Targeted inactivation of the androgen receptor gene in murine proximal epididymis causes epithelial hypotrophy and obstructive azoospermia.

The epithelial lining of the epididymal duct expresses the androgen receptor (Ar) along its entire length and undergoes rapid and profound degeneration when androgenic support is withdrawn. However, experiments involving orchidectomy with systemic testosterone replacement, and testicular efferent duct ligation, have indicated that structural and functional integrity of the initial segment cannot be maintained by circulating androgen alone, leaving the role of androgen in this epididymal zone unclear. We addressed this question in a mouse model with intact testicular output and selective Ar inactivation in the proximal epididymis by creating double-transgenic males carrying a conditional Ar(loxP) allele and expressing Cre recombinase under the promoter of Rnase10, a gene specifically expressed in proximal epididymis. At 20-25 d of life, on the onset of Rnase10 expression, Ar became selectively inactivated in the principal cells of proximal epididymis, resulting in epithelial hypoplasia and hypotrophy. Upon the subsequent onset of spermiation, epididymal obstruction ensued, with the consequent development of spermatic granulomata, back pressure-induced atrophy of the seminiferous epithelium, orchitis, and fibrosis of the testicular parenchyma. Consistent with these findings, the mice were infertile. When the effect of Ar knockout on gene expression in the proximal epididymis was compared with that of efferent duct ligation and orchidectomy, we identified genes specifically regulated by androgen, testicular efferent fluid, and both. Our findings demonstrate that the development and function of the epididymal initial segment is critically dependent on direct androgen regulation. The phenotype of the produced knockout mouse provides a novel model for obstructive azoospermia.

It is well established that the normal development and functional integrity of the epididymis are dependent on androgen action. The epididymis does not develop in the absence of normal testicular testosterone production (1), and if androgen action in adult age is eliminated by orchidectomy (OE) or antiandrogen treatment, the consequence is epididymal weight loss secondary to disappearance of spermatozoa from the lumen as well as cytoplasmic shrinkage and apoptosis of epididymal cells (1). If testosterone levels are restored after castration, the regressive changes in the distal caput, corpus, and cauda epididymidis are reversed (2) but not in the initial segment (IS), which has cast doubt on androgen dependence of this epididymal zone, whose normal function is crucial for the functional maturation of sperm. Efferent duct ligation (EDL) experiments, producing clear hypotrophic changes in the IS, have emphasized the importance of factors present in the testicular efferent fluid (i.e. lumicrine regulation) in the maintenance of structural and functional integrity of the IS (2). In agreement with this contention, a recent study on androgen replacement of regressed rat epididymis after OE showed that cell proliferation was increased in all epididymal regions except for the IS (3). However, this type of experiment cannot exclude the possibility that combined action of androgen and lumicrine factors is required for the maintenance of IS. Of note, the androgen receptor (Ar) gene is strongly expressed in the IS (4), but as stated above, its function still remains unclear.

As a novel approach to elucidate the role of androgen action in epididymal function and in particular in the IS, we used the targeted gene ablation technique, crossing the Ar mice (5) with a novel Cre deletor mouse expressing Cre recombinase under the promoter of Rnase10, a gene with specific expression in the proximal epididymis (6). The results obtained clearly demonstrate a role for direct androgen action in the proximal epididymis.

Materials and Methods

Reagents

Unless otherwise specified, all chemicals and solvents used were obtained from Sigma-Aldrich Co. Ltd (Poole, UK). PCR reagents were from QIAGEN (Valencia, CA).

Animals

All experimental work involving animals was carried out with approval of the local ethics committee and in line with national regulations. Mice of the wild-type (WT) strain C57BL/6 were supplied by Charles River Laboratories (Margate, Kent, UK). Heterozygous B6.Rnase10 males were produced, as described in Supplemental Information, published on The Endocrine Society's Journals Online web site at http://endo.endojournals.org, and Fig. 1, and used for the production of proximal epididymis-specific Ar knockout mice by crossing to congenic B6.Ar homozygous females (5). Genotypes of transgenic mice were determined by PCR with allele-specific primers. The iCre sequence was identified with the following pair of primers: 5′-CAGGCAGGCCTTCTCTGAAC-3′ and 5′-CATTCTCCCACCATCGGTGC-3′, yielding an amplification product of 481 bp (annealing temperature 58 C). The intact Rnase10 allele was identified with 5′-TGGGGAATGTGAGGAGAAGG-3′ and 5′-GCCAAGTGCCAGACCTTCTG-3′ primers (product size 383 bp; annealing temperature 58 C). Carriers of floxed (Ar), WT, and null alleles of Ar gene located on the X chromosome were identified as described before (5). EDL and OE were carried out under isoflurane anesthesia, with postoperative analgesia provided by ip administration of buprenorphine.

