Protein Phosphatase 4 (PPP4) Prevents Female Reprogramming in the Mouse Testis After Sex Determination and Protects Male Fertility

Background: Impairment of lineage specication and function of gonadal somatic cells can lead to disorders of sexual development (DSDs) and fertility defects in humans. However, little is known about the function of protein phosphatases in testis development. Results: We showed that protein phosphatase 4 (PPP4) could maintain SOX9 expression in Sertoli cells and play an essential role in Sertoli cell lineage maintenance and male fertility. Conditional deletion of Ppp4c, a PPP4 catalytic subunit gene, caused the reprogramming of Sertoli cells to granulosa-like cells postnatally by inducing ectopic expression of FOXL2, which in turn led to testicular BTB structure damage, germ cell loss and ultimate testis to ovary-like gland transformation. Conclusion: Reprogramming of Sertoli cells due to absence of PPP4 may help explain the etiology of disorders of sexual differentiation and male infertility.


Introduction
In mammals, the testes in males or the ovaries in females are derived from the bi-potential gonads. Development of a testis or ovary is dependent on the differentiation of gonadal somatic cells, which is regulated by sex-determining genes. The Sry gene is transiently expressed in the somatic cells of undifferentiated XY gonads from E10-E12 in mice. This gene activates SOX9 and is essential for directing Sertoli cell differentiation in bi-potential gonads and testis formation (ref. 1,2). In XX gonads, which lack SRY expression, the somatic cells differentiate into granulosa cells (ref. [3][4][5], which is regulated by FOXL2 (ref. 6, 7) and R-spondin1 (RSPO1)/WNT4-β-catenin (CTNNB1) signaling pathway (ref. [8][9][10]. It is especially noteworthy that Sertoli (or granulosa) cell fate, once speci ed, is not permanent but instead needs to be constantly reinforced. For example, loss of FOXL2 function in the adult ovary allows granulosa cells to transdifferentiate into Sertoli cells (ref. 6), while loss of DMRT1 in Sertoli cells of the adult testis allows their transformation into granulosa cells (ref. 11). Proper differentiation of these cell types de nes the somatic cell environment that is essential for germ cell development, hormone production, and establishment of the reproductive tracts. Impairment of lineage speci cation and function of gonadal somatic cells can lead to DSDs in mammals including humans (ref. 12).
The maintenance of differentiated Sertoli cell fate is critical for the formation of testicular cord and testis development. As the precursor of seminiferous tubules, testis cord formation begins with the aggregation of Sertoli cells. These Sertoli cells change their surface matrix, recognize and adhere to each other, and surround germ cells to form the testis cord (ref. 13). Then, Sertoli cells enter into the differentiation process, which includes a cessation of proliferation, alterations in protein expression and transcription, and functional maturation (ref. 14). Mature Sertoli cells create the blood-testis barrier (BTB) to provide microenvironments for spermatogenesis and secrete many functional products to nourish germ cells and organize the events of spermatogenesis (ref. 15,16). Thus, any abnormality in the population and function of Sertoli cells will result in aberrant spermatogenesis and eventually infertility.
