Cryptorchidism in Mice with an Androgen Receptor Ablation in Gubernaculum Testis

Abstract

Androgens play a critical role in the development of the male reproductive system, including the positioning of the gonads. It is not clear, however, which developmental processes are influenced by androgens and what are the target tissues and cells mediating androgen signaling during testicular descent. Using a Cre-loxP approach, we have produced male mice (GU-ARKO) with conditional inactivation of the androgen receptor (Ar) gene in the gubernacular ligament connecting the epididymis to the caudal abdominal wall. The GU-ARKO males had normal testosterone levels but developed cryptorchidism with the testes located in a suprascrotal position. Although initially subfertile, the GU-ARKO males became sterile with age. We have shown that during development, the mutant gubernaculum failed to undergo eversion, a process giving rise to the processus vaginalis, a peritoneal outpouching inside the scrotum. As a result, the cremasteric sac did not form properly, and the testes remained in the low abdominal position. Abnormal development of the cremaster muscles in the GU-ARKO males suggested the participation of androgens in myogenic differentiation; however, males with conditional AR inactivation in the striated or smooth muscle cells had a normal testicular descent. Gene expression analysis showed that AR deficiency in GU-ARKO males led to the misexpression of genes involved in muscle differentiation, cell signaling, and extracellular space remodeling. We therefore conclude that AR signaling in gubernacular cells is required for gubernaculum eversion and outgrowth. The GU-ARKO mice provide a valuable model of isolated cryptorchidism, one of the most common birth defects in newborn boys.

In most mammals, the male gonads descend from their original embryonic intraabdominal position into the scrotum. Cryptorchidism, or hidden testis, defined as the absence of one or both testes from the scrotum, is the most common birth defect, affecting 1–4% of all newborn boys with an even higher incidence (up to 30%) in prematurely born boys (1, 2). In the majority of cases, cryptorchidism is presented as an isolated abnormality and the most common diagnosis is unilateral cryptorchidism in which one testis is located caudal to the external inguinal rings. Infertility and an increased risk of germ cell cancer in adulthood are strongly linked to cryptorchidism, and it has been proposed that all three conditions are the result of common testicular dysgenesis syndrome (3); however, recent reviews suggest that there is little evidence of shared causes between these disorders (4, 5).

Testicular descent, the movement of the testis during development, can be generally divided into two phases (6, 7). The transabdominal phase occurs in humans between wk 10 and 15 of gestation and in mice between embryonic days (E) 14.5 and E16.5. This stage is characterized by the descent of the gonad from the original pararenal location into a low abdominal position. Experimental data derived from transgenic mouse models indicate that this phase of testis descent is androgen independent and controlled by the peptide hormone insulin-like 3 (INSL3), produced in testicular Leydig cells (8, 9). The second, inguinoscrotal phase of testicular descent is believed to be androgen dependent; it is characterized by the movement of the testes from an intraabdominal position, through the inguinal canal, and across the pubic region to the scrotum. In humans, inguinoscrotal descent is finalized before birth, and in mice, the testes and epididymides move inside the scrotum within the first 2 wk of neonatal development (2, 10). Although the morphological changes taking place during testicular descent have been described (6, 11), we know remarkably little about the cellular targets of hormones, the local cell signaling pathways, and the auto- or paracrine differentiating stimuli activated by androgens during testicular descent.

Understanding the mechanisms of testicular maldescent in androgen receptor (AR)-compromised cryptorchidism is hampered by an absence of comparable animal models. Complete ablation of the AR causes testicular feminization (Tfm) in male mice, characterized by the presence of hypoplastic gonads located in a low abdominal position (12). The Tfm males have female external genitalia, do not develop a scrotum, and have a low level of testosterone and therefore are not suitable for the analysis of inguinoscrotal testicular descent. In this study, a conditional deletion of Ar using a Cre/loxP approach allowed us to study the significance of AR signaling in the masculinization of different organs. Notably, none of the reported testis- or epididymis-specific AR mutants exhibited cryptorchidism, suggesting that androgen signaling in other organs was essential for testicular descent (13–18). The objective of the present study was to demonstrate that the functional AR in the caudal genital ligament (gubernaculum) is necessary for normal testicular descent. We describe here the mouse model of cryptorchidism caused by an AR deficiency in the gubernaculum (GU-ARKO). Such males failed to form the processus vaginalis and the testes were located in a suprascrotal position. The developmental abnormalities in GU-ARKO males and possible local signaling mechanisms involved in mediating hormonal stimuli in the gubernaculum during testicular descent were analyzed.

