Adenomatous Polyposis Coli (APC) Is Essential for Maintaining the Integrity of the Seminiferous Epithelium

Abstract

Sertoli cells provide the microenvironment necessary for germ cell development and spermatogenesis; disruption of Sertoli cell morphology or function can lead to germ cell aplasia, which is observed in testicular dysgenesis syndrome. Mutation of the adenomatous polyposis coli (APC) gene has been associated with various human cancers, including testicular cancer, but its involvement in nonmalignant testicular pathologies has not been reported. We have developed a mouse model (APC^cko) that expresses a truncated form of APC in Sertoli cells. Despite normal embryonic and early postnatal testicular development in APC^cko mice, premature germ cell loss and Sertoli cell-only seminiferous tubules were observed in mutant testes without affecting Sertoli cell quiescence, apoptosis, or differentiation, which were confirmed by the absence of both proliferating cell nuclear antigen, DNA strand breaks, and anti-Müllerian hormone, respectively. We show that mutant Sertoli cells lose their apical extensions, which would normally enclose germ cells during various stages of spermatogenesis, and were unable to maintain the blood-testis barrier because of disrupted expression of junctional proteins. We also observed an up-regulation of Snail and Slug, markers suggestive of epithelial-mesenchymal transition in the Sertoli cells, but tumorigenesis was not observed. No comparable phenotype was observed with Sertoli cell-specific loss-of-function mutations in β-catenin, leading us to speculate that truncation of APC in Sertoli cells results in progressive degeneration of the seminiferous tubules by a mechanism that disrupts the integrity of Sertoli cell junctions independently of APC-regulated β-catenin activities and leads to development of a Sertoli cell-only phenotype.

Mutations in adenomatous polyposis coli (APC), a multifunction tumor suppressor, are associated with development of various human cancers, including colon, liver, ovarian, endometrial, and testicular cancers (1–4). The most prominent role played by APC in the cell is to inhibit Wnt/β-catenin signaling by at least three different mechanisms. APC with axin, casein kinase 1, and glycogen synthase kinase 3β increases phosphorylation of β-catenin, which subsequently undergoes ubiquitination and proteasomal degradation. APC also decreases the availability of nuclear β-catenin by controlling the nuclear/cytoplasmic ratio of β-catenin. Lastly, nuclear APC binds to β-catenin and hampers activation of its transcriptional target genes in conjunction with the T-cell factor (TCF)/lymphoid enhancer factor (LEF) family of transcription factors (1, 5).

Although APC is a well-known antagonist of the Wnt signaling pathway, it also interacts with several other proteins in different signaling pathways to control various basic biological processes, including cell proliferation, cell fate determination, cell adhesion, and cell polarity (1, 5). Recent studies have suggested that APC plays an important role in cell adhesion and migration, which is intricately linked with its tumor promoting activities (1, 5). For example, mislocalization of oocytes in Drosophila egg chambers, which is indicative of a cell adhesion defect, was observed in APC mutants but not after deletion of axin and activation of β-catenin, suggesting a specific role for APC in cellular adhesion (5, 6). Similarly, restoration of full-length APC in a colon cancer cell line, SW480, increased cell adhesion and decreased cell migration by inducing changes in the cellular distribution of β-catenin and E-cadherin (7). In the colon, deletion of APC first induces changes in crypt architecture and later affects cell proliferation and survival in a colon cancer mouse model (8). Additionally, overexpression of APC in intestinal epithelial cells leads to defects in cell migration and adhesion without affecting the distribution of β-catenin and cell cycle dynamics (9).

Wnt signaling and, by extension, APC play an important role in normal reproductive tract development (10–14). Wnt7a deletion inhibits Müllerian duct regression and leads to retention of the embryonic female reproductive tract tissues in males (13). Dysregulated activation of β-catenin in the Müllerian duct mesenchyme also causes focal Müllerian duct retention and differentiation in males (15), suggesting that a fine balance of Wnt/β-catenin signaling is needed for normal reproductive tract development. For example, overexpression of Wnt4 disrupts testicular vasculature, suppresses male-specific steroidogenesis in mice, and induces sex reversal in human males (16). Disruption of naked cuticle 1, an inhibitor of Wnt pathway, causes nuclear accumulation of β-catenin in the late stages of testicular germ cells and defective spermatogenesis in male mice (12), suggesting that stage-specific control of Wnt/β-catenin signaling is required for normal spermatogenesis. A detailed analysis of APC in testicular development or function has not been reported. In this study, we show that APC is essential for the maintenance of Sertoli and germ cell junctions and that testis-specific, conditional truncation of APC resulted in premature germ cell loss and defective spermatogenesis, without affecting Sertoli cell cycle quiescence or inducing tumorigenesis.

Results

Generation of a conditionally truncated APC in testis

Wnt signaling pathway members play important roles in regulating functions of multiple cell types in testes (12, 16), but the effects of APC mutation in the testis have not been described. Complete APC knockout is embryonic lethal at embryonic day (E)6.5 (17, 18). Therefore, to understand the role of APC in testicular biology, particularly in germ and Sertoli cell interactions, we generated mice (Fig. 1A, a) with conditional deletion of exon 14 of the APC gene, APC^cko (19), using Amhr2-Cre (20), which is expressed and induces efficient recombination in Sertoli cells of murine testis (10, 21). Mutations in exon 14 are observed in human patients with familial adenomatous polyposis syndrome (22), supporting the physiological relevance of this mutation in human disease. Similar to other APC knockout mouse models (17, 18), homozygous deletion of exon 14 of APC gene is lethal, suggesting that its loss causes abnormalities in APC functions, including its ability to regulate cellular β-catenin levels (19). APC protein is 2842 amino acids long, and loss of exon 14 results in formation of a truncated 580-amino acid protein (19, 23).

Fig. 1.

