Mouse Pramel1 regulates spermatogonial development by inhibiting retinoic acid signaling during spermatogenesis

Mammalian spermatogenesis is a highly coordinated and dynamic process where spermatogonial stem cells (SSCs) differentiate into functional spermatozoa (Griswold, 2016). Precise gene regulation in SSCs determines their biological activities to ensure that large numbers of spermatozoa are produced continuously throughout reproductive life. During spermatogenesis, germ cells undergo three major transitions, including the undifferentiated to differentiating transition of spermatogonia, meiosis of spermatocytes and spermiogenesis of spermatids, which are all regulated by retinoic acid (RA) and its receptors (RARs) (Endo et al., 2017). RA is periodically synthesized by Sertoli and germ cells, and regulates genes associated with germ cell differentiation, proliferation and apoptosis (Epping et al., 2005; Zhu et al., 2018).

In neonatal mice, the first transition of germ cells occurs when prospermatogonia (also known as gonocytes) begin to respond to the first RA pulse at postnatal day 3 (P3) and progressively migrate from the center of seminiferous cords to the basement membrane (Clermont, 1972; Nagano et al., 2000; Law et al., 2019) – a process known as homing (Bellve et al., 1977; Clermont and Perey, 1985). The underlying mechanisms of germ cell homing remains poorly understood (Singh et al., 2013; Xu et al., 2015; Yin et al., 2021). During the homing process, most of prospermatogonia becomes the undifferentiated spermatogonial population, which includes SSCs and non-stem cell progenitors. The remaining prospermatogonia differentiate directly to A2 spermatogonia and initiate the first round of spermatogenesis (de Rooij and Russell, 2000; Yoshida et al., 2006; Law et al., 2019). It has been suggested that cell fate was predetermined before birth, and cell-fate committed prospermatogonia have a heterogeneous capacity to respond to RA (Law et al., 2019; Velte et al., 2019). However, little is known about how the number of these three subtypes of germ cells are balanced initially to establish continuous and nonsynchronous spermatogenesis.

During the first round of spermatogenesis, A2 spermatogonia divide into B spermatogonia, becoming preleptotene spermatocytes (Pl) in response to the second RA pulse at P8-9, followed by meiotic initiation (Anderson et al., 2008; Drumond et al., 2011). Pachytene spermatocytes emerge at P14 with the third RA pulse, and round spermatids form around P21 in response to the fourth RA pulse (Endo et al., 2017). The first round of spermatogenesis concludes around P35 with spermatozoa released into the lumen of seminiferous tubules, eventually reaching the cauda epididymis by P41. Although, theoretically, an A2 spermatogonia would produce 128 spermatozoa, significantly fewer are produced in reality (Fayomi and Orwig, 2018; Kluin et al., 1982; de Rooij, 2001). Remarkably, an early wave of germ cell apoptosis appears around P14-P21 during the first round of spermatogenesis (Mori et al., 1997; Wright et al., 2007), due to a surplus of A2-A4 spermatogonia undergoing cell competition (de Rooij and Janssen, 1987; de Rooij, 2001). The mechanism behind this surplus undergoing apoptosis remains a mystery.

RA signaling involves the binding of RA to a nuclear receptor (RAR), forming a complex with retinoid X receptor (RXR) (Dawson and Xia, 2012). This complex modulates transcription by binding to DNA at a RA response element in enhancer regions of RA-responsive genes (Chambon, 1996; Duester, 2008), recruiting co-repressors or co-activators to inhibit or induce transcription (Mark et al., 2015; Griswold, 2016). One of the co-repressors of RA identified in cancerous cells and embryonic stem cells (ESCs) is known as preferentially expressed antigen in melanoma (PRAME) (Epping et al., 2005). PRAME was first discovered as a tumor antigen expressed in human melanoma cells (Ikeda et al., 1997). During evolution, the eutherian-specific PRAME gene family, which amplifies to over 60 copies on autosomes in various species, has transposed to and amplified on the rodent X and the bovid Y chromosome (Birtle, et al., 2005; Chang et al., 2011). Members of the PRAME gene family encode leucine-rich repeat (LRR) proteins (Wadelin et al., 2010) and belong to a group of cancer/testis antigens (CTAs), which are expressed in testis, ovary and a variety of cancers (Nettersheim et al., 2016; Kern et al., 2021). Like other LRR proteins, PRAME folds into a horseshoe shape in their tertiary structure, providing a versatile structural framework for protein–protein interactions in diverse molecular recognition processes, including signal transduction (Kobe and Deisenhofer, 1994; Epping et al., 2005; Wadelin et al., 2010).

The mouse Prame gene family is predominantly expressed in the germline at different developmental stages throughout the life cycle (Kern et al., 2021). Several members of the Prame family are expressed in ESCs and primordial germ cells (PGCs) during embryonic development to maintain stem cell pluripotency (Casanova et al., 2011; Graf et al., 2017; Napolitano et al., 2020). After birth, many Prame members are expressed during gametogenesis, with some being specific to the testis (Wang et al., 2001; Mistry et al., 2013) or ovary (Dadé et al., 2003; Minami et al., 2003; Monti and Redi, 2009), whereas others are expressed in both male and female gonads (Mistry et al., 2013). Studies on Pramex1 (Lu et al., 2020) and Pramef12 (Pramel13; Wang et al., 2019b), using gene knockout (KO) approaches, have shown their importance in spermatogonia (including SSCs) proliferation and differentiation. However, it remains unclear whether the Prame family functions through RA signaling during spermatogenesis.

The mouse PRAMEL1 proteins aggregate to form clusters of protein complexes that are enriched in the nucleus and cytoplasm (especially in germ granules) of spermatogenic cells (Mistry et al., 2013; Liu et al., 2021). The findings from this study on global and conditional Pramel1 KO mice suggest that PRAMEL1 plays a crucial role in regulating the responsiveness of cell-fate committed prospermatogonia to RA, ensuring a dynamic balance between undifferentiated and differentiating spermatogonia during the establishment and maintenance of spermatogenesis.