Materials and Methods

Production of mice with conditional deletion of AR

All animal studies were approved by the Institutional Animal Care and Use Committees at the Baylor College of Medicine and Florida International University. The mice with Cre-mediated deletion of AR were produced by crossing females with the floxed Ar allele Ar^tm1.1Verh/Ar^tm1.1Verh (Ar^fl/fl) with transgenic males hemizygous for Tg(Rarb-cre)1Bhr (19), Tg(ACTA1-cre)79Jme/J (23), and Tg(Tagln-cre)1Her/J (24). The Rarb-cre mice were kindly provided by Dr. Richard Behringer (University of Texas M. D. Anderson Cancer Center, Houston, TX); the last two strains were from The Jackson Laboratory (Bar Harbor, ME). Genotyping was performed as previously described for each mutant using DNA isolated from the ear clips. At least three mice with the same genotype were analyzed in each experiment.

Hormone levels

The free testosterone level was determined in testicular homogenates of postnatal day (P) 3 testes and in adult male serum. The testes were isolated, minced on ice with a mechanical homogenizer in 150 μl of PBS, and sonicated for 90 sec to ensure cell disruption. After centrifugation (16,000 × g for 1 min) to remove cell debris, the tissue homogenate was stored at −80 C until analysis. Five wild-type and six GU-ARKO males were used in this analysis. The blood from adult males was drawn by cardiocentesis from six Tg(Rarb-cre), Ar^fl/Y cryptorchid and seven Ar^fl/Y wild-type males euthanized at 4 months of age. The serum was collected after centrifugation at 3000 × g for 15 min. The testosterone and LH levels were determined in the University of Virginia Center for Research in Reproduction Ligand Assay and Analysis Core (University of Virginia, Charlottesville, VA) using mouse RIA and two-site sandwich immunoassay, respectively.

Analysis of transgenic Cre expression using LacZ staining

To evaluate expression of the Cre transgene, we crossed females homozygous for ROSA26-LacZ reporter (Gtrosa26^tm1Sor) (34) with Tg(Rarb-cre)1Bhr. The bottom parts of the double-transgenic Tg(Rarb-cre), Rosa26-LacZ, and wild-type Rosa26-LacZ control animals at PI were frozen in Tissue-Tek optimum cutting temperature medium, sectioned at 12–15 μm, and stained by using a β-galactosidase kit (Cell Signaling Technology, Inc., Danvers, MA) and counterstained with eosin.

Histology and immunohistochemistry (IHC)

The mouse organs were collected, fixed, and embedded in paraffin, and 7-μm frontal and sagittal sections were cut. The IHC was performed using the following antibodies: AR (Santa Cruz Biotechnology, Santa Cruz, CA,); desmin (Sigma-Aldrich, St. Louis, MO); smooth muscle actin (Sigma-Aldrich); and Ki67 (Thermo Scientific, Fremont, CA). For negative controls, normal rabbit IgG or mouse IgG (Vector Laboratories, Burlingame, CA) was used at appropriate primary antibody dilutions. Detection was performed using a Vectastain ABC (avidin-biotin-peroxidase) kit (Vector Laboratories) as recommended. The color was developed with diaminobenzidine as chromogen. Samples were counterstained with Harris hematoxylin. The analysis of cell proliferation in processus vaginalis walls of the P3 males was performed in three GU-ARKO and four wild-type littermates in at least 10 viewpoints by counting the percentage of Ki67-positive cells. Stained slides were examined with a Carl Zeiss Axio A1 microscope (New York, NY), and images were captured by an AxioCam MRc5 CCD camera (Carl Zeiss). For immunofluorescence detection, sections were incubated with a goat anti-mouse secondary antibody conjugated with Alexa Fluor 488 (Invitrogen, Carlsbad, CA) to produce green fluorescence and a goat anti-rabbit Alexa Fluor 555 to produce red fluorescence. Fluorescent images were captured using an Olympus BX61 motorized fluorescence microscope (Center Valley, PA). Due to a large size in some cases (indicated in the figure legends), the presented images were manually consolidated from several images of the same area.

RNA isolation and real-time quantitative RT-PCR

Total RNA was isolated from mouse tissues using the RNeasy kit (QIAGEN, Valencia, CA) according to the manufacturer's protocol. cDNA was synthesized using an oligo(deoxythymidine) primer and RETROscript kit (Ambion, Austin, TX). A Q-PCR SybrGreen real-time assay on an Eppendorf Mastercycler ep realplex instrument (Eppendorf, Westbury, NY) was used for the real-time quantitative RT-PCR (qRT-PCR). Gapdh expression was used for normalization of SybrGreen data. The relative fold change in mRNA level was calculated by the comparative cycle threshold (2^–ΔΔCt) method. All primer sequences are available upon request.

