Absence of Inhibin Alpha and Retinoblastoma Protein Leads to Early Sertoli Cell Dysfunction

Sertoli cells, the support cells of mammalian spermatogenesis, are regulated by a number of nuclear factors and express retinoblastoma (RB) tumor suppressor protein. We hypothesized that RB is an important mediator of Sertoli cell tumorigenesis in inhibin α knockout (Inha KO) mice. In our previous mouse studies, we found that conditional knockout (cKO) of Rb in Sertoli cells caused progressive Sertoli cell dysfunction. Initially, loss of RB had no gross effect on Sertoli cell function as the mice were fertile with normal testis weights at 6 weeks of age, but by 10–14 weeks of age, mutant mice demonstrated severe Sertoli cell dysfunction and infertility. Although double knockout (dKO) of Rb and Inha did not result in exacerbation of the tumorigenic phenotype of Inha-null mice, we found that the dKO mice demonstrate an acceleration of Sertoli cell dysfunction compared to Rb cKO mice. Specifically, in contrast to Rb cKO mice, Inha/Rb dKO mice showed signs of Sertoli cell dysfunction as early as 4 weeks of age. These results demonstrate that RB is not essential for Sertoli cell tumorigenesis in Inha KO mice but that loss of Inha accelerates the infertility phenotype of Rb cKO mice.


Introduction
Mammalian spermatogenesis is dependent on the proper functioning of Sertoli cells, the somatic support cells of germ cell maturation. During embryogenesis, immature Sertoli cells emerge at E10.5 and form the seminiferous tubules [1]. They continually replicate prenatally and postnatally and finally enter replicative senescence at P12-17 when they differentiate into their mature form [2]. During puberty, Sertoli cells must form extensive intercellular junctions to support an expanding population of differentiating germ cells undergoing the process of spermiogenesis, which involves extensive morphological differentiation of round spermatids into mature spermatozoa [1]. At 5-6 weeks of age, mice produce spermatozoa from the first wave of spermatogenesis, and previous studies suggest that this first wave of spermatogenesis is significantly different from later waves [3,4,5]. At this point, Sertoli cells must once again adapt because, unlike the first wave of spermatogenesis, subsequent waves of spermatogenesis not only depend on forming new cellular junctions but also depend on removing old ones [1]. Our previous work in this field indicated that retinoblastoma protein (RB) is essential to maintain cell cycle quiescence and junctional remodeling in mature adult Sertoli cells [6].
RB is an important intracellular regulator of the cell cycle. RB is a repressor of E2F transcription factors, which function to promote progression from G1 to the S phase of the cell cycle [7]. For the cell cycle to progress, RB is inactivated via phosphorylation by cyclin dependant kinases (CDKs). RB and RB pathway components are frequently altered in many human cancers [8], but little is known about the role of RB in Sertoli cell tumorigenesis.
Inhibins are a:b heterodimeric members of the transforming growth factor b (TGFb) superfamily and act as competitive antagonists of activins, which are b:b homodimeric TGFb ligands [9]. Targeted deletion of the inhibin a subunit (Inha) in mice revealed that loss of inhibin leads to the development of gonadal sex cord-stromal tumors, which originate from Sertoli cells in the male [10]. Further studies using Inha-null mice deficient in either follicle stimulating hormone (FSH) and/or luteinizing hormone (LH) revealed that these tumors were gonadotropin-dependent [11,12,13], but the intracellular regulators of Sertoli cell tumorigenesis remained unknown [14]. Subsequent studies showed that RB pathway components, cyclin D2 (Ccnd2) and cyclin dependant kinase 4 (Cdk4), are increased in Inha knockout (KO) tumors while the CDK inhibitor, p27 (Cdkn1b), is decreased [15]. Double knockout (dKO) studies in our laboratory revealed that p27 [15] and cyclin D2 [16] are important modifiers of the Inha KO phenotype as Inha/Cdkn1b dKOs had accelerated cancer development, whereas Inha/Ccnd2 dKOs had attenuated tumor formation.
Based on these findings, we hypothesized that RB is essential for the progression of Inha KO tumorigenesis and, similar to Inha/ Cdkn1b dKO, that Inha/Rb dKO mice would demonstrate rapid tumor formation. We endeavored to carefully characterize this double knockout of Rb and Inha in the male to better understand the role of RB in Sertoli cell cycle control and tumorigenesis.
Unexpectedly, the loss of RB on an Inha deficient background does not worsen disease progression. However, our findings show that loss of inhibin accelerates Sertoli dysfunction in the Sertoli cellspecific RB knockout mice [6].