Defective spermatogenesis and seminiferous tubular degeneration in APC^cko mice. A, a, Genetic cross used to developed APC^cko mice; b, PCR performed to detect APC flox and recombined alleles with equal amounts of genomic DNA template isolated from the indicated cells and tissues of APC^cko mice. B, a, IHC shows that APC is expressed in Sertoli cell (arrowhead), spermatids (blue arrowhead), and Leydig cells (arrow) of 12-wk-old control testes. B, c, Higher magnification image of rectangular area in B, a, showing APC expression in Sertoli cell cytoplasm (arrowheads) and apical extensions (arrow). B, b and d, Loss of APC expression in Sertoli (arrow heads) but not in germ (blue arrowhead) and Leydig cells (arrow) of 12-wk-old mutant testes. Graph shows mean testicular (C, a) and seminal vesicles (C, c) weights of control and APC^cko mice. C, b, Gross image of control and APC^cko adult testes. n equals number of mice used for each group, and error bars represent sem. Asterisk indicates statistically significant difference (P < 0.0001). D, Testes from control and mutant mice were analyzed by hematoxylin and eosin (H&E) at the indicated ages. Four-week testes from controls (D, a) had largely normal germ cell arrangement and morphology, with some mutant testes (2/5) (D, b and c) showing a few tubules with early germ cell loss and a SCO phenotype. Arrowheads in D, c, indicate Sertoli cell nuclei. Decreased tubular and luminal width (*) occasionally accompanied with abnormal accumulation of cells in the tubular lumen (arrow) was observed in 7-wk-old mutant testes (D, e and f) compared with controls (D, d). By 12 wk, complete or partial germ cell loss, and abnormal cell accumulation, was observed throughout mutant testes (D, h and i). D, i, Higher magnification image of rectangular area outlined with black dotted line in D, h, showing SCO tubule (*) and tubule with abnormal accumulation of cells in the center (arrow, arrowheads: Sertoli cells). Normal germ cell arrangement and spermatogenesis was observed in adult control testes (D, g). Scale bars, 50 um.

To confirm that Amhr2-Cre causes faithful recombination of the APC allele, we performed PCR to detect the flox (430 bp) and recombined (500 bp) APC alleles in testes and tails collected from APC^cko mice. We observed the recombined allele in testes but not in tails, confirming faithful expression of Amhr2-Cre (Fig. 1A, b). We also isolated genomic DNA from Sertoli and interstitial cells of APC^cko mice and observed a strong band by PCR for the recombined allele in Sertoli cells and a weaker band in interstitial cells (Fig. 1A, b), which is consistent with mainly Sertoli cell-specific expression of Amhr2-driven β-galactosidase expression in the knockin Amhr2-LacZ mouse (Supplemental Fig. 1A, published on The Endocrine Society's Journals Online web site at http://mend.endojournals.org) (10, 21). Next, we examined the expression of APC in control and mutant testis using an antibody directed to the C-terminal end of APC, which was detected in the spermatids (Fig. 1B, a, blue arrowhead), interstitial (Fig. 1B, a, arrow), and Sertoli (Fig. 1B, a, arrowheads) cells of control APC^flox/flox testes. A higher magnification view shows APC protein expression in the cytoplasm (Fig. 1B, c, arrowheads) and apical extensions (Fig. 1B, c, arrows) of control Sertoli cells. In APC^cko testes, APC protein expression in spermatids (Fig. 1B, a–d, blue arrowhead) and interstitial cells (Fig. 1B, a–d, arrow) was similar to controls. However, APC protein expression is lost or suppressed in Sertoli cells of mutant testes (Fig. 1B, d). Similarly, Colnot et al. (23) developed another mouse model with deletion of exon 14 of the APC gene and showed loss of APC protein expression in early intestinal lesions using an antibody directed to the C terminus of APC. Collectively, these data strongly suggest that Amhr2-Cre-induced recombination of the APC allele occurs mostly in the Sertoli cells of the APC^cko testes. Previous reports have also shown inefficient recombination by Amhr2-Cre in the Leydig cells (10, 21, 24, 25).

Gross examination of the male reproductive tract revealed no differences in embryonic and early testicular development of APC^cko mice (data not shown). The testicular weight at different stages of development revealed no difference in testicular weight in younger, 4- to 7-wk-old control APC^flox/flox and mutant APC^cko mice (Fig. 1C, a). However, at 12 wk, mutant APC^cko testes were much smaller compared with controls (Fig. 1C, b), and the weight of mutant testes was approximately 1/5th of the controls (Fig. 1C, a). Even though Amhr2-Cre is also expressed in the Leydig cells, no difference in seminal vesicles weight was observed between control and mutant animals (Fig. 1C, c), suggesting that testosterone levels were normal in mutant and control animals and that expression of APC^cko does not affect Leydig cell development or function. To confirm that the Leydig cells are functionally normal, we performed immunohistochemical examination of 3β-hydroxysteroid dehydrogenase (a Leydig cell marker) and observed no difference in the expression pattern between controls and mutant mice (data not shown).

Defective spermatogenesis, abnormal germ cell differentiation, and premature release of germ cells in APC^cko mice

To determine the cause of the decreased testicular size and weight in APC^cko mice, we performed histological examination of control and mutant testes at different developmental stages. At 4 wk of age, seminiferous tubules in control and mutant testes showed similar morphology and normal arrangement of germ and Sertoli cells (Fig. 1D, a, and Supplemental Fig. 1). However, at this stage of development, some APC^cko mice (2/5) testes showed deterioration in a few tubules and germ cell loss (Fig. 1D, b and c). Examination of 7-wk-old testes revealed decreased tubular and luminal width accompanied by abnormal accumulation of Sertoli and germ cells in the lumen of seminiferous tubules from APC^cko mice (Fig. 1D, e and f) compared with controls (Fig. 1D, d). By 12 wk, germ cells were absent in most tubules of APC^cko testes (Fig. 1D, h and i). Additionally, some of the Sertoli cells were displaced from their normal location at the periphery of the seminiferous tubules toward the lumen (Fig. 1D, h and i). In contrast, normal germ cell development was observed in adult control mice (Fig. 1D, g). Abnormally large multinucleated giant cells were also present in some seminiferous tubules of adult mutant mice, suggesting abnormal germ cell differentiation (Supplemental Fig. 1).