Expression microarray analysis

Gene expression profiles were analyzed using the Illumina MouseRefseq-8 Expression BeadChip platform (Illumina, San Diego, CA). The cremasteric sac samples from four 4-month-old cryptorchid Tg(Rarb-cre), Ar^fl/Y males and four wild-type Ar^fl/Y males derived from the same litters were used for RNA isolation. The array experiments were performed in the Microarray Core Laboratory at the University of Texas Health Science Center (Houston, TX), using standard Illumina protocols. Data were analyzed using BeadStudio software (Illumina). Ingenuity IPA 8.5 Pathway Analysis software (Ingenuity Systems, Mountain View, CA) was used to identify pathways enriched by the genes differentially expressed in wild-type vs. cryptorchid tissues.

Western immunoblotting

SDS-PAGE and Western immunoblotting were performed using protein extracts isolated from the cremasteric sacs of five GU-ARKO and five wild-type (Ar^fl/Y) 6-wk-old littermates. After 4–15% gradient SDS-PAGE, the proteins were electrotransferred to a nitrocellulose membrane (Invitrogen). Mouse monoclonal β-tubulin antibody (Millipore, Carlsbad, CA) and rabbit anti-smooth muscle actin antibody (Abcam, Cambridge, MA) were used as primary antibodies with the appropriate horseradish peroxidase-conjugated secondary antibody (Promega, Madison, WI). Supersignal West Pico chemiluminescence kit (Thermo Scientific) was used to detect target proteins.

Statistical analysis

Student t test for two groups and one-way ANOVA for multiple group comparisons were used to assess significance of differences. Differences were expressed as mean ±sem; P < 0.05 was considered significant. All analyses were performed using the GraphPad Software package (GraphPad Software, La Jolla, CA).

Results

Deletion of the AR in the gubernaculum causes cryptorchidism

To generate a conditional deletion of the Ar floxed allele in the gubernaculum, we used a Tg(Rarb-cre) transgene. Cre in Tg(Rarb-cre) transgene is driven by the retinoic acid receptor 2 promoter and is expressed in the mesonephric mesenchyme and its derivatives, including different cellular components of the gubernaculum (9, 19). The expression of the transgene was previously detected in E14.5 gubernaculum and in adult cremasteric sacs (9, 19). To establish the pattern of Cre expression in the gubernaculum of newborn males, we analyzed male mice with Tg(Rarb-cre) and a ROSA26-LacZ reporter transgene. The expression of β-galactosidase (LacZ) in ROSA26 mice is activated upon the Cre-mediated excision of the floxed stop cassette; and thus, only the cells in which the Cre transgene is expressed and their progenitor cells are stained in a LacZ assay. Cre activity was detected in various gubernacular cells of newborn Tg(Rarb-cre), ROSA26 males (Fig. 1). The strongest expression was detected in the mesenchymal cells of the gubernacular bulb and the epithelial cells of the gubernacular ligament. Much lower, if any, expression was seen in the muscle cells of the abdominal wall or processus vaginalis, suggesting a possible different cellular ontogeny of the muscle cells. Some LacZ activity was also detected in the epididymis, vas deferens, and Leydig cells of the testis. We then mated the homozygous Ar^fl/fl females with Tg(Rarb-cre) males to obtain F1 males with a conditional inactivation of AR, restricted by Cre expression.

Fig. 1.

Expression analysis of the Cre transgene in a Tg(Rarb-cre), Gtrosa26^tm1Sor gubernaculum at P1. The expression of Cre recombinase leads to an activation of β-galactosidase activity of the ROSA26 allele and blue staining. A, No staining was detected in control Gtrosa26^tm1Sor animals at P1 in gubernaculum (gub) or other organs. B, Strong expression of Cre is detected in the mesenchymal cells of the gubernacular bulb (gb), epididymis (ep), vas deferens (vd), and testicular Leydig cells (t). cm, Differentiating cremaster muscles; wa, abdominal wall. C, Higher magnification of gubernacular bulb, gb. Scale bar, 100 μm.

All Ar^fl/Y, Tg(Rarb-cre) males had an external male phenotype (Fig. 2A). However, their testes were located inside the abdominal cavity, cranial to the inguinal canal in a suprascrotal position. The cryptorchid testes were protruding into the abdominal wall and were located below a thin muscle layer under the skin (Fig. 2Ae). Whereas in anesthetized wild-type males the temporarily retracted testes could be easily pushed into the scrotum manually (Fig. 2Ab), it was not possible to do in mutants (Fig. 2Ac). Thus, the conditional inactivation of Ar in the gubernaculum (GU-ARKO) caused cryptorchidism. Contrary to the female external phenotype in Tfm males (12), all parts of the male reproductive system (testis, epididymis, seminal vesicles, vas deferens, prostate, and penis) were developed in the GU-ARKO mice. The comparison of total body size and seminal vesicle weight in 10- to 12-wk-old males did not reveal significant differences between the GU-ARKO and wild-type (Ar^fl/Y) littermates (Table 1). The testis and epididymis weight in mutant males was reduced to 50% of that of the wild-type males (Table 1 and Fig. 2Af and 2Ag). The GU-ARKO males also had well-developed mammary nipples, suggesting the inactivation of AR by Cre expression in these organs and hence the importance of AR in nipple development (Fig. 2Aa).

