E2F1 regulates testicular descent and controls spermatogenesis by influencing WNT4 signaling

The testis must be fully descended into the scrotum during development for normal spermatogenesis to occur. Malposition of the testis in the inguinal canal or abdomen (cryptorchidism) is one of the most common birth defects in humans and is a proven risk factor for infertility and testicular tumors (Hadziselimovic and Herzog, 2001; Lipshultz et al., 1976; Pike et al., 1986). Spermatogenesis is a continuous iterative process within the seminiferous tubules involving somatic Sertoli cells, as well as mitotic, meiotic and differentiating germ cells. Within the testis, the seminiferous tubules are divided into the basal and adluminal compartments by the blood-testis-barrier (BTB), formed by specialized tight junctions between adjacent Sertoli cells (Russell, 1977). The spermatogonial stem cells (SSCs), as well as the proliferating and differentiating spermatogonia up until the pre-leptotene stage of meiosis, remain on the outside of the BTB in the basal compartment of the seminiferous tubules. Meiosis I and II produces the haploid spermatids that undergo spermiogenesis and subsequent spermiation, which occurs in the adluminal compartment protected by the BTB (Russell, 1977). Disruption of these developmental and functional processes, and/or the complex intercellular interactions that maintain testicular architecture, can result in spermatogenic failure.

E2F transcription factor-1 (E2F1) is a transcription factor that displays opposing actions in cellular regulation, depending upon cellular characteristics and microenvironment. E2F1 promotes cell cycle progression by regulating the expression of genes required for DNA synthesis at the G1/S boundary, therefore assisting in DNA repair and contributing to the induction of apoptosis through its interactions with TP53 or TP73. Whether E2F1 leads a cell to undergo proliferation or apoptosis depends upon the molecular characteristics of the cell (Polager and Ginsberg, 2009; Wu et al., 2009). E2F1 executes most of its biological functions through transcriptional activation by upregulating the expression of genes involved in cell cycle, DNA synthesis and replication, checkpoint control, DNA damage repair, apoptosis, autophagy, self-renewal, development and differentiation (Müller et al., 2001; Polager et al., 2002, 2008; Yeo et al., 2011). E2F1 expression is regulated by the direct binding of retinoblastoma-1 (RB1) and by post-translational modifications involving protein degradation, phosphorylation or acetylation (Meng and Ghosh, 2014; Mundle and Saberwal, 2003). RB1 binding to E2F1 prevents the interaction of E2F1 with the transcription machinery of the cell. Like RB1, E2F1 is highly expressed in spermatocytes but, unlike RB1, E2F1 is not expressed in Sertoli cells (Rotgers et al., 2015; Yan et al., 1997).

Infertile men have a higher incidence of microdeletions and microduplications in E2F1, and a subset of these men also have cryptorchidism (Jorgez et al., 2015; Rocca et al., 2019). Mouse models of E2f1 deletion or overexpression display similar phenotypes to those observed in humans. Mice lacking E2f1 exhibit testicular atrophy with aging due to a loss of spermatogonia, as well as an increased testicular tumor incidence (Hoja et al., 2004; Rotgers et al., 2015; Yamasaki et al., 1996). Conversely, transgenic mice overexpressing E2f1 are infertile due to testicular atrophy caused by TP53-independent apoptosis of germ cells (Holmberg et al., 1998). These results indicate that both deletion and overexpression negatively impact spermatogenesis. E2F1 is a master regulator of cellular processes with a role that extends beyond cell cycle regulation, proliferation and apoptosis. Interestingly, the mechanism by which E2F1 regulates germ cell function is unknown, yet it is required to maintain male fertility. In addition, the role of E2F1 in testicular descent has never been investigated. This study shows that E2F1 plays an important role in testicular descent, as well as in germ cell survival to some extent through the WNT4 signaling.

Previously, our laboratory reported that a subset of infertile men with microdeletions, microduplications and single nucleotide polymorphisms (SNPs) in E2F1 displayed spermatogenic failure and cryptorchidism (Jorgez et al., 2015). Additional, copy number variants (CNVs) in E2F1 associated with infertility, cryptorchidism and testicular tumors were reported by others (Avila et al., 2013; Rocca et al., 2017, 2019; Wanderley et al., 2005). Even though microduplications in E2F1 are more common than microdeletions in cryptorchid men, in this study the testicular position of E2f1-null mice was evaluated as both E2f1 deletion and overexpression in mice negatively impacts spermatogenesis (Hoja et al., 2004; Holmberg et al., 1998; Rotgers et al., 2015; Yamasaki et al., 1996). In 95% of E2f1-null mice, the testicular descent was abnormal, with significant increases in bilateral (69%) and unilateral (26%; Fig. 1D) cryptorchidism above the inguinal ring compared with wild-type (WT) mice (4.5%, Fig. 1A). One of 76 E2f1-null mice analyzed lacked a left testis (monorchism) with the contralateral testis being hypertrophic with a weight of 60.7 mg [twice the size of the average weight (29.57 mg) of age-matched E2f1-null mice]. Cryptorchid E2f1-null mice displayed an underdeveloped cremasteric sac containing less striated muscle and connective tissue sheaths and decreased amounts of collagen (Fig. 1E) compared with WT mice (Fig. 1B). Transmission electron microscopy (TEM) of E2f1-null mice revealed that the gubernacular ultrastructure exhibited abnormal organelles, including a dilated endoplasmic reticulum, autophagosomes and irregularly shaped mitochondria with disorganized cristae (Fig. 1F-H) that is reminiscent of the ultrastructural abnormalities seen in cremasteric muscle of cryptorchid boys (Tanyel et al., 2001). Mast cells were commonly present (Fig. 1I), which are reportedly linked to the inflammation and fibrosis that occurs with cryptorchidism (Mechlin and Kogan, 2012). None of these anomalies were observed in WT mice (Fig. 1C). One 12-month-old E2f1-null mouse developed a testicular B-cell lymphoma (Fig. S1).