Mice with global knockout (gKO) of Pramel1 were generated using a CRISPR/Cas9 approach that resulted in a 50 bp deletion of exon 3 and a codon shift in the Pramel1 gene (Fig. 1A), and the absence of the PRAMEL1 protein in the gKO testis (Fig. S1A,B). Pramel1 gKO mice developed normally except for distinct reproductive phenotypes observed in males (Fig. 1B-I). At the juvenile age of 14 days (postnatal day 14, or P14), testes of Pramel1 gKO males were significantly larger (20%) than wild-type controls, but this difference was resolved by P60. Mature testes of Pramel1 gKO at age of 4 months and 1 year were 11% and 28% smaller, respectively, than those in wild-type mice (Fig. 1B and Fig. S1C). The average diameter of round seminiferous tubule cross-sections in the Pramel1 gKO mice were larger at P14 and P21, but smaller at P120 and P365 when compared with wild-type mice (Fig. 1C and Fig. S1D). Although the number of Sertoli cells was similar between the Pramel1 gKO and wild-type mice (Fig. 1D,E), the number of germ cells was increased in the enlarged seminiferous tubules in young gKO males (Fig. 1D,F). We speculated that the germ cell increases in the young gKO testis would lead to a higher sperm count at the end of the first round of spermatogenesis. As expected, we observed a 31% increase in epididymal sperm count at P41 in gKO males compared with wild-type males (Fig. 1G and Fig. S1E). Surprisingly, no difference was observed in the sperm count between the Pramel1 gKO and wild-type mice at P60, but a significant decrease was observed at P120 (−19%) and P365 (−43%) (Fig. 1G and Fig. S1E). Taken together, our data suggest that sperm production increased during the first round of spermatogenesis but decreased after 2 months of age in mature Pramel1-deficient mice.

We conducted two mating tests to examine the fecundity of spermatozoa produced from the young (P35-P50) and mature (P60-P240) Pramel1 gKO males. The young gKO males showed a significantly higher litter size (43%) than wild-type males (Fig. 1H and Fig. S1F,G), indicating functional sperm production and increased fecundity during the first round of spermatogenesis. However, for mature males, Pramel1 gKO mice did not follow the expected standard curve (Fig. 1H) (Flurkey et al., 2007). Instead, their litter size continuously declined over 6 months of breeding, with a significantly reduced litter size in the third and fourth litters compared with wild-type males (Fig. 1H). Collectively, these data suggest that Pramel1 deletion significantly affected male mouse fecundity and provides a unique model with which to study spermatogenesis in different age groups.

The age-dependent phenotypes in the Pramel1 gKO males prompted us to analyze a conditional Pramel1 KO (cKO) mouse line generated by the Cre-LoxP system with Stra8 (stimulated by retinoic acid 8)-iCre (Fig. 1A). As the Stra8 expression is testis specific, and induced by the initial RA signaling at P3, the Stra8-iCre recombinase is activated in differentiating spermatogonia leading to Pramel1 deletion in the testis from P3 onwards (Fig. S1A,B) (Sadate-Ngatchou et al., 2008). A comparison between the Pramel1 gKO and cKO mice allowed us to test whether Pramel1 is involved in RA signaling during the first round of spermatogenesis.

As shown in Fig. 1B-I, the phenotypes on testis size and sperm production of the young Pramel1 cKO mice (<P41) were like those of the floxed and wild-type controls that were significantly smaller than that of the Pramel1 gKO mice. The Pramel1 gKO testis (at P7-P120) had consistently ∼7% of seminiferous tubule cross-sections with disorganized layers of germ cell loss (Fig. 1D,I). Remarkably, all affected seminiferous tubule cross-sections in the Pramel1 gKO testis at P7 were found to be Sertoli cells-only (SCO) tubules. The frequency of SCO tubules gradually decreases during the development of the testis and, by P35, SCO tubules have disappeared, indicating a process of recovery (Fig. 1I). In contrast, we did not observe any SCO tubules in Pramel1 cKO mice. Instead, we observed disorganized seminiferous tubule cross-sections in the cKO males with a low frequency (<1%) at P7, which then gradually increased with age (3% at P21) to 7% by P35, matching the frequency seen in the gKO mice (Fig. 1I). These results indicate that both global and conditional deletion of Pramel1 affect germ cell development and their associated epithelium structure in about 7% of tubules, but the severity differed in young males during the first round of spermatogenesis.

Unlike the young males, the mature Pramel1 cKO mice were like the mature Pramel1 gKO males with a significantly smaller testis, lower sperm count and smaller litter size compared with the controls at the age of 4 months (Fig. 1B,G,H). By P365, no difference was found between the Pramel1 cKO and the control mice (Fig. 1B,C,G-I). These findings suggest that the phenotypic differences observed in young males and similarities observed in mature males between the Pramel1 gKO and cKO mice could be due to the response of germ cells to the initial RA pulse at P3 in the neonatal testis.

To test the involvement of PRAMEL1 in RA signaling, we injected all-trans RA (atRA) or DMSO (vehicle control) in wild-type, Pramel1 cKO and gKO mice at P2, and analyzed testis index and sperm count at P41 (Fig. 2A). For the DMSO-treated groups, the testis index was similar among Pramel1 cKO and gKO and wild-type mice (Fig. 2B and Fig. S2A), whereas the sperm count was 33% higher in the Pramel1 gKO mice than in cKO and wild-type mice at P41 (Fig. 2C and Fig. S2B). With atRA treatment, the testis index was increased by 6% in the Pramel1 cKO and by 12% in the gKO mice compared with wild type. The sperm count increased 2.3-fold and 3.5-fold in the cKO and gKO mice, respectively. These data indicate that the time point when Pramel1 was deleted from the genome, either before or at P3, affected the response of the animals to the exogenous atRA, and the Pramel1 gKO mice appeared to be more sensitive to the exogenous atRA.

To further determine whether the SCO phenotype observed in Pramel1 gKO mice was the consequence of the changes in the RA signaling, we injected atRA, WIN18,446 (which inhibits biosynthesis of RA) or DMSO in P2 mice and performed a whole-mount immunofluorescence staining on seminiferous tubules at P5 (Fig. 2D,E and Fig. S2C). In this experiment, Id4-eGfp⁺ reporter mice were bred with Pramel1 gKO mice to produce the Id4-eGfp⁺Pramel1 gKO mice. A previous study on EGFP intensity of Id4-eGfp transgenic mice at neonatal stage demonstrated that SSCs were enriched in ID4-EGFP^Bright cells, that intermediate progenitors were likely ID4-EGFP^Mid and that differentiating spermatogonia were ID4-EGFP^Dim or ID4-EGFP negative (Oatley et al., 2011; Law et al., 2019). We applied TRA98 antibody to label all germ cells in the Id4-eGfp⁺ reporter mice before analyzing the SCO segments in seminiferous tubules. In the control mice, SCO segments were not found in DMSO or the RA inhibitor treatments, but were clearly observed in the atRA treatment (Fig. 2D,E), suggesting that the exogenous atRA, i.e. a higher than the normal RA level in the wild-type mice, could lead to a SCO phenotype. Among the Id4-eGfp⁺Pramel1 gKO mice, the SCO segments were present in the DMSO-treated mice as expected, but disappeared in the RA inhibitor-treated mice. Furthermore, the atRA treatment led to an increased frequency of SCO in the Id4-eGfp⁺Pramel1 gKO testis. These data indicate that: (1) injection of exogenous atRA in neonatal wild-type mice resulted in a similar phenotype (i.e. SCO) to the Pramel1 gKO mice; (2) the SCO phenotype in the Pramel1 gKO mice was rescued upon treatment with the RA inhibitor; and (3) the frequency of SCO segments was related to the RA level.