Fig. 2.

Bilateral cryptorchidism in GU-ARKO [Ar^fl/Y, Tg(Rarb-cre)] mice. Aa, The adult wild-type and GU-ARKO males. The mammary gland nipples are well developed in mutant males. Ab and Ac, In euthanized males the testes can be manually pushed into the scrotum in wild-type (Ab) but not in GU-ARKO males (Ac). Note the absence of pigmented scrotal skin in GU-ARKO males (Ac). Ae, Mutant testes are located in a suprascrotal position, protruding through the abdominal wall, whereas in wild-type Ar^fl/Y littermates (Ad), the testes were fully descended into the scrotum in the 2-month-old animals shown. Note the poorly differentiated inguinal scrotum channel, scrotum sac (dashed brackets), and the reduced testis and vas deferens size in GU-ARKO mice (Af and Ag). B, Spermatogenesis in 2- (Ba and Bb) and 6-month-old (Bc and Bd) wild-type and GU-ARKO males. Note the abnormal seminiferous tubules in older mutants. Depletion of sperm in a mutant epididymis (Be vs. Bf). Abnormal development of the cremasteric sacs from GU-ARKO 1-month-old males (Bc vs. Bf). Bg and Bh, The mutant cremasteric sac is smaller and contains poorly organized muscle cells. Bc and Bf were consolidated from two images of the same section. a, Anus; pg, preputial glands; cm, cremasteric muscle; cs, cremasteric sac; ep, epididymis; gc, gubernacular cord; p, penis; sm, epididymal smooth muscle; sp, sperm; sv, seminal vesicles; t, testis; vd, vas deferens. *, Seminiferous tubule lumen. Scale bar, Ba, Bb, Bc, Bd, Be, Bf, 100 μm; Bg and Bh, 500 μm.

Table 1.

Reproductive organ weights and sperm count in 10- to 12-wk-old wild-type and GU-ARKO male mice

	Body weight (g)	Testis (mg)	Epididymis (mg)	Seminal vesicles (mg)	Cremasteric sac (mg)	Sperm count (mln/ml)
Wild type	27.4 ± 1.0 (8)	211 ± 7 (8)	81.5 ± 6.4 (8)	303.5 ± 32.8 (8)	53.1 ± 3.3 (8)	14.3 ± 3.3 (5)
GU-ARKO	24.3 ± 1.3 (7)	105 ± 4 (7)a	34.3 ± 7.8b	276 ± 32.2 (7)	29.1 ± 3.2 (7)a	0.9 ± 0.4 (5)b

Values are mean ± sem; number in parentheses is the number of animals analyzed; mln, millions.

^a P < 0.001.

^b P < 0.01.

The fertility of the GU-ARKO males was analyzed in crosses with wild-type females. When the young 6- to 12-wk-old males were used, few successful matings were recorded. The older, 4- to 12-month-old, GU-ARKO males were completely infertile. The spermatozoa number decreased significantly in the epididymal semen of the 10- to 12-wk-old GU-ARKO males (Table 1) without an appreciable increase in the abnormal motility or shape (data not shown). The histological examination of the 2-month-old GU-ARKO testis did not reveal noticeable differences with the wild-type males (Fig. 2Ba and 2Bb). In older animals, the degeneration of spermatogenesis was found with increased vacuolization of Sertoli cells, characteristic for a cryptorchid testis. There was a reduced number of sperm in the mutant epididymis (Fig. 2Bc–f). The size of the adult cremasteric sac was significantly reduced (Table 1); it was more collagenous and had visibly less muscle tissue in appearance (Fig. 2Bg and 2Bh).

No differences in serum testosterone (38.6 ±9.1 vs. 41.7 ±10.1 ng/dl) or LH (0.3 ±0.1 vs. 0.3 ±0.1 ng/ml) concentrations between 4-month-old wild-type and GU-ARKO males were found. A reduction of AR was also detected in adult cauda epididymis and vas deferens (data not shown), in agreement with the expression of Cre in these organs (Fig. 1).