Although spermatogenic failure was described in E2f1-null mice (Rotgers et al., 2015; Yamasaki et al., 1996), how E2f1 causes spermatogenic failure is still unknown. As E2F1 CNVs play an important role in male fertility (Jorgez et al., 2015; Rocca et al., 2019), it is necessary to understand the mechanism by which E2f1 regulates spermatogenesis. Testes of E2f1-null mouse were characterized at different ages. E2f1-null mouse testes were significantly lighter (P<0.001, 50%) at 3 weeks and 3 months of age, and by 7 months of age, the E2f1-null testes weighed up to 72% less than those of WT mice (Table 1). Seminiferous tubule diameter and epithelial thickness were significantly reduced at all ages in E2f1-null mice (Fig. 2J; Table 1). Testicular architecture in the seminiferous tubules was analyzed and categorized as normal histology (all germ cell layers present in the appropriate position and cellularity), abnormal (loss or abnormal position of germ-cells) or Sertoli cell only (SCO, no germ cells present). No differences were observed in the percentage of normal seminiferous tubules (95-100%) in WT mice at any age (Fig. 2A-C). However, as early as 3 weeks of age, E2f1-null testes had 21.2% abnormal seminiferous tubules (decreased or germ cell depleted; Fig. 2E,K). The decline in germ cell numbers in E2f1-null mice increased as they aged with 27.3% of seminiferous tubules being abnormal at 3 months (Fig. 2F,K) and 79.4% of seminiferous tubules being abnormal by 7 months of age (Fig. 2G,K). In addition, adult E2f1-null mice (7 months of age) showed evidence of Leydig cell hyperplasia (Fig. 2G).

During normal spermatogenesis, dynamic modulation of tight and adherens junctions are important for Sertoli-Sertoli and Sertoli-germ cell interactions to facilitate migration of developing germ cells from the basal to adluminal compartments of the seminiferous tubules (Wong et al., 2008). In E2f1-null mice, Sertoli cell nuclei were frequently located towards the lumen of the seminiferous tubules with associated germ cell disorganization (Fig. 2E-G). As the assembly of tight junctions is impacted by metallo-serine-proteinase inhibitors, the expression of Timp1 and Serpina3n genes was investigated (Fig. 3D). As with the Rb1-null mouse testes (Nalam et al., 2009), Timp1 and Serpina3n expression was upregulated in E2f1-null mice. Sertoli cell disruption within the seminiferous epithelium results in major structural abnormalities to the BTB, including mislocalization of crucial junctional and cytoskeletal elements, and leads to disruption of barrier function. A biotin tracer, used as a marker of BTB stability, was injected into the interstitial space of the testis to define the relative permeability of the seminiferous tubules. In control mice, the biotin tracer was present in the interstitial spaces and basal compartment but excluded from the adluminal compartment (Fig. 4A-C). In contrast, the biotin tracer passed through the BTB and entered the adluminal compartment in more than 60% of the seminiferous tubules in the testes of 3-month-old E2f1-null mice (Fig. 4E-H). Testicular staining was absent in control mice injected with PBS (Fig. 4D). The loss of BTB integrity may be associated with the upregulation of genes involved in tight and adherens junctions.

To further define the effect of E2f1 deficiency, immunohistochemistry using ZBTB16 (commonly known as PLZF, a marker of undifferentiated spermatogonia), and SOX9 (a marker of Sertoli cells) was performed at various times after birth (Fig. 5). No differences in the numbers of ZBTB16⁺ and SOX9⁺ cells were present in WT and E2f1-null mouse testes before the initiation of puberty (Fig. 5I). Yet, by 3 months of age, a significant decrease was observed in only the number of ZBTB16⁺ cells in E2f1-null mouse testes (P<0.001), indicating the presence of fewer undifferentiated spermatogonia in the seminiferous tubules after puberty. This decline in undifferentiated spermatogonia cell number correlated with the absence of ZBTB16⁺ cells in 53% of the seminiferous tubules cross-sections in E2f1-null mice (Fig. 5J; P<0.001). These results indicate that E2f1 deficiency negatively affected testicular germ cell development and function, eventually leading to germ cell depletion.

Mature spermatozoa were reduced in the epididymis of E2f1-null mice (Fig. 2H) compared with WT mice (Fig. 2D), correlating with the significant decrease in epididymal sperm motility and density (Table 1). Hormonal values [follicle-stimulating hormone (FSH), luteinizing hormone (LH) and testosterone] varied among mice and no significant difference was identified between WT and E2f1-null males at any age (Table 1).