Analysis of the expression of germ cell nuclear antigen (GCNA), a germ cell marker, in both control and mutant testes (n = 3) was also performed to confirm germ cell loss in the mutant testes. Strong GCNA staining was observed in spermatogonia located in the basal compartment, and weak GCNA staining was observed in spermatocytes and spermatids in controls (Fig. 2A, a, c, and e). In APC^cko testes, GCNA staining confirmed the progressive germ cell loss and abnormal germ cell differentiation phenotype (Fig. 2A, b, d, f, and g) that we had observed by histology in Fig. 1. We counted GCNA-positive cells in 12-wk-old mutant testes (n = 3) as previously described (10) and found that nearly 40% of tubules were completely devoid of GCNA-positive germ cells, and 20% of the tubules contained fewer than five GCNA-positive germ cells. Interestingly, GCNA-positive cells were also observed in adult APC^cko epididymides (Fig. 2A, h), suggesting that immature germ cells were being released from the mutant tubules. We examined the epididymides from control and APC^cko mice and observed mature spermatozoa in both control and mutant epididymides collected from 7-wk-old mice (Fig. 2B, a and b). However, examination of 12-wk-old mutant epididymides revealed the presence of immature round germ cells instead of mature spermatozoa (Fig. 2B, c and d), suggesting that the integrity of the junctional complexes between Sertoli and germ cells was compromised.

Fig. 2.

Progressive germ cell loss in APC^cko testes. A, GCNA (red) expression was analyzed in testes and epididymides by IF at the indicated ages. Nuclei were stained with DAPI (blue). Strong GCNA staining marks spermatogonia, and spermatocytes/spermatids are shown with weak staining. A, a, c, and e, GCNA-specific staining in control testes revealed normal arrangement and presence of full complement of germ cells in seminiferous tubules. Intense staining for GCNA was observed in spermatogonia present in the basal compartment, and weak staining was observed in more mature germ cells nearer to the adluminal compartment. A, b, d, and f, Fewer GCNA-positive cells were present in mutant testes compared with controls. In addition, intensely GCNA-positive cells were also abnormally present in adluminal compartment. A, g, Abnormal giant GCNA-positive germ cells were also observed in adult mutant testes. A, h, GCNA staining in adult APC^cko epididymides showed presence of immature germ cells. White dotted lines outlines the seminiferous tubules. B, Histology of the epididymides at the indicated ages by hematoxylin and eosin (H&E) staining shows mature sperm were present in control (B, a) and mutant epididymides (B, b) of 7-wk-old mice and in 12-wk-old control epididymides (B, c). Immature round germ cells were present in 12-wk mutant epididymides (B, d). Scale bars, 50 um.

Normal Sertoli cell differentiation and maintenance of Sertoli cell cycle quiescence in APC^cko mice

Sertoli cells differentiate and completely exit the cell cycle by postnatal d 14 in mice to create the microenvironment essential for normal germ cell development (10, 26, 27). Because we observed germ cell loss and abnormal germ cell differentiation in APC^cko mice, we examined the cell cycle and differentiation status of the Sertoli cells by colocalization of sex-determining region Y-box-containing gene 9 (SOX9) (a Sertoli cell-specific marker) and proliferating cell nuclear antigen (PCNA) (a proliferation marker) in adult APC^cko and control testes (n = 3 per each). In control adult testes, SOX9-positive cells were negative for PCNA staining, and only spermatogonial cells were positive for PCNA staining (Fig. 3A), indicating that the Sertoli cells were mature and exited the cell cycle. Examination of mutant adult testes revealed that although the occasional SOX9-negative germ cell was proliferating, SOX9-positive Sertoli cells were also PCNA-negative (Fig. 3B), suggesting that Sertoli cells maintain cell cycle quiescence in APC^cko testes. Next, we performed terminal deoxynucleotidyl transferase 2′-deoxyuridine, 5′-triphosphate nick end labeling (TUNEL) staining to detect cell death and determined that Sertoli cells in mutant adult testes were not undergoing apoptosis to any degree different from that in controls (Fig. 3, C and D). To ascertain status of Sertoli cell maturation, we performed immunostaining for anti-Müllerian hormone (AMH) [also known as Müllerian-inhibiting substance (MIS)], a known marker for immature Sertoli cells (10), and did not observe AMH expression in either control or APC^cko adult testes (Fig. 3, E and F). AMH was observed in PND1 Sertoli cells, which was used as a positive control (Fig. 3E, inset).

Fig. 3.

Normal Sertoli cell differentiation and maintenance of Sertoli cell quiescence in APC^cko murine testes. A and B, IF staining of SOX9 (red, Sertoli cell marker) and PCNA (green, marker of proliferating cells) is shown in control and mutant testes at 12 wk postnatal. Note that SOX9 (arrowheads)-positive cells are negative for PCNA (arrows). C and D, TUNEL staining (red) on control and mutant testes revealed no staining in Sertoli cells (arrowheads). However, some germ cells (arrow) were positive for TUNEL staining. E and F, Sertoli cells of control and mutant testes showed no staining for AMH (a marker of immature Sertoli cells). Inset in E shows AMH staining in Sertoli cells of 1-d-old testis, which was used as a positive control. Nuclei are stained with DAPI (blue). Scale bars, 50 um.