Development of testicular maldescent in GU-ARKO mice

To analyze the effect of AR deficiency in GU-ARKO mice, we compared the testicular descent in mutant and wild-type littermates at P1, P3, and P12 (Fig. 3). The morphological differences between the two groups were already clearly visible in newborn males. In wild-type males, the processus vaginalis began to form through the outgrowth and invagination of the gubernacular bulb into the caudal region (10, 20). At that stage, the epididymis, connected to the body wall through the mesenchymal gubernacular cord, began to move into the forming inguinal sac. The inverted rims of the gubernacular bulb formed the walls of the processus vaginalis and a dense population of AR mesenchymal cells was present in the outgrowing gubernacular bulb (Fig. 4Aa). In mutant GU-ARKO males, the formation of the processus vaginalis was delayed, and the epididymis and testis were located intraabdominally under the bladder (Fig. 3, 4A). At P3, the wild-type processus vaginalis was significantly increased with the epididymis fully located inside of it. The processus vaginalis of GU-ARKO littermates was only marginally increased in size; both epididymis and testis were only slightly caudally repositioned (Fig. 3). No significant difference in cell proliferation levels in the walls of the forming cremasteric sacs in P3 GU-ARKO and wild-type males interrogated with the Ki67 antibody was found (27.7 ±3.4 vs. 32.5 ±2.2%).

Fig. 3.

Failure of gubernacular invagination in GU-ARKO male mice. The lines show the extent of processus vaginalis. bl, Bladder; gb, gubernacular bulb; pg, preputial gland; pv, processus vaginalis; sc, scrotum; wa, abdominal wall; other abbreviations are as in Fig. 2. The presented images were consolidated from two to six images from the same section. Scale bar, 500 μm.

Fig. 4.

Immunohistological localization of AR in wild-type and GU-ARKO. A, IHC for AR in the wild-type and GU-ARKO males indicates an ablation of Ar expression in the gubernacular bulb (gb, dotted line) of newborn (P1) and 12-d-old (P12) wild-type and GU-ARKO processus vaginalis littermates. Inward growth of gubernaculum in GU-ARKO mutant (Ab and Ad). An asterisk shows the development of the hypocellular region in the mutant gubernacular bulb. Aa was consolidated from two images, Ac, from four images, and Ad, from six images from the same section. B, Immunofluorescence analysis of AR (red), muscle cell marker desmin (green), and nuclei (blue fluorescent) staining showing loss of AR immunostaining in gubernaculum and in processus vaginalis of P3 GU-ARKO males. Bc and Bd, High magnification of processus vaginalis walls. Bc, Nuclei of the muscle cells (arrowheads) are negative for AR staining, whereas mesenchymal cells (arrows) are positive for AR in wild-type processus vaginalis. Bd, Mutant processus vaginalis is negative for AR apart from the few cells with weak AR expression (arrows). Note the strong AR staining in epididymal epithelium in A (brown) and B (red). Scale bar, A, Ba, and Bb, 100 μm; Bc and Bd, 10 μm.

By P12, the distal end of the wild-type processus vaginalis reached the end of the scrotum; the epididymis and the testis were located inside the developed scrotum. In GU-ARKO males, the processus vaginalis reached about half of the distance achieved by their wild-type counterpart, and only the cauda epididymis was located inside of it. The mutant testis remained in the abdominal cavity. Part of the preputial glands moved into the scrotal area, further preventing gubernacular invagination. We did not detect any remains of the mammary tissues in mutant males, which were previously suggested as the reason for testis maldescent in AR deficiency (21). Interestingly, the GU-ARKO gubernacular bulb still grew significantly in size, but inward (Fig. 4, Ad and Bb).

The pattern of AR ablation in the GU-ARKO animals was confirmed by IHC analysis (Fig. 4). In wild-type newborn males, strong AR expression was detected in the mesenchymal cells of the gubernacular bulb, peritoneal epithelium, and walls of the processus vaginalis (Fig. 4A). The GU-ARKO males had a sharply reduced number of AR-positive cells in all parts of the P1 gubernaculum (Fig. 4A), with few cells showing staining, perhaps due to a mosaic expression of Cre. The significant reduction of AR-positive cells was also visible in stromal cells of the cauda epididymis (Fig. 4Ac and 4Ad). Histological sections revealed less organized muscle cells in the walls of the GU-ARKO processus vaginalis and intense staining for desmin in the mutant gubernacular bulb in P12 males (Fig. 4B). The muscle cells of the cremasteric sac did not show significant staining for AR (Fig. 4B).

We then analyzed the level of free intratesticular testosterone in P3 testes from GU-ARKO and wild-type male siblings (Fig. 5A). No difference was detected between the two groups. Furthermore, the expression of Stra8, P450c17, Hsd17b6, and Hsd17b3 genes involved in testosterone production did not change in GU-ARKO neonatal testes when compared with wild-type littermates (Fig. 5B). The qRT-PCR analysis of Ar expression in P3 gubernaculum showed a significant down-regulation (Fig. 5C). The expression of Pax7, a muscle stem cell marker, was up-regulated, suggesting a delay in muscle cell differentiation in the GU-ARKO gubernaculum. The skeletal muscle α1-actin, Acta1, was significantly down-regulated in GU-ARKO gubernacula. No significant differences were detected by qRT-PCR in the expression of smooth muscle actin, Acta2, or INSL3 receptor Rxfp2 in mutant tissues, previously suggested to be a mediator of AR signaling in abnormal gubernacular development in LH-deficient mice (22).

