺ samples, respectively. Single-cell transcriptomes were then filtered based on quality control metrics (Fig. S3D,E), including selection for those with <300 genes detected to remove low-quality transcriptomes and >6.5% mitochondrial gene representation to remove dead cells, and exclusion of transcriptomes with >50,000 genes to remove potential doublets. Germ cells in each library were then defined based on expression of well-known specific marker genes, such as Ddx4 (Gallardo et al., 2007), Dazl (Nicholas et al., 2009) and Pou5f1 (Yoshimizu et al., 1999) (Fig. S3F). After filtering, a total of 5160 germ cells were analyzed further (2568 and 2592 in control and in Rb1-cKO^Blimp1;tdTomato⁺, respectively).

scRNA-seq analysis revealed highly consistent transcriptomic signatures of triplicate samples based on modified Pearson's correlation coefficients of 0.988-0.996 among control libraries and 0.990-0.995 among Rb1-cKO^Blimp1;tdTomato⁺ libraries. Based on the results above, which demonstrated the central role of Rb1 in cell cycle arrest in vivo, we first investigated whether gene expression profiles from scRNA-seq reflected in vivo measurements. To do so, CellCycleScoring analysis was performed between control and Rb1-cKO^Blimp1 samples, which predicted cell cycle phase based on relative gene expression signatures. This analysis revealed that a significantly greater (P≤0.05) proportion of prospermatogonia were in S/G₂/M in Rb1-cKO^Blimp1;tdTomato⁺ males compared with controls (Fig. 4A). The expression of Ki67 is generally upregulated across S/G₂/M (Miller et al., 2018); thus, cell cycle predictions from scRNA-seq are consistent with the in vivo Ki67 staining (Fig. 3).

Owing to the asynchronous nature of germ cell development and heterogeneity of gene expression in wild-type cells at the embryonic/fetal age points of E12.5-18.5 (Law et al., 2019; Nguyen et al., 2019, 2020), we first aimed to explore whether disruption of entry to mitotic arrest on a normal timeline alters transcriptome profiles throughout the prospermatogonial population. Using a method of comprehensive integration, we aligned the six libraries followed by uniform manifold approximation projection (UMAP) and graph-based clustering to capture germ cell heterogeneity at E14.5 (Fig. 4B,C). A total of five clusters were identified using shared nearest neighbor modularity optimization with a resolution of 0.2. Under this resolution, clusters were well-distinguished from each other based on differential gene expression (Fig. 4D; gene list in Table S1). Annotation of marker genes suggested that clustering was influenced by cell cycle phase. For example, Ccne1 and Rrm2, upregulated in cluster 0, function in the G₁/S transition (DeGregori et al., 1995), and Ube2c and Ube2s, upregulated in cluster 1, promote mitotic exit (Sivakumar and Gorbsky, 2015) (Fig. S3G). Consistent with differential gene expression analysis, CellCycleScoring predicted that 38.0% and 48.9% of germ cells in cluster 0 were in phases G₀/G₁ and S, respectively, whereas the cells in the cluster 2 were primarily in G₂/M (Fig. S3H). Taken together, clustering effectively defined germ cell heterogeneity such that differences between control and cKO would not be masked by population diversity.

Using the clustering approach described above, we then defined differentially expressed genes (DEGs) between control and Rb1-cKO^Blimp1;tdTomato⁺ males in the context of germ cell heterogeneity (see gene list in Table S2). Transcripts associated with cell cycle regulation were among the top Gene Ontology (GO) terms in clusters 0, 1, 2 and 3 (based on P≤0.01) (Table 1). In particular, genes associated with promotion of the G₁/S transition were upregulated in Rb1-cKO^Blimp1;tdTomato⁺ within each of the clusters (Fig. 4E). As a result, a greater proportion of germ cells transitioned from G₀/G₁ to S phase with Rb1 inactivation across all clusters based on CellCycleScoring (Fig. 4F). Collectively, these data indicate that prospermatogonia arrest in G₀/G₁ and in the absence of RB1 the G₁/S checkpoint is bypassed leading to an inability for entry into mitotic arrest on a normal timeline.

Collectively, the phenotypic impacts of delaying entry to mitotic arrest by conditionally inactivating Rb1 in the developing male germline suggested that a period of proliferative quiescence is important during the prospermatogonial stage to program fate commitment in postnatal development. To explore this critical time further, we first sought to construct a temporal map of cellular processes that occur in normal prospermatogonia during mitotic arrest. Using this map, we then cross-referenced perturbations that occurred when Rb1 was inactivated to identify cellular functions that may be dependent on mitotic arrest.