We conducted a reciprocal co-IP experiment with anti-PRAMEL1 and anti-RARα antibodies in the wild-type testis to validate the involvement of PRAMEL1 in the RA signaling. The results revealed that PRAMEL1 and RARα form a protein complex, whereas the YBX2 protein (serving as a negative control) is not present in this complex (Fig. 2F). This suggests that PRAMEL1 is physically associated with RARα in the testis, confirming the previous finding from a cancer cell study (Epping et al., 2005). Collectively, our data suggest that PRAMEL1 is involved in the RA signaling pathway during the initiation of spermatogenesis in the neonatal testis.

To determine whether Pramel1 is involved in the transition of prospermatogonia in response to the first RA pulse in the neonatal testis, we performed dual staining on testis cross-sections using PLZF (a marker for SSCs and progenitors) (Filipponi et al., 2007) and STRA8 (a marker for A1-B spermatogonia) (Qing et al., 2008). No differences in germ cell number were observed between wild-type and Pramel1 gKO testes at P2 (Fig. 3A,B). However, after the first RA pulse, significant differences in the number of germ cells were observed from P3 onwards. The proportion of differentiating spermatogonia that were positive (+) for both PLZF and STRA8 markers was 80-168% higher, and the proportion of undifferentiated spermatogonia (PLZF+ STRA8−) was 45-50% lower in Pramel1 gKO mice than in wild type at P3-P6 (Fig. 3B). Furthermore, the total number of spermatogonia in Pramel1 gKO was 35-45% lower than in wild-type testis from P3 to P6. These results were validated by qRT-PCR using P6 testis RNA samples where the undifferentiated spermatogonial markers Thy1, Id4 and Plzf (Filipponi et al., 2007; Reding et al., 2010; Sun et al., 2015) were downregulated, whereas differentiating spermatogonial markers Kit and Stra8 were upregulated in Pramel1 gKO mice (Fig. 3C). In addition, the dissociated germ cells of Id4-eGfp⁺Pramel1 gKO and the control Id4-eGfp⁺ testis were labeled with DDX4 (which is a marker specific to germ cells) and analyzed by flow cytometry. Among all DDX4-positive cells, the proportion of ID4-EGFP^Bright and ID4-EGFP^Mid cells was 14% lower, whereas the ID4-EGFP⁻ cells were 16% higher in Pramel1 gKO mice than in the wild-type control (Fig. 3D), indicating that deletion of Pramel1 tilts the balance between undifferentiated and differentiating spermatogonia toward differentiation in the neonatal testis.

The presence of SCO tubules in Pramel1 gKO, but not in cKO, testis suggests that the SCO phenotype was associated with the response of germ cells to the first RA pulse at P3 during the establishment of spermatogenesis. To determine when and how SCO tubules were formed in the neonatal testis, we performed whole-mount immunofluorescence staining with TRA98 and SOX9 antibodies on dissociated seminiferous tubules of wild type, and Pramel1 cKO and gKO from P2 to P6. No differences were found between KO and wild-type mice at P2 (Fig. 4A-D). By P3, SCO segments were clearly observed in the Pramel1 gKO, but not in wild-type and cKO testes (Fig. 4A). The length of the SCO segment along the development of seminiferous tubules was gradually enlarged with age (Fig. 4A). In addition, compared with the wild-type and Pramel1 cKO mice, germ cell count from the seminiferous tubule cross-sections was decreased in the Pramel1 gKO by 21%, 41% and 17% at P3, P5 and P6, respectively (Fig. 4B). As the loss of germ cells occurred during the time period of germ cell homing, and coincides with a reduction in the number of undifferentiated spermatogonia in the Pramel1 gKO testis (Fig. 3B), we speculated that the global deletion of the Pramel1 may impair the process of germ cell homing and promote cell apoptosis, leading to the formation of SCO tubules. To test this hypothesis, we analyzed the location of every TRA98⁺ germ cell on the seminiferous tubule cross-sections to identify homed spermatogonia (Fig. 4C). Homed cells were defined as those situated near the basement membrane and residing amidst the Sertoli cells, whereas non-homing cells are positioned within the lumen of the seminiferous tubules. We observed that the percentage of homed germ cells was similar in wild-type (46%) and Pramel1 gKO (45%) mice at P2. However, upon responding to the first RA pulse, most germ cells (76%) were homed in the wild-type testis, whereas only 41% of germ cells were homed in the Pramel1 gKO testis at P3 (Fig. 4C,D and Fig. S3). As the development progressed to P4 and P5, the number of homed germ cells was slightly increased in wild-type mice (77-89%), but it significantly increased in the Pramel1 gKO testis (72-80%) (Fig. 4D).

To investigate whether the expression of genes essential for spermatogonia homing was affected, we performed a qRT-PCR analysis on three genes (Gfra1, Ccr1 and Ccl3) for the testes of gKO and wild-type mice at P5. The migration of prospermatogonia relies on the coordination of multiple signaling pathways, including GDNF and CCL3 pathways, with prospermatogonia expressing GFRα1 and CCR1 as their respective receptors (Hofmann and McBeath, 2022). Meanwhile, CCL3 is secreted by Sertoli cells. As expected, the expression of both Gfra1 and Ccr1 from prospermatogonia was lower (∼50%) in the gKO mice compared with wild type, while Ccl3 expression from Sertoli cells remained unchanged (Fig. 4E). These data indicate that the germ cell homing process was affected by the global Pramel1 deletion.

To confirm whether lack of homed germ cells was caused by apoptosis, we conducted whole-mount P3 testis staining with TRA98 and cleaved-CASP3 (c-CASP3, a marker for cell apoptosis) antibodies (Porter and Jänicke, 1999). As shown in Fig. 4F, germ cells on either end of the SCO region were not homed but underwent apoptosis, confirming that the lack of homed germ cells and forming of SCO region was caused by apoptosis of germ cells that failed in the homing process. To further investigate the germ cell type of apoptotic cells adjacent to the SCO region, we stained the whole-mount seminiferous tubules, isolated from the Id4-eGfp⁺ and the Id4-eGfp⁺Pramel1 gKO mice, for c-CASP3. The results clearly showed that germ cells that failed in homing and underwent apoptosis at P3 were ID4-EGFP^Mid, i.e. progenitors, in the Id4-eGfp⁺Pramel1 gKO testis (Fig. 4G). Accordingly, we concluded that the SCO tubules in the young Pramel1 gKO testis likely originated from apoptosis of progenitors due to their failure in the homing process.