To define their fertility, sexually mature E2f1-null males (6 weeks old) were mated with age-matched WT females over a 6 month period (Table 2). Each WT male sired six litters throughout the test period; however, of the ten E2f1-null males evaluated, one male was infertile, and the remainder of the cohort was subfertile (Fig. 2I). During the first 2 months of breeding, no differences were identified in E2f1-null male fecundity. After the second litter was born, there was a significant increase in the length of time to achieve a subsequent pregnancy. By this time, three of the E2f1-null males had become infertile. Only two of the ten E2f1-null males produced a fifth litter, and none produced a sixth litter, reflecting their progressive decline in fertility and fecundity culminating in infertility. The overall time to pregnancy was significantly different between the two groups of male mice (35.04±17.07 versus 24.20±2.89 days; P<0.01). E2f1-null males had a significant decrease in the number of pups produced per litter compared with those of WT control (5.96±2.84 versus 8.74±2.92 pups; mean±s.d., P<0.01). This decreased in fertility with age in E2f1-null males positively correlated with a significant 30% decrease in testicular weight and a 52% decrease in sperm motility compared with E2f1- null males at 3 and 7 months of age (Table 1).

To elucidate the effect of E2f1 deficiency on spermatogenesis, testicular gene expression in E2f1-null and WT mice was defined at 3 and 7 months of age (Table S1). The expression levels of 66 genes were studied. Genes were selected based on testicular expression and E2f1 interaction, as determined by literature review (Boyer et al., 2012; Choi et al., 2004; Devgan et al., 2005; Liu et al., 2014; Nalam et al., 2009). At 3 months, 18.5% (6.2% upregulated and 12.3% downregulated) of the selected genes assessed had greater than twofold altered gene expression, including those involved in the regulation of the cell cycle, testicular descent, testicular function, tissue remodeling and those in the Wnt-signaling pathway. The magnitude of gene expression alteration was identical at different ages in 49 of the genes; however, progressively altered gene expression associated with the severe dysregulation of 11 genes was present at 7 months of age.

Because E2f1 is an important regulator of the cell cycle, its deficiency impacted expression of the cell cycle genes measured, as well as other genes within its transcriptional regulatory network (Fig. 3A). Expression levels of several genes encoding inhibitors of cell cycle progression were unaffected (Trp53 and Cdkn1a), whereas others were only moderately downregulated (Rb1). Cyclin proteins (Ccnd1, Ccnb1 and Ccndbp1) were significantly downregulated in E2f1-null testes. Additionally, Ncaph, an E2f1 cell cycle target gene altered in testes of Rb1-null mice, was also highly downregulated (Nalam et al., 2009). No significant changes in the expression of apoptotic genes were present in adult mice (Fig. 3B).

As E2f1-null mice have cryptorchidism, the expression of female gonad sex determination genes was analyzed. Rspo1 and Foxl2 showed no difference, whereas Wnt4 was significantly increased (Table S1, Fig. 3D). As E2F1 represses WNT4 in several cell lines (Devgan et al., 2006; Suzuki et al., 2011), the expression of Wnt4 and several members of the Wnt-signaling pathway were defined (Fig. 3C). Testicular Wnt4 expression was significantly increased at all ages in all E2f1-null mice with an average 20-fold increase in Wnt4 expression in 7-month-old E2f1-null mice (Fig. 3D). The expression of two other Wnt ligands (Wnt5a and Wnt6) was significantly increased. Expression of Ctnnb1 was downregulated in E2f1-null mouse testes (Fig. 3C). The decreased Ctnnb1 gene expression correlated with decreased CTNNB1 protein in E2f1-null mice (Fig. 6D). WT mice expressed CTNNB1 in the basal compartment of the seminiferous tubules at the Sertoli-germ cell intercellular junction (Fig. 6C).

WNT4 is a secreted protein involved in female sex differentiation that needs to be repressed for testicular development (Jameson et al., 2012). Female mouse embryos lacking Wnt4 develop testicular structures at birth (Ottolenghi et al., 2007; Vainio et al., 1999). On the contrary, in male embryos, loss of Wnt4 signaling results in delayed testicular development. These data support a crucial role for Wnt4 in the sex-specific patterning of the bipotential gonad for both sexes (Jeays-Ward et al., 2004, 2003). WNT4 expression pattern was further investigated in E2f1-null mouse testes. At birth, the mouse testis contains only undifferentiated type A1 spermatogonia. By postnatal day (P)3, differentiation begins through a series of mitotic divisions into more advanced spermatogonial stages. By P8-10, spermatocytes are observed in the leptotene phase of meiosis. By P12, pachytene spermatocytes are present. Round spermatids (postmeiotic cells) are observed by P20 (McCarrey, 2013; Nebel et al., 1961). At P1, WNT4 was highly expressed in the cytoplasm of gonocytes, peritubular and interstitial cells in WT mouse testes (Fig. 6A,B). At P5, WNT4 was highly expressed in the cytoplasm of spermatogonia type A, peritubular and interstitial cells in WT mouse testis (Fig. 6B). At P7, proliferating spermatogonia develop within the seminiferous epithelium and WNT4 expression was limited to the interstitial and peritubular cells (Fig. 6E). At P14, WNT4 was expressed in the interstitial cells (Fig. 6F). By adulthood, WNT4 was almost undetectable in the testes, with the exception of some variable expression present in interstitial cells (Fig. 6H, Fig. 7A). There was non-specific staining observed in some of the sperm in WT testes (Fig. 7A), which could be attributed to non-specific IgG binding to the sperm surface (Richards and Witkin, 1984). Similar to WT mice from P7, age matched E2f1-null mice expressed WNT4 in the peritubular and interstitial cells but not in the germ cells (Fig. 6I). However, unlike WT mice, WNT4 continued to be highly expressed in both peritubular and interstitial cells in P14 and P21 E2f1-null mice (Fig. 6J,K). Adult E2f1-null mice expressed WNT4 in the interstitial cells, and 67% of the mice (n=12) showed evidence of germ cell expression in both the membrane and cytoplasm (Fig. 6L, Fig. 7B), suggesting that WNT4 silencing in the seminiferous tubules is absent in E2f1-null mice.