Sertoli cells lose their apical extensions and are displaced from the basement membrane in APC^cko mice testes

Sertoli cells are mesoepithelial cells located at the basement of seminiferous tubules at regular intervals and at a fixed ratio of roughly 30–50 germ cells per Sertoli cell, a ratio that is required for successful spermatogenesis (10, 28). To examine the arrangement of Sertoli cells in APC^cko mice testes, we performed coimmunostaining for vimentin and SOX9. In controls, Sertoli cells were regularly spaced and normally arranged at the base of the seminiferous tubules, and vimentin staining revealed the apical extensions of Sertoli cells projecting toward the lumen of the tubules (Fig. 4A, a, c, and e). In contrast, Sertoli cells in some tubules of 4-wk-old APC^cko mice testes showed loss of apical extensions (Fig. 4, A, b, and Supplemental Fig. 2). At later stages, Sertoli cells of mutant testes lose their apical extensions and were often observed in the lumen of the seminiferous tubules by 12 wk (Fig. 4A, d and f), leading to the disruption of normal tubular arrangement, which could contribute to the germ cell loss phenotype observed in APC^cko mice. To confirm that loss of Sertoli cells apical extensions in APC^cko mice was not due to germ cell depletion, we examined the colocalization of mouse vasa homolog (mvh) and vimentin in 7-wk-old mutant testis and observed loss of apical extensions in tubules with nearly a full complement of germ cells (Fig. 4B, a–d).

Fig. 4.

Loss of typical Sertoli cell arrangement and apical extensions in APC^cko testes. Expression of vimentin (green, staining perinuclear areas and apical extensions) and SOX9 (red, Sertoli cell nuclei) to mark the Sertoli cells was observed by IF in testes at the indicated ages. A, a, c, and e, vimentin and SOX9 costaining revealed normal regular arrangement of the Sertoli cells at the periphery of the seminiferous tubules. Vimentin staining also marks the apical extensions (arrowheads) of the Sertoli cells projected toward lumen of seminiferous tubules. A, b, In 4-wk-old mutant testes, loss of apical extensions (arrows) of Sertoli cells was observed in some tubules. However, Sertoli cells were equally spaced and located on the periphery of the seminiferous tubules. A, d and f, Later during development (7–12 wk) of mutant testes, most of the Sertoli cells lose their apical extensions (arrows). Displaced Sertoli cells are indicated with an asterisk in F. A spoke-like arrangement pattern (arrowheads) was observed for α-tubulin expression by IF in controls, which represents the microtubules of the Sertoli cells cytoskeleton (A, g, i, and k). However, the spoke-like pattern was disrupted (arrows) in mutant testes (A, h, j, and l). White dotted line marks the basement membrane of the seminiferous tubules. Nuclei are stained with DAPI (blue). B, Colocalization of vimentin and mvh in control and mutant testes (7 wk old) showing that loss of apical extensions of Sertoli cells in mutant testes occurs even when germ cell were present. C, IF for α-tubulin (red) and GCNA (green) in 7-wk-old control and mutant testes. Scale bars, 50 um.

APC has been shown to regulate cell migration and polarization by affecting the stability of microtubules (29). An extensive network of microtubules is present in the Sertoli cells and plays an important role in migration of germ cells toward the lumen of seminiferous tubules during spermatogenesis (30). For example, the exposure of murine and rat testis to the microtubules-specific toxins causes severe germ cell loss (31). Because we also observed germ cell loss in APC^cko mice testes, we examined whether expression of APC^cko is affecting Sertoli cell microtubules by analyzing the expression of tyrosinated α-tubulin, a known marker of microtubules in Sertoli cells (Fig. 4). In control testes, a spoke-like arrangement pattern was observed for α-tubulin staining. The tyrosinated α-tubulin staining showed that Sertoli cell microtubules are located perpendicular to the basement membrane of seminiferous tubules and extend throughout the length of the cells (Fig. 4A, g, i, and k). In contrast, diffuse α-tubulin staining was observed, and the spoke-like arrangement of microtubules was disrupted in APC^cko testes (Fig. 4A, h, j, and l), suggesting that the microtubular network in the Sertoli cells of mutant testes is disorganized. To confirm that changes in Sertoli cell microtubules were not a result of germ cell loss, we examined 7-wk-old testes of control and mutant mice. We chose to examine testes at this stage because some of the mutant tubules still maintained a normal germ cell arrangement. Distinct localization of strong GCNA and tyrosinated α-tubulin in 7-wk-old mice testes confirms that Sertoli cells microtubules are affected irrespective of germ cell loss in APC^cko testes (Fig. 4C).

Changes in the expression pattern of tight junction (TJ)-associated and adherens junction (AJ)-associated proteins at the blood-testis barrier (BTB) site and disruption of integrity of BTB in APC^cko testes

Sertoli-basement membrane, Sertoli-Sertoli, and Sertoli-germ cells interconnections define the basic framework of the seminiferous tubules and contribute to the formation of BTB, which provides an immunoprotective environment for postmeiotic germ cell development (26, 32). Unlike blood-retinal and blood-brain barriers, TJ, AJ, and desmosomal junctions coexist and contribute to the formation of BTB (32). Because disruption of various junctional proteins has been shown to induce Sertoli misplacement and germ cell loss in murine testes (33), we analyzed the expression of AJ (β-catenin, N-cadherin) and TJ [zonula occludens 1 (ZO-1)] proteins in control and mutant testes at different developmental stages. Similar to previous observations (34), β-catenin protein expression was observed in the basal and lower adluminal compartment at the BTB of control testes (Fig. 5A, a, c, and e). In contrast, nuclear accumulation of β-catenin in Sertoli cells accompanied by loss of membranous β-catenin expression in adjoining areas was observed in APC^cko testes by 12 wk (Fig. 5A, b, d, and f).

Fig. 5.