To build a developmental trajectory of prospermatogonia, we first integrated publicly available scRNA-seq datasets from isolated prospermatogonia spanning from E12.5 to SSC establishment at P3.5 (GEO accession numbers GSE119045 and GSE124904; Fig. 5A,B). Because of the asynchronous nature of germline development, single-cell transcriptomes were ordered using pseudotime analysis in Monocle, which accurately places the trajectory analyses with respect to developmental ages. Next, we performed differential gene expression analysis through pseudotime and identified six broad groups with distinct gene expression patterns (Fig. 5C, gene list located in Table S1). The expression of genes associated with cell cycle regulation (Group 2, Fig. 5C,D) declined early along the trajectory and were upregulated towards the end, consistent with the cell cycle progression of prospermatogonia in vivo. In addition, well-described processes, such as DNA methylation (Group 3), piRNA metabolism (Group 3) and cell migration (Group 6) were temporally regulated along the trajectory in a manner consistent with in vivo progression (Fig. S4). Collectively, our pseudotime analysis accurately recapitulated the in vivo germ cell developmental trajectory.

Interestingly, we also identified indications of metabolic shifts along the germ cell trajectory (Group 4, Fig. 5C,D). Genes associated with oxidative phosphorylation (OXPHOS) transiently increased in expression among quiescent prospermatogonia before decreasing during the period of cell cycle re-entry. To begin to assess whether metabolic shifts are disrupted in prospermatogonia that do not initiate mitotic arrest on a normal timeline, we applied filtering to the scRNA-seq database of E14.5 Rb1-cKO^Blimp1 and control prospermatogonia to standardize for the percentage of mitochondrial gene expression (Table S3), thereby allowing for focused assessment on nuclear-derived OXPHOS-related mRNAs. With cell cycle progression being disrupted in Rb1-cKO prospermatogonia (Fig. 5E, Fig. S5A and gene list in Table S1), genes associated with OXPHOS were also significantly downregulated across the entire population at E14.5 compared with controls (Fig. 5E, Figs S5B, S6A, and gene list in Table S1). These findings indicate that the normal transition to OXPHOS is influenced by cell cycle arrest, although additional biochemical experimentation is needed to fully support this concept. In addition, expression of genes that drive glycolysis, such as Eno1, were found to be upregulated in Rb1-cKO prospermatogonia compared with Rb1-sufficient control cells (Fig. S7), thus further supporting the notion of altered OXPHOS activity.

Surprisingly, expression of genes associated with male meiosis were not only detected along the germ cell developmental trajectory in wild-type mice, but also transiently upregulated around E16.5 (Fig. 5D). This observation is unexpected given that male germ cells enter meiosis postnatally unless misplaced outside seminiferous cords during fetal development (Zamboni and Merchant, 1973). However, with Rb1 inactivation, the expression of genes associated with male meiosis were upregulated at E14.5 (Fig. 5E, Figs S5C, S6B, and gene list in Table S1) and no indications of aberrant germ cell colonization were observed (Fig. 3). To assess the ramifications of disrupted entry into quiescence on a normal timeline that should have occurred at E14.5 in vivo, we assayed for transcript abundance and protein expression for the meiotic marker gene Stra8 (Zhou et al., 2008) in testes of Rb1-cKO^Blimp1 mice at E16.5. The relative abundance of Stra8 transcripts at E16.5 was significantly higher (P≤0.05) in Rb1-cKO prospermatogonia compared with controls (Fig. 5F). Strikingly, immunostaining of testis cross-sections revealed a significant (P≤0.05) increase by ∼18-fold in the percentage of STRA8⁺ prospermatogonia in Rb1-cKO^Blimp1 testes at E16.5 compared with controls (Fig. 5G,H), and by E18.5 the leveled increase in Rb1-cKO cells had dropped by ∼50% but was still greater compared with controls (Fig. S7A,B). The ectopic expression of STRA8 and significant increase in expression of genes related to meiosis indicate that inhibition of meiotic entry could be compromised in Rb1-cKO prospermatogonia. Overall, these findings suggest that the period of mitotic arrest in prospermatogonia facilitates a metabolic shift to an OXPHOS state and inhibition of meiotic initiation, which likely influence or accompany proper establishment of the SSC pool in postnatal development (Fig. 6).