We then performed a time-course study on germ cell apoptosis in prepubertal and mature testes using TUNEL assay (Fig. 5A) (Gavrieli et al., 1992). We found that the average number of apoptotic (TUNEL⁺) cells/tubule on testis cross-sections was higher in the Pramel1 gKO mice than wild type at P7, P35, P60 and P120, but it was lower in the Pramel1 gKO mice at P14 and P21. The expected early wave of germ cell apoptosis observed in wild-type mice (Mori et al., 1997), which peaked around P14-P21 (Fig. 5B,C), disappeared in Pramel1 gKO mice, suggesting that Pramel1 plays a role in the spermatogonial survival in prepubertal testes.

Results from the TUNEL assay raised the question of whether the Pramel1 gKO mice have a surplus of A2-A4 spermatogonia to initiate the early wave of germ cell apoptosis. To address this, we applied dual staining with TRA98 or PLZF and c-CASP3 on P21 testis cross-sections to identify the type of germ cells that underwent apoptosis. As shown in Fig. 5D, a high frequency of apoptotic cells (c-CASP3⁺) was observed in the wild-type testis, supporting the TUNEL results. Among these apoptotic cells, most were TRA98⁺ and PLZF⁻, implying that they were differentiating/differentiated spermatogonia in wild-type mice. In contrast, a low frequency of c-CASP3⁺ apoptotic cells were observed in the Pramel1 gKO testis, most of which were PLZF⁺ undifferentiated spermatogonia. Our data suggest that disappearance of the early wave of apoptotic germ cells in the Pramel1 gKO testis was due to the lack of surplus of A2-A4 spermatogonia during the first round of spermatogenesis.

In our Id4-eGfp⁺ mouse model, we found that ID4-EGFP^Bright cell-enriched segments (underlined in green in Fig. 6A) along the seminiferous tubules could be used as an early indicator of the seminiferous cycles. The average length of the cycles was substantially (83%) longer in the gKO testis than in wild-type control at P7. There were two additional segments within each seminiferous cycle: a TRA98 highly labeled region (Fig. 6A, underlined in red); and a ID4-EGFP^Mid and TRA98 moderately labeled region (Fig. 6A, underlined in orange). In the Pramel1 gKO mice, a SCO region was constantly adjacent to the ID4-EGFP^Bright region at one end and to the TRA98-highly labeled region at the other end (Fig. 6A). The TRA98 highly labeled segment was 77% longer than the wild type (Fig. 6B). Germ cell density in TRA98 highly labeled segments of the gKO mice was similar to the wild type but decreased by 20% in the ID4-EGFP^Bright regions and ID4-EGFP^Mid regions (Fig. 6C), indicating that germ cell loss in the neonatal Pramel1 gKO testis occurred mainly in ID4-EGFP^Bright and ID4-EGFP^Mid regions. Collectively, these data suggest that deletion of Pramel1 affects the initial establishment of seminiferous cycle, and the increased length of TRA98 highly labeled segments could be the reason for high sperm production during the first round of spermatogenesis in the Pramel1 gKO mice.

To further understand how stages of the seminiferous epithelium cycle is initiated, we investigated the histology of seminiferous tubules at P21 (Fig. 6D) and P35 (Fig. S3A), and analyzed the cellular association by TRA98 and YBX2 (a spermatocyte and spermatid-specific marker) staining on testis cross-sections (Chowdhury, 2014). We focused on stage IX tubules as germ cell transition occurs around stage IX after each RA pulse. Three layers of germ cells were present in the wild-type testis at P21, including undifferentiated and differentiated spermatogonia (layer 1, TRA98⁺ cells close to the basement membrane), preleptotene spermatocytes (layer 2, TRA98⁺ cells further away from the basement membrane) and pachytene spermatocytes (layer 3, YBX2⁺ cells) (Fig. 6D). We found that germ cell loss in the seminiferous epithelium of the Pramel1 gKO mice could occur in three layers (layers 1-3), two layers (layers 2-3) or one layer (layer 3) in responding to the RA pulses (Fig. 6D). As these germ cells were missing at the corresponding time immediately after each RA pulse, and the number of missing layers of germ cells was correlated with the number of RA pulses, it was evident that Pramel1 was involved in the establishment of the seminiferous epithelium cycle.

Next, we analyzed the percentage of seminiferous stage I to stage XII based the testis cross-sections at P35 when the seminiferous epithelium is fully developed. The frequency of stage VII-VIII in the gKO mice was 79% higher, but stage I and stage IX-X were 19% and 36% lower, respectively, than the wild-type control (Fig. 6E). The increase in the stage VII-VIII frequency was consistent with the increased length of TRA98 highly labeled segments, implying that a longer segment of seminiferous tubule in the first seminiferous epithelium cycle was committed to produce the first cohort of spermatozoa in the gKO testis.

We noticed that the SCO phenotype was partially recovered (after P41) from the Pramel1 gKO mice (Fig. 1I, Fig. S4). Although the SCO tubules disappeared, similar patterns (Fig. 6D) and frequencies (Fig. 1I) of disorganized seminiferous tubules with germ cell loss were still present in the Pramel1 gKO mice after P41 (Fig. S4). These data suggest that absent layer(s) of germ cells in the partially recovered SCO tubules would be kept in the Pramel1-deficient testis throughout life, which further confirmed that Pramel1 is important for the proper establishment of the seminiferous epithelium cycle during spermatogenesis.

To further investigate whether Pramel1 deficiency affects subsequent rounds of spermatogenesis in mature mice, we performed two sets of experiments using in situ and in vivo approaches. In the in situ experiment, ID4 and STRA8 staining on testis cross-sections revealed that the number of undifferentiated spermatogonia in the Pramel1 gKO mice was significantly smaller than that in the wild type at P35 and P120 (Fig. 7A,B). In the in vivo experiment, a scrotal heat-stress (HS) treatment was performed on the wild-type and Pramel1 gKO mice at the age of 2 months. Disrupted spermatogenesis was expected to be gradually recovered after HS treatment (Rockett et al., 2001), and the recovery rate would reflect the ability of undifferentiated spermatogonia to re-establish spermatogenesis. As expected, the HS treatment led to a significant reduction in testis weight/index for both wild-type and the Pramel1 gKO males at post-treatment day 5 (HSD5) (Fig. 7C,D). The testis weight/index was then gradually recovered in wild-type mice, but was not recovered in Pramel1 gKO at HSD14 and HSD38. Furthermore, sperm count was significantly decreased in Pramel1 gKO mice at HSD38 (Fig. 7E). These data imply that the Pramel1 gKO testes had a slower rate to re-establish spermatogenesis due to the disruption in the maintenance of undifferentiated spermatogonial population. Therefore, we concluded that PRAMEL1 is involved in the maintenance of undifferentiated spermatogonial populations during spermatogenesis.