Fig. 5.

Intratesticular free testosterone concentration and qRT-PCR analysis of gene expression in neonatal GU-ARKO males. A, Intratesticular free testosterone concentration in wild-type (n = 5) and GU-ARKO (n = 6) P3 testes. No significant difference was detected. B, Gene expression in P3 wild-type and GU-ARKO testis. No statistically significant differences were detected (n = 5 in each group). C, Gene expression in P3 wild-type and GU-ARKO gubernacula revealed significant down-regulation of Ar and Acta1 genes and up-regulation of Pax7. *, P < 0.05 (n = 3 in each group). Gapdh gene expression was used for normalization.

Deletion of AR in smooth or striated muscle cells in the gubernaculum does not affect testicular descent

The invagination of the gubernacular bulb gives rise to the processus vaginalis and the cremasteric sac (20). The muscle cells forming the cremasteric sac differentiate at the outer rims of the gubernacular bulb. The question then arises of whether androgen signaling directly affects the differentiation of cremaster muscle cells and whether the inactivation of AR in muscle cells will disrupt testicular descent. Using the same approach as described above, we produced mice with an Ar^fl/Y, Tg(ACTA1-cre) and Ar^fl/Y, Tg(Tagln-cre) genotype. The expression of Cre in these mice was controlled by ACTA1, skeletal muscle α1-actin, gene promoter specific for striated muscles (23), or transgelin (Tagln, smooth muscle protein 22-α) gene promoter, specific for smooth muscle cells (24). Both Acta1 and Tagln genes and the Cre-transgenes were expressed in the gubernacular ligament (25). Indeed, the analysis of genomic DNA from Ar^fl/Y, Tg(ACTA1-cre) and Ar^fl/Y, Tg(Tagln-cre) adult cremaster muscles indicated a deletion of the Ar floxed allele (Fig. 6). The intensity of the PCR band corresponding to the deleted Ar allele was lower in the smooth muscle-specific Cre than that in Ar^fl/Y, Tg(Rarb-cre) or Ar^fl/Y, Tg(ACTA1-cre), confirming that the adult cremaster muscle contains mainly striated muscle cells (26). The anatomical examination of males at 1 month of age did not reveal any difference in size or position of the testes or any appreciable differences in cremasteric sac development in the mutant animals when compared with the wild-type controls. The testes were fully descended by that time in both mutants. Thus, an ablation of androgen signaling in muscle cells does not affect gubernaculum development or testicular descent.

Fig. 6.

Deletion of the Ar^fl allele in the cremasteric sacs of mice containing different Cre transgenes. The upper PCR band identifies a nonrecombinant Ar^fl allele, and the lower band is a deleted Ar allele (Ar^ko).

Altered extracellular signaling and the shift from striated to smooth muscle composition are the main outcomes of AR deletion in the gubernaculum

To evaluate the consequences of AR ablation on gene expression in the GU-ARKO gubernacula, we performed a whole-genome expression analysis of the mutant and wild-type cremasteric sacs in 4-month-old males. Total RNA was analyzed using the Illumina Expression BeadChip platform. Both the qRT-PCR (data not shown) and array data indicated a significant down-regulation of Ar expression in the mutant samples, suggesting an efficient ablation of the gene in the cremasteric sac. The full list of the genes misexpressed in the mutant can be found in Supplemental Table 1, published on The Endocrine Society's Journals Online web site at http://mend.endojournals.org. In total, 120 genes with more than a 2-fold difference (P < 0.01) were misexpressed in the mutants, of which 63 were down-regulated. Of all 8852 genes detected on the array with P < 0.01 significance, 982 genes were known to be a direct or secondary targets of androgen signaling (27). Such genes were significantly overrepresented among the genes with at least a 2-fold difference in expression in the GU-ARKO cremasteric sacs (27 of 120, P = 0.0002) (Supplemental Table 1).

The functional analysis of the GU-ARKO misexpressed probe set was performed using Ingenuity IPA 8.5 Pathway Analysis software. The Gene Ontology molecular functions and Gene Ontology cellular components were mainly related to the extracellular matrix composition and space, signaling, and receptor binding. Among them, hyaluronan synthases, HAS1 and HAS2, were down-regulated, whereas the expression of renin was significantly up-regulated in adult mutant organs. The markers of smooth muscle cells (ACTA2, ACTG2, MYH11, MYLK, TAGLN) were significantly up-regulated, whereas the skeletal muscle heavy polypeptide 4 myosin (MYH4) was down-regulated. No differences were found in the expression of the Rxfp2, Hoxa10, Hoxa11, or Arid5b genes, mutations of which cause cryptorchidism in mice (28).