Although a period of mitotic arrest and then re-initiation of cell cycle progression in prospermatogonia are events known to occur during genesis of the male germline, the underlying regulatory mechanisms driving these steps and the functional importance of this were not known. Here, and in previous studies (Yang et al., 2013), we developed mouse models to address these gaps in knowledge by eliminating the master cell cycle regulator RB1 in prospermatogonia prior to the normal timing of when entry and exit of mitotic arrest should occur. Previous studies have established that the prospermatogonial population is heterogeneous with subsets possessing characteristic transcriptome profiles that are linked with functional fates to either form the foundational SSC pool or transition to a differentiating state directly (Kluin and de Rooij, 1981; Law et al., 2019). In prospermatogonia lacking Rb1 at the onset of their formation, entry into quiescence fails to occur on a normal timeline, thus leading to apoptosis of all subsets and subsequent germline ablation (findings from the present study). In contrast, when the prospermatogonial population enters quiescence on a normal time frame but re-initiation of cell cycle progression is disrupted, only the SSC pool fails to develop (Hu et al., 2013; Yang et al., 2013), thus suggesting that proper timing of mitotic re-entry is crucial for locking in SSC fate within prospermatogonia. Interestingly, we found that the prospermatogonial population still undergoes mitotic arrest even in the absence of RB1 activity but on a delayed time frame (E18.5) compared with the normal situation (E16.5), potentially due to compensatory activity of other cell cycle regulators, such as cyclin D, Cdk4/6, Cdkn2a and Cdkn2b. Regardless, the delay in timing of mitotic arrest leads to prospermatogonial apoptosis, which effectively ablates the germline in males. Thus, the act of entering mitotic arrest and doing so at the proper time in development are crucial factors in genesis of the spermatogenic lineage.

A wave of apoptosis in the prospermatogonial population is known to occur during perinatal development and has been speculated to eliminate aberrant cells (Bejarano et al., 2018; Nguyen et al., 2019; Wang et al., 1998). Indeed, recent studies demonstrated that prospermatognia with developmentally defective genetic and epigenetic integrity are eliminated by apoptosis in the mouse (Nguyen et al., 2020). Based on results of the current study that showed massive apoptosis of prospermatogonia lacking RB1 activity and therefore abnormal cell cycle regulation during fetal development, we speculate that correct timing of mitotic arrest is an indicator of proper fitness that saves cells from apoptosis and disruption of this timing leads to elimination of cells that would normally contribute to the spermatogenic lineage in postnatal life.

Proper cell cycle regulation is paramount to the development and continuity of all cellular lineages that support homeostasis. The germ cell lineage is the eternal link between generations and in males the genesis relies on proper transition between PGC, prospermatogonial and SSC states. Using mouse models, previous studies have tracked the proliferation kinetics and cell cycle status underlying the PGC stage of development. Around E6.25, PGC fate is initiated in a subset of epiblast cells (∼40 precursor cells) that rapidly proliferate to ∼25,000 cells by E13.5 (Tam and Snow, 1981), and full germ cell commitment is locked in after migration to the genital ridge has completed (Nicholls et al., 2019). Thus, it is interesting that PGCs develop normally in Rb1-cKO^Blimp1 males, particularly when considering that proliferation rates among this population from E6.25 to E13.5 are not constant. Specifically, PGCs arrest in G₂ from E7.5 to E9.5 (Seki et al., 2007), and, based on gene expression profiles, cell cycle regulation closely resembles that of embryonic stem cells (ESCs) at E11.5, except for strict regulation of the G₁ phase (Sorrentino et al., 2007). ESCs lack a G₁/S checkpoint and can enter S phase rapidly following the previous M phase, and although Rb1 is expressed by ESCs, the related pathway is not active (LeCouter et al., 1996; Rohwedel et al., 1996). Similar regulation appears to occur among PGCs from E9.5 onward, such that cell cycle progression is regulated independently of RB1 and is collectively more ESC-like. Consistent with this notion, we found that RB1 is hyperphosphorylated and inactive among germ cells from E12.5 to E13.5, but as germ cells enter mitotic arrest RB1 becomes a central regulator of quiescence.