Animal Care and Treatment
Mice were maintained on a 129SvEv/C57BL/6 background and housed with unlimited access to food and water and exposure to 12 h:12 h light:dark cycles in accordance with the standards of the Association for Assessment and Accreditation of Laboratory Animal Care. The Institutional Animal Care and Use Committee at the Baylor College of Medicine approved the study under approval number AN-716. Survival curves were generated as described [13]. Mice were weighed weekly from 4 weeks to a maximum of 26 weeks to monitor for symptoms of cancer cachexia and sacrificed when body weights decreased to less than 16.0 g. For serum collection, mice were anesthetized, and blood was collected by cardiac puncture. Microtainer tubes (BD) were utilized for serum isolation and sent to the University of Virginia Center for Research in Reproduction Ligand Assay and Analysis Core (http://www.healthsystem.virginia.edu/internet/crr) for detection of FSH, LH, testosterone (T), and estradiol (E2).

Morphological and Histological Analysis
Immediately after cardiac puncture, mice were euthanized, and the desired tissues were harvested and weighed. In general, one testis/tumor was placed into fixative and the other was frozen for DNA (220uC) and RNA (280uC in RNAlater, Qiagen) analysis. Tissues were fixed in 10% neutral buffered formalin (Harleco) or Bouin's fixative (Sigma) prior to paraffin embedding. Tissue embedding, sectioning, and staining for periodic acid Schiff and hematoxylin were performed by the Histology Core of the Department of Pathology of Baylor College of Medicine.

RNA Extraction and Quantitative PCR
RNA was extracted from 4 week-old testes using the RNeasy Mini Kit (Qiagen, Valencia, CA). For quantitative PCR, the isolated RNA (n = 3 for each genotype) was converted to cDNA using SuperScriptIII (Invitrogen). Primers (E2f1, Cdkn2a, Timp1) used in a SYBR green based qPCR assay were previously reported [6,22]. Bbc3 was assayed with a Taqman probe (Mm00519268_m1, Applied Biosystems). Quantitative PCR reactions were performed using Gapdh as an endogenous control for relative quantification [22].

Statistical Analysis
Statistical analysis utilized JMP 8.0.1 software (SAS Institute). Statistical significance was determined by one-tailed t test assuming unequal variance for two sample comparison and by one-way analysis of variance (ANOVA) followed by Tukey's honestly significant difference (HSD) test for multiple sample comparisons. Survival curves were compared using the log-rank test. Groups were considered not significantly different from one another if p.0.05.
After we had confirmed that the Rb allele was recombined in our dKO mice, all groups of mice were weighed weekly starting at 4 weeks of age to monitor disease progression since Inha KO mice experience death secondary to cancer cachexia [23]. We monitored these groups until they reached 26 weeks of age and found that while 100% of control and Rb cKO mice survived until the end of the observation period, only 8% of Inha 2/2 and 10% of Inha/Rb dKO mice survived to that age ( Figure 2A). The difference in mortality between Inha 2/2 and Inha/Rb dKO mice was not significant (p.0.05 by log-rank test), and both groups had 50% survival at 14 weeks of age (Figure 2A).
Disease progression was also monitored by examining other parameters. Tumor burden of the combined gonads was determined at 6 weeks, 10 weeks, and end of life. Inha/Rb dKO mice did not differ significantly from Inha 2/2 mice at every age examined ( Figure 2B, p.0.05). Gonadal tumors in Inha 2/2 mice secrete an excess of activins, which signal through type 2A activin receptors (ACVR2A) to cause a cachexia wasting syndrome characterized by decreased body weights and liver weights [23,24]. An examination of body weights from 4-10 weeks of age showed no significant differences between Inha 2/2 and Inha/Rb dKO ( Figure 3A, p.0.05). Similarly, liver weights from 6 weeks, 10 weeks, and end of life were also not significantly different between these groups ( Figure 3B   Serum hormone levels were also measured in 10 week-old mice, due to their implications in disease progression [12,13,25]. No significant differences were detected between Inha 2/2 and Inha/Rb dKO levels of FSH, LH, or testosterone (T) ( Table 1). The only significant difference detected was in estradiol (E2) levels, which were significantly decreased in Inha/Rb dKO mice as compared to Inha 2/2 mice (Table 1). Although prior studies have indicated that loss of estrogen receptors in Inha-null tumors prevented early tumorigenesis and death [25], we saw no functional consequence of this decreased estradiol level as disease progression in Inha/Rb dKO mice did not differ significantly from Inha 2/2 mice, as outlined previously.