We analyzed the expression of smooth muscle actin in the cremasteric sacs of P3 and adult GU-ARKO males and their wild-type littermates (Fig. 7). In neonatal gubernacular an increased staining for ACTA2 was detected, perhaps due to a failure of gubernacular inversion. An increased expression of this marker was detected in adult mutant scremasteric sacs by IHC and Western blot hybridization, suggesting a shift toward smooth muscle myogenic differentiation in GU-ARKO cremasteric sacs.

Fig. 7.

Increased expression of smooth muscle actin in a P3 gubernaculum and in a 6-wk-old GU-ARKO cremasteric sac. A, The ACTA2-positive cells in inverted wild-type noninverted mutant gubernaculum (gb, arrowheads). Scale bar, 100 μm. B, IHC showing an increase in staining for smooth muscle actin (dark brown, arrowheads) in a GU-ARKO cremasteric sac. The wild-type image was consolidated from two images of the same section. ep, Epididymis; gb, gubernaculum; vd, vas deferens. Scale bar, 500 μm. C, Western blot analysis of two wild-type and two GU-ARKO cremasteric sacs using smooth muscle actin (ACTA2) antibody. β-Tubulin was used as the loading control.

Discussion

The majority of human testicular maldescent cases are caused by abnormalities during the inguinoscrotal phase, when the testes move from a low abdominal position into the scrotum. It has been suggested that androgen signaling plays a critical role at this stage by controlling the differentiation of the gubernaculum, processus vaginalis, and cremasteric sac. Indeed, prenatal and early neonatal exposure to antiandrogenic and estrogenic compounds causes cryptorchidism (21, 29, 30). Similarly, complete androgen resistance due to mutations in AR causes male testicular feminization syndrome, with an external female phenotype and small hypoplastic testes located in a low abdominal position (12). The use of a Cre/loxP conditional approach in mice allowed the targeting of Ar in specific organs or cells. It has been previously (16) shown that the selective inactivation of Ar in different testicular cells might result in infertility; however, it did not cause cryptorchidism, suggesting an importance of AR signaling in organs other than the testis during testicular descent. Here we present a mouse model that allows the study of mechanisms of AR signaling in inguinoscrotal testis descent and the consequences of cryptorchidism on germ cell development. We have shown that inactivation of AR in cells derived from the mesonephric mesenchyme, which includes different cells of the gubernaculum, leads to an abnormal cremasteric sac development and testicular maldescent. Thus, the gubernaculum is an essential target of androgen signaling in testicular descent.

The analysis of Rarb-cre in ROSA26 mice showed Cre expression in fetal Leydig cells. Consequently, Ar was deleted in neonatal GU-ARKO Leydig cells, yielding the question of whether such mutants had abnormal testosterone production. Previously, mice with a Leydig cell specific knockout of the Ar gene were produced using cre inserted into anti-Műllerian hormone receptor 2 gene (Amhr2-cre) (15, 16). Such mice had hypotestosteronemia caused by reduced expression of several key steroidogenic enzymes, including 17β-hydroxysteroid dehydrogenase-3, 3β-hydroxysteroid dehydrogenase-6, and P450c17. It was later demonstrated that the Amhr2-cre transgene expression is not limited to Leydig cells; it is also expressed in both fetal and adult Sertoli cells (31). Thus, a combination of AR deletions in different testicular components might be a reason for hypotestosteronemia in Amhr2-cre, Ar^fl/Y males. As shown here, the GU-ARKO mutants had normal levels of testosterone in both neonatal testis and adult serum, indicating that the deletion of AR in Leydig cells does not, in fact, affect testosterone production. The deletion of AR in GU-ARKO males was also detected in stromal cells in the cauda epididymis and vas deferens. This might suggest there is a common origin of all these cells from mesonephric mesenchymal progenitors; however, a more detailed cell fate and spatiotemporal analysis of Rarb-cre transgene expression is needed.

The cryptorchid animals produced viable sperm and were able to sire pups until 3 months of age. The analysis of testis sections in older males revealed the appearance of abnormal seminiferous tubules with arrested spermatogenesis and vacuolization of the Sertoli cells. The epididymal sperm count was significantly decreased in mutant males, suggesting the causative role of an abnormal testis position in cryptorchidism-induced infertility. There might be an additional contribution to infertility in GU-ARKO males due to abnormal function of the epididymis and vas deferens, in which AR deletion in stromal cells was detected.