Sertoli cells are a central target for the regulation of spermatogenesis. In mammals, spermatogenesis employs an elaborate regulatory mechanism, which is controlled by a multitude of regulators, including hormones, growth factors, endotoxins, and proin ammatory cytokines (ref. [11][12][13][17][18][19]. These regulators induce a series of alterations into appropriate receptors that are transduced into the interior of the cells by means of post-translational modi cations, mainly (de)phosphorylations (ref. 20). Protein phosphorylation is accomplished by the concerted action of protein kinases and protein phosphatases that insert or remove, respectively, the γ-phosphate of ATP into the target amino acid, making protein phosphorylation a reversible process (ref. 21). The importance of protein kinases in spermatogenesis is re ected in infertility phenotypes observed when knocked out, for example SRC family kinases (SFKs) (ref. 22 superfamily including seven members (PP1, PP2A, PP2B, PP4, PP5, PP6, and PP7). PPPs dephosphorylate serine and threonine residues, which are responsible for more than 90% of all the dephosphorylations that occur in a cell. PPP4 regulates many cellular functions independently of other related protein phosphatases in the PPP family (ref. 30). PPP4 often functions as a heterotrimeric complex consisting of one evolutionarily conserved catalytic subunit (PPP4C) that associates with two types of regulatory subunits, which are ubiquitously expressed in all cell types (ref. 31). Emerging evidence indicates that PPP4 may play a role in the DNA damage response and regulation of chromatin activities. Several cellular signaling routes, including NF-kappaB JNK pathway, apoptotic signaling and the target of rapamycin (TOR) pathways appear to be regulated by PPP4 (ref. [32][33][34][35][36][37]. However, the physiological role of PPP4 in mammalian gonads remains unclear.
In the present study, we investigated the function of the PPP4 during testis development in mice. We found that constitutive inactivation of PPP4, by deletion of its catalytic subunit Ppp4c exon 3 in Sertoli cells after sex determination, caused the reprogramming of Sertoli cells to granulosa-like cells postnatally, which in turn led to dysgenesis of the testes, testes to ovary-like gland transformation and male infertility. These ndings report the critical role of PPP4 in preventing female reprogramming in the mammalian testis after sex determination and protecting male fertility.

Results
Generation of Ppp4c conditional knockout (Ppp4c ox ) mouse strain Using CRISPR/Cas9 technology, a Ppp4c allele in which exon 3 is anked by loxP sites (Fig. 1A-D) was introduced into the mouse germ line. Expression of Cre recombinase results in deletion of exon 3 and frame-shift mutation (Fig. 1E) and generates an allele (Ppp4c Δ ) leading to inactivation of PPP4C protein.
Germ-line generation of the Ppp4c Δ allele was achieved by using the oocyte-speci c Zp3-Cre transgene.

Conditional knockout of Ppp4c in Sertoli cells results in impaired testis development and infertility in mice
To address the role of PPP4 in Sertoli cells and male reproduction, we introduced the Sertoli cellexpressed Amh-Cre transgene into the Ppp4c ox strain to obtain Ppp4c ox/ ox ,Amh-Cre males (Ppp4c Amh-cKO). A previous study reported that Cre recombinase was speci cally expressed in Sertoli cells in Amh-Cre mice at E14.5 (ref. 38).
We found that Ppp4c ox/ ox ,Amh-Cre animals were fully viable. No gross abnormalities of external genitalia were observed in 2-month-old Ppp4c ox/ ox ,Amh-Cre males ( Fig. 2A). However, the testes size from the Ppp4c ox/ ox ,Amh-Cre was dramatically smaller (Fig. 2B, C), and testes weight was only ≈4% of that of control littermates (Fig. 2D). The rest of the reproductive tract including vas deferens, epididymis, and seminal vesicles was observed in the mutants (Fig. 2B). Thus, the regression of the Müllerian duct system and Wol an duct differentiation were largely normal. Although the body weight was not changed between mutant and control mice (Fig. 2E), the ratio of testes weight/body weight was signi cantly reduced in Ppp4c ox/ ox ,Amh-Cre males (Fig. 2F). Then, we evaluated the impact of Ppp4c-de ciency in Sertoli cells on male fertility using a successive breeding assay. Adult Ppp4c ox/ ox ,Amh-Cre or littermate control male mice were mated with wild-type female mice. The probability of observing a vaginal plug was not obviously different between the two groups of mice (data not shown). However, after a total of 87 matings, control mice sired 12 pups per litter, whereas the Ppp4c ox/ ox ,Amh-Cre male mice did not produce any progeny ( Fig. 2G and H). These results indicated that Ppp4c-de ciency in Sertoli cells could injure testes development and induce male infertility in mice.