WNT4 is a negative regulator of SSCs as it acts downstream of CTNNB1 to downregulate SSC activity and increase germ cell apoptosis, suggesting an important role for WNT4 in the postnatal testes (Boyer et al., 2012). WNT4 and E2F1 are independently associated with germ cell regulation. To distinguish between the effect of E2f1 deficiency and Wnt4 overexpression on testicular function in vivo, a double-null E2f1 and Wnt4 mouse was generated. To circumvent Wnt4-null lethality at birth (Stark et al., 1994), mice with the Wnt4 gene flanked by loxP site (Wnt4^f/f) (Kobayashi et al., 2011) were crossed with E2f1-null mice (Field et al., 1996). To achieve recombination in germ cells, Stra8-Cre mice, which express Cre only in male germ cells at P3 (Sadate-Ngatchou et al., 2008), were used to produce E2f1^−/−;Wnt4^f/f;Stra8^cre/+ double knockout mice. To determine whether lack of Wnt4 alone in the germ cells has an effect on fertility, Stra8-Cre mice were crossed with Wnt4^f/f mice, herein called Wnt4^f/f;Stra8^cre/+. The expression of WNT4 in the testes of WT, E2f1-null, Wnt4^f/f;Stra8^cre/+ and E2f1^−/−;Wnt4^f/f; Stra8^cre/+ was analyzed with the investigator blinded to the genotype. WNT4 expression was absent in germ cells of WT, Wnt4^f/f;Stra8^cre/+ and E2f1^−/−;Wnt4^f/f;Stra8^cre/+ adult testes (Fig. 7A,C,D); a finding in contrast to the WNT4 expression noted in E2f1-null testes (Fig. 7B). This result, showing effective recombination of Wnt4 allele in germ cells of E2f1^−/−;Wnt4^f/f;Stra8^cre/+ mice, correlates with the results showing very low expression of Wnt4 mRNA in WT and E2f1^−/−;Wnt4^f/f;Stra8^cre/+ mice, and a high expression in E2f1-null testes (Fig. 8J).

The fertility of E2f1^−/−;Wnt4^f/f;Stra8^cre/+ mice was assessed after mating 6-week-old E2f1^−/−;Wnt4^f/f;Stra8^cre/+ male mice (n=5) with WT females (n=5) for 6 months, in which litter number, size and frequency were compared between groups (Fig. 8D,E). All E2f1^−/−;Wnt4^f/f;Stra8^cre/+ mice were fertile. Their fertility was comparable to WT mice (similar number of litters and pups produced in both groups) and significantly different from E2f1-null mice. E2f1^−/−;Wnt4^f/f;Stra8^cre/+ testes, although cryptorchid, were bigger than those in E2f1-null (64.16±8.71 mg versus 41.11±3.95 mg; mean±s.d., P<0.001), but smaller than those in WT mice (64.16±8.71 mg versus 99.47±8.36; mean±s.d., P<0.001) (Fig. 8A,F). Most E2f1^−/−;Wnt4^f/f;Stra8^cre/+ testes had normal seminiferous tubules (Fig. 8C) in contrast to E2f1-null (Fig. 8B). E2f1^−/−;Wnt4^f/f;Stra8^cre/+ epididymis were similar to those analyzed in WT mice and bigger than E2f1-null epididymis (Fig. 8G). When comparing the sperm of E2f1^−/−;Wnt4^f/f;Stra8^cre/+ with WT mice, sperm density was similar, but the motility was significantly lower in E2f1^−/−;Wnt4^f/f;Stra8^cre/+ mice (26.63±13.01% versus 51.85±16.19; mean±s.d., P<0.001; Fig. 8H,I); both WT and E2f1^−/−;Wnt4^f/f;Stra8^cre/+ sperm density and motility were significantly higher than E2f1-null mice. In addition, E2f1^−/−;Wnt4^f/f;Stra8^cre/+ mice displayed a testicular gene expression profile comparable to WT (Fig. 8J,K). The spermatogonial marker, CCND1 (Beumer et al., 2000), was expressed in all mice analyzed (Fig. 7E-H) indicating the presence of proliferative spermatogonia in both E2f1-null and E2f1^−/−;Wnt4^f/f;Stra8^cre/+ testes. However, a decrease in CCND1 spermatogonia was observed in E2f1-null testes, but not in WT or E2f1^−/−;Wnt4^f/f;Stra8^cre/+ mice (Fig. 7M). These immunohistochemistry results correlate with the qPCR findings (Fig. 8J). Also, the decline in undifferentiated ZBTB16⁺ spermatogonia cell number seen in E2f1-null seminiferous tubules (Fig. 5J, Fig. 7F) was not observed in WT, Wnt4^f/f;Stra8^cre/+, or E2f1^−/−;Wnt4^f/f;Stra8^cre/+ testes (Fig. 7I,K,L,N).