PRAME has been recently renamed as ‘PRAME nuclear receptor transcriptional regulator’. PRAME exerts its biological functions via regulation of its downstream targets, such as RAR, p53, p21, Bcl2, TRAIL, Hsp27 and S100A4, leading to the control of several cellular processes, including cell proliferation, differentiation, apoptosis and growth arrest (Al-Khadairi and Decock, 2019; Xu et al., 2020). Numerous studies in cancer cells have investigated the interaction between PRAME and RA/RAR, demonstrating that PRAME acts as a repressor of RAR signaling, resulting in the downregulation of apoptosis and cell differentiation, and the upregulation of cell proliferation in cancer cells (Epping et al., 2005; Zhu et al., 2018). The Prame family is known to confer ESCs with a resistance to RA signaling, enabling the maintenance of pluripotency (Casanova et al., 2011; Graf et al., 2017; Napolitano et al., 2020). Outside embryonic development, Prame family members are predominantly expressed in the germline at different developmental stages throughout the life cycle (Kern et al., 2021). However, the mechanism underlying the function of PRAME in germ cells is not well understood. In the present study, we provide evidence that mouse Pramel1 plays a role in spermatogonia development through RA signaling. We observed that upregulation of RA signaling, which is achieved by either administering exogenous RA or knocking out Pramel1, resulted in distinct phenotypes characterized by regional SCO tubules and elevated sperm production during the first round of spermatogenesis. The SCO phenotype could be rescued by the administration of the RA inhibitor, suggesting that Pramel1 functions in germ cells by inhibiting RA/RAR signaling. Furthermore, we successfully confirmed the interaction between PRAMEL1 and RARα in the wild-type testes through our co-immunoprecipitation experiments. These findings have important implications for future research on the role of the Prame family of genes in spermatogenesis, and provide valuable insights into the shared molecular mechanism between gametogenesis and oncogenesis (Old, 2001; Simpson et al., 2005).

To date, two members of the Prame family, Pramex1 and Pramef12, have been investigated during spermatogenesis (Wang et al., 2019b; Lu et al., 2020). Pramex1-deficient mice are fertile with a reduced testis size and sperm count, affecting particularly the development of pachytene spermatocytes (Lu et al., 2020). On the other hand, Pramef12 mutants display significantly smaller testes and SCO tubules, and are infertile. Disruption of Pramef12 impairs early spermatogenic maintenance and causes a decline in undifferentiated spermatogonial populations over time, indicating that Pramef12 plays a vital role in SSC maintenance (Wang et al., 2019b). In this study, we investigated the dynamic changes of the undifferentiated and differentiated spermatogonial populations in the Pramel1-deficient mice. We found that Pramel1 deficiency led to a decline of undifferentiated spermatogonial in both young and mature mice with a distinctive phenotype. Collectively, our results suggest that Pramel1 plays a role in the maintenance of undifferentiated spermatogonial populations.

Regional SCO tubules were found in young male mice with Pramex1 cKO and Pramel1 gKO, but not with Pramel1 cKO, suggesting that these two genes are not functionally redundant. After the first round of spermatogenesis, these SCO tubules partially recovered in both Pramex1- and Pramel1-deficient mice, although the underlying mechanism for this recovery process is unknown. We hypothesize that this may be related to the self-renewal of SSCs and their ability to migrate along the seminiferous tubules. Because only A_single stem cells can colonize empty stretches of seminiferous tubules to re-establish spermatogenesis (de Rooij, 2001), labeling A_single cells with an EGFP reporter and tracking their migration into the SCO regions could be one way to study the recovery process in future research.

Despite the reproductive phenotypes, the Pramex1 cKO, and Pramel1 gKO and cKO male mice apparently produce functional spermatozoa, suggesting that both Pramex1 and Pramel1 are not essential for spermatogenesis. Based on the observations made in this study and on the results obtained for other members of the Prame gene family, such as Pramel7 and Gm12794c (Pramel19; Graf et al., 2017; Napolitano et al., 2020), it is believed that various members of the Prame family may perform distinct roles at different stages of germ cell development as a fine-tuner of RA signaling. These roles may include the regulation of germ cell proliferation, differentiation and apoptosis. Although each member of the family may have a minor or non-essential function in spermatogenesis, as a whole, the family collaborates to regulate the three significant transitions of spermatogenic cells that are activated by RA (Endo et al., 2017).

Our co-immunoprecipitation results revealed that PRAMEL1 interacts with RA/RAR in germ cells, confirming the previous finding from cancer cells (Epping et al., 2005; Zhu et al., 2018). Through this interaction, PRAMEL1 regulates the RA responsiveness of cell-fate committed prospermatogonia, balancing undifferentiated and differentiating spermatogonia populations when germ cells respond to the first RA pulse at P3. A schematic diagram is given in Fig. 8 to explain the role of this interaction for germ cell development. In the wild-type mice, prospermatogonia have completed homing, and stem cell niches are formed for the SSC-fate prospermatogonia around P3 (Bellve et al., 1977; Orwig et al., 2002; Yoshida, 2018). According to Law et al. (2019), A2 spermatogonia are predestined for the first round of spermatogenesis, whereas the initial progenitors are predetermined for the second round. On the other hand, SSCs are preset for successive rounds of spermatogenesis. This sequential process establishes the foundation for uninterrupted spermatogenesis throughout the reproductive life (Fig. 8A).

The progenitors are typically RA-nonresponsive in the wild-type neonatal testis (Velte et al., 2019), but they were clearly responsive to the first RA pulse in the Pramel1 gKO mice (Fig. 3; Fig. 8A), underwent apoptosis and failed to home (Fig. 4). We reasoned that the epithelial space normally occupied by the progenitors is empty, leading to the SCO phenotype (Fig. 8B) and the lack of sperm production in the second round of spermatogenesis (Fig. 8A), validated by sperm count at P46 (Fig. S5). The empty space in SCO tubules provides an opportunity for the expansion of the differentiating spermatogonia (Fig. 6A) into the neighboring SCO region at P3-P7, resulting in a significantly longer segment of stage VII-VIII tubule at P35 (Fig. 6E). It mitigates germ cell competition and saves A2-A4 spermatogonia to eliminate the early wave of germ cell apoptosis (Fig. 5B,C), resulting in more sperm production during the first round of spermatogenesis (Fig. 8A). The number of undifferentiated spermatogonia was significantly decreased in both young (Fig. 3A,B) and mature (Fig. 7) gKO mice, leading to a reduction in sperm production in subsequent rounds of spermatogenesis, and a low fecundity (Fig. 8A).