Fig. 1.

Targeted insertion of iCre into the Rnase10 locus. A, Diagrammatic representation of the targeting strategy showing the wild-type Rnase10 locus (top panel), the targeting vector (middle panel), and the modified allele (bottom panel). The gene comprises two exons with alternative transcription start sites; the coding sequence shown as a filled box resides entirely in the second exon. The targeting vector introduces the iCre-Neo cassette instead of the first eight nucleotides of exon 2, which include the Rnase10 translation initiation site (shown in bold font); the endogenous splice acceptor is retained, whereas the iCre open reading frame is headed by a strong Kozak consensus sequence. The positive selection cassette, PGK-Neo, is flanked by minimal FRT sites (shown as shaded triangles) enabling its subsequent removal. Both iCre and Neo are appended with sequences containing a polyadenylation signal (pA). Bold lines represent stretches of homology; arrows indicate annealing sites for the PCR primers used for initial screening of geneticin-resistant embryonic stem (ES) cell clones. Also shown are diagnostic restriction sites and genomic probes for Southern blot analysis. B, Southern blot analysis of the heterozygous ES cell clone used for blastocyst injection. Genomic DNA digested with NheI and BstZI was successively hybridized to radiolabeled probes as shown in A. C, RT-PCR analysis of iCre and Rnase10 transcripts in the proximal epididymis of the Rnase10 male. Total RNA was isolated individually from segments I, II, III, and IV (lanes 1–4) and subjected to RT-PCR. iCre transcript was identified with the pair of specific primers detailed in Materials and Methods; oligonucleotides used for the detection of Rnase10 mRNA were 5′-GTATGGAGAGCAGTTTCTGG-3′ and 5′-CCGTTACAGACAGCTTTGAC-3′. RNA samples were not treated with deoxyribonuclease I; therefore, control reactions were performed with the omission of reverse transcriptase (−RT).

Histological analysis

Reproductive tracts were dissected out, fixed in neutral buffered formalin, dehydrated in ethanol, cleared in Histoclear, and embedded in molten paraffin. Ten-micrometer sections were cut, deparaffinized on poly-l-lysine coated slides, and either used for immunohistochemistry or stained with hematoxylin and eosin.

Before immunohistochemical detection of the androgen receptor (AR), antigen retrieval was carried out by boiling tissue sections in 10 mm sodium citrate buffer (pH 6.0) supplemented with 0.05% Tween 20. After blocking in 8% BSA in PBS for 1 h at room temperature, sections were incubated overnight at 4 C with rabbit anti-AR IgG PG-21 (Millipore, Watford, Hertfordshire, UK) at a dilution of 1:500 in 1% BSA in PBS. After incubation with secondary antibody, biotin-labeled antirabbit IgG (ABC kit; Vector Laboratories, Peterborough, UK), at a dilution of 1:100 in PBS for 30 min, avidin-conjugated horseradish peroxidase was applied (ABC kit; Vector Laboratories) for another 30 min. Horseradish peroxidase was then detected using the diaminobenzidine chromogen substrate kit (Vector Laboratories). Once stain had developed, nonstained cell nuclei were counterstained with hematoxylin. Stained sections were dehydrated through ascending concentrations of ethanol, cleared in Histoclear, and finally mounted in dibutyl phthalate/xylene.