Nuclear accumulation of β-catenin in the Sertoli cells of APC^cko mice. A, Expression of β-catenin in testes was analyzed by IF at the indicated ages. A, a, c, and e, In control testes, membranous β-catenin (green) was mainly observed at the site of the BTB (arrowhead). β-Catenin staining was observed in Sertoli cell nuclei (arrow) and absent at the BTB (*) in mutant testes (A, b, d, and f). Nuclei are stained with DAPI. B, TCF1 and LEF1 protein expression in control and mutant testes. Arrowheads represent Sertoli cell nuclei. C, Colocalization of SOX9 and β-catenin, and GATA1 and β-catenin in control and mutant testes. Arrowheads are pointed toward Sertoli cells. D, Coimmunostaining of β-catenin with vimentin, αSMA, and mvh in APC^flox/flox and APC^cko testes. Arrowheads are pointed toward Sertoli cells. Scale bar, 50 um.

TCF1 and LEF1 are downstream targets of Wnt/β-catenin signaling (1, 35). We examined the expression of these transcription factors in control and mutant testes (Fig. 5B). In control testes, very weak signal was detected in Sertoli cells (Fig. 5B, a and c, arrowheads) for TCF1 and LEF1. In contrast, TCF1 and LEF1 were strongly expressed in Sertoli cells of the mutant testes confirming that active Wnt/β-catenin signaling occurred in Sertoli of the mutant testes (Fig. 5B, b and d). Consistent with well-accepted paradigm that all cells with APC mutations and deletions need not show nuclear accumulation of β-catenin (36, 37), only a few Sertoli cell nuclei showed high levels of nuclear β-catenin staining in APC^cko testes (Fig. 5A). Additionally, coimmunostaining of β-catenin with SOX9, GATA1, vimentin (Sertoli cells markers), α-smooth muscle actin (αSMA) (a peritubular myoid cell marker), and mvh confirmed that strong nuclear β-catenin staining is limited to a few Sertoli cells (Fig. 5, C and D). Weak expression of SOX9 was observed in Sertoli cells with nuclear β-catenin (Fig. 5C), consistent with a previous report showing β-catenin signaling suppresses SOX9 expression in Sertoli cells (38). However, cells with nuclear β-catenin were positive for GATA1 and vimentin confirming their identity as Sertoli cells (Fig. 5, C and D). Colocalization of nuclear β-catenin in cells with mvh or αSMA was never observed (Fig. 5D and Supplemental Fig. 2), and in the interstitium, nuclear β-catenin was never observed in a Leydig cells, as detected by steroidgenic factor 1 (SF-1) expression (Supplemental Fig. 2).

Similar to β-catenin expression, ZO-1 and N-cadherin were observed in the basal compartment at the site of the BTB in control testes (Fig. 6A, a, e, i, c, g, and k). However, ZO-1 and N-cadherin proteins were observed diffusely and away from the BTB site (Fig. 6A, b, f, j, d, h, and l) in APC^cko testes, suggesting that the integrity of BTB was compromised. We performed a biotin tracer assay on control and mutant testes to check the permeability of BTB. In control testes, biotin permeability is limited to the basal compartment of tubules and to interstitial cells (Fig. 6B, a, b, and c). However, in APC^cko testes, the biotin tracer passed through the BTB and was present in the adluminal compartment (Fig. 6B, d, e, and f), confirming that the BTB was compromised by the expression of APC^cko.

Fig. 6.

Changes in the expression pattern of adherens (N-cadherin) and TJ (ZO-1) proteins accompanied with loss of integrity of the BTB in APC^cko testes. A, a, e, i, c, g, and k, In control testes, ZO-1 and N-cadherin were localized by IF at the BTB site close to the basement membrane. In contrast, ZO-1 and N-cadherin proteins diffuse away from the BTB and basement membrane in mutants (A, b, f, j, d, h, and l). Biotin tracer was injected into the interstitial space of 8-wk-old control and APC^cko testes (n = 3). Biotin was limited to the basal and interstitial compartment of control testes (B, a, b, and c). In contrast, biotin penetrated to the adluminal compartment of APC^cko testes (B, d, e, and f). Actin cytoskeleton of seminiferous tubules was stained with phalloidin (red). White dotted lines represents the basal membrane of the seminiferous tubules. Nuclei are stained with DAPI. Scale bar, 50 um.

Mislocalization of E-cadherin and up-regulation of epithelial-mesenchymal transition (EMT) markers in mutant testes

Wnt signaling has been shown to promote EMT in cancer cell lines by affecting Snail and Slug expression (39) The forced expression of Snail and Slug affects E-cadherin expression and induces EMT in ovarian cancer cell lines (40). APC is known to regulate cell migration and adhesion in a colon cancer cell line by affecting localization and association of junctional proteins such as E-cadherin and β-catenin (7). Although Sertoli cells are mesoepithelial in origin, mature Sertoli cells exhibit a more epithelial phenotype and are stationed at the periphery of tubules, connected to the basement membrane (26, 41), Because Sertoli cells in APC^cko testes were displaced to the lumen of tubules and lose their epithelial characteristics, we suspected that mutant Sertoli cells undergo EMT and, thus, examined expression of EMT markers (E-cadherin, Snail and Slug) by performing coimmunostaining with GCNA. E-cadherin is mainly expressed at the BTB site in the basal compartment and also in apical extensions of Sertoli cells in control testes (Fig. 7A). However, in mutant testes, perinuclear punctate staining for E-cadherin was detected, and expression of E-cadherin was lost in the apical extensions of Sertoli cells (Fig. 7B). Slug expression was higher in the Sertoli cell nuclei of APC^cko testes than in controls (Fig. 7, C and D), and Snail expression also appeared marginally increased in the APC^cko testes compared with controls (Fig. 7, E and F). We also confirmed that up-regulation of Snail and Slug was observed in the Sertoli cells of mutant testes by costaining these EMT markers with vimentin (Fig. 7, G–R). Although we cannot rule out that Snail and Slug might be up-regulated in germ cells as well, Western blot analyses of control and mutant testes confirmed increased Snail and Slug and decreased E-cadherin protein expression in mutant mice compared with controls (Fig. 7S). Although we never observed tumor formation in mutant mice, collectively, these findings suggest that Sertoli cells in APC^cko adult testes might be undergoing EMT.

Fig. 7.

Abnormal E-cadherin localization and increased EMT markers (Snail and Slug) expression in mutant testes. Germ cells are stained with GCNA (red), and nuclei are counterstained with DAPI (blue). A and B, In control testes, E-cadherin IF is localized at the BTB (arrowhead) in Sertoli-Sertoli or Sertoli-germ cells AJ and cytoplasmic extensions (arrow) of the Sertoli cells. In mutant testes, E-cadherin protein expression was suppressed (*) and was only observed in some cells in a perinuclear punctate pattern (arrowhead). C and D, In control testes, expression of Slug is observed in spermatocytes and spermatids and limited to the adluminal compartment. In mutant testes, increased Slug expression was observed in the Sertoli cells located in basal compartment. Asterisk mark seminiferous tubule with Slug expression but without any GCNA-positive germ cells. E and F, A minimal increase in Snail expression was observed in adult mutant testes compared with controls. Nuclei are stained with DAPI. G–R, Costaining of Snail and Slug with vimentin in 12-wk-old control and mutant testes. S, Western blot analyses for E-cadherin, Snail and Slug in control and mutant testes. Scale bar, 50 um.

Normal E-cadherin expression and seminiferous tubule morphology with conditional deletion of β-catenin (Ctnnb1^cko) in mouse testis

β-Catenin is a component of AJ as well as a key mediator of the canonical Wnt signaling (42). At present, it is unclear if the same pool of β-catenin contributes to its membranous and Wnt signaling functions or whether cells maintain two different pools of β-catenin, each contributing to its two different but equally essential functions (42). Because APC regulates the bioavailability of β-catenin and we observed loss of membranous β-catenin adjacent to the cells with nuclear β-catenin in APC^cko testes (Fig. 5A), we hypothesized that the phenotype observed in APC^cko testes might be because of disrupted junctional complexes caused by loss of membranous β-catenin concomitant with translocation of most cytoplasmic β-catenin to the nuclei of the Sertoli cells. This loss of membranous β-catenin might also be the reason for changes in the expression pattern of E-cadherin in APC^cko testes that we observed in Fig. 7. To test these possibilities, we analyzed another mouse model, in which β-catenin is deleted in Sertoli cells using Amhr2-Cre as described in our previous study (43). Histological examination of adult (>12 wk) control and Ctnnb1^cko mice testes revealed no abnormities in their testicular morphology (Fig. 8, A and B). β-Catenin protein expression analyses confirmed loss of β-catenin at BTB site in Ctnnb1^cko mice testes (Fig. 8, C and D). To test whether loss of β-catenin induces changes in expression pattern of E-cadherin, we analyzed and observed no changes in expression of E-cadherin between control and mutant mice (Fig. 8, E and F). These findings suggest that loss of membranous β-catenin is not the cause of the phenotype observed in APC^cko testes.

Fig. 8.

Normal testicular development in β-catenin deleted mice (Ctnnb1^cko). A and B, Hematoxylin and eosin (H&E) staining showed normal morphological arrangement of germ and somatic cells in control and mutant testes. C and D, IF of β-catenin showed staining in control and absence in mutant Sertoli cells. E and F, Normal E-cadherin expression was observed in control and mutant testes. Nuclei are stained with DAPI. Scale bar, 50 um.

Discussion

Sertoli cells are located at the periphery of seminiferous tubules and rest on the basement membrane, where their interactions with germ cells are essential for normal spermatogenesis (26). Inter-Sertoli cell junctions contribute to the formation of the BTB and divide seminiferous tubules into two different compartments, basal and adluminal (32). Spermatogonia and preleptotene germ cells are located in the basal compartment, whereas meiotic germ cells are increasingly adluminal during spermatogenesis (32). The BTB is highly dynamic, because it continuously needs to restructure to accommodate the migration of germ cells from the basal to the adluminal compartment (44). Disruption of junctional proteins causes defects in Sertoli-germ cell adhesion and the BTB and ultimately leads to germ cell loss and mislocalization of the Sertoli cells (33). In this study, we have shown that APC is an important regulator of Sertoli cell junctions and that its deletion causes their disruption, the displacement of the Sertoli cells toward the lumen of seminiferous tubules, and germ cell loss. The striking similarities in the testicular phenotype observed in junctional proteins knockout mouse models (33) and APC^cko mice strongly suggest that the phenotype observed in APC^cko testes is due to disruption of Sertoli and germ cell junctions.

APC deletions are known to induce tumor formation in several organs by inducing alterations in cell survival, proliferation, and differentiation (1). However, in this study, conditional deletion of exon 14 of the APC gene had no affect on Sertoli cell cycle quiescence and differentiation, and testicular tumors were not observed. In contrast, expression of an activated allele of β-catenin by deleting exon 3 of β-catenin (Ctnnb1^ex3−/+) with Amhr2-Cre maintains Sertoli cells in an immature, proliferative state (10). In Ctnnb1^ex3−/+ testes, immature Sertoli cells were unable to provide the proper microenvironment to support germ cell development, and abnormal accumulation of germ cells occurred in the lumen of seminiferous tubules (10). By comparison, Sertoli cells of APC^cko mice support normal spermatogenesis during early development, and the majority of defects observed in APC^cko testes appeared after 7 wk of postnatal development. These findings further highlight that abnormities observed in APC^cko testes are caused by the effect of mutated APC on Sertoli cell junctions.