Aberrant histology of testes and non-obstructive azoospermia (NOA) of Ppp4c ox/ ox ,Amh-Cre males To investigate the functions of PPP4 in spermatogenesis, the histology of testes from control and Ppp4c ox/ ox ,Amh-Cre mice was examined. Histologically, Ppp4c ox/ ox ,Amh-Cre testes from 2-monthold animals bore no resemblance to age-matched control testes. Mutant testes completely lacked the normal tubular architecture observed in control testes (Fig. 3A, Band C) and consisted primarily of scattered cells and only some tubular architecture (Fig. 3D, E andF). In addition, mature spermatozoa with crescent-shaped heads were identi ed in the control epididymis (Fig. 3G). In contrast, no spermatozoa were found in the Ppp4c ox/ ox ,Amh-Cre epididymis (Fig. 3H). This aberrant histology exhibited in Ppp4c ox/ ox ,Amh-Cre males was very similar to human NOA, which is one of the major causes of male infertility in humans.
Ectopic expression of FOXL2 protein in the testes of Ppp4c ox/ ox ,Amh-Cre mice To examine the cell types of aberrant Ppp4c ox/ ox ,Amh-Cre testes, the expression of several germ and somatic cell marker proteins was examined by immunostaining and western blotting. Germ cell-speci c marker MVH was detected in the germ cells of the control testes (Fig. 4A, arrow), whereas no MVH signal was detected in the Ppp4c ox/ ox ,Amh-Cre testes (Fig. 4B). This result indicated that the germ cell was completely lost in the PPP4C-de cient testes. In normal testes, steroidogenic enzyme 3β-HSD was speci cally expressed in Leydig cells with a strong signal (Fig. 4C, arrow). However, in Ppp4c ox/ ox ,Amh-Cre testes, partial cells showed strong signals of 3β-HSD (Fig. 4D, arrow), but more disorganized cells showed weak signals (Fig. 4D, arrowhead). Normally, SOX9 (Fig. 4E, arrow) and WT1 (Fig. 4G, arrow) proteins are speci cally expressed in Sertoli cells of seminiferous tubules. However, almost no SOX9 signals were detected in the Ppp4c ox/ ox , Amh-Cre testes except for some remnant tubular architecture (Fig. 4F). In comparison, more weak signals of WT1 were observed in the mutant testes (Fig. 4H, arrow).
In mammalian females, 3β-HSD shows a low level in granulosa cells and oocytes of fetal mouse ovaries, but a higher level after theca cell recruitment and formation of the rst antral follicles (ref. 12). Additionally, WT1 is initially expressed in somatic cells of bi-potential gonads. And in adults, its expression is maintained in ovarian granulosa cells and testicular Sertoli cells. However, SOX9 is exclusively expressed in Sertoli cells of testes, not in somatic cells of the ovary. Considering the above facts, we surmised that the WT1-positive or 3β-HSD-positive cells may be granulosa-like cells or theca-like cells.
To test these hypotheses, we rst examined gonads of Ppp4c ox/ ox ,Amh-Cre males for the presence of FOXL2, a female-speci c transcription factor expressed in the nuclei of granulosa cells and theca cells(ref. 11,39,40), two somatic cell types of the ovarian follicle (Fig. 4K). No FOXL2 signals were found in control testes (Fig. 4I). However, abundant cytoplasmic FOXL2-positive cells were present within mutant testes (Fig. 4G, arrow). Additionally, western blot results also showed FOXL2 expression in Ppp4c ox/ ox ,Amh-Cre testes with a larger size compared to that of the normal ovary (Fig. S1). Previous studies reported that post-translational regulation (including phosphorylation) of FOXL2 could change its subcellular localization from normal nuclear distribution to cytoplasmic mislocalization (ref. [41][42][43][44]. So cytoplasmic location and larger amount of FOXL2 protein may be a result of its phosphorylation. Besides, western blot results showed that aromatase protein CYP11A1 was strongly expressed in mutant gonads, which was robustly expressed in granulosa cells of the ovary (Fig. S1). These results indicated that loss of the PPP4C in mouse Sertoli cells activated FOXL2 and reprogrammed testicular somatic cells into granulosa-like cells and theca-like cells.