Quantitative real-time RT-PCR analysis

For the isolation of total RNA, fresh tissues (initial segments of epididymides) were processed using the QIAGEN RNeasy minikit, and RNA concentration and purity were determined by measuring OD260 and OD280 on a spectrophotometer. After deoxyribonuclease I treatment (Invitrogen, Carlsbad, CA), first-strand cDNA synthesis was carried out with random primers using the high-capacity cDNA reverse transcription kit (Applied Biosystems, Carlsbad, CA). Real-time PCR was performed with SYBR Green I dye and primers as detailed in Supplemental Table 1, with annealing at 55 C. Detection of a single amplicon was confirmed by a dissociation curve at the end of real-time PCR cycle. Cycle threshold values were analyzed as described (7) using Gapdh mRNA as internal control.

Statistical analyses

The numeric data are presented as mean ± sem. One-way ANOVA, followed by Bonferroni post hoc test, was used for the statistical analyses. A P < 0.05 was considered statistically significant.

Results

Generation of mice with disrupted AR in proximal epididymis

For the inactivation of the Ar specifically in the proximal epididymis, we used the binary Cre-loxP transgenic technology and produced a novel Cre deletor mouse, Rnase10 (Fig. 1 and Supplemental Materials and Methods). Ar is located on the X chromosome, and therefore, to produce males carrying both the floxed Ar exon 2 and the Rnase10 allele, female Ar mice were crossed to Rnase10 males. Male progeny of the desired genotype (Ar, Rnase10) were identified by RT-PCR with allele-specific primers (see Materials and Methods) and used for phenotype analysis. Ablation of Ar was specific for the proximal epididymis of the double-transgenic males [termed proximal epididymis-specific androgen receptor knockout mouse (ProxE-ARKO)], as demonstrated by immunohistochemistry with an anti-AR antibody (Fig. 2B and Supplemental Fig. 1).

Fig. 2.

Morphological appearance of the proximal epididymis in ProxE-ARKO mice. A, Dorsal view of the proximal epididymides of WT (left) and ProxE-ARKO (right) littermates at 30 dpp; the arrowheads indicate points of entry of the superior epididymal artery demarcating the IS distally; hypoplasia of the knockout proximal epididymis is demonstrated. Bar, 1 mm. B (immunostaining with anti-AR), The proximal segments (I-III) of the epididymides of 40-d-old ProxE-ARKO (right) and WT littermates (left) are shown; arrows indicate the reduced volume of principal cells devoid of AR protein immunoreactivity in the knockout specimens. Bar, 50 μm.

The expression of Rnase10 starts at around 17 d postpartum (dpp) (6) coincidentally with the differentiation of IS. Although Rnase10 has previously been described to be expressed exclusively in the epididymal IS (6), the zone of Ar inactivation in the ProxE-ARKO mice appeared also to include the whole of segment II, continuing into segment III (Fig. 2B and Supplemental Fig. 1). The spatial pattern of Ar inactivation in the ProxE-ARKO epididymis was found to correspond to the area of endogenous Rnase10 expression as confirmed by RT-PCR (Fig. 1C). The zone of Ar inactivation began abruptly in the margin between the efferent ducts and IS and waned gradually distally, without a distinct border (Supplemental Fig. 1). Immunohistochemical analysis of proximal epididymides from prepubertal mice showed that in the ProxE-ARKO males AR began to disappear from the nuclei of epithelial cells between d 20 and 25 of life (Supplemental Fig. 2). We found that during this time window, proximal epithelium began to differentiate from simple cuboidal into columnar pseudostratified. The pseudostratification is a result of appearance of different cell types within the epithelium, and we observed ablation of AR in nuclei of the middle layer, i.e. in differentiating principal cells that started to express Rnase10 (Supplemental Fig. 2). Consequent to the disappearance of the AR, there was a loss of epithelial height, whereas further development of the IS was aborted. Accordingly, the principal cell-specific glutamate transporter excitatory amino acid carrier 1 (solute carrier family 1) (8), localized in stereocilia in IS and in microvilli in the more distal parts of the epididymis, was absent in the knockout IS but present in corpus and cauda (Supplemental Fig. 3). Ar expression remained intact in the rest of the Wolffian duct-derived structures, as well as in the testes and accessory sex glands (Fig. 3), and similar to that detected in WT control mice (Supplemental Fig. 4).

Fig. 3.