In previous studies, we discovered that inactivation of Rb1 in prospermatogonia of mice during the normal fetal quiescence period did not disrupt the first round of spermatogenesis in postnatal life, but the germline was summarily lost when subsequent rounds should commence (Hu et al., 2013; Yang et al., 2013). This phenotype implies that the foundational SSC pool required to regenerate the spermatogenic lineage following a round of differentiation was lacking. This observation begs the question of why the prospermatogonial subset normally fated to give rise to SSCs does not do so in the absence of RB1 to regulate cell cycle progression. Our recent studies indicate that prospermatogonia programmed during the quiescent period to adopt an SSC fate re-initiate cell cycle progression in neonatal development prior to the subset that is fated to initiate differentiation directly (Law et al., 2019). Thus, it is possible that inactivating Rb1 and disrupting the timing of cell cycle re-entry led to apoptosis of the SSC-fated prospermatogonial subset or a differentiating fate was induced directly in these cells. An intriguing alternative possibility is that prospermatogonia normally fated in late fetal development to form the foundational SSC pool instead adopted a progenitor fate directly in neonatal development and therefore did not either contribute to the first round of spermatogenesis or function as stem cells to support continual spermatogenesis. This notion is supported by our recent predictions, built from scRNA-seq profiles of prospermatogonial subsets, which suggest the existence of at least three different prospermatogonial fate trajectories during neonatal development that are programmed during the fetal period (Law et al., 2019). It is possible that activation of the programming laid down during fetal development is intimately linked with the timing of cell cycle re-entry in the neonatal period; exploring this possibility will require future experimentation.

A major unanswered question in the field of male germ cell biology has been what occurs within prospermatogonia during the quiescent period? To address this and define how disrupted entry compromises prospermatogonial survival, we compared transcriptome profiles from cells with and without RB1 activity at E14.5. Thus, we were able to compare gene expression profiles in prospermatogonia that had initiated quiescence on a normal developmental timeline and prospermatogonia that had not entered quiescence at the same point in development. Interestingly, pathway analysis uncovered that genes associated with meiosis were upregulated and genes associated with OXPHOS were downregulated in RB1-deficient prospermatogonia. These findings suggest that inhibition of meiotic initiation and a shift in metabolic activity normally occurs in prospermatogonia during the quiescent period. However, additional experimentation with biochemical assays are needed to fully assess metabolic shifts during prospermatogonial development. Considering the relatively large number of cells that are needed to carry out metabolomic analyses, exploring changes in OXPHOS activity throughout developmental time in more depth will require future technological advances.

In corroboration of altered expression of meiosis-associated genes, a significantly greater proportion of prospermatogonia immunostained for the meiotic marker STRA8 in Rb1-cKO mice compared with controls at E16.5. Previous studies have reported a similar relationship between mitotic arrest and meiotic inhibition in spermatogonia that are deficient for the transcription factor BNC2 (Vanhoutteghem et al., 2014), suggesting that the two forms of cell division are tightly coupled in the germline.

Metabolism is known to be a driver of cellular bioenergetics but is also associated with regulation of cell signaling and the epigenome (Folmes et al., 2012). Indeed, plasticity in metabolic pathways allows stem cells to meet the divergent demands of self-renewal and differentiation. Also, ensuring a glycolytic state facilitates the induction and maintenance of pluripotency in induced pluripotent stem cells and ESCs (Folmes et al., 2012). In addition, glycolysis is the primary bioenergetic process in SSCs and impacts their regenerative capacity (Chan et al., 2014; Helsel et al., 2017; Lord and Nixon, 2020). Furthermore, evidence from studies with embryonal carcinoma cells indicate that activation of OXPHOS activity triggers loss of pluripotency and transition to a differentiating state (Vega-Naredo et al., 2014). In addition, increased OXPHOS activity has been reported to occur during both PGC and SSC differentiation (Guo et al., 2017; Hayashi et al., 2017; Lord and Nixon, 2020; Varuzhanyan et al., 2019). Thus, metabolic shifts between functional states occur in many cell types but the mechanisms underpinning the process are largely unknown. Here, we found that in prospermatogonia when initiation of the quiescent period is disrupted by Rb1 inactivation, altered expression of genes associated with OXPHOS activity also occurs. This finding suggests that, similar to other types of stem cells, changing states of development align with a metabolic shift.