PPP4C maintains SOX9 and suppresses FOXL2 expression in postnatal Sertoli cells Next, we examined the time course of this aberrant testis development and the timing of FOXL2 induction. At postnatal day 7 (P7), although seminiferous tubules were present in Ppp4c ox/ ox ,Amh-Cre testes, partial tubules showed damaged architecture (Fig. S2). PPP4C and SOX9 double staining experiments showed that all SOX9-positive Sertoli cells in Ppp4c ox/ ox ,Amh-Cre testes were PPP4Cnegtive, indicating speci c deletion of PPP4C in Sertoli cells of mutant mice (Fig. S3). In control testes at P7, Sertoli cells strongly expressed SOX9, whereas FOXL2 was undetectable ( Fig.5A-D, arrow).
Surprisingly, in Ppp4c ox/ ox ,Amh-Cre testes at P7, we found some intratubular cells with typical Sertoli cell features including tripartite nucleoli, co-expressed SOX9, and weak FOXL2 ( By 1-3months Ppp4c ox/ ox ,Amh-Cre testes increasingly lost normal tubular architecture and few SOX9positive cells remained and most remnant cells strongly expressed FOXL2 (Fig. S4). Histological analysis of mutant gonads is shown in Fig. S2. These results showed that foetal loss of PPP4C caused postnatal Sertoli cells to lose the male-promoting SOX9 and instead express the female-promoting FOXL2.

Ppp4c deletion in Sertoli cells results in damaged formation of BTB and functional change of Sertoli cells
The above experiments found partially damaged architecture in Ppp4c ox/ ox ,Amh-Cre testes at P7. One of the major functions of Sertoli cells is to form the structure of the BTB, which, when disrupted, results in germ cell death and spermatogenic defects. To test whether the formation of BTB was damaged in Ppp4c-de cient testes, we detected the expression of gap junction protein Connexin 43 (CX43) which is a BTB-constituted protein that not only modulates the BTB integrity (ref. 45), but also maintains the homeostasis of the BTB via its effects on tight junction reassembly (ref. 16). Here, in the seminiferous epithelium of control males, CX43 was found to be immunolocalized between Sertoli cells and spermatogonia/primary spermatocytes ( Fig. 6A-C, G-I, arrow). In contrast, no immunostaining at all was detected in seminiferous tubules from Ppp4c ox/ ox , Amh-Cre mice at P7, only few interstitial cells displaying weak CX43 signals (Fig. 6D-F, arrowhead), indicating that neither Sertoli cells nor spermatogonia were able to synthesize CX43 protein. Besides, at 2-month-mutant testes, although CX43 signals were found in remnant cells, they displayed several non-continuous punctate dots signals like that in the control (Fig. 6J-L, arrow). Western blot results also showed that expression of CX43 was decreased in mutant testes compared with the control at 2 months (Fig. S1). In addition, we also found that stem cell factor (SCF), a paracrine growth factor normally produced by Sertoli cells, showed aberrant increased secretion in intratubular cells and interstitial cells of mutant testes compared with the control (Fig. S5). These results indicated that damaged formation of BTB and functional change of Sertoli cells emerged in Ppp4c ox/ ox ,Amh-Cre testes after PP4 inactivation.
Ppp4c deletion in Sertoli cells results in partial Sertoli cell apoptosis and further massive germ cell death We considered that integrity disruption of the BTB would result in germ cell death and spermatogenic defects, then, cellular apoptosis in the testes was analyzed with TUNEL staining. The apoptotic cells were increased signi cantly in the testes of Ppp4c ox/ ox ,Amh-Cre testes at P7 (Fig. 7D, arrow), compared with the control (Fig. 7A, arrow). However, no obvious apoptotic difference was observed between the testes of Ppp4c mutant (Fig. 7E and F) and control ( Fig. 7B and C) at 1-2 months. Furthermore, we also detected another apoptosis marker protein cleavage of caspase 3 (Fig. S6). We found that apoptotic signals in mutant testes were dramatically increased relative to the control at P7. Next, we analyzed the cell types of abundant apoptotic cells existing in Ppp4c ox/ ox ,Amh-Cre testes at P7. Co-staining experiments showed that a few TUNEL-positive cells expressed Sertoli cell-speci c protein SOX9 (Fig. S7), and most TUNELpositive cells expressed germ cell marker protein MVH compared with the control (Fig. S8). These results suggested that Ppp4c deletion in Sertoli cells at an early stage of development would result in partial Sertoli cell apoptosis, further massive germ cell death and spermatogenic defects. CTNNB1 is not involved in Foxl2 expression in Ppp4c ox/ ox ,Amh-Cre testes Our previous study showed that, by constitutive activation of Ctnnb1 in Sertoli cells, accumulation of CTNNB1 protein in the nuclei of Sertoli cells led to the transformation of testis Sertoli cells to ovarian granulosa-like cells by inducing Foxl2 expression (ref. 8). Here we found that CTNNB1 protein was localized at the plasma membrane, not the cell nucleus, of Sertoli cells and germ cells in Ppp4c ox/ ox ,Amh-Cre testes (Fig. 8D-F, arrow) at P7, similar to control testes ( Fig. 8A-C, arrow). At 2 months, although CTNNB1 protein was dramatically decreased in Ppp4c-mutant testes (Fig. 8J-L, arrow) compared with the control (Fig. 8G-I, arrow), only weak signals were found at the plasma membrane of scattered cells, suggesting that CTNNB1 protein may play a cell adherent role in Ppp4c-mutant testes. These ndings indicated that the transformation of testis Sertoli cells to ovarian granulosa-like cells by inducing Foxl2 expression in Ppp4c ox/ ox ,Amh-Cre testes is not a result of the dysregulation of CTNNB1 expression.
Phosphorylation modi cation of histone protein is altered in Ppp4c ox/ ox ,Amh-Cre testes PPP4 belongs to the phosphoprotein phosphatases (PPPs) superfamily and evidence indicates that PPP4 may play a role in the regulation of chromatin activities. Here, western blot results showed that the histone H3 and H2A protein levels were decreased in the Ppp4c ox/ ox ,Amh-Cre testes compared with the control testes (Fig. S1). Surprisingly, the expression of pH3 and γH2AX was completely absent in the mutant testes, indicating that protein phosphorylation modi cations of histones were changed. These results suggested that PPP4 may be involved in the aberrant testis development and the transformation of testis Sertoli cells to granulosa-like cells through regulation of chromatin activities. Here we show that sexual fate was regulated by protein phosphatases in the testis: loss of the catalytic subunit of protein phosphatases 4 (PPP4C) in mouse Sertoli cells activates FOXL2 and reprograms Sertoli cells into granulosa-like cells. PPP4-de ciency in Sertoli cells caused the dysfunctions of Sertoli cells including the loss of expression of Sertoli cell-speci c genes, damage of BTB formation, and dysgenesis of the testes. In this environment, theca cells form from Leydig cells probably due to effect of reprogrammed granulosa cells, germ cells appeared feminized but ultimate lose. Thus, PPP4 was essential for maintaining mammalian testis determination, and competing regulatory networks maintain gonadal sex long after the fetal choice between male and female.

Discussion
It is reported that adhesion and transcription factor CTNNB1 regulates differentiation of granulosa cells in the absence of SRY, and overactivation of CTNNB1 in the presence of SRY leads to granulosa formation prior to sex determination (ref. 58). And our previous study found CTNNB1 overactivation in Sertoli cells after sex determination can also direct the transformation of testis Sertoli cells to ovarian granulosa-like cells by inducing FOXL2 expression (ref. 8). In the Ctnnb1-activating mouse model, CTNNB1 protein was accumulated in the nuclei of Sertoli cells. However, in Ppp4c-mutant testes, CTNNB1 protein was not localized at cell nuclei but at the plasma membrane of Sertoli cells and germ cells similar to control testes, and the expression of CTNNB1 protein was also not affected at P7. These ndings suggested that the transformation of testis Sertoli cells to ovarian granulosa-like cells by inducing FOXL2 expression in Ppp4c ox/ ox ;Amh-Cre testes was not a result of the dysregulation of CTNNB1 expression.
In addition, the phenotype of DMRT1-de ciency in Sertoli cells mice (ref. 11) is reminiscent of the phenotype in Ppp4c ox/ ox ;Amh-Cre mice. However, in the PPP4C-de cient mouse model, the FOXL2 expression was identi ed earlier than in the DMRT1-de cient mice and the tubular structure was completely disrupted at 1 month. The phenotype was more severe than that of the DMRT1-de cient mouse model. Thus, the dysgenesis of the testes in Ppp4c Amh-cKO mice was also probably not related to DMRT1.
Inactivation of PPP4 causes the reprogramming of Sertoli cells to granulosa-like cells, which may result from the PPP4C or PPP4 molecule itself suppressing SOX9 expression directly or indirectly and in turn inducing FOXL2 expression in Sertoli cells. Besides, evidence indicates that PPP4 may play a role in the regulation of chromatin activities. Correspondently, we found that the histone H3 and H2A protein levels were decreased in the Ppp4c ox/ ox ;Amh-Cre testes. Even the expression of pH3 and γH2AX was completely absent in the mutant testes, indicating that protein phosphorylation modi cations of histones were changed. Therefore, another mechanism of this reprogramming in PPP4c-de cient mice was probably related to protein phosphorylation modi cation of PPP4 on key genes involved in testis development and sex reversal. Therefore, the downstream target genes through which the PPP4 or PPP4C acts to regulate the lineage transition between Sertoli and granulosa-like are still unclear and require further investigation.

Conclusion
The analysis presented here demonstrates that protein phosphatases are essential for Sertoli cell lineage maintenance and male fertility. Protein phosphatases PPP4 could maintain SOX9 expression in Sertoli cells directly or indirectly. Deletion of PPP4 in Sertoli cells after sex determination causes the reprogramming of Sertoli cells to granulosa-like cells postnatally, which in turn leads to testicular BTB structure disruption, germ cell loss and ultimate testes to ovary-like gland transformation. Moreover, because many genes implicated in this study are evolutionarily conserved, similar antagonism between PPP4 and FOXL2 for control of gonadal sex may therefore be extended to other mammals. Reprogramming due to inactivation of PPP4 also may help explain the etiology of disorders of sexual differentiation and male infertility.

Mice
All animal work was carried out in accordance with the protocols approved by the Institutional Animal Care and Use Committee at the Institute of Zoology, Chinese Academy of Sciences (CAS). All mice were maintained in a C57BL/6;129/SvEv mixed background. DNA isolated from tail biopsies was used for genotyping. Genotyping was performed by PCR as described previously (ref. 59).

Tissue collection and histological analysis
Testes were dissected from Ppp4c-Amh cKO and control mice immediately after euthanasia and xed in 4% paraformaldehyde for up to 24 h, stored in 70% ethanol, and embedded in para n. Five-micrometerthick sections were cut and mounted on glass slides. After depara nization, the sections were processed for hematoxylin-eosin (H&E) staining, immunohistochemistry (IHC) or immuno uorescent analysis.

Western blot analysis
The Western blot assays were performed as previously described (ref. 59). The protein lysates (15 µg) were separated via SDS-PAGE and electro transferred to a nitrocellulose membrane. The membrane was scanned using the ODYSSEY Sa Infrared Imaging System (LI-COR Biosciences, Lincoln, NE, USA).

Statistical analysis
Experiments were repeated at least three times. Three to ve control or mutant testes were randomly used for immunostaining or western blot at each time point. The quantitative results are presented as the mean ± SEM. Statistical differences were calculated by two-tailed unpaired t-test for two datasets. P values < 0.05 (*) or 0.01(**) were considered to be signi cant.  Gross phenotype of 2-months-old Ppp4c ox/ ox, Amh-Cre males. Normal external genitalia of Ppp4c ox/ ox, Amh-Cre males (A), reproductive tracts from control male and mutant male displaying reduced testes sizes but normal development of both vas deferens, seminal vesicle and epididymis in mutant (B and C), normal body weight (E) but severely reduced weight of testes (D) and ratio of testes/body weight (F) from Ppp4c ox/ ox, Amh-Cre males (N≥5). Using a successive breeding assay, control mice sired 12 pups per litter, whereas the Ppp4c ox/ ox, Amh-Cre male mice did not produce any progeny (G and H) (N=5). Ep, epididymis, tes, testis, vd, vas deferens, sv, seminal vesicle. All statistical data are represented as means ± SEM. **p<0.01.  Identi cation of remnant cells and ectopic expression of FOXL2 protein in the testes of Ppp4c ox/ ox, Amh-Cre mice at adulthood. Germ cell-speci c marker MVH was observed in the testes of control mice (A, arrow), whereas no MVH signal was found in the Ppp4c ox/ ox, Amh-Cre testes (B). In normal testes, steroidogenic enzyme 3β-HSD was speci cally expressed in Leydig cells with a strong signal (C, arrow).
Although, in mutant testes, partial cells showed strong signals of 3β-HSD (D, arrow), more disorganized cells showed weak signals (D, arrow head). SOX9 (E, arrow) and WT1 (F, arrow) proteins were speci cally expressed in Sertoli cells of seminiferous tubules at normal levels. No SOX9 signal was detected in the Ppp4c ox/ ox, Amh-Cre testes (F). However, several weak WT1 signals were observed in the mutant testes (H, arrow). In normal adult ovary, FOXL2 protein was expressed in nuclei of granulosa cells (K, arrow) and theca cells (K, arrowhead), which did not exist in control testes (I). However, abundant cytoplasmic FOXL2 signals were observed within remnant cells of adult mutant testes (G, arrow).  Differential CX43 expression between the testes obtained from control and Ppp4c ox/ ox, Amh-Cre mice at P7. In the seminiferous epithelium of control males, CX43 was found to be immunolocalized between Sertoli cells and between Sertoli cells and spermatogonia/primary spermatocytes at P7 (A-C, arrow). In contrast, no immunostaining at all was detected in seminiferous tubules from Ppp4c ox/ ox, Amh-Cre mice at P7, only few interstitial cells displaying weak CX43 signals (D-F, arrowhead). In the testes of control males at 2 months, CX43 expression was more predominant at the basal ES (ectoplasmic specialization) region of seminiferous epithelium (G-I, arrow). In the testes of at 2-month-mutant testes, although CX43 expression was found in remnant cells, they displayed several discontinuous punctate signals (J-L, arrow) different from that of the control. PPP4c deletion in Sertoli cells results in cell apoptosis in the mutant testes by TUNEL staining. Apoptotic cells in the seminiferous epithelium were detected by TUNEL staining (positive cells are green). The apoptotic cells were increased signi cantly in the testes of Ppp4c ox/ ox, Amh-Cre testes at P7 (D, arrow), compared with controls (A, arrow). However, no obvious apoptotic difference was observed between the testes of ppp4c mutant (E and F) and control testes (B and C) at 1month and 2 months.

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download. supplementary.rar