Spermatogenesis arrest caused by conditional deletion of Hsp90α in adult mice

Androgens are essential steroid hormones in initiation and maintenance of spermatogenesis as well as in the development of secondary male sex maturation. Nevertheless, it is still controversial whether the androgen receptor (AR) of germ cells is indispensable for spermatogenesis (Lyon et al., 1975; Johnston et al., 2001). Studies using germ cell-specific AR deficient mice showed almost normal spermatogenesis as well as fertility (Tsai et al., 2006). The AR gene in these mice is completely deleted in all germ cells beyond the pachytene stage; however, it remains unknown whether it is deleted in spermatogonia (Tsai et al., 2006; Wang et al., 2009).

Heat shock protein (Hsp) 90 is one of the most highly expressed cytosolic molecular chaperones, comprising 1% of the total cellular protein even in non-stressed conditions. It interacts with several hundred client proteins, simultaneously recruiting its own battery of co-chaperones, to facilitate their correct folding, transport, and degradation, thus maintaining normal cellular functions in ATP-dependent manner (Taipale et al., 2010; Echeverria and Picard, 2010). There are two distinct isoforms of Hsp90, Hsp90α and Hsp90β, whose homology is nearly 86% at the amino acid level. Both isoforms are thought to function in various tissues in a mostly redundant manner.

We previously generated Hsp90α-null mice to investigate the role of Hsp90α in antigen-cross presentation. During these studies we became aware that Hsp90α male mice were infertile and found that this resulted from the lack of sperm formation (Ichiyanagi et al., 2010; Imai et al., 2011). But before that, Grad et al. reported that Hsp90α gene trap mutant mice, in which Hsp90α is deleted throughout the entire life of the mouse, developed testicular atrophy caused by apoptosis of spermatocytes (Grad et al., 2010). In this report, there was a further intriguing observation that in one of the two distinct mutant lines the expression of Hsp90α was greatly reduced initially but resumed around day 45. Nonetheless, the mice still remained infertile (Grad et al., 2010). These results strongly imply a critical role of Hsp90α in establishing the first wave of spermatogenesis before puberty, and this in turn led us to the question of whether Hsp90α is required for the maintenance of normal spermatogenesis after testicular maturation is completed.

Based on the outcome of the study of Grad et al., we designed our mutant mice to be tamoxifen-inducible via the loxP-cre system, thus allowing Hsp90α-deletion in adult mice. We observed severe apoptosis occurring in germ cells beyond the pachytene stage, results that led us to conclude that Hsp90α is indispensable for the maintenance of germ cell development even in the mature adult testis. Moreover, we found that expression of AR is absent or reduced in spermatogonia of Hsp90α KO testes, whereas it is strongly expressed in WT testes. Our results indicate that the AR of spermatogonia, which is specifically chaperoned by Hsp90α, is critical for the initiation and maintenance of spermatogenesis.

The EIIa-cre mouse carries a cre transgene under the control of the adenovirus EIIa promoter that targets expression of Cre recombinase to the early mouse embryo (JAX Mice Database – 003724 B6.FVB-Tg(EIIa-cre)C5379Lmgd/J). Breeding the mice to loxP-flanked Hsp90α mice (Hsp90aa 1^neo) (Fig. 1A) produced mice of three genotypes – Hsp90aa Exon 9,10-floxed mice (Hsp90aa 1^flox), mice with a deletion of neo and Exons 9 and10 on chromosome (Hsp90aa 1⁻) (Fig. 1A) and neo-floxed mice (not shown). Complete Hsp90αKO (Hsp90aa 1^−/−) mice were obtained by interbreeding the Hsp90aa 1^−/+ mice (Imai et al., 2011). We obtained a total of 375 F1 mice and the genotypic ratio of the offspring was Hsp90aa 1^+/+:Hsp90aa 1^−/+: Hsp90aa 1^−/− = 113:194:68. The number of homozygous Hsp90aa 1^−/− offspring was significantly lower than the expected Mendelian ratios, suggesting that Hsp90aa 1^−/− embryos survive by chance, not in every case, possibly through compensatory contribution of the other isotype of Hsp90, Hsp90β.

By breeding the Hsp90α^flox (Hsp90aa 1^flox) to B6CreER^T2 mice, we obtained inducible mutant mice, thus, Hsp90α^flox/flox/CreER^T2 and Hsp90α^flox/+/CreER^T2. CreER^T2 mice were generated by knock-in of a fusion gene composed of Cre recombinase and the ER^T2 domain of human estrogen receptor gene with three mutations (G400V/M542A/L544A) into the Rosa 26 region (Seibler et al., 2003). Binding of 4-hydroxy tamoxifen to CreER^T2 on cell surface causes relocation of the fusion protein to the nucleus, where it deletