Combined Loss of the GATA4 and GATA6 Transcription Factors in Male Mice Disrupts Testicular Development and Confers Adrenal-Like Function in the Testes

Abstract

The roles of the GATA4 and GATA6 transcription factors in testis development were examined by simultaneously ablating Gata4 and Gata6 with Sf1Cre (Nr5a1Cre). The deletion of both genes resulted in a striking testicular phenotype. Embryonic Sf1Cre; Gata4^flox/flox Gata6^flox/flox (conditional double mutant) testes were smaller than control organs and contained irregular testis cords and fewer gonocytes. Gene expression analysis revealed significant down-regulation of Dmrt1 and Mvh. Surprisingly, Amh expression was strongly up-regulated and remained high beyond postnatal day 7, when it is normally extinguished. Neither DMRT1 nor GATA1 was detected in the Sertoli cells of the mutant postnatal testes. Furthermore, the expression of the steroidogenic genes Star, Cyp11a1, Hsd3b1, and Hsd17b3 was low throughout embryogenesis. Immunohistochemical analysis revealed a prominent reduction in cytochrome P450 side-chain cleavage enzyme (CYP11A1)- and 3β-hydroxysteroid dehydrogenase-positive (3βHSD) cells, with few 17α-hydroxylase/17,20 lyase-positive (CYP17A1) cells present. In contrast, in postnatal Sf1Cre; Gata4^flox/flox Gata6^flox/flox testes, the expression of the steroidogenic markers Star, Cyp11a1, and Hsd3b6 was increased, but a dramatic down-regulation of Hsd17b3, which is required for testosterone synthesis, was observed. The genes encoding adrenal enzymes Cyp21a1, Cyp11b1, Cyp11b2, and Mcr2 were strongly up-regulated, and clusters containing numerous CYP21A2-positive cells were localized in the interstitium. These data suggest a lack of testis functionality, with a loss of normal steroidogenic testis function, concomitant with an expansion of the adrenal-like cell population in postnatal conditional double mutant testes. Sf1Cre; Gata4^flox/flox Gata6^flox/flox animals of both sexes lack adrenal glands; however, despite this deficiency, males are viable in contrast to the females of the same genotype, which die shortly after birth.

In most mammals, inheritance of the Y chromosome assures commitment to a male fate. Sex determination becomes realized at midgestation through the expression of the Y chromosome testis-determining gene Sry. SRY-dependent activation of the transcription factor sex-determining region Y-box 9 (SOX9) orchestrates a cascade of events leading to differentiation of the Sertoli cell population that guides the conversion of the bipotential embryonic gonad into testes rather than ovaries (1, 2). After sex determination, the testis forms two separate compartments, the testicular cords and the interstitial region. The interstitial region lies outside of the testis cords and contains several cell types, most notably the steroidogenic fetal Leydig cells (2). Normal development of fetal Leydig cell progenitors depends on paracrine signaling instructions emanating from the Sertoli cells to initiate steroidogenesis (3). The master regulator steroidogenic factor 1 (SF1) (SF1/NR5A1/Ad4BP, henceforth SF1) is at the helm of the steroidogenic expression program in several endocrine organs, including the testis, where it is the first genetic marker that gives steroid-synthesizing cells their distinctive identity and controls their metabolism, proliferation, and survival (4).

In vertebrates, 6 GATA transcription factors act as key regulators of the development of multiple tissues. Two of these proteins, GATA4 and GATA6, are expressed in the somatic cells of the embryonic testis (5). Early in gonadal development, GATA4 in association with its cofactor FOG2/ZFPM2 (friend of GATA/zinc finger protein multiple 2) acts to promote sex determination and testis differentiation (6). The Cre-LoxP loss-of-function genetic approach has been applied to clarify the role of GATA4 in testis differentiation using testis-specific Cre drivers to direct Gata4 gene deletion (7, 8). Sf1Cre; Gata4^flox/flox males develop partially descended small testes, exhibit a short anogenital distance, and are infertile. The morphology of the Sf1Cre; Gata4^flox/flox testis cords is irregular, with numerous immature Sertoli cells being observed within them. The expression of Dmrt1, one of the key transcription factors in the male pathway (9, reviewed in Ref. 10), is absent throughout embryogenesis (8). Sf1Cre (11) effectively deleted Gata4 as early as embryonic day (E)11.5–E12.5 in the precursors of Sertoli and Leydig cells (8). In contrast, in Amrh2Cre; Gata4^flox/flox males, no obvious defects were observed during embryonic or early postnatal testis development, and the external genitalia and testicular descent were normal. Adult Amrh2Cre; Gata4^flox/flox males develop age-dependent infertility, accompanied by testicular atrophy and vacuolization of the seminiferous tubules (7). Amhr2Cre is expressed in fetal Sertoli cells and in Sertoli and Leydig cells postnatally (12); however, the extent of deletion in Sertoli vs Leydig cells varied depending upon the gene studied (7, 12). Therefore, it is possible that the absence of a prenatal testicular phenotype is the result of a delayed or mosaic Amhr2Cre-mediated recombination in the fetal testes (reviewed in Ref. 13).

Although the involvement of GATA4 in regulating Sertoli cells is incontrovertible, the cell-autonomous role of this protein in the steroidogenic interstitial cells is less clear. XY GATA4-null embryonic stem cells are unable to differentiate into Leydig cells (14); however, interstitial cells expressing Leydig steroidogenic enzymes develop normally in mice deficient in the GATA4 protein (8). The presence of Gata6 in the developing mouse testis has been long documented (5, 15), but no specific regulatory function has been assigned to GATA6 in any testicular lineage. Given that GATA6 is coexpressed with GATA4 in the testis, it is unknown whether their functions completely overlap or whether GATA6 plays an independent role in testis development. To address these questions, we carried out a deletion of both Gata4 and Gata6 in the mouse embryonic testis. Here, we report that these proteins exhibit several overlapping functions in the Sertoli and Leydig cells of the testis.

Materials and Methods

Generation of mouse strains

Procedures involving live animals were approved by the Institutional Animal Care and Use Committees of University of Florida. The Gata4^flox/flox and Gata6^flox/flox “flox” mice were obtained from The Jackson Laboratory repository. The transgenic Sf1Cre mice (a gift from late Dr Parker) harbor Sf1 (Nr5a1) regulatory elements driving Cre expression within a bacterial artificial chromosome (BAC) (11). Strains carrying Sf1Cre-mediated deletions were produced by crossing flox mice with Sf1Cre-containing animals, followed by backcross to generate homozygous deletions. Sf1Cre; Gata6^flox/flox mice are fertile, but Sf1Cre; Gata4^flox/flox mice are sterile (16). Therefore, Sf1Cre; Gata4^flox/+ Gata6^flox/flox males were backcrossed with “double flox” Gata4^flox/flox Gata6^flox/flox females to generate conditional double mutants (Sf1Cre; Gata4^flox/flox Gata6^flox/flox). Gata4^flox/flox Gata6^flox/flox animals were used as experimental controls. The mice were maintained in a mixed 129/C57BL/6 genetic background. The primers used for genotyping (Integrated DNA Technologies) are listed in Supplemental Table 1.

First-strand cDNA synthesis and quantitative RT-PCR (qPCR)

Gonad-mesonephros complexes (for E13.5) and testes were collected at different stages of development (E15.5 and E18.5 and postnatal day [PND]4, PND9, and PND47) from controls and Sf1Cre; Gata4^flox/flox Gata6^flox/flox animals for RNA extraction. The conditions are described in Supplemental Materials and Methods. The primers used (Integrated DNA Technologies) are listed in Supplemental Table 2.

Immunofluorescence (IF)

Testes were collected from control and Sf1Cre; Gata4^flox/flox Gata6^flox/flox animals (n = 3 from each genotype) at different stages of development (E13.5, E15.5, and E18.5 and PND4, PND7, and PND30). IF experiments were carried out as previously described (16, 17). The primary antibodies and experimental conditions are listed in the supplemental antibody table.

Hematoxylin and eosin (H&E) staining

Testes from controls and Sf1Cre; Gata4^flox/flox Gata6^flox/flox mice (n = 2 from each genotype) were harvested at PND7, PND17, and PND30 for histological analysis. Tissue sections were processed as previously described (17).

Immunohistochemistry

Immunohistochemical reactions were performed with the ImmPRESS polymerized reporter enzyme staining system kit (Vector Laboratories, Inc), which uses peroxidase for detection. The procedure is described in detail in Supplemental Materials and Methods.

Intratesticular testosterone concentration

The intratesticular testosterone concentration was determined using the competitive Cayman's testosterone enzyme immunoassay kit (Cayman Chemical Co), following the manufacturer's guidelines. The procedure is described in the Supplemental section.

Bromodeoxyuridine (BrdU) incorporation and Terminal deoxynucleotidyl transferase dUTP Nick End Labeling (TUNEL) assays

These procedures are described in Supplemental Materials and Methods.

Whole-mount in situ hybridization

The procedure is described in Supplemental Materials and Methods.

Results

Absence of doublesex and mab-3-related transcription factor 1 (DMRT1) expression in the E13.5 Sertoli cells of Sf1Cre; Gata4^flox/flox Gata6^flox/flox testis

In the testis, GATA4 is already present in the somatic cells at E10.5 (8, 18, 19). Extending earlier observations (5, 15), we show that GATA6 is detected in the Sertoli and interstitial cells of control testis at E13.5 (Figure 1A). GATA4 and GATA6 are coexpressed in the Sertoli cells and in some interstitial (presumably Leydig cells) and coelomic epithelial cells (Figure 1A). Sf1Cre-mediated recombination is highly effective in the embryonic testis (compare Figure 1, A and F), and expression of the GATA4 and GATA6 proteins was no longer detectable in the somatic cells of Sf1Cre; Gata4^flox/flox Gata6^flox/flox testes as early as E13.5. As described previously (8, 20), residual coelomic epithelial cells in the double mutant testis remained positive for GATA4 or GATA6, with some of these cells expressing both proteins (Figure 1F). The efficiency of Sf1Cre in achieving the deletion of Gata genes remained high on all subsequent embryonic and PNDs examined (compare Figure 2, A and D and G and J, for E15.5 and E18.5, respectively, and figures 4 and 5 PND4 and PND30, respectively, below).

Figure 1.

Gene expression analysis of E13.5 control and Sf1Cre; Gata4^flox/flox Gata6^flox/flox testes. Representative sections of control (A–E) and Sf1Cre; Gata4^flox/flox Gata6^flox/flox (F–J) testes at E13.5. Testicular sections were stained with antibodies against GATA4 (green) and GATA6 (red) (A and F); DMRT1 (green) and SF1 (red) (B and G); AMH (green) and SOX9 (red) (C and H); the pluripotent germ cell marker OCT3/4 (green) and WT1 (red) (D and I); and the universal germ cell marker MVH (red) (E and K). Nuclei were stained with DAPI (blue). Scale bars represent 100 μm. TC, testicular cords. K, Quantitative analysis of gene expression in Sf1Cre; Gata4^flox/flox Gata6^flox/flox testes at E13.5. The examined genes were Amh, Dhh, Dmrt1, Mvh, and Sox9. The results are shown as the mean ± SEM of the fold change relative to controls for at least 4 biological replicates (n = 4), with significance considered at **, P < .01.

Figure 2.

The analysis of somatic gene expression at E15.5 and 18.5 in Sf1Cre; Gata4^flox/flox Gata6^flox/flox testes. A–L, Representative images of testicular sections from controls (A–C and G–I) and Sf1Cre; Gata4^flox/flox Gata6^flox/flox mice (D–F and J–L) at E15.5 (A–F) and E18.5 (G–L). The sections were stained for GATA4 (green) and GATA6 (red) (A, D, G, and J); DMRT1 (green) and SF1 (red) (B, E, H, and K); and AMH (green) and SOX9 (red) (C, F, I, and L). Nuclei were stained with DAPI (blue). Scale bars represent 100 μm. M and N, Gene expression analysis via qPCR in Sf1Cre; Gata4^flox/flox Gata6^flox/flox testes at E15.5 (M) and E18.5 (N). The examined transcripts were Amh, Dhh, Dmrt1, Mvh, Sf1, Sox9, Gata4, and Gata6. The results are graphed as the mean ± SEM of the fold change relative to controls, from n = 5 for E15.5 and n = 4 for E18.5 biological replicates, with significance considered at *, P < .05; **, P < .01; and ***, P < .001.

In mice, DMRT1 is expressed in the genital ridge in both sexes until approximately E14.5, when it becomes testis specific and is detected in both Sertoli and germ cells (9, 21). IF experiments revealed that DMRT1 is expressed in both the Sertoli cells (by colocalization with SF1) and gonocytes of testes (Figure 1B). In contrast, the only cells expressing DMRT1 in E13.5 Sf1Cre; Gata4^flox/flox Gata6^flox/flox testis were germ cells, whereas the Sertoli cells were devoid of DMRT1 staining (Figure 1G). A similar pattern of expression for DMRT1 was observed in subsequent stages of embryonic development (compare Figure 2, B and E and H and K, for E15.5 and E18.5, respectively). Accordingly, gene expression analysis via quantitative reverse transcription-polymerase chain reaction also revealed significant down-regulation of Dmrt1 (P < .01) in all embryonic stages evaluated (Figures 1K and 2, M and N).

In males, anti-Müllerian hormone (AMH) is responsible for the regression of the Müllerian ducts and is secreted by fetal and early postnatal Sertoli cells (reviewed in in Ref. 22). The expression of Amh in mice begins at E11.5 (22–24). During embryogenesis (E13.5 to E18.5), AMH was expressed by the Sertoli cells of both the controls and the double mutant testes (Figures 1, C and H, and 2, C and F and I and L). Early in development (E13.5), the expression of Amh in Sf1Cre; Gata4^flox/flox Gata6^flox/flox testes was no different from that in controls. In contrast, Amh expression was significantly up-regulated (P < .01) in the double mutant testes at E15.5 (Figure 2M); this trend continued at E18.5, although it was not significant (Figure 2N).

Similar to AMH, the SOX9 transcription factor is expressed by the pre-Sertoli cells and is a major protein promoting their subsequent differentiation (25, 26). SOX9 is first detectable in the bipotential gonad, and at E11.5, its expression becomes notably up-regulated in the testes and down-regulated in the ovaries (1). Previous work demonstrated that Amh expression is directly controlled by SOX9 through its binding site in the Amh promoter (27). SOX9 was immunolocalized to the Sertoli cells, with no detectable changes in the pattern of expression in Sf1Cre; Gata4^flox/flox Gata6^flox/flox testes compared with the controls in all embryonic stages examined (Figures 1, C and H, and 2, C, F, I, and L). Similarly, quantitative assessment of Sox9 expression did not reveal any significant changes in double mutant testes relative to the controls (Figures 1K and 2, M and N). Another important signaling molecule produced by Sertoli cells is the desert hedgehog (DHH) protein. DHH is required for the differentiation of steroidogenic fetal Leydig cells (28, 29). In mice, Dhh expression in the testis is detected at E11.5 and continues throughout embryogenesis (6, reviewed in Ref. 3). The expression of Dhh in Sf1Cre; Gata4^flox/flox Gata6^flox/flox testes was normal (Figures 1K and 2, M and N).

Abnormal testis cord architecture and decreased numbers of gonocytes in Sf1Cre; Gata4^flox/flox Gata6^flox/flox embryonic testis

At E13.5, we observed no notable difference in the overall number of primordial germ cells (by IF staining for the mouse vasa homolog [MVH], the pluripotent germ cell marker, POU domain, class 5, transcription factor 1 (OCT3/4), and via qPCR) between control and double mutant testis (compare Figure 1, D, E, I, and J, respectively). However, an irregular distribution of gonocytes in the disorganized testis cords of the Sf1Cre; Gata4^flox/flox Gata6^flox/flox testis was already prominent. A dramatic reduction in the overall number of gonocytes became apparent in subsequent stages of embryonic development (E15.5 and 18.5) (compare Figure 3, C and G and D and H). Accordingly, significant down-regulation of the Mvh transcript was detected at both E15.5 and E18.5 (Figure 2, M and N).

Figure 3.

Decrease in gonocyte numbers and loss of the testis architecture in Sf1Cre; Gata4^flox/flox Gata6^flox/flox males. Representative sections of control (A, C, E, and G) and Sf1Cre; Gata4^flox/flox Gata6^flox/flox (B, D, F, and H) testes at E15.5 (A–D) and E18.5 (E–H). The sections were stained for AMH (green) and the universal germ cell marker MVH (red) (A, B, E, and F); and Laminin (green) and MVH (red) (C, D, G, and H). Nuclei were stained by DAPI (blue). Scale bars represent 100 μm. TC, testis cords.

The smaller size of Sf1Cre; Gata4^flox/flox Gata6^flox/flox testes compared with the control organs was notable at the earliest stage we analyzed, E13.5 (Figures 1, A and F, and 2, G and J). To determine whether cell proliferation is compromised in the double mutant testes, we used BrdU DNA labeling. Numerous BrdU-labeled cells were observed in both and Sf1Cre; Gata4^flox/flox Gata6^flox/flox testes at E15.5 (Supplemental Figure 1, A–C and F–H) and E17.5 (Supplemental Figure 1, D, E, I, and J). Colocalization of BrdU-positive cells with the Wilms' tumor 1 (WT1) protein showed that somatic (mostly Sertoli) cells proliferate normally in both genotypes at E15.5 (Supplemental Figure 1, D and H). The ratio of BrdU-labeled cells to 4′,6-diamidino-2-phenylindole (DAPI)-positive cells did not differ between the control and double mutant testes at E17.5 (Supplemental Figure 1K)

In contrast, analysis of cell death using TUNEL staining revealed more apoptotic nuclei in Sf1Cre; Gata4^flox/flox Gata6^flox/flox testes (Supplemental Figure 2, B and C) than in controls (Supplemental Figure 2A) at E15.5. Numerous apoptotic nuclei were localized proximal to the coelomic epithelium at both embryonic points examined (E15.5 and E17.5) (Supplemental Figure 2, B and E). Similarly, a greater number of gonocytes (identified based on colocalization with MVH) was undergoing cell death in the double mutant testes (Supplemental Figure 2C).

GATA1 is not expressed in the Sertoli cells of Sf1Cre; Gata4^flox/flox Gata6^flox/flox testes

The GATA1 protein figures prominently in hematopoietic development and is required for normal erythroid and megakaryocytic development (reviewed in Ref. 30). GATA1 is absent from the developing gonad but becomes robustly expressed in the testis shortly after birth. Curiously, the Sertoli cells of the postnatal testis are the only known extrahematopoietic site of Gata1 expression (31, 32, reviewed in Refs. 33, 34). It has previously been reported that GATA1 is dispensable for Sertoli cell function and for the expression of a number of testis-specific genes (35, 36). We considered the possibility that upon the deletion of both Gata4 and Gata6, Gata1 could display a compensatory function for Sertoli gene expression. Unexpectedly, we observed that after the deletion of Gata4 and Gata6, GATA1 expression in Sertoli cells did not commence as normal in the postnatal double mutant testis (compare Figures 4, B and C and G and H, and 5, D and H). The absence of Gata1 expression was corroborated through qPCR (P < .001) (Figures 4K and 5I). Thus, our model allows the analysis of testis gene expression in the absence of all 3 GATA factors. To verify that the absence of GATA1 alone is insufficient to exert changes in the somatic or germ cells, we examined gene expression in the Gata1 transgenic model in which the transgene rescues Gata1 expression exclusively in the hematopoietic cell compartment of the Gata1-null mice, but nowhere else in the animal (ensuring the survival of the otherwise lethal Gata1 gene deletion) (35). We observed no differences in the expression patterns of the GATA4, GATA6, and H2AX proteins (Supplemental Figure 3); AMH expression was also not elevated in the absence of GATA1.

Figure 4.

GATA1 expression is lost in Sf1Cre; Gata4^flox/flox Gata6^flox/flox testes at PND4. Representative images of testicular sections from controls (A–E) and Sf1Cre; Gata4^flox/flox Gata6^flox/flox mice (I–M) at PND4. The sections were stained for GATA4 (green) and GATA6 (red) (A and F); GATA1 (green) and WT1 (red) (B and G); DMRT1 (green) and SF1) (red) (D and I); and AMH (MIS; green) and SOX9 (red) (E and J). C and H, Higher magnifications of B and G, respectively. The arrow in I points to the few DMRT1 and SF1 double-positive cells in the double mutant testis. Scale bars represent 100 μm (B, F, G, I, and J), 50 μm (A, D, and E), or 20 μm (C and H). K, Examination of gene expression via qPCR in Sf1Cre; Gata4^flox/flox Gata6^flox/flox testes at PND4. The examined transcripts were Amh, Dmrt1, Gata1, Mvh, Sox9, Gata4, and Gata6. The results are shown as the mean ± SEM of the fold change relative to controls from at least n = 3 biological replicates, with significance considered at **, P < .01 and ***, P < .001.

Figure 5.

Persistent AMH expression in Sf1Cre; Gata4^flox/flox Gata6^flox/flox testes after PND7. Representative images of testicular sections from controls (A–D and J–N) and Sf1Cre; Gata4^flox/flox Gata6^flox/flox mice (E–H and O–S) at PND7 (A–H) or PND30 (J–S). PND7 sections were stained with H&E (A and E) and with antibodies against AMH (green) and phosphorylated histone family protein H2A (G-H2AX) (red) (B and F); or against GATA1 (green) and SOX9 (red) (D and H). C and G, Higher magnifications of B and F, respectively. PND30 sections were stained with H&E (J and O) and with antibodies against GATA4 (green) and GATA6 (red) (L and Q); AMH (green) and MVH (red) (M and R); or against DMRT1 (green) and SF1 (red) (N and S). DAPI (blue) was used for nuclear staining. K and P, Higher magnifications of J and O, respectively. In panels R and S, arrows point to the remaining spermatogonia in Sf1Cre; Gata4^flox/flox Gata6^flox/flox testes. Scale bars represent 200 μm (A, E, J, and O), 100 μm (F, H, L–N, and Q–S), 50 μm (B, D, K, and P), or 20 μm (C and G). I and T, Analysis of gene expression via qPCR in Sf1Cre; Gata4^flox/flox Gata6^flox/flox testes at PND9 (I) and PND47 (T). The analyzed genes were Amh, Dmrt1, Foxl2, Gata1, Mvh, Sox9, Gata4, and Gata6. The bar graphs represent the mean ± SEM of the fold change relative to controls from at least n = 3 biological replicates, with significance considered at *, P < .05; **, P < .01; and ***, P < .001.

Similarly, the DMRT1 protein was virtually absent in the Sertoli cells of postnatal Sf1Cre; Gata4^flox/flox Gata6^flox/flox testes (compare Figures 4, D and L, and 5, N and S). The DMRT1-positive cells remaining in double mutant testes were mostly spermatogonial cells; only rare DMRT1-positive cells were observed among the Sertoli cells (double DMRT1; SF1-positive) (Figure 4I, arrow). In agreement with these results, gene expression analysis revealed significant down-regulation of Dmrt1 expression (P < .001) at all postnatal stages evaluated (Figures 4K and 5T). It has been reported that loss of Dmrt1 expression in the Sertoli cells (but not germ cells) leads to sex reversal, defined by ectopic expression of the female-specific transcription factor FOXL2 in the Sertoli cells of PND28 mouse testes (37). Although the expression of the Dmrt1 gene was dramatically down-regulated and protein staining was virtually absent in Sf1Cre; Gata4^flox/flox Gata6^flox/flox Sertoli cells at all developmental stages examined, IF experiments did not detect FOXL2-positive cells in the PND30 double mutant testis (data not shown). Although qPCR at PND47 detected a tendency for the increased expression of Foxl2, it was not statistically significant (P = .07) (Figure 5T).

Continuous expression of AMH and atypical distribution of spermatogonia in Sf1Cre; Gata4^flox/flox Gata6^flox/flox testes after PND7

AMH is expressed in the mouse testis throughout embryonic development until birth, when its expression begins to decline. We noted that at PND7 AMH protein is sharply reduced in the Sertoli cells of control testes (Figure 5, B and C) and becomes completely absent by PND30 (Figure 5M). In contrast, the expression of AMH in the Sertoli cells of Sf1Cre; Gata4^flox/flox Gata6^flox/flox testes remained high postnatally (Figure 5, F and R, for PND7 and PND30, respectively). qPCR experiments verified the IF data and corroborated significant up-regulation of Amh expression at PND9 (P < .05) (Figure 5I) and PND47 (P < .01) (Figure 5T). No postnatal changes in the expression of its major regulator SOX9 (27), determined either via IF at PND7 (Figure 5, D and H) or qPCR at PND9 (Figure 5I), were observed in double mutant testes. These results are similar to those obtained at the prenatal time points we have evaluated.

Postnatal Sf1Cre; Gata4^flox/flox; Gata6^flox/flox testes are remarkably underdeveloped (compare Figure 5, A and J and E and O, respectively). In particular, at PND30, the diameter of the seminiferous tubules is markedly smaller in the double mutant testis than in controls (compare Figure 5, K and P), with fewer spermatogonia and Sertoli (somatic) cells being observed within each testicular cord. Furthermore, in the postnatal testes, localization of the spermatogonia adjacent to the basement membrane of the testis cords is evident (G-H2AX-positive cells in Figure 5, B and C; cells with large purple nuclei in Figure 5K; and DMRT1-positive cells in Figure 5N), whereas in PND7 Sf1Cre; Gata4^flox/flox Gata6^flox/flox testes, only rare spermatogonia migrated to the basement membrane (Figure 5, F and G). The numbers of germ cells continued to decline in conditional double mutant testes, and at later postnatal stages, the expression of Mvh became significantly lower (P < .01) (Figure 5T); very few spermatogonia were detected by IF (Figure 5, R and S, arrows).

The steroidogenic pathway is compromised in Sf1Cre; Gata4^flox/flox Gata6^flox/flox fetal testes

Interstitial Leydig cells are responsible for the production of testosterone that ensures the persistence of the Wölffian ducts and stimulates their subsequent differentiation into organs of the male reproductive tract (reviewed in Ref. 38). Several enzymes have been implicated in the synthesis of testosterone from its precursor cholesterol, including the steroidogenic acute regulatory protein (STAR), cytochrome P450 side-chain cleavage enzyme (CYP11A1), 17α-hydroxylase/17,20 lyase (CYP17A1), 3β-hydroxysteroid dehydrogenase (3βHSD), and 17βHSD (38). Two populations of Leydig cells are recognized in rodents: the fetal population, which arises after sex determination and declines shortly after birth, and the adult population, which emerges during the first 2 weeks after birth and remains throughout adulthood (3, 39). Unlike adult Leydig cells, the fetal Leydig population is presumed to synthesize testosterone in a pituitary-independent manner (40).

Immunohistochemical assessment of steroidogenic enzymes in control testes showed robust staining corresponding to the CYP11A1, 3βHSD, and CYP17A1 proteins in the interstitial fetal Leydig cells at E15.5 and E18.5 (Figure 6, A–C and J–L, respectively). In contrast, in the Sf1Cre; Gata4^flox/flox Gata6^flox/flox testes, we observed a major reduction in the number of cells expressing CYP11A1 and 3βHSD at both embryonic developmental stages examined (Figure 6, D and E and M and N). Moreover, only occasional CYP17A1-positive cells were immunolocalized in the double mutant testis at E15.5 (Figure 6, F and I), which became completely undetectable at E18.5 (Figure 6O). qPCR confirmed the significant down-regulation of all of the steroidogenic genes examined, at both E13.5 and E15.5 (Figure 6, P and Q), in the Sf1Cre; Gata4^flox/flox Gata6^flox/flox testes. In agreement with these results, whole-mount in situ hybridization experiments performed at E15.5 demonstrated the same strong downward trend of Cyp11a1, Cyp17a1, and Hsd17b3 RNA expression in the double mutant testes (Supplemental Figure 4). Similarly, at E18.5, most of the steroidogenic genes were down-regulated, including Cyp11a1 (P < .01), Hsd3b1 (P < .01), and Hsd17b3 (P < .001); only Hsd3b6 was significantly up-regulated (P < .001) (Figure 6R).

Figure 6.

Analysis of steroidogenic enzyme expression during embryogenesis in Sf1Cre; Gata4^flox/flox Gata6^flox/flox testes. Sections of control (A–C and J–L) and Sf1Cre; Gata4^flox/flox Gata6^flox/flox (D–F and M–O) testes at E15.5 (A–F) and E18.5 (J–O) were stained for CYP11A1 (A, D, J, and M), 3βHSD (B, E, K, and N), and CYP17A1 (C, F, L, and O). G–I, Higher magnifications of D–F, respectively. Note the reduced number of CYP11A1-positive (D and M), 3βHSD-positive (E and N), and CYP17A1-positive (F and O) cells in Sf1Cre; Gata4^flox/flox Gata6^flox/flox testes at both developmental stages. Scale bars represent 100 μm (A–F and J–O) and 50 μm (G–I). P–R, qPCR analysis of changes in the expression of Star, Cyp11a1, Hsd3b1, Hsd3b6, Hsd17b3, and Insl3 in Sf1Cre; Gata4^flox/flox Gata6^flox/flox testes at E13.5 (P), E15.5 (Q), and E18.5 (R). The results are plotted as the mean ± SEM of the fold change relative to controls from at least n = 3 biological replicates for E13.5 and E15.5 and n = 4 biological replicates for E18.5, with significance considered at *, P < .05; **, P < .01; and ***, P < .001.

Leydig cells also express insulin-like factor 3 (Insl3), a peptide hormone that is critical for testicular descent (41; reviewed in Refs. 22, 42). In mice, Insl3 has been detected as early as E12.5 (43). Quantitative analysis of Insl3 expression in Sf1Cre; Gata4^flox/flox Gata6^flox/flox testes revealed a significant reduction of the transcript throughout embryogenesis (E13.5, P < .05; E15.5, P < .001; and E18.5, P < .05) (Figure 6, P–R), which may explain the undescended intraabdominal position of the double mutant testes proximal to the kidneys (Supplemental Figure 5, A and B). This phenotype is similar, but more severe than that of the Sf1Cre; Gata4^flox/flox males, in which the testes partially descend (8); however, distinct from Sf1Cre; Gata6^flox/flox males in which testicular development is normal (Supplemental Figure 6).

Increased expression of steroidogenic genes in Sf1Cre; Gata4^flox/flox Gata6^flox/flox postnatal testes

Before birth, a profound decrease in the steroidogenic competence of the Sf1Cre; Gata4^flox/flox Gata6^flox/flox double mutant testis is observed (Figure 6). In contrast, CYP11A1- and 3βHSD-positive cells become abundant in the interstitial region of the double mutant testis at PND4 (Figure 7, D and E, respectively). Gene expression analysis via qPCR confirmed the significant up-regulation of the expression of the steroidogenic genes Star (P < .001), Cyp11a1 (P < .01), and Hsd3b6 (P < .001) in double mutant testes (Figure 7M).

Figure 7.

Analysis of steroidogenic enzymes in the postnatal Sf1Cre; Gata4^flox/flox Gata6^flox/flox testis. Sections of control (A–C and J–L) and Sf1Cre; Gata4^flox/flox Gata6^flox/flox (D–F and M–O) testes at PND4 (A–F) or PND30 (G–L) were stained for CYP11A1 (A, D, G, and J), 3βHSD (B, E, H, and K), and CYP17A1 (C, F, I, and L). Scale bars represent 200 μm (G–L) and 100 μm (A–F). M and N, qPCR analysis of changes in the expression of Lhr, Star, Cyp11a1, Hsd3b1, Hsd3b6, and Hsd17b3 in Sf1Cre; Gata4^flox/flox Gata6^flox/flox testes at PND4 (M) and PND47 (N). The bar graphs represent the mean ± SEM of the fold change relative to controls from at least n = 3 biological replicates for both developmental stages, with significance considered at *, P < .05; **, P < .01; and ***, P < .001. O, Intratesticular testosterone concentrations (pg/mL) in controls (black bar) and Sf1Cre; Gata4^flox/flox Gata6^flox/flox (gray bar) animals at PND ≥ 120. The bar graph shows the mean concentration adjusted per mg of testicular tissue ± SEM from n = 3 biological replicates of each genotype. The data were analyzed using Student's t test, with significance considered at ***, P < .001.

Abundant CYP11A1- and 3βHSD-positive cells were also present in the testes of the Sf1Cre; Gata4^flox/flox Gata6^flox/flox animals at PND30; however, the distribution of steroidogenic cells was notably different from that at PND4. Most the CYP11A1- and 3βHSD-positive cells were clustered proximal to the coelomic epithelium (Figure 7, J and K, respectively), with only scattered CYP11A1- and 3βHSD-positive cells being localized in the interstitial region. Additionally, qPCR experiments conducted at PND47 (Figure 7N) showed that although the expression of some markers of steroidogenic cells (Cyp11a1 and Hsd3b6) did not differ from that in control testes, others (Star and Hsd3b1) were significantly up-regulated in the double mutant testes (P < .01 and P < .05, respectively).

Intriguingly, Hsd3b6 expression has been associated with adult Leydig cells (44, 45; however, see Ref. 46). Because we observed premature (E18.5) up-regulation of Hsd3b6 as well as an increase in the steroidogenic cell population in Sf1Cre; Gata4^flox/flox Gata6^flox/flox testes at PND4, we evaluated the possibility that adult Leydig cells appear precociously in the double mutant testes. Normal adult Leydig cell function is dependent on pituitary LH and requires expression of the LH receptor, Lhr (40). We examined the expression of Lhr via qPCR and found it to be down-regulated in the double mutant testes at both PND4 and PND47 (P > .01 and P > .05, respectively). Additionally, similar to the embryonic points we evaluated, no CYP17A1-positive cells were observed in the Sf1Cre; Gata4^flox/flox Gata6^flox/flox testes at PND4, with such cells only rarely being detected at PND30 (Figure 7, F and L), and the expression of Hsd17b3 was significantly lower (P < .001) at PND4 and PND47 (Figure 7, M and N). Furthermore, testosterone synthesis was significantly reduced in the double mutant testes, as assessed via ELISA (Figure 7O), and testosterone-responsive tissues, such as seminal vesicles and submaxillary glands, were severely affected in Sf1Cre; Gata4^flox/flox Gata6^flox/flox males (Supplemental Figure 5, C–E). In summary, these data suggest that premature differentiation of adult Leydig cells in Sf1Cre; Gata4^flox/flox Gata6^flox/flox animals is an unlikely explanation for the increased activity of the selected steroidogenic genes observed in the double mutant testes.

Overexpression of adrenal genes and clusters of CYP21A2 cells in conditional double mutant testes

It has long been proposed that steroidogenic adrenocortical and testis cells are derived from a common progenitor population of the adrenogonadal primordium (4, 47, 48). However, it was only recently demonstrated that the fetal mouse testis harbors a limited number of cells that express Cyp11b1 and Cyp21a1, which are genes that encode enzymes required for corticosteroid synthesis (49, 50). Interestingly, Sf1Cre; Gata4^flox/flox Gata6^flox/flox animals do not develop adrenal glands (S.G. Tevosian, E. Jiménez, H.M. Hatch, T. Jiang, D.A. Morse, S.C. Fox, M.B. Padua, manuscript submitted); however, males survive and live normal lifespans, in contrast to their female littermates, which die shortly after birth (17).

We hypothesized that steroidogenic gene expression in the testes of Sf1Cre; Gata4^flox/flox Gata6^flox/flox animals stems from the expansion of their adrenal-like population. As early as PND4, we detected overexpression of the adrenal genes Mc2r (P < .001), Cyp21a1 (P < .001), Cyp11b1 (P < .01), and Cyp11b2 (P < .01) in Sf1Cre; Gata4^flox/flox Gata6^flox/flox testes (Figure 8I), and the same trend was observed at later stages (Figure 8J). Histological analysis of the double mutant testis at PND17 revealed the presence of clusters of hypertrophic cells localized in the interstitial region, proximal to the coelomic epithelium (Figure 8F, arrowheads). CYP21A2, a key enzyme common to the synthesis of the adrenocortical hormones corticosterone and aldosterone, was similarly immunolocalized in the interstitial region at PND17 and PND30, within the cells clustered in the subepithelial zone (Figure 8, G and H, arrows). We concluded that the steroidogenic expression observed in the testes of the Sf1Cre; Gata4^flox/flox Gata6^flox/flox animals is derived not from the fetal or adult Leydig cells but from the expanded adrenocortical-like population.

Figure 8.

Adrenocortical genes are overexpressed in Sf1Cre; Gata4^flox/flox Gata6^flox/flox testes as early as PND4. Representative sections of control (A–D) and Sf1Cre; Gata4^flox/flox Gata6^flox/flox (E–H) testes at PND17 (A–C and E–G) or PND30 (D and H) were stained with H&E (A and E) and for CYP21A2 (C, D, G, and H). B and F, Higher magnifications of A and E, respectively. The arrowheads in F indicate a cluster of hypertrophic cells localized in the interstitial region. Scale bars represent 200 μm (A, C–E, G, and H) or 100 μm (B and F). I–J, Quantitative changes in the expression of the adrenal transcripts Mc2r, Cyp21a1, Cyp11b1, and Cyp11b2 in Sf1Cre; Gata4^flox/flox Gata6^flox/flox testes at PND4 (I) and PND47 (J). The results are graphed as the mean ± SEM of the fold change relative to controls from at least n = 3 biological replicates, with significance considered at **, P < .01 and ***, P < .001.

Discussion

Previously, we and others demonstrated that the GATA4 transcription factor is required for the normal development and function of the reproductive organs of both sexes, ie, the testes (7, 8) and ovaries (16, 51). Now, we show that the deletion of both GATA transcription factors GATA4 and GATA6 within the somatic compartment of the testis reveals a synergistic function for these proteins in testis differentiation. Male of the Sf1Cre; Gata4^flox/flox Gata6^flox/flox genotype develop small, nondescended testes, with irregular testis cords, and only a low number of gonocytes/spermatogonia are found at puberty. Not surprisingly, these conditional double mutant males are sterile.

Our data suggest that the reduction in the size of the double mutant testes is caused by an imbalance between cell proliferation and cell death. Although the proportion of proliferating cells in embryonic Sf1Cre; Gata4^flox/flox Gata6^flox/flox testes does not differ from that in controls, a greater number of apoptotic nuclei were detected in both the somatic and germ cells of double mutant testes. The precocious death of gonocytes at E17.5 is likely to be the main reason for the low number of spermatogonia in the adult testes. It is possible that the survival of gonocytes/spermatogonia was negatively affected by the disorganization of the testis cords in the mutants. Disorganized testis cords are known to disrupt the positioning and interaction of Sertoli cells and gonocytes/spermatogonia within them (52). In addition, there is experimental evidence suggesting the importance of Leydig cells in maintaining testis cord structure and ensuring germ cell survival (29, 53–55). Sf1Cre; Gata4^flox/flox Gata6^flox/flox testes are devoid of fetal and adult Leydig cells. It is possible that in addition to the viability, the development of the germ cells in Sf1Cre; Gata4^flox/flox Gata6^flox/flox testes is also affected. However, we did not specifically assess the status of germ cell differentiation in the double mutant testes beyond their ability to initiate G-H2AX expression.

The transcription factor SOX9 is a key regulator of Sertoli cell differentiation (reviewed in Refs. 1, 56). The levels of SOX9 were not affected in Sf1Cre; Gata4^flox/flox Gata6^flox/flox testes. However, we observed abundant apoptotic nuclei near the coelomic epithelium in the embryonic double mutant testes. Sertoli cells differentiate from precursors derived from the coelomic epithelium (57) that express the GATA4 protein (18). Sf1Cre-mediated loss of GATA proteins may preferentially affect the viability of the transitional Sertoli cell progenitors.

Unlike SOX9, DMRT1 and GATA1 were strongly down-regulated in the Sertoli cells of Sf1Cre; Gata4^flox/flox Gata6^flox/flox testis. We previously showed that DMRT1 is lost from the Sertoli cells of Sf1Cre; Gata4^flox/flox testes but only during embryogenesis (8). This pattern differs from that in the double mutant testis, where somatic DMRT1 is also undetectable postnatally (eg, at PND47), suggesting a role for GATA6 and/or GATA1 in the postnatal expression of Dmrt1. The GATA1 testis-specific promoter element contains a conserved GATA site (58), and it is likely that GATA1 is a direct target of GATA4 and GATA6 in Sertoli cells.

In contrast, AMH is highly up-regulated in the postnatal Sf1Cre; Gata4^flox/flox Gata6^flox/flox testis. It has been proposed that GATA proteins are required for the regulation of Amh expression (59). Here, we show that AMH is expressed in the Sertoli cells of the embryonic testis in the absence of GATA4 and GATA6 and is ectopically expressed in the adult testis in the absence of all 3 GATA proteins (GATA1, GATA4, and GATA6). We conclude that Amh gene expression does not require GATA function in males.

Interestingly, previously described transgenic male mice overexpressing AMH (MT-hAMH) exhibit a low number of mature Leydig cells and significant reduction of serum testosterone; hence, their virilization is incomplete (60, 61). These characteristics resemble the phenotype of the Sf1Cre; Gata4^flox/flox Gata6^flox/flox males, in which the external genitalia were underdeveloped (data not shown) and the concentration of intratesticular testosterone was dramatically reduced. We also showed that the expression of Hsd17b3, the enzyme responsible for testosterone synthesis, was significantly down-regulated. However, in MT-hAMH animals, Lhr expression is increased 5-fold, some steroidogenic enzymes are down-regulated, and the diameter of the seminiferous tubules and spermatogenesis are normal (61). Thus, the MT-hAMH testicular phenotype is distinctly different from that observed in the Sf1Cre; Gata4^flox/flox Gata6^flox/flox testes. Hence, it is highly unlikely that the up-regulation of AMH in the Sf1Cre; Gata4^flox/flox Gata6^flox/flox testes is solely responsible for their phenotype.

In Sf1Cre; Gata4^flox/flox Gata6^flox/flox testes, 2 distinct patterns of the expression of genes encoding steroidogenic enzymes could be distinguished: embryonic and postnatal, with both patterns differing from those in the controls. In the embryonic double mutant testes, a strong decline in the expression of most steroidogenic enzymes is observed (with only Hsd3b6 being overexpressed). In contrast, the same steroidogenic gene set is up-regulated in the Sf1Cre; Gata4^flox/flox Gata6^flox/flox testis right after birth. We explored the possibility of precocious differentiation of adult Leydig cells in double mutant testes based on the early up-regulation of Hsd3b6, which is known to be expressed in adult, but not fetal Leydig cells (44, 45). This possibility was found to be inconsistent with the overall gene expression pattern in the early postnatal testis of the double mutants. For example, LH is required for normal adult Leydig cell function, and the LH receptor is up-regulated in adult Leydig cells (40). However, Lhr was down-regulated in the postnatal Sf1Cre; Gata4^flox/flox Gata6^flox/flox testes. Moreover, in addition to the expression of Hsd3b6 in adult Leydig cells, recent work revealed robust Hsd3b6 expression in the adrenal glands (46). Yamamura et al also established that another Hsd3b isoform, Hsd3b1, is expressed much more efficiently by adrenocortical cells (46) compared with the Leydig cells (45, 46). Hsd3b1 expression was increased in the postnatal Sf1Cre; Gata4^flox/flox Gata6^flox/flox testes. In addition, the Sf1Cre; Gata4^flox/flox Gata6^flox/flox testes expressed common enzymes required for the androgenic and glucocorticoid/mineralocorticoid pathways (Star, Cyp11a1 and Hsd3b6, and Hsd3b1), whereas the level of Hsd17b3, the gene encoding the key enzyme for testosterone synthesis, was significantly reduced.

In summary, these data suggest that the cellular clusters found in Sf1Cre; Gata4^flox/flox Gata6^flox/flox testes expressing steroidogenic enzymes are not Leydig cells, but an adrenocortical-like population. These clusters are likely derived from the expansion of the rare adrenal-like cells present in the developing testis (49, 50). In the normal testis, the significance of the presence of these cells is currently unknown. It has been hypothesized that these cells are merely misallocated to the testes during the separation of the adreno-gonadal primordia (49). However, a role for these cells in normal testis development cannot be excluded.

In contrast to the androgen synthesis pathway, which is notably compromised in the Sf1Cre; Gata4^flox/flox Gata6^flox/flox testis, the corticosteroid and mineralocorticoid pathway is fully active in the testis of these animals, with overexpression of the adrenal enzymes Cyp21a1, Cyp11b1, Cyp11b2, and Mcr2 being observed. This is the most parsimonious explanation for the normal lifespan of the Sf1Cre; Gata4^flox/flox Gata6^flox/flox males, whereas their female littermates all die within 2 weeks after birth (17). Intriguingly, human patients with congenital adrenal hyperplasia develop testicular adrenal rest tumors that express the adrenal cortex-specific genes CYP11B1, CYP11B2, and MC2R (62). However, and distinct from Sf1Cre; Gata4^flox/flox Gata6^flox/flox testis, testicular adrenal rest tumors also express RNA for HSD17B3 and INSL3 (62).

Recently, Pihlajoki et al (64) described Sf1Cre; Gata6^flox/flox mice, in which the Gata6 gene was deleted using an Sf1Cre line of mice generated previously (63), FVB-Tg(Nr5a1-cre)2Lowl/J) that differs from the Sf1Cre mouse line used in this work (11). These animals had no obvious testicular phenotype but developed small adrenal glands with compromised steroidogenic adrenal function (64). Interestingly, the adrenal glands of the Sf1Cre; Gata6^flox/flox mice expressed gonadal-like transcripts, such as Amhr2, Inha, Inhba, and Inhbb (64). Gonadectomy of Sf1Cre; Gata6^flox/flox males led to an increase in the adrenal expression of Amhr2, Lhcgr, and Cyp17 (64). Taken together, these data suggest a role for GATA4 and GATA6 in establishing and maintaining the characteristic steroidogenic cell identities of gonads and adrenals.