Histological appearance and immunohistochemical detection of AR protein in reproductive organs of ProxE-ARKO mice. At 40 dpp: A, testis (S, Sertoli cells; L, Leydig cells); B, efferent ducts; C–I, epididymis segments IV-X; at 60 dpp: J, vas deferens; K, seminal vesicle; L, coagulating gland; M, ampullary gland; N, ventral prostate; O, dorsal prostate. Bar, 50 μm (counterstaining with hematoxylin). Immunodetection of AR in WT male reproductive tract is shown in Supplemental Fig. 4.

Fertility of ProxE-ARKO males

Ten knockout males that had reached maturity were mated with a total of 30 wild-type CD1 females over a period exceeding 4 months. Even though vaginal plugs were regularly observed, no pregnancies occurred. Two females were killed early on the day of mating, and upon inspection of the uterine contents, only few immotile sperm were found. Macroscopic examination of the male reproductive organs showed obstruction of the epididymis at the level of caput, revealing obstructive azoospermia as the cause of infertility.

Morphological findings

Under a dissecting microscope, a clear hypoplasia in IS of the ProxE-ARKO mice was observed at 30 dpp (Fig. 2A). Microscopic examination at around the time of onset of spermiation (39–40 dpp) revealed a substantial decrease in height of the epithelium of epididymal segments I and II (Fig. 2B). This was mainly due to a reduction in the number and volume of cytoplasm of the principal cells, which manifested as both dilatation of the intertubular space and as a relative increase in luminal diameter. In some specimens, the accumulation of spermatozoa in the affected segments of the epididymal duct was observed already at this stage (Supplemental Fig. 5). Further buildup of spermatozoa resulted in severe dilatation and finally complete occlusion of the duct (Supplemental Fig. 5D).

As a consequence of the interruption of flow of testicular fluid, severe distension developed in the efferent ducts, rete testis and seminiferous tubules. Testes examined at the age of 4 months were increased in size apparently secondary to fluid retention (Fig. 4B). The seminiferous tubules were at various stages of degeneration (Fig. 4, D and E), some with numerous mononuclear cells in the lumen. A significant proportion of the tubules had lost all cells of the germinal epithelium (Fig. 4E) and contained intraepithelial cysts (Fig. 4F), consistent with backpressure testicular atrophy; some males were found to have developed unilateral orchitis (Fig. 4F). Epididymal epithelium distal to the point of occlusion displayed signs of degeneration with the formation of intraepithelial cysts and accumulation of hyaline-like material in the lumen (Fig. 4G). Spermatic granulomata were frequently observed in obstructed epididymides (Fig. 4, A and H).

Fig. 4.

Consequences of Ar ablation in the proximal epididymis. A, ProxE-ARKO epididymis on the right shows accumulation of inspissated spermatozoa in the hypoplastic proximal part (arrow) and a spermatic granuloma in the caudal region (arrowhead); a wild-type control epididymis is shown on the left. Bar, 2 mm. B, Testicular outflow obstruction results in dilatation of seminiferous tubules and increased testis size (asterisk) in the ProxE-ARKO male (right); wild-type control is shown on the left. Bar, 3 mm. C–H, Hematoxylin and eosin. C, Control testis. Bar, 100 μm. D and E, Progressive dilated atrophy of germinal epithelium; lumen of seminiferous tubules is indicated with asterisks. Bar, 100 μm. F, Orchitis and interstitial fibrosis (crosses) in ProxE-ARKO testis; asterisks indicate seminiferous tubules, some containing intraepithelial cysts. Bar, 200 μm. G, Degenerative changes in cauda epididymidis. Bar 150 μm. H, Focus of granulomatous inflammation in distal caput epididymidis; arrow indicates spermatozoa in the center of spermatic granuloma. Bar, 150 μm.

Effects of AR knockout on epididymal gene expression

To address the molecular consequences of Ar ablation, the transcription of six genes with predominant expression in the principal cells of the IS was measured by quantitative RT-PCR. Specific mRNA levels in the ProxE-ARKO mice were compared, besides WT controls, with those in WT mice subjected to EDL or OE (Fig. 5). EDL and OE were carried out at 20–25 dpp (shortly after the onset of Rnase10 expression and differentiation of IS), whereas gene expression was measured at 35–40 dpp, around the time of onset of spermiation and before the development of ductal occlusion in the ProxE-ARKO males. In this way we were able to discriminate the dependence of expression of the genes on endocrine androgen effect (reduced in PoxE-ARKO), lumicrine factors (reduced in EDL), or either one of the former (reduced in OE).

Fig. 5.

A, Expression of Rnase10, Etv4, Srd5a1, Ros1, Araf, and Bcl2l15 in the IS of intact WT, ProxE-ARKO (ARKO), and EDL and OE wild-type mice. Each bar is the mean ± sem of measurements from three animals. Statistical significances: ***, P < 0.001; *, P < 0.05. NS, Not significant. The micrographs at the bottom (B) show representative views of proximal epididymis of the four experimental groups. Bar, 30 μm.

Rnase10 is a ribonuclease-like protein with unknown function, being expressed mainly in the IS (6), and Etv4 is a member of the polyoma virus enhancer activator 3 family, shown to be expressed in the rat epididymis with highest level in IS (9). Srd5a1 encodes the enzyme steroid 5α-reductase type I that converts testosterone to 5α-dihydrotestosterone (DHT), and it is also predominantly expressed in the IS (10). Ros1, a protooncogene tyrosine kinase receptor, is expressed in the proximal epididymis, and its knockout mice present with agenesis of the IS (11). Another protooncogene, Araf, encodes a serine/threonine kinase and is expressed mainly in the principal epithelial cells of the murine caput epididymidis (12). Bcl2l15, also termed Bfk, is a novel member of the Bcl2 gene family, and its expression has been found to be confined to principal cells of IS (13).

Rnase10 and Etv4 expression were similarly, and nearly totally, lost in all experimental groups, indicating that the proper expression of this gene requires both endocrine androgen action and lumicrine factors. Srd5a1 expression responded in similar fashion but displayed higher residual activity that apparently represents constitutive expression independent of androgen or lumicrine regulation. Ros1 expression persisted in ProxE-ARKO mice but was lost in the other two groups, indicating that it is mainly regulated by lumicrine factors, however, with a considerable constitutive component. The expression patterns of Araf and Bcl2l15 were similar, with reduced levels in the AR knockout and OE groups but with persistent expression after EDL, indicating that the expression was androgen but not lumicrine factor dependent.

Discussion

The development of obstruction in the proximal epididymis as a consequence of targeted Ar deletion provides strong evidence for a direct role of androgens in the regulation of this epididymal zone. Direct evidence for such androgen function has hitherto remained elusive. Besides the epididymal obstruction and decreased height of the proximal epididymal epithelium, alterations in IS gene expression were detected that were specific for the Ar ablation.

Androgen dependence of the epididymis, in terms of both its development and maintenance, has long been known. However, earlier studies on EDL and OE, with and without testosterone supplementation, have emphasized the importance of exocrine testicular luminal factors in regulation of the proximal portion of epididymis, and particularly of the IS. In rats and mice, interruption of the flow of testicular fluids by EDL results in striking involution of the IS, with less dramatic, yet clear regressive changes more distally (14–16). Even more extreme involution of the epididymis is observed after bilateral OE, and administration of testosterone, even at supraphysiological doses, fails to prevent regression of the IS (2), thus implying a relative androgen independence of this epididymal zone. Findings of the present study, however, suggest the androgen dependence of IS. Even though the epithelium of the proximal zones of the ProxE-ARKO mice is exposed to testicular fluid before the development of obstruction, it closely resembles in histological appearance the postcastration epithelium (Fig. 5B). This similarity does not rule out the role of testicular factors other than testosterone in the regulation of the proximal epididymis; rather it suggests a permissive role for androgens.