Previous studies have shown that Wnt/β-catenin signaling can have stage-dependent effects on male gonadal development (10, 24, 38, 45). For example, activation of β-catenin using SF1-Cre, which is expressed after E11.5 in somatic cells of gonad, causes sex reversal of XY males to females (45). In these mice, the Sertoli cell lineage is suppressed and is accompanied by the loss of expression of the Sertoli cell marker genes (SOX9 and AMH), suggesting that activation of β-catenin suppresses male-determining pathways. Stabilization of β-catenin in E13.5 XY gonads disrupts testicular cord formation and Sertoli cell fate determination (38); these mice later developed testicular tumors (46). In contrast, early testicular development appears normal in Ctnnb1^ex3−/+ mice developed using Amhr2-Cre (10). Nuclear accumulation of β-catenin in Sertoli cells start at postnatal d 7 in Ctnnb1^ex3−/+ mice testes and Sertoli cell fate determination, as well as testicular cord formation, are normal in these mice at that time. However, when Sertoli cells normally exit the cell cycle (26), differentiation of Amhr2-Cre;Ctnnb1^ex3−/+ Sertoli cells was compromised and they continued to proliferate (10). In contrast to the phenotypes observed with constitutively activated β-catenin, the APC^cko Sertoli cells differentiated normally and stopped proliferating as evidenced by the lack of AMH and PCNA expression, even though nuclear β-catenin expression was up-regulated (Fig. 5). It was only after spermatogenesis had begun, at approximately 4 wk postnatal, that failure of the truncated APC to maintain junctional complexes led to the observed phenotype.

Many of the tubules in the APC^cko mice resembled those found in patients with Sertoli cell only (SCO) syndrome, the origin of which has been associated with abnormal functions of Sertoli cells (47). Although deletion and duplication of the APC has been associated with testicular dysgenesis syndrome (TDS) and other testicular abnormities (48), its loss in SCO has heretofore not been described. Additionally, mutations and hypermethylation of the APC gene has been observed in human testicular yolk sac tumor patients (2), and a progressive decrease in male fertility was noted in APC^min/+ mice, a well-studied mouse model of colon cancer (49). However, the mechanisms involved in APC-mediated infertility were not reported. Interestingly, examination of the contralateral testes of human testicular cancer patients showed varying degrees of germ cell loss and the presence of SCO tubules, suggesting that SCO tubules might be a precursor stage for the development of cancer (50). Although we observed SCO in APC^cko mice, we never observed testicular tumors suggesting otherwise, at least in this model system. This study has highlighted the role of APC in maintenance of testicular AJ and spermatogenesis, and further examination of the status of APC in TDS patients should shed more light on its role in TDS development.

The evidence for Amhr2-Cre expression in postnatal Sertoli cells in this report, which we and others have previously shown elsewhere (10, 21, 24, 25), appears inconsistent with other studies suggesting expression of Amhr2-Cre is limited to postnatal Leydig cells only (51, 52). AMHR2, the type II receptor for AMH/MIS (27), is expressed in Müllerian duct mesenchyme and is required for Müllerian duct regression in male embryos (53, 54). Amhr2 mRNA is also expressed in Sertoli cells (13, 53, 55, 56) and in postnatal Leydig cells (57–59). In situ hybridization studies showed Sertoli cell-specific expression of AMHR2 in embryonic mice testes (13, 55). Although, LacZ expression driven by Amhr2 promoter (Amhr2-LacZ) faithfully phenocopies Amhr2 mRNA expression in the Müllerian duct mesenchyme embryonically, Amhr2-LacZ was not detected in E12.5 to E18.5 male gonads (21). In the postnatal testis, Amhr2-LacZ expression at 2, 7, 14, 42 d old and adult (more than 6-wk-old testes) was specific to Sertoli cells at 14 d postnatal through adult testes (21). Similar to Amhr2-LacZ, Amhr2-Cre was developed by insertion of Cre recombinase at exon 5 of the Amhr2 gene (20, 21). Amhr2-Cre expression has been shown in both Sertoli and Leydig cells of the mice testis using yellow fluorescence protein or LacZ reporter mice (10, 24, 43, 52). However, Amhr2-driven Cre recombination in Sertoli and Leydig cells has been inconsistently reported (10, 24, 25, 52, 60). For example, Jeyasuria et al. (52) used Amhr2-Cre to delete SF1 and argued the Amhr2-Cre causes recombination mainly in Leydig cells of the testis, even though Amhr2-Cre driven LacZ expression was observed in both Leydig and Sertoli cells. However, others and we have shown Amhr2-Cre to activate β-catenin by deleting exon 3 of β-catenin gene and observed nuclear accumulation only in Sertoli but not in Leydig cells of the mouse testis (10, 24). No genomic recombination and nuclear β-catenin accumulation was observed in purified Cyp17A1-positive Leydig cells, and comparable testosterone concentrations were observed in control and mutant animals (10, 24). Recently, Kyrönlahti et al. (25) studied mice with conditional deletion of GATA4 using Amhr2-Cre and observed Sertoli but not Leydig cell-specific decrease or loss of GATA4 protein expression in mutant testis. In this study, we only observed loss of APC protein expression in Sertoli cells of the mutant APC^cko mice testes. Additionally, although accumulation of nuclear β-catenin, an indicator of APC loss, was observed in the Sertoli cells of the mutant testis, colocalization of β-catenin and SF1 showed no nuclear accumulation of β-catenin in Leydig cells of the testis (Supplemental Fig. 2). Furthermore, examination of TCF1 and LEF1 also revealed no differences in expression of these transcriptional proteins in Leydig cells of mutant and control testis (Supplemental Fig. 2). The lack of nuclear β-catenin accumulation in Leydig cells could be attributed to inefficient Cre-mediated deletion of the Apc flox allele. Alternatively, the normal expression of β-catenin in Leydig cells might not be sufficient to allow its accumulation and nuclear localization. In either case, although our studies and those of others conclusively show Amhr2-Cre activity in Sertoli cells, they do not preclude Amhr2-Cre expression and activity in Leydig cells. Thus, although we did not detect any evidence of a Leydig cell phenotype, we cannot rule out a contribution of Leydig cell-specific deletion of Apc to the phenotype.