Wnt4^f/f;Stra8^cre/+ testicular weight was similar to WT mice (93.42±6.46 mg versus 94.54±13.46; mean±s.d., Fig. 8A). The fertility of Wnt4^f/f;Stra8^cre/+ mice was assessed after mating 6-week-old Wnt4^f/f;Stra8^cre/+ male mice (n=4) with WT females (n=4) for 6 months. Wnt4^f/f;Stra8^cre/+ male mice were fertile with an average time between litters of 27.50±5.06 days, and 8.3±2.5 pups per litter (mean±s.d.). The fertility of Wnt4^f/f;Stra8^cre/+ male was not significantly different to WT male mice (time between litters was 24.20±2.89 days and pups per litter 8.74±2.92; mean±s.d.).

Cryptorchidism is the most common congenital anomaly in full-term newborn males, affecting ∼6% of newborns, and commonly leads to infertility (Barthold and González, 2003; Hadziselimovic and Herzog, 2001; Lipshultz et al., 1976). Unilateral (13%) or bilateral (89%) cryptorchid men without orchidopexy (surgical placement of the testes in the scrotum) are at high risk of developing azoospermia due to spermatogenic arrest (Hadziselimovic and Herzog, 2001). We previously demonstrated that E2F1 gene-dosage changes are present in infertile and cryptorchid males (Jorgez et al., 2015). Subsequently, Rocca et al. (2019) showed that CNVs in E2F1 predispose men to infertility (Rocca et al., 2017, 2019). In their cohort of infertile men (n=343), 3.5% had CNVs in E2F1, with 50% of them having a history of cryptorchidism (Rocca et al., 2019). Infertility in mice lacking or overexpressing E2f1 has been reported previously (Hoja et al., 2004; Holmberg et al., 1998; Rotgers et al., 2015; Yamasaki et al., 1996). However, the presence of cryptorchidism in these mouse models was apparently never assessed. Occasionally when subjecting mouse models to detailed phenotypic analysis of the genitourinary tract, birth defects, such as cryptorchidism, are frequently overlooked, especially when the testis has completed the first phase of descent and sits superior to the inguinal ring and does not descend into the scrotum. Other examples, where outside laboratories identified genitourinary birth defects not mentioned by the publications describing the testicular phenotype, include mouse models deficient in CRKL (Guris et al., 2001; Haller et al., 2017). Consistent with the presence of cryptorchidism in males with E2F1 gene-dosage changes (Jorgez et al., 2015; Rocca et al., 2017, 2019), 95% of E2f1-null mice had testes located above the inguinal ring, implying an effect on the second phase of testicular descent. Dysfunction of a master regulator, such as E2f1, could affect different stages of testicular descent, by regulating genes required for testicular descent, such as Insl3, Ar, Lgr8, Hoxa10, Amh, Dmrt1 and Fst (Barthold et al., 2008; Bugrul et al., 2019; Feng et al., 2004; Harris et al., 2010; Kolon et al., 1999; Raymond et al., 1999; Simoni et al., 2008; Tannour-Louet et al., 2010, 2014). Promoter analysis of the genes mentioned above revealed the presence of multiple putative E2f1 binding sites in the promoter regions of both human and mouse genes (data not shown). Androgens are essential for the inguinoscrotal second phase of testicular descent (Husmann and McPhaul, 1992). E2f1 and androgen receptor (Ar) genes functionally collaborate and interact (Altintas et al., 2012). However, as observed in cryptorchid Rxfp2-null mice (Kaftanovskaya et al., 2011), E2f1 could affect testicular descent by regulating novel signaling pathways, such as the Wnt/β-catenin pathway.

Although cryptorchidsm was a common occurence in E2f1-null mice, one E2f1-null mouse also showed the rare defect of left-sided monorchism. Monorchism in humans often occurs on the left side, with contralateral testicular hypertrophy (Laron and Zilka, 1969). The presence of cryptorchidism and monorchism in E2f1-null mice suggests an important role for E2f1 in testicular descent and development (Fig. 9). Although rare, an older E2f1-null mouse had a testicular tumor. The C57Bl/6 background of the E2f1-null mice may explain the low incidence of testicular tumors observed in this study compared with the studies in the 129/Sv inbred strain, in which differences in the balance of cell cycle regulators caused increased testicular tumor susceptibility (Cook et al., 2011). Future experiments could include the generation of E2f1-null mice in a 129/Sv background, especially as CNVs in E2F1 were significantly increased (6.5%) in a cohort of 261 men with testicular tumors (Rocca et al., 2017).