Features of germline development are highly conserved among mammals, and stringent maintenance of genetic integrity is crucial for proper inheritance regardless of species (Barton et al., 2016; Hamer and de Rooij, 2018; Murphey et al., 2013). During germ cell development, apoptosis that occurs at both fetal and postnatal ages is believed to select for the most robust cells to persist into adulthood and generate gametes that will pass genetic information to the next generation (Bejarano et al., 2018; Nguyen et al., 2020; Wang et al., 1998). The mechanisms for biosensing of germ cell fitness are largely unknown, but apoptosis is generally required to maintain reproductive fitness through the removal of developmentally incompetent or ‘undesirable’ cells, thereby ultimately ensuring maintenance of gamete quality (Aitken et al., 2011; Nguyen et al., 2020; Runyan et al., 2006). In the current study, we discovered that disruption of the normal developmental timing of when prospermatogonia enter quiescence leads to apoptosis and complete germline ablation. We postulate that, in addition to meiotic inhibition and metabolic shift, there are likely other key layers of programming being laid down during the quiescent period in prospermatogonia to ensure proper fate specification and fitness in postnatal life, and an ingrained biosensing mechanism eliminates cells that fail to initiate mitotic arrest on a correct developmental timeline.

Mice were maintained under specific pathogen-free (SPF) conditions at 20-26°C and 40-70% humidity in the animal core of Washington State University. All animal procedures were approved by the Washington State University Institutional Animal Care and Use Committee (IACUC). Rb1^flox (Rb1^fl; stock #026563), Blimp1-Cre (Blimp1-Cre^Tg; stock #008827) and Rosa26-tdTomato^fl_STOP_fl (stock #007909) transgenic mice were obtained from Jackson Laboratories. Primer sequences for genotyping have been previously described or are presented in Table S4. The day of mating plug detection was designated as E0.5 and the first day following birth was designated as P0.5.

To generate Rb1-cKO models, Rb1^fl/fl female mice were first crossed with Blimp1-Cre male mice to obtain Cre^Tg;Rb^fl/+ males that were then crossed with Rb1^fl/fl females to obtain Cre^Tg; Rb1^fl/− (Rb1-cKO). To generate multi-transgenic Rb1 cKO males, Rb1^fl/fl;Rosa26-tdTomato^fl_STOP_fl founders were first obtained through mixed crosses of mice that carried combinations of each transgene. Rb1^fl/fl;Rosa26-tdTomato^fl_STOP_fl females were then crossed with Blimp1-Cre^Tg;Rb1^fl/+ males to generate Rb1 cKO males with fluorescence-labeled germ cells.

Embryonic gonads and postnatal testes were fixed in 4% paraformaldehyde (PFA; Sigma-Aldrich, 158127) in PBS for 2 h at 4°C or Bouin's solution (Sigma-Aldrich, 15990-01) for 24 h at room temperature. Fixed samples were either stored in 70% ethanol at 4°C before paraffin embedding or dehydrated in 15% sucrose and frozen in optimal cutting temperature medium (Sakura, 4583). For all tissue blocks, 5 µm sections were cut using a Leica RM2255 automated microtome or Leica CM1950 cryostat.

For histomorphological analysis, paraffin sections from samples fixed in Bouin's solution were adhered to glass slides followed by incubation in xylene for dewaxing, rehydration by incubation in reducing concentrations of ethanol, and staining with Hematoxylin (Sigma-Aldrich, HHS16) and Eosin (Ricca, 2845-16). For immunofluorescence staining, antigen retrieval was achieved by incubation in an acidic (pH 6.0) solution of 10 mM sodium citrate (Amresco, 0101) and 0.05% Tween-20 (Sigma-Aldrich, P1379) for 20 min at 96°C. Non-specific antibody binding was blocked by incubating sections in a solution containing 2% bovine serum albumin (BSA; Thermo Fisher Scientific, BP1600) and 0.1% Triton X-100 (Sigma-Aldrich, T8787) in PBS at room temperature for 2 h. Sections were then incubated overnight at 4°C with primary antibody: rabbit anti-RB1 (1:150, Cell Signaling Technology, 9313), rabbit anti-phospho-RB1(Ser807/811) (1:100, Cell Signaling Technology, 9308), rat anti-TRA98 (1:500, Abcam, ab82527), rabbit anti-Ki67 (1:300, Abcam, ab15580), rabbit anti-cleaved caspase-3 (1:150, Cell Signaling Technology, 9661) or rabbit anti-STRA8 (1:4000, donated from Dr Mike Griswold, Washington State University, USA). Sections were washed in PBS and then incubated with a secondary antibody: TRITC-conjugated goat anti-rat IgG (1:500, Thermo Fisher Scientific, A11006) or FITC-conjugated goat anti rabbit IgG (1:500, Thermo Fisher Scientific, A11055, A11008) or Alexa Fluor 647-conjugated goat anti rabbit IgG (1:500, Thermo Fisher Scientific, A21245) at room temperature for 2 h. Sections were again washed in PBS and then coverslips mounted with medium containing DAPI (Vector Laboratories, H-1200). Images were captured using an Olympus IX51 or Leica DMi8 microscope.