Previous studies have revealed that the establishment of epithelial cycles in seminiferous tubules occurs before birth (Yoshida et al., 2006; Hu et al., 2013). The cell-fate committed germ cells migrate in clusters (Law et al., 2019) and occupy specific cylindrical segments within the prefigured epithelial cycle during homing (Timmons et al., 2002; Yoshida et al., 2006). However, it is unclear which segment is occupied by SSCs, progenitors or A2 spermatogonia, and how these homed cells contribute to the establishment of stages I to XII of the seminiferous epithelial cycle during testis development. In the current study, we found that the undifferentiated spermatogonia were enriched periodically along the seminiferous tubules, with the differentiating spermatogonia in between, signifying an early indication of seminiferous cycles in the neonatal testis (Fig. 8B, left). As mentioned earlier, disrupting Pramel1 altered responsiveness of initial progenitors to the first RA pulse, leading to cell apoptosis and SCO in 7% of seminiferous tubules (Fig. 8B, right). This incidence matches the frequency of stage VII tubules (∼7% in testis cross-sections at all ages) (Hu et al., 2013), suggesting that initial progenitors migrate into spaces predetermined for stage VII during the homing process. Therefore, our findings suggest that PRAMEL1 may contribute to germ cell homing and the establishment of stages of seminiferous epithelial cycles during the initiation of the first round of spermatogenesis.

In summary, our gene-knockout approach revealed that Pramel1 regulates germ cell proliferation, differentiation and apoptosis by inhibiting RA/RAR signaling, akin to its role in cancer cells (Epping et al., 2005; Zhu et al., 2018). Although not essential for spermatogenesis, Pramel1 plays a crucial role in regulating RA responsiveness of cell-fate committed prospermatogonia, maintaining a balance between undifferentiated and differentiating spermatogonia during the initial round of spermatogenesis. Our data suggest that Pramel1 more significantly affects progenitors than other subtypes of germ cells in young males, and contributes to maintaining undifferentiated spermatogonial populations in mature mice. Overall, PRAMEL1 acts as a fine-tuner in RA signaling to ensure the proper establishment of the first and subsequent rounds of spermatogenesis.

All animal procedures were approved by the Animal Care and Use Committees of the Penn State University (protocol 46391). The animals were given free access to food and water under a 12 h light and 12 h dark cycle. The founder mice had a genetic background of C57BL/6 and were produced using either a CRISPR/Cas9 approach or the traditional Cre-loxP system. The wild-type (C57BL/6) male and female mice, and the stimulated by retinoic acid 8 (Stra8)–iCre [B6.FVB–Tg(Stra8-iCre)1Reb/LguJ] male mice were purchased from the Jackson Laboratory (017490).

The CRISPR/Cas9 system was used to generate Pramel1 gKO mice. The guided RNAs (5′-GGAGCTGCGTATGTACTGCA-3′ and 5′-CTTAACTTCGCCCCTTACTT-3′) target the 3rd exon of Pramel1 and delete 50 bp of sequence that would lead to a codon shift for the rest of the sequence. The truncated Pramel1 protein was produced as the results of codon shift in the Pramel1 gKO mice. The transgenic mice had a 50 bp deletion of Pramel1 exon 3 (GTATGTACTGCATCAGTAATCCTGTCTGCCTGCTTAACTTCGCCCCTTAC). The five founder mice (two male and three female heterozygous mice) were used to mate in order to produce homozygous global knockout mice.

The Id4-eGfp⁺ reporter mouse line was generated and is described in a previous report (Chan et al., 2014). To generate multi-transgenic Id4-eGfp⁺Pramel1 gKO mice, the Id4-eGfp⁺ transgenic mice were crossed with Pramel1 gKO mice to produce F1 heterozygotes. F1 x F1 mating was used to produce the F2 homozygous mice.

In the Pramel1 cKO mice, exons ∼2-4 were selected as the conditional knockout regions. The two LoxP sites were inserted in the genome as shown in Fig. 1B. The alleles with a genotype of Pramel1^fl/+ were intercrossed to produce homozygous females (Pramel1^fl/fl), which were then bred with Stra8–cre males to produce heterozygous males (Stra8-cre; Pramel1^fl/+). The heterozygous male offspring carrying the Stra8-cre transgene intercrossed with Pramel1^fl/fl females to generate homozygous Pramel1 cKO males (Stra8-cre; Pramel1^fl/fl). The Pramel1 is in chromosome 4. In the Pramel1 cKO males, the floxed Pramel1 allele (exons 2∼4) was deleted only after germ cells responded to RA at postnatal day 3 (P3). The homozygous Pramel1^fl/fl male mice without Stra8–cre transgene served as control animals (floxed) in this study.

The genomic DNA isolated from tail snips were used for PCR to determine the genotypes of mice. The PCR primers sequences and annealing temperature (Tm) are listed in Table S1. The Pramel1 gKO primer pair, the floxed for cKO primer pair, the Stra8-cre for cKO primer pair and the Id4-eGfp⁺ mice primer pair were all applied with 60°C for the annealing temperature. All the PCR reactions were performed in 20 μl with 10-50 ng genomic DNA, 2.5 μM of each primer and 0.5 unit of Taq polymerase (BIO–21105, Bioline). PCR products were resolved on 1.5% agarose gels with ethidium bromide in 1×Tris-acetate-EDTA buffer and imaged with a GelDoc XR+ Image System.

Testes from Pramel1 gKO, Pramel1 cKO, floxed and wild-type mice at the age of P7, P14, P21, P35, P41, P60, P120 and P365 were collected and weighed (n=3∼5). The testis index [(bilateral) testicular weight (g)/body weight (g)×100] is testis weight normalized to body weight. The caput and cauda epididymis of P41 and cauda epididymis of P60, P120 and P365 mice were minced in 1 ml PBS. Spermatozoa were allowed to swim out of the minced epididymides during a 15 min incubation at 37°C, which resulted in a homogeneous sperm suspension after pipetting the PBS and spermatozoa gently. To count the spermatozoa, the suspension was diluted with pure water and the number of spermatozoa was counted with a hemocytometer (Konno et al., 2016).

To evaluate whether the global and conditional deletion of the Pramel1 gene would affect the fertility of the mice, short- and long-term breeding tests were performed. In the short-term breeding test, juvenile male Pramel1 gKO and wild type (at P35) were co-caged with mature CD1 wild-type virgin female mice (8 weeks of age) for 15 days. For the long-term breeding test, the mature Pramel1 gKO, Pramel1 cKO, floxed and wild-type male mice (8 weeks of age) were continuously mated with CD1 wild-type virgin (8 weeks of age) female mice for 6 months. One male and one female were housed in a cage. The mating (plug) rate and litter size were recorded. At least seven mating cages (n=6 or 7) were set up for each genotype.