Mice with inguinal cryptorchidism can be fertile depending upon strain background (Kaftanovskaya et al., 2013). This observation is similar to the phenotype found in human males with cryptorchidism, as only 13% of men with treated unilateral undescended testis (UDT) have spermatogenic failure (Hadziselimovic et al., 2011; Hadziselimovic and Herzog, 2001; Verkauskas et al., 2019). In mouse models of cryptorchidism, the higher testicular temperature may contribute to the infertility phenotype in part, as occurs in some cases of human cryptorchidism. Nevertheless, for some genetic and genomic defects causing cryptorchidism, it is the gene-dosage effect on the cellular processes in the seminiferous tubules, rather than solely the elevated temperature due to the malposition of the testis in the pelvis or abdomen, that contributes to the infertility (Haller et al., 2017; Tannour-Louet et al., 2014). Thus, like these human and mouse models, a late meiotic arrest, the typical histopathology of cryptorchid boys, was not observed in E2f1-null mice. This observation suggests a direct effect of these gene-dosage changes on the process of spermatogenesis, rather than the known effects of elevated temperature due to cryptorchidism on spermatogenesis. Similar to what is observed in humans, although elevated testicular temperatures may contribute to the spermatogenic failure in E2f1-null mouse, it may not be the main or only cause of the spermatogenic failure. Multiple cellular processes required for normal spermatogenesis converge on the evolutionarily conserved protein, E2F1. First, control of the cell cycle during spermatogenesis is essential to guarantee self-renewal and proper differentiation of germ cells. Cells respond to mitogenic stimuli and will only proceed through the phases of the cell cycle if they pass the first restriction checkpoint. RB1 (restriction point switch) is hypo-phosphorylated in resting G0 cells, forming a complex with E2F1. In late G1, cyclin-dependent kinases phosphorylate RB1, leading to its dissociation from E2F1, and thereby promoting the expression of genes under the control of E2F1 (Chellappan et al., 1991). Because the cell cycle plays a crucial role in spermatogenesis, dysfunction of RB1 and E2F1 have substantial effects on fertility (Hu et al., 2013; Jorgez et al., 2015; Rotgers et al., 2015; Yang et al., 2013). E2F1 expression in mitotic spermatogonia is gradually diminished as spermatogonia differentiate and enter the meiotic pathway, suggesting a function during the proliferative phase of spermatogenesis (Rotgers et al., 2015). The significant decrease in ZBTB16⁺ cells in E2f1-null adult males could indicate that E2F1 is necessary for spermatogonial maintenance, and the loss of spermatogonia likely contributed to the progression of the E2f1-null phenotype. E2f1 binding sites, which are present in the promoter regions, include expression of several genes necessary for cell cycle progression, such as those required for DNA replication (e.g. DHFR and RNR) and cyclin-dependent kinase activity (e.g. cyclins D, E, and A). Mutation of E2f1 binding sites in a subset of these promoters demonstrates the importance of E2f1-mediated interaction in regulating gene expression (Slansky and Farnham, 1996). Downregulation of genes involved in DNA synthesis/cell cycle control in E2f1-null mice suggests that loss of E2f1 in the testes causes cell cycle impairment and germ cell loss. Second, regulation of matrix metalloproteinase genes by E2F1 is important (Johnson et al., 2012). E2F1 is not expressed in Sertoli cells; however, its role in the integrity of the BTB could be through the regulation of Timp1 and Serpina3n, which play important roles in the tubular remodeling of cell-cell junctions. In addition, the defect in the BTB could be a consequence of cryptorchidism. In rats with unilateral cryptorchidism, BTB protein levels are similar in both testes, but their distributions are altered in the UDT affecting its function and contributing to spermatogenic failure (Kato et al., 2020). The cell-cell communications between the germ cells and Sertoli and Leydig cell compartments are important for spermatogenesis (Mruk and Cheng, 2004), and the altered cell-cell interaction in E2f1-null mice plays an important role in their testicular phenotype.

E2f1-null testes showed a progressive loss of spermatogonia, but the mechanism causing this germ cell loss is unknown. Wnt signaling is activated in stress situations causing cell death; in particular, Wnt4 is upregulated after testicular exposure to genotoxic agents, phthalates and retinoic acid (Pećina-Šlaus, 2010; Spade et al., 2019; Xu et al., 2014). Ectopic activation of WNT4/CTNNB1 signaling in Sertoli cells downregulates SSC activity (Boyer et al., 2012). E2f1 directly represses Wnt4 in the brain and in keratinocytes by binding to functional E2f1 binding sites in Wnt4 (Devgan et al., 2006; Suzuki et al., 2011). E2f1-null mice expressed WNT4 in the peritubular and interstitial cells in pubertal stages and in germ cells in adult mice, suggesting that the mechanism to silence WNT4 in these cells is absent and the ectopic expression of WNT4 could contribute to the downregulation of SSC activity in the testes, indicating that E2f1 could be a repressor of Wnt4 in the testes (Fig. 9). Thus, developmental events affecting the cell cycle and post-developmental events resulting in inappropriate upregulation of WNT4 could lead to SCO, as WNT4 is secreted in a paracrine manner to amplify its own signal. Wnt signaling molecules elicit a major impact on tissue homeostasis when expressed at the right time and dose (Laine et al., 2013). To define the relative importance of the Wnt4 pathway in E2f1-related testicular dysfunction, rescue experiments were performed. As E2f1-null mice express WNT4 in germ cells and E2F1 is only expressed in germ cells of the seminiferous tubules, we generated double null mice lacking E2F1 and WNT4 in germ cells (E2f1^−/−;Wnt4^f/f;Stra8^cre/+ mice). E2f1^−/−;Wnt4^f/f;Stra8^cre/+ mice were fertile with normal sperm counts, indicating that E2f1 testicular function involves the Wnt4-signaling pathway. Transgenic mice overexpressing human WNT4 generated in the CB6F1 background were fertile, despite their defects in testicular vasculature and testosterone biosynthesis (Jordan et al., 2003). Whether the mice generated in the CB6F1 background represent the true phenotype of WNT4 testicular overexpression in mouse remains uncertain as Wnt4 transgenic founders in the C57BL/6 background were infertile (Jordan et al., 2003). E2f1^−/−;Wnt4^f/f;Stra8^cre/+ testes were larger than those in E2f1-null but did not reach the normal size and weight of those observed in WT mice, suggesting that the dual Wnt4-E2f1 deficiency did not fully overcome the consequences of E2f1 deficiency alone.