In summary, we have shown that expression of a truncated form of APC, which is one of the most common mutations of the gene found in humans (19), causes seminiferous tubule degeneration and failure of spermatogenesis in mice. We have also shown evidence that the phenotype is the result of disrupted Sertoli-Sertoli and Sertoli-germ cell junctions and a compromised BTB, suggesting a mechanism for SCO syndromes in humans.

Materials and Methods

Mice genetics and husbandry

All protocols involving animal experimentation were approved by the Institutional Animal Care and Use Committee at Massachusetts General Hospital. Mice used in this study were housed under standard animal housing conditions and maintained on a mixed genetic background (C57BL/6;129/SvEv). The following mice strains: Amhr2^tm3(cre)Bhr (Amhr2-Cre, obtained from Richard Behringer) (20), APC^flox/flox (obtained from National Cancer Institute, Rockville, MD) (19), and Ctnnb1^flox/flox (obtained from The Jackson Laboratory, Bar Harbor, ME) (61) were crossed to generate Amhr2^tm3(cre)Bhr;APC^flox/flox, Amhr2^tm3(cre)Bhr;APC^flox/+, APC^flox/flox, Amhr2^tm3(cre)Bhr;Ctnnb1^flox/flox, and Ctnnb1^flox/flox mice. Hereafter, these mice strains Amhr2^tm3(cre)Bhr;APC^flox/flox and Amhr2^tm3(cre)Bhr;Ctnnb1^flox/flox will be referred as APC^cko and Ctnnb1^cko mice, respectively. Amhr2-LacZ mice (21) were obtained with the permission of Richard Behringer. Tail biopsies were used to perform genotyping in this study using standard PCR protocols. Gross pictures of male gonads were taken using a Nikon SMZ1500 microscope with an attached Spot camera (Diagnostic Instruments, Sterling Heights, MI).

Isolation of Sertoli and interstitial cells

The cells were isolated using protocols described previously (38, 62). The genomic DNA was isolated using DNeasy kit (QIAGEN, Valencia, CA). PCR conditions and primers are described in our previous study (35).

Histological analyses, immunofluorescence (IF), and immunohistochemistry (IHC)

The male reproductive tracts were collected from control and mutant mice at different developmental stages. The testes were fixed in Bouin's and 4% paraformaldehyde solution for histological and IF/IHC examination. The detailed protocols for histological and IF/IHC are previously described (10, 15). The primary and secondary antibodies used in this study were β-catenin (BD Transduction Laboratories, San Jose, CA); E-cadherin, AMH/MIS, PCNA, vimentin, and GATA-1 (Santa Cruz Biotechnology, Inc., Santa Cruz, CA); TCF1 and LEF1 (Cell Signaling Technology, Danvers, MA); tyrosine α-tubulin, β-catenin, and αSMA (Sigma, St. Louis, MO); Slug, Snail, and mvh (Abcam, Cambridge, MA); APC (Neomarkers, Fremont, CA); GCNA1 (a gift from George Enders) (63); SOX9 (Chemicon, Billerica, MA); SF1 (was a kind gift from Ken Morohashi, Kyushu University, Japan) (64); ZO-1 and N-cadherin (Developmental Studies Hybridoma Bank, Iowa City, IA); Alexa Fluor 568 phalloidin and Alexa Fluor second antibodies (Invitrogen, Carlsbad, CA); and Biotinylated donkey antimouse or antirabbit antibody F(ab)₂ (Jackson ImmunoResearch, West Grove, PA).

TUNEL staining

A detailed procedure for TUNEL staining is described in Ref. 10. Briefly, testes were collected from three different control and mutant mice and fixed overnight at 4 C in 4% paraformaldehyde. The staining to detect the level of apoptosis was performed on 5-um sections (n = 3 per group) as per manufacturer's protocol provided with cell death detection kit (Roche, Indianapolis, IN). The nuclei were counterstained with 4′,6-diamidino-2-phenylindole (DAPI) solution. All images were taken at same gain and magnification for further analyses.

Biotin tracer assay

The biotin tracer assay was performed to test the permeability of BTB as previously described (65) with minor modifications. Eight-week-old APC^flox/flox and APC^cko mice (n = 3 per group) were anesthetized. The testes of these animals were gently exposed, and a small opening was created in the tunica albuginea using a fine forceps. Afterwards, 50 μl of freshly prepared 10 mg/ml EZ-Link Sulfo-NHS-LC-Biotin (Thermo Scientific, Rockford, IL) in PBS with 1 mm CaCl₂ was injected in interstitial space of one testis, and PBS containing 1 mm CaCl₂ was injected in the contralateral control testis. Animals were euthanized after 30 min, and the testes were snap frozen for further processing; 5-μm-thick cryosections of these testes were incubated with streptavidin Alexa Fluor 488 and Alexa Fluor 568 phalloidin (Invitrogen) for 1 h at room temperature. The nuclei were counterstained with DAPI solution.

Western blot analyses

Testes (n = 3 per group) from three different APC^flox/flox and APC^cko mice were collected and protein extract was prepared in radio-immunoprecipitation assay buffer. Protein concentration was determined by the Bradford assay, and equal amounts of protein were loaded for polyacrylamide gel electrophoresis and Western blotting. β-Actin was used as a loading control. The following antibodies were used in Western blot analyses: E-cadherin (BD Biosciences, Billerica, MA), Slug, Snail (Abcam), and β-actin (Neomarkers).

Statistical analysis

Statistical analysis was performed using Prism software (GraphPad Software, La Jolla, CA). The unpaired t test was used to calculate statistical differences between groups, and P values of less than or equal to 0.05 were considered statistically significant.