Testes from the Pramel1 gKO, Pramel1 cKO, floxed and wild-type mice at P7, P14, P21, P35, P41, P60, P120 and P365 (n=∼3-5), were fixed in Bouin's solution (Sigma-Aldrich). They were embedded in paraffin wax and sectioned (5 μm), followed by a standard procedure for Hematoxylin and Eosin (H&E) staining at the histology laboratory in the Microscopy and Cytometry Facility at the Huck Institutes of the Life Sciences, Pennsylvania State University (Mistry et al., 2013). To classify the seminiferous tubules, all seminiferous tubules on cross-sections of the entire testis were examined and categorized into normal tubules and abnormal tubules based on the structure of the seminiferous epithelium, layer and number of germ cells. For mice at or older than P60, at least 100 randomly obtained cross-sections of seminiferous tubules per mouse (bilateral testis) were examined. Hematoxylin and Eosin staining images were captured using an Olympus BX51 microscope. The diameter of seminiferous tubules was measured across the minor axis of their profiles in the images using Olympus cellSens (Ver.2.2) imaging software of an Olympus BX51 microscope. The stages (I-XII) of seminiferous tubules were classified under the microscope, according to morphological criteria based on cellular association and spermatid development (Russell et al., 1993).

Testes from the Pramel1 gKO, Pramel1 cKO and wild-type mice at P2, P3, P4, P5, P6, P7, P14, P21, P35, P120 and P365 (n=∼3-5) were fixed in Bouin's solution overnight, embedded in paraffin wax and sectioned at 5 µm. Slides of testis were dewaxed twice in xylene (10 min each) and then immersed in sequential ethanol baths (100% twice for 10 min, then 95%, 70% and 50% for 5 min each). Antigen retrieval was carried out at 95°C for 15 min in citrate buffer (pH 6.0). After antigen retrieval, testis sections were blocked in 10% donkey serum in Tris-buffered saline (TBS) supplemented with 0.1% Tween 20 (TBS–T) followed by incubation overnight in a humid chamber at 4°C with one or two of the following primary antibodies diluted 1:100 in dilution buffer (1% bovine serum albumin and 1% normal donkey serum in TBS-T): rabbit anti-SOX9 polyclonal antibody (ab5535, Millipore), monoclonal rat anti-TRA98 (73–003, AS ONE International), monoclonal mouse anti-YBX2 (sc-393840, Santa Cruz Biotechnology), monoclonal mouse anti-PLZF (D-9 sc-28319, Santa Cruz Biotechnology), polyclonal rabbit anti-STRA8 (AFF-DF13234, Affinity), monoclonal mouse anti–ID4 (B-5, sc-365656, Santa Cruz Biotechnology) or polyclonal Rabbit anti-cleaved-CASPASE-3 (Asp175) (9661, Cell Signaling). The negative controls for each experiment were incubated with dilution buffer without primary antibody. After incubation, the sections were washed three times in TBS-T (10 min each) and incubated for 1 h at room temperature with the corresponding secondary antibody: donkey anti-rabbit immunoglobulin G (IgG) Alexa Fluor 555 (A31572, Thermo Fisher Scientific), donkey anti-mouse IgG Alexa Fluor 488 (A21206, Thermo Fisher Scientific), donkey anti-mouse IgG Alexa Fluor 555 (A31570, Thermo Fisher Scientific), donkey anti-rat IgG Alexa Fluor 488 (A21208, Thermo Fisher Scientific), donkey anti-rat IgG Alexa Fluor 555 (A48270, Thermo Fisher Scientific) diluted at 1:200 in dilution buffer. For dual staining, two secondary antibodies were mixed in one dilution buffer. Finally, the sections were washed in TBS-T three times for 5 min each, mounted in SlowFade Gold Antifade Mountant with 4′,6-diamidino-2-phenylindole (DAPI; S36938, Thermo Fisher Scientific) and analyzed by fluorescence microscopy using an Olympus BX51 microscope. Over 100 randomly obtained seminiferous tubules from each mouse were used for counting Sertoli cells and various germ cells. Because the Sertoli cell number per tubule varies significantly depending upon the shape of seminiferous tubules, the cell numbers were also counted randomly in three 49305 μm² area (three times for each area) and n=∼3-5 for each group.

Freshly separated seminiferous tubules were fixed in 4% PFA overnight at 4°C. The fixed tissues were washed three times with PBS-T (PBS+1% Triton X-100) with slow rocking for 10 min each, at room temperature. The tissues were dehydrated in a graded series of ethanol (50%, 70%, 95% and 100% for 10 min each). After dehydration, the tissues were rehydrated in a graded series of ethanol (100%, 95%, 70% and 50% for 10 min each). Next, the tissues were washed four times in PBS-T for 20 min each. After washing, tissues were blocked for 1 h in 1% BSA+0.2% non-fat dry milk powder in PBS supplemented with 0.3% Triton X-100 (blocking buffer) followed by incubation overnight in a humid chamber at 4°C with one or two of the following primary antibodies diluted 1:50 in blocking buffer: rabbit polyclonal anti-SOX9 (ab5535, Millipore), monoclonal rat anti-TRA98 (73-003, AS ONE International), polyclonal rabbit anti-cleaved caspase 3 (Asp175) (9661, Cell Signaling), monoclonal mouse anti–ID4 (B-5, sc-365656, Santa Cruz Biotechnology) or custom-made rabbit anti-PRAMEL1 antibodies overnight at 4°C. Tissues were washed four times in PBS-T (20 min each) and the secondary antibodies were incubated with the tissues for 1 h at room temperature. After washing in PBS-T three times for 20 min each, tissues were mounted with SlowFade Gold Antifade Mountant with 4′,6-diamidino-2-phenylindole (DAPI; S36938, Thermo Fisher Scientific) and analyzed by fluorescence microscopy using an Olympus BX51 and IX83 confocal microscope. Z-stacks were flattened using Olympus cellSens (Ver.2.2) imaging software. Germ cell number and the length of seminiferous regions were counted and measured based on the flattened images (n=∼3-5).