As E2F1 regulates the expression of genes involved in a wide range of cellular processes, it is possible that E2F1 activates genes involved in DNA quality control during the critical pre-meiotic period. Thus, loss of E2F1 could lead to a mechanism that prevents the replication of damaged DNA by eliminating germ cells, ultimately resulting in spermatogenic failure. In addition, E2f1 null mice have an increase in testicular Gdnf which could play a role in testicular defects. GDNF-overexpressing mice are infertile with small testes due to a progressive testicular atrophy (Meng et al., 2000). The presence of a testicular tumor in one of our older E2f1-null mice and the defects in the BTB suggest that Gdnf dysregulation could be involved in the pathology of E2f1-null mice.

Many animal models exhibit male infertility, but translation of these findings to clinical diagnosis of infertile men has not necessarily impacted patient diagnosis, in part because of the complexity of the signaling pathways and the numerous genes required for fertility (Matzuk and Lamb, 2002, 2008). In humans, forward-genetics-based studies used array comparative genomic hybridization and next generation sequencing, and studies of families and/or well-defined spermatogenic morphologic and functional anomalies, to identify gene defects associated with human male infertility (reviewed by Punjani and Lamb, 2020; Wang et al., 2020). E2f1-null mice recapitulate the phenotypes observed in men with E2F1 gene-dosage changes that include infertility, cryptorchidism and testicular tumors (Jorgez et al., 2015; Rocca et al., 2017, 2019), demonstrating the importance of E2F1 for testicular descent, function and health. Germ cell integrity is compromised in E2f1-null mice, not only as a result of cryptorchidism, but also because of the microenvironment disturbance in the seminiferous tubules that is attributed to downregulation of cell cycle gene expression, increase of trophic factors, such as Gdnf, disruption of cell interactions and an inability to repress the inhibitory effect of WNT4 (Fig. 9). Repression of Wnt4 is essential for testicular development based upon the observation that when Wnt4 expression is not inhibited due to loss of either Fgf9 or Fgfr2, male-to-female sex reversal occurs (Jameson et al., 2012). The current study indicates that in the adult testis, Wnt4 must be repressed by E2f1 for normal testicular function as the rescue of spermatogenesis and fertility observed in the E2f1^−/−;Wnt4^f/f;Stra8^cre/+ mice indicates that expression of Wnt4 is responsible for testicular degeneration, germ cell loss and infertility in mice. In our E2f1^−/−;Wnt4^f/f;Stra8^cre/+ mouse, testes are still above the inguinal canal as Wnt4 was only ablated in testicular germ cells. Future experiments should be conducted to elucidate the mechanism by which E2f1 influences testicular descent to determine whether it is also Wnt4 dependent.

The mice employed in these studies were used with approval and oversight of the Baylor College of Medicine (BCM) Institutional Animal Care and Use Committee. E2f1-null (E2f1^tm1Meg) mice were obtained from Dr Weei-Chin Lin at BCM. E2f1-null mice (Field et al., 1996) were cross-bred with C57BI/6-mice for several consecutive generations to obtain a uniform genetic background. Wnt4fl/fl (Wnt4^{tm1.1Bhr/BhrEiJ}) mice (Kobayashi et al., 2011) were obtained from Dr Richard Behringer (University of Texas MD Anderson Cancer Center, USA). Stra8-Cre [Tg(Stra8-icre)^1Reb/J] mice (Sadate-Ngatchou et al., 2008) were purchased from The Jackson Laboratory. Mating cages were set up in tandem and litter frequency and size were recorded for up to 6 months. Before being euthanized, mice were weighed and then anesthetized via direct inhalation of isoflurane. Serum was obtained through direct cardiac puncture on anesthetized mice, followed by centrifugation in Microtainer serum separation tubes for 10 min at 1000 g. Testis location was defined during dissection and organ weights were recorded. An RNeasy Mini Kit (Qiagen) was used for RNA isolation. Sperm density and motility were determined using sperm obtained from the minced cauda epididymis placed in modified human tubal fluid (HTF) medium (Irvine Scientific). Live sperm were spread onto a slide and classified as motile or immotile. The results were expressed as percentage of motile sperm for density, the HTF medium was diluted 1:20 and sperm were counted twice using a hematocytometer. Tissues were fixed in Bouin's solution before embedding. Mouse FSH, LH, and testosterone were measured by the University of Virginia Ligand Core Facility.