Loss of Inha accelerates the progressive Sertoli cell dysfunction observed in Sertoli cell-specific RB deficient mice
Since we could not discern any striking differences between Inha 2/2 and Inha/Rb dKO mice using the criteria described above, we sought to examine mice from younger stages of disease progression. Histology of 4 week-old and 6 week-old Inha/Rb dKO testes did not show increased tumorigenesis when compared to age-matched Inha 2/2 testes (data not shown). However, when we examined the non-tumorigenic areas that contained intact seminiferous tubules, we observed an acceleration of the progressive infertility phenotype of the Rb cKO mice [6]. In Rb cKO mice, histological signs of Sertoli cell dysfunction, such as vacuolization, are not observed until roughly 8 weeks of age ( Figure 4, top row middle inset) [6]. Loss of the tubular lumen, severely decreased tubular widths, and severe loss of germ cells generally occurs at 10 weeks of age ( Figure 4, bottom row middle inset) [6]. In Inha/Rb dKO testes, there were prominent signs of vacuolization ( Figure 4, top row right panel, arrows) in 4 week-old tubules that were not observed in either Inha 2/2 or Rb cKO testes ( Figure 4, top row). However, 4 week-old Inha 2/2 , Rb cKO, and  Inha/Rb dKO testes showed similar germ cell compositions (Figure 4, top row). In 6 week-old Inha/Rb dKO testes, there were signs of severe Sertoli cell dysfunction, such as loss of tubular lumens, decreased tubular widths, and loss of advanced germ cells, that resembled much older Rb cKO mice (Figure 4, bottom row). Examination of 6 week-old Inha 2/2 and Rb cKO testes revealed multiple tubules with full complements of germ cells, including elongating spermatids, and intact tubular widths and lumens (Figure 4, bottom row). Some Inha 2/2 tubules showed loss of elongating spermatids (Figure 4, asterisk), but overall, the seminiferous tubules of 6 week-old Inha 2/2 and Rb cKO mice showed far less dysfunction than 6 week-old Inha/Rb dKO mice.
To investigate the cause of the accelerated Sertoli cell dysfunction in Inha/Rb dKO mice, we explored changes in mRNA levels of genes that were implicated in the pathogenesis of the Rb cKO phenotype [6]. We used RNA from whole testes of 4 week-old mice that exhibited only minor histological differences from control mice to examine select gene changes. Genes related to Rb cKO apoptotic defects (E2f1, Cdkn2a, and Bbc3/PUMA) and differentiation defects (Timp1) were all significantly increased in Inha/Rb dKO testes as compared to control ( Figure 5).