Experiments in rats have indicated that the main mode of action of testosterone in caput epididymis is through its conversion into DHT via the amplification pathway operated by type I steroid 5α-reductase (10, 17). Although the microsomal fraction of this enzyme found throughout the epididymis appears to be regulated by circulating testosterone (18), 5α-reductase activity associated with the nuclear fraction is markedly and selectively down-regulated in IS after unilateral OE or EDL and cannot be maintained by exogenous testosterone (19, 20). A hypothetical model was therefore proposed that the proximal epididymis is regulated by DHT whose local formation from testosterone, i.e. by steroid 5α-reductase, is in turn controlled by lumicrine factors (21). Whereas this model may apply to the rat, the situation in the mouse appears to be somewhat different. Although the administration of DHT after orchidectomy can restore the morphology of murine IS to some extent (22), the genital structures and fertility of males with 5α-reductase type I knockout (23), and combined type I and II knockout (24), are largely uncompromised, with the exception of smaller prostates and seminal vesicles. The urogenital tissues (prostate, seminal vesicles, coagulating glands) of the double-knockout mice were found to accumulate high levels of testosterone, indicating that conversion of testosterone to DHT largely functions as signal amplification mechanism and is redundant in the presence of sufficient testosterone levels (24). The phenotypic mismatch between steroid 5α-reductase knockout males and those lacking the Ar (present study) therefore supports the contention that direct androgen action, not through the local production of DHT under lumicrine control in the IS, is sufficient to maintain normal function of the proximal zones of the epididymis.

Although the disruption of Ar in the proximal epididymis resulted in the current study in severe hypoplasia of the epithelium and obstruction, it is curious that the latter does not occur in the Ros1 knockout males with complete IS agenesis (11). Instead, these males are infertile because of impaired sperm volume regulation during their epididymal transit, resulting in swollen spermatozoa displaying angulated tails and inability to pass through the uterotubal junction (25, 26). The reason for this phenotype mismatch may be simply mechanical. In the Ros1 knockout, the length of the tubule lined with underdeveloped epithelium is considerably shorter than that in the ProxE-ARKO mice, in which gene inactivation extends down to zone 3 and occurs at a later postnatal stage in a partly differentiated and therefore longer tubule. Although clogging of spermatozoa in this dead space may also be promoted by the altered luminal fluid affecting the physicochemical properties of the sperm cell surface, epididymal obstruction also ensues in a proximal epididymis with less complete, mosaic inactivation of the Ar in principal cells, as demonstrated in a study using a FoxG1-Cre mouse to inactivate Ar in the epididymis (27).

The differential response patterns of the epididymal genes to the experimental manipulations allowed us to conclude that the proximal epididymis expresses genes that are predominately under direct androgen regulation, requiring functional Ar expression; such genes were Araf and Bcl2l15. There were also genes regulated by both androgen and lumicrine factors (Rnase10, Etv4, and Srd5a1). In this group, the lumicrine action of androgen present in testicular efferent fluid at concentrations up to 60-fold of those in blood plasma (26) remains a possibility. On the other hand, existence of hitherto unidentified lumicrine regulators distinct from androgen is indicated by the profile of Ros1 expression.

The Cre-loxP technology used in this study allows spatial control over inactivation of the Ar gene, its timing being determined by the developmental onset of Rnase10 expression, i.e. from d 17 postpartum by RT-PCR (6). In our immunohistochemical characterization of the prepubertal epididymal epithelium in ProxE-ARKO males, we found that the development of the proximal segment of the epididymis was aborted once the epithelial cells had reached the differentiation stage at which they began to express Rnase10. Therefore, the definitive IS never formed in ProxE-ARKO males, thus indicating an essential role of androgen in proper development of the proximal epididymis. In this context, a possibility exists that the low levels of Etv4, Araf, Srd5a1, and Bcl2l15 mRNA found in the ARKO group at 35–40 dpp may be due to the fact that Ar inactivation occurred in principal cells before the developmental onset of transcription of these genes. Gene expression analysis of prepubertal ProxE-ARKO epididymides is therefore warranted to test this possibility. On the other hand, cessation of Rnase10 expression is a clear indication of direct androgen dependence of gene expression in principal cells, whose continued presence in ProxE-ARKO epididymis is signified by unaltered expression of principal cell-specific Ros1.

In conclusion, in contrast to previous studies that have emphasized the importance of lumicrine factors in the regulation of the proximal epididymis, our current studies provide convincing evidence for a direct role of androgen action. The phenotype of Ar knockout in proximal epididymis, i.e. obstructive azoospermia, is intriguing, and provides a hypothesis for clinical studies on disturbed androgen action in the epididymis as the pathogenetic mechanism of this form of male infertility.