Single-cell suspensions were generated from E14.5 testes as described previously. Briefly, tissue was incubated in a solution of 5 ml 0.25% trypsin-EDTA (Thermo Fisher Scientific, 25200-056) and 2 ml deoxyribonuclease I solution (7 mg/ml; Sigma-Aldrich, DN25) with occasional pipetting at 37°C for a total of 10 min. Trypsin digestion was quenched by addition of fetal bovine serum (FBS; Thermo Fisher Scientific, 10438-034). The cell suspension was then passed through a 40 µm cell strainer and centrifuged at 600 g for 7 min at 4°C. Cells were then washed in DPBS-S [PBS with 1% FBS, 10 mM HEPES (Sigma-Aldrich, H0887), 1 mM sodium pyruvate (Thermo Fisher Scientific, 11360-070), 1 mg/ml glucose (Sigma-Aldrich, G7528), 1% penicillin/streptomycin (Thermo Fisher Scientific, 15140-122)]. tdTomato⁺ germ cells were isolated using an SH800 FACS machine (Sony Biotechnology). Germ cells were washed and resuspended in a solution of 0.04% non-acetylated BSA in PBS. The resulting cell suspension was loaded into a Chromium Controller (10x Genomics) and single-cell cDNA libraries were generated using v2 chemistry according to the manufacturer's protocol. A total of six libraries were pooled at proportions netting equal read depth and sequenced in the same lane on an Illumina HiSeq 4000 (Genomics & Cell Characterization Core Facility, University of Oregon, USA).

Raw base call files were demultiplexed using the 10x Genomics Cell Ranger pipeline and aligned to the mouse mm10 transcriptome. Six libraries totaling 7585 cells were imported into R (version 3.6.0) and analyzed with the Seurat 3.0 package. Using anchor integration, libraries were merged, normalized, scaled, and dimensionally reduced with default settings. For clustering, statistically significant principal components (PCs) were chosen based on a ‘JackStraw’ score (P<0.05). Determination of DEGs was performed with default settings. Single-cell trajectory analysis was constructed on integration of scRNA-seq data generated in the current study with GEO datasets GSE124904 and GSE119045 using the Monocle 2 package based on the top 1000 DEGs (Table S2, P≤0.05) from the 11 clusters generated in Seurat 3.0. Gene Ontology assessment was conducted using the DAVID Bioinformatics Resource v6.8 (https://david.ncifcrf.gov/summary.jsp).

Assessment of apoptotic cells was achieved by scoring of cleaved CASPASE-3-positive germ cells in cross-sections of testes. Briefly, paraffin cross-sections mounted on glass slides were processed for co-immunofluorescence staining of cleaved CASPASE-3 and TRA98 as described above and examined by fluorescent microscopy at 20× magnification. For each testis, three different cross-sections were examined, and digital images were captured. In each image, the number of germ cells and cleaved CASPASE-3-positive germ cells per seminiferous cord/tubule were counted to determine the percentage of the population that was apoptotic.

Total RNA was isolated using Trizol reagent (Thermo Fisher Scientific, 155996026) and reverse transcription was performed using a SuperScript III First-Strand kit (Thermo Fisher Scientific, 18080-051). Relative abundance of Stra8 mRNA was measured using a 7500 Fast PCR System (Applied Biosystems) with Maxima SYBR Green/ROX qPCR Master Mix (Thermo Fisher Scientific, K0221) and normalization to mRNA levels of the housekeeping gene Actb. Primer sequences are listed in Table S4. Melt curve analysis was used to confirm the generation of a single amplicon and relative transcript levels were determined using the 2^−ΔΔct method.

All quantitative data are presented as mean±s.e.m. and differences between experimental groups were determined statistically using the Student's t-test (independent-samples t-test) or nonparametric test (Mann–Whitney U-test) function of SPSS software (IBM Corporation, version 19.0) when the data were of normal distribution or non-normal distribution, respectively.

The peer review history is available online at https://journals.biologists.com/dev/article-lookup/doi/10.1242/dev.194571