Discussion
Whereas absence of p27 in Inha KO mice accelerates the formation of Sertoli cell tumors [15], our present studies indicate that RB plays a minor or redundant role in Inha KO tumorigenesis. By following the disease progression of Inha/Rb dKO mice as compared to Inha 2/2 mice, we determined that loss of RB does not significantly affect a number of parameters associated with Inha KO tumorigenesis, including survival, tumor burden, body weight, or liver weight. However, our findings do show that the progressive Sertoli dysfunction exhibited in Rb cKO mice [6] is accelerated by the additional loss of inhibin a.
Activin levels are pathologically increased in Inha-null mice [23] and are the major cause of tumorigenesis in these mice. This is suggested by studies in which decreased levels of free activins decreased tumor progression and disease severity [26]. Also, double knockout of Inha and Smad3, the activin-responsive transcription factor, resulted in attenuation of Sertoli cell tumorigenesis [20,27]. These studies suggest that pathologically increased activins are in large part responsible for Sertoli cell tumorigenesis in Inha-null mice. Why is it that Sertoli cells react by becoming tumorigenic since the entire organism is exposed to increased activins in the bloodstream? Previous studies conducted in our lab indicated that gonadotropins, especially FSH, are essential for the progression of Inha KO tumorigenesis in male mice [11,12]. Perhaps, it is the gonadotropin-responsiveness of Sertoli cells that makes them uniquely susceptible to pathological activin signaling. FSH signaling in Sertoli cells can activate the Ras-related mitogen activated signaling pathway involving ERK1 (mitogen-activated protein kinase 3) and ERK2 (mitogen-activated protein kinase 1), which subsequently upregulate cyclin D1 [28]. CDK4/cyclin D complexes and other Ras-related kinases modulate the phosphorylation of SMAD3 [29,30,31], and this modulation of SMAD3 causes its downstream signaling to change from tumor suppressive to oncogenic [29,30]. We were very surprised at the minor effect of loss of RB on the disease progression in Inha-null mice. However, since our hypothesis was based on double knockout models that suggested that RB pathway components, specifically p27 and cyclin D2, modulate Inha KO tumorigenesis, we speculate that crosstalk between the RB and activin/FSH pathways were responsible for our previous results. In our model (Figure 6), the pathway of FSH signaling is crucial for establishing crosstalk between signaling of the components of the RB pathway of cell cycle control (cyclin D1/2) and activin signaling to result in tumorigenesis; however, retinoblastoma protein itself is not crucial for oncogenesis in this model. It will be important in the future to study the expression levels of activin pathway components and RB pathway components (e.g., cyclin D1) in our Rb/Inha dKO mouse to further clarify if our model of Inha KO tumorigenesis is correct.
Retinoblastoma protein is a major determinant of Sertoli cell maturation, and mice with Sertoli cell-specific depletion of RB have progressively dysfunctional testes after puberty [6]. In light of this knowledge, it is quite interesting that deletion of Rb on an Inhanull background causes acceleration of the phenotype of progressive Sertoli cell dysfunction. This could be due to increased FSH signaling in Inha-null mice as FSH levels are not increased in Rb cKO mice until older ages [6]. Additionally, activation of SMAD3 in Inha-null Sertoli cells may cause an elevation of tumorigenic gene expression that may subsequently increase E2F1-and p53-target genes ( Figure 6). We propose that these pathways converge to affect Sertoli cell function in Inha/Rb dKO mice.
Previously, we proposed that RB was important for Sertoli cell differentiation because of its interactions with androgen receptor [6]. The current studies raise the possibility that the pathological expression of cell cycle genes on an RB-deficient background are Figure 6. Proposed mechanism of Sertoli cell tumorigenesis in Inha 2/2 mice and acceleration of dysfunction in Inha/Rb dKO mice. Activins signal through type 1 and 2 activin receptors (ACVR1/2) to phosphorylate and activate SMAD3. Deletion of inhibin a not only leads to loss of competitive inhibition at the level of ACVR1/2 receptors, but it also leads to a pathological increase in activin levels. In our model, excess activin and FSH signaling synergistically activate SMAD3 tumorigenic pathways. Although the current studies suggest RB is not an important modifier of the inhibin a KO phenotype, we speculate that FSH mediates crosstalk between the RB pathway and inhibin a KO tumorigenic pathway by activating cyclin D1/2, which also modulates SMAD3 activity to affect tumor development and progression. We propose that SMAD3 activation synergizes with loss of RB in Inha/Rb dKO Sertoli cells thus resulting in the upregulation of E2F1-and p53-target genes. Solid line, direct interaction; dotted line, indirect interaction; dashed line, nuclear translocation. doi:10.1371/journal.pone.0011797.g006 also directly related to Sertoli cell differentiation. Strikingly, we found that Timp1, a gene important to the inhibition of collagenremodeling that is highly expressed in immature Sertoli cells, was upregulated in Inha/Rb dKO mice. Timp1 was also highly upregulated in Sertoli cell-specific Rb cKO mice. The early upregulation of Timp1 in Inha/Rb dKO testes as compared to Rb cKO testes suggests that differentiation genes like Timp1 may be related to cell cycle pathways downstream of RB and SMAD3. Elucidation of our proposed model will be vital to better understanding the functions and regulation of the Sertoli cell.