We assessed the fertility of mice by longitudinal breeding (6 months). Each WT or knockout male was housed with one WT female for 6 months beginning at 6 weeks of age. During this period, we recorded the number of litters and the number of pups produced from each breeding pair. After 6 months of mating, male mice were euthanized, and testes and epididymidies were collected and their weights recorded.

cDNA was prepared using the High Capacity Reverse Transcription kit (Applied Biosystems). qPCR was performed using Taqman gene expression assays (Applied Biosystems) listed in Table S1 and normalized to Gapdh. All genes were tested in duplicate at least three times in independent plate runs. A minimum of six independent mice per group per gene were analyzed. A challenge when performing gene expression studies in the testis is the selection of the best gene to use for normalization due to cell heterogeneity in the testis. Accordingly, qPCR was normalized to the carefully selected housekeeping genes, Gapdh (role in carbohydrate metabolism, Fig. 3, ABI#4352339E) and β-actin (involved in the cytoskeleton, data not shown, ABI#4310881E). Using these two reference genes, no difference was observed in gene expression. Plates were run and analyzed in the QuantStudio 12K Flex Real-Time PCR System. Statistical analysis was performed using the Relative Expression Software Tool (REST) from Qiagen.

Testes and epididymidies were fixed for 5 h in Bouin's solution. Sections (5 µm) were cut and stained with Periodic acid–Schiff (PAS). An Olympus BX51 microscope equipped with a DP73 high-performance Peltier cooled digital color camera was used for microphotography. The Olympus CellSens software was used to operate the imaging system and determine the diameter of the seminiferous tubules, lumen and epithelium thickness as described previously (Garcia et al., 2014). To perform immunohistochemistry, the avidin-biotin-immunoperoxidase complex technique that uses diaminobenzidine (DAB) as a substrate was used to generate a brown-colored polymeric oxidation product (Vector Labs) in paraffin-embedded testicular samples. Primary antibodies against E2F1 (Santa Cruz Biotechnology, sc-22820), CCND1 (Abcam, 134175) and WNT4 (Abcam, 150596) were used at a 1:100 dilution. CTNNB1 (BD Biosciences, 610153) antibody was used in a 1:25 dilution using the Mouse on Mouse (M.O.M.) Immunodetection Kit (Vector Laboratories). Hematoxylin was used as a counterstain. To perform immunofluorescence, slides were dehydrated, the antigen retrieved via sodium citrate baths, and incubated with primary antibodies against PLZF (R&D Systems, AF2944) and SOX9 (Cell Signaling Technology, 82630). Slides were then incubated with donkey anti-goat IgG Alexa Fluor 488- and anti-rabbit IgG Alexa Fluor 594-conjugated secondary antibodies (Life Technologies, A11055 and A21207, respectively). Following secondary antibody conjugations, slides were mounted with Vectashield Mounting Medium with DAPI (Vector Laboratories, H-1200). Primary antibodies against PLZF and SOX9 were used at dilutions of 1:100 and 1:150, respectively. All secondary antibodies were used at 1:500.

Tissue was dissected from animals and immediately placed in cold drops of modified Karnovsky's fixative (Graham et al., 1965) in 0.1 M cacodylate (pH 7.4) for 24 h at 4°C. After several 0.1 M cacodylate buffer rinses, the tissue was osmicated for 1 h, followed by a dehydration sequence of progressively more concentrated ethanol. Tissue was then infiltrated with a progressively higher ratio of embedding resin to ethanol and given three changes of pure resin: one change overnight, then two changes for 3-4 h each. Tissue was oriented for sectioning and embedded in Spurr's Low Viscosity resin (Spurr, 1969), then polymerized at 62°C for 3 days. Ultra-thin sections were cut with a Diatome Ultra45 knife, using a Leica U7 ultra-microtome. Sections were collected on 150 hex-mesh copper grids, stained with saturated uranyl acetate in 50% ethanol and counterstained with Reynolds' lead citrate (Reynolds, 1963). The stained sections were viewed on a Hitachi H7500 TEM and images were captured using an AMT XR-16 digital camera and AMT Image Capture software (v602.600.51).

Permeability of the BTB was assessed with a biotin tracer as described previously (Meng et al., 2005). Briefly, 3-month-old WT and E2f1-null mice were anesthetized and injected with 50 μl of 10 mg/ml EZ-Link-NHS-LC-Biotin freshly dissolved in PBS containing 1 mM CaCl (Thermo Scientific, 21336) into the interstitium of testes. The testes were removed after 30 min and frozen on dry ice. Optimal cutting temperature compound blocks were generated and frozen sections were cut. For localization of the biotin tracer, testis sections (5 μm) obtained from different levels (close to the poles and equatorial area) were incubated with streptavidin-fluorescein conjugate (1:3000; Invitrogen, SA1001) for 30 min at room temperature. The sections were rinsed with PBS and mounted in Vectashield containing DAPI. An Olympus BX51 fluorescence microscope, equipped with a DP73 high-performance Peltier cooled digital camera, was used to take microphotographs. The Olympus CellSens software was used to analyze the images.

All experiments were performed independently at least three times, with two replicates per experiments. Statistical analyses were performed using GraphPad Prism 8.0. All confidence intervals presented are at a 95% confidence level. All tests were two tailed.

We thank Dr Francesco DeMayo for providing key reagents; Dr Weei-Chin Lin for providing the E2f1-null mice; Dr Richard Behringer for providing the Wnt4-Flox mice; Ms Carolyn Schum and Ms Adlai Nelson for helpful editing of the manuscript; and Ms Kailand Johnson for technical assistance.