The most crucial step affected by AR deficiency appeared to be the failure of the gubernaculum to undergo invagination, leading to the formation of the processus vaginalis. Two mechanisms of AR-compromised cryptorchidism have been previously proposed. First, it was suggested that the persistence of mammary gland tissues in males treated with antiandrogens might prevent the gubernacular outgrowth (21). However, we did not find any mammary tissues in the scrotal area of GU-ARKO males that would interfere with an eversion of the gubernaculum. Second, the stimulatory role of AR signaling was proposed in the genitofemoral nerve release of calcitonin gene-related peptide, providing a chemotactic gradient to guide gubernacular cell migration (2). However, the genetic targeting of Calca did not cause cryptorchidism (32). Although we did not analyze the expression of the Rarb-cre transgene and the deletion of AR in GU-ARKO neural tissues, our data indicate that the deletion of Ar in gubernacular cells, derived from the mesonephric mesenchyme rather than an indirect effect of androgens through neurotransmitters (33), was essential in the inguinoscrotal maldescent in GU-ARKO males.

We have established that the genes regulated by androgens, or participating in androgen receptor signaling, were statistically overrepresented among misregulated genes in the adult GU-ARKO cremasteric sacs. Importantly, an expression analysis revealed a significant number of genes involved in cell signaling, extracellular space, and matrix composition. The extracellular matrix plays a crucial role in regulating tissue development and function, mainly through the specific arrangement of macromolecules such as collagens, proteoglycans, glycosaminoglycans, and glycoproteins. Hyaluronan, the glycosaminoglycan, is one of the main components in the adult gubernaculum. The analysis revealed a decreased expression of hyaluronan synthases, HAS1 and HAS2, in the GU-ARKO cremasteric sac, suggesting a transcriptional down-regulation of these genes. In the newborn gubernaculum, the AR-positive mesenchymal cells were concentrated in the gubernacular bulb. The secretion of hyaluronan and other extracellular molecules by these cells may be important for the formation of increased hypocellularity in the scrotal region of the gubernacular bulb floor.

Gene expression analysis also revealed altered expression in a number of muscle-specific markers. The cremaster muscle is mainly composed of striated muscle bundles with some smooth muscle dispersed between them (26). We have shown that in neonatal GU-ARKO gubernaculum, the expression of the early muscle developmental marker Pax7 was increased, whereas the expression of skeletal muscle actin was decreased, suggesting the inhibition of normal muscle maturation in mutants. The adult GU-ARKO cremasteric sacs showed an increase in expression of smooth muscle markers and a decreased expression of striated muscle markers. This was consistent with the morphological changes observed in cryptorchid animals, in which the striated cremaster muscle organization was noticeably abnormal, and the size of the cremasteric sac was significantly reduced. The selective ablation of Ar in smooth or striated muscle cells using a Cre/loxP approach did not interfere with the testis descent or cremaster muscle development, suggesting that the androgen stimuli did not affect the differentiation or function of muscle cells directly. We therefore suggest that the deficiency of AR and the absence of AR-induced paracrine signals from the gubernacular cells on the myoblast may be responsible for the abnormal muscle differentiation. Notably, an altered expression of muscle related genes have been previously detected in a rat strain with inherited cryptorchidism (25) and in rats treated with antiandrogen flutamide (30). We also observed an abnormal myogenesis in INSL3/RXFP2-deficient gubernacula (9). Previously, using the same Cre transgene, we have shown that the conditional ablation of Notch1 or β-catenin genes (7) as well as an Rxfp2 deletion led to a significant reduction of AR-positive cells in the gubernacular bulb and subsequent failure of the gubernaculum to undergo invagination. Thus, the presence and normal function of AR-positive cells in the gubernaculum appear to be necessary for the normal formation of the processus vaginalis and testicular descent.

In summary, Cre/loxP gene targeting allowed us to produce mice with an AR deficiency in the gubernacular ligament (GU-ARKO). Such males developed all male reproductive organs, had normal hormonal milieu, but exhibited low intraabdominal cryptorchidism, demonstrating that the gubernaculum is a target organ for hormonal signaling during testis descent. A significant number of genes involved in extracellular signaling, matrix composition, and muscle differentiation were misregulated in the GU-ARKO cremasteric sac, a derivative of the gubernacular bulb. Importantly, an ablation of Ar in the gubernacular smooth or striated muscle cells using specific Cre transgenes did not affect testis descent, suggesting an indirect effect of AR signaling on myogenesis of the cremaster muscle and cremasteric sac. The GU-ARKO mouse provides an important in vivo model of AR-compromised cryptorchidism. Further analysis of this model might reveal the mechanisms of testicular descent and cryptorchidism-induced infertility.