The western blot protocol essentially followed a previous method (Liu et al., 2017). Briefly, proteins from testis tissue were extracted using CelLytic MT Cell Lysis Reagent (C3228, Sigma-Aldrich) with protease and phosphatase inhibitors (1860932 and 1862495, Thermo Fisher Scientific). The testis protein extracts were combined with 4× Laemmli sample buffer (1610747, BioRad) and 10× Bolt sample reducing agent (B0009, Life Technologies), then denatured by boiling at 90°C. Denatured protein extracts were separated using BioRad Stain Free Gels (4568044, BioRad) and electro-transferred to polyvinylidene difluoride membranes (IPVH00010, Millipore). The membranes were blocked in 5% non-fat dried milk in TBST, incubated overnight at 4°C with the anti-RARα (sc-515796, Santa Cruz Biotechnology), anti-PRAMEL1 and anti-YBX2 at a dilution of 1:500, washed with TBST and incubated with anti-rabbit/mouse IgG, HRP-linked secondary antibody (7074S/7076S, Cell Signaling Technology, 1:1000 dilution) for 1 h at room temperature. The reactive proteins were detected using SuperSignal West Femto Maximum Sensitivity Substrate (34095, Thermo Fisher Scientific). Western blot data were analyzed using the BioRad ChemiDoc Imaging System.

The Pierce Crosslink Magnetic IP/co-IP Kit (88805, Thermo Fisher Scientific) was used for the co-immunoprecipitation experiment. Custom-made rabbit anti-PRAMEL1 or monoclonal mouse anti-RARα antibodies were bound to protein A/G magnetic beads, and the testis proteins were added to the beads and incubated overnight at 4°C. The protein complexes were then eluted from the beads for western blotting with anti-PRAMEL1 or anti-RARα antibodies, respectively.

Wild-type, Pramel1 cKO, gKO, Id4-eGfp⁺ reporter line and Id4-eGfp⁺Pramel1 gKO mice received i.p. injections of 10 μl/g body weight of 2.5 mg/ml all-trans RA (Sigma-Aldrich) or 10 μl/g body weight of 5 mg/ml WIN18,446 (sc-295819; Santa Cruz Biotechnology) in 10% DMSO-H₂O (n=∼3-4) at P2. Vehicle control mice received 10% DMSO-H₂O as a control in this experiment. At P5, testes of mice were collected and the seminiferous tubules were isolated. At P41, mice were euthanized to harvest testis and epididymis.

For heat-stress (HS) experiments, 2-month-old wild-type and the Pramel1 gKO mice were bathed in a 42°C water bath for 20 min (Wang et al., 2019a), and testis samples were collected after HS treatment at 5 days, 14 days and 38 days (D5, D14 and D38; n=3). The control groups were 2- and 3-month-old mice without HS treatment.

TRlzol reagent (Invitrogen) and Superscript III First-Strand Synthesis System (Invitrogen) were used to extract RNA, and reverse transcription was performed as described by the manufacturer. The extraction and qRT-PCR was performed according to the previous report (n=3) (Mistry et al., 2013). Sequence-specific primers used for qRT-PCR amplification of mouse Thy1, Id4, Plzf, Kit, Stra8, Gfra1, Ccr1, Ccl3 and Actb are shown in Table S1.

The experiment was performed as previously described (Law et al., 2019). Briefly, the neonatal testes and tunica of Id4-eGfp⁺ and Id4-eGfp⁺Pramel1 gKO mice (n=3) were removed in HBSS and incubated for 10 min at 37°C in 0.25% trypsin/EDTA (Thermo Fisher Scientific) with 2 mg/ml deoxyribonuclease I (Sigma-Aldrich). During the incubation, the solution was gently pipetted up and down. Fetal bovine serum (FBS, 10%) was used to quench the trypsin digestion and a solution of 1% FBS, 10 mM HEPES, 1 mM sodium pyruvate, 1 mg/ml glucose, 100 unites/ml penicillin and 100 µg/ml streptomycin in PBS (Thermo Fisher Scientific or Sigma-Aldrich) was used to wash and resuspend the cells after centrifuging at 600 g for 7 min at 4°C. To ensure a single-cell suspension is achieved, the cell suspension was passed through a 40 µm pore size cell strainer. The cell suspensions were then incubated with rabbit anti-DDX4 polyclonal antibody (ab13840, Abcam) at a dilution of 1:100 for 30 min on ice, gently washed three times and incubated with donkey anti-rabbit immunoglobulin G (IgG) Alexa Fluor 555 (A31572; Thermo Fisher Scientific) for another 30 min on ice. The cells were washed three times before analysis on a Guava EasyCyte Plus System (CytoSoft 5.3 software) and data were processed with FlowJo software 10.8.1 (Becton Dickinson). Unstained wild-type testis suspensions at equivalent ages were used to determine background fluorescence. For each sample, EGFP fluorescent signal was divided into thirds as ID4-EGFP^Bright, ID4-EGFP^Mid and ID4-EGFP^Dim populations, as in previous studies (Law et al., 2019). The EGFP-spermatogonia were gated from DDX4 expression.

Apoptosis was evaluated using an In Situ Cell Death Detection Kit, Fluorescein (11684795910, Roche) according to the manufacturer's protocol. In brief, testis sections were incubated with 0.1% sodium citrate with 0.1% Triton X–100 for 10 min at room temperature and washed in PBS. Sections were then incubated with the TUNEL (terminal deoxynucleotidyl transferase dUTP nick end labeling) reaction mixture at 37°C for 1 h. Finally, the sections were washed in PBS three times for 5 min each and mounted SlowFade Gold Antifade Mountant with DAPI. TUNEL staining images were acquired using a fluorescence Olympus BX51 microscope. NC experiments were performed without the TUNEL reaction mixture during incubation. Approximately 300 seminiferous tubules were analyzed, and TUNEL+ cells were counted for the wild-type and the Pramel1 gKO mice at the age P7, P14, P21, P35, P60 and P120 (n=3).

The data were previously analyzed using the normality test (Shapiro–Wilk test) and equal variance test (Brown–Forsythe) using Sigma Plot 12.0 (Statistical Software). After meeting the assumptions of normally distributed data and homogeneity of variance, the difference in treatment levels was evaluated for significance using a one-tailed Student's t-test. Data are expressed as the mean±s.e.m. A value of P≤0.05 was considered statistically significant.

The authors thank Dr Francisco Diaz for his insightful discussions and his help in managing the mouse colonies. We thank Melissa Oatley for her assistance with Id4-eGfp mouse genotyping and EGFP cell flow cytometric analysis, and Chandlar Kern for her critical reading of the manuscript. We also express our gratitude to the Animal Resource Program at Penn State, as well as the Transgenic Mouse and Microscopy and Cytometry Facilities at the Huck Institutes of the Life Sciences, Penn State, for their training of animal procedures and provision of equipment for this research project.

The peer review history is available online at https://journals.biologists.com/dev/lookup/doi/10.1242/dev.201907.reviewer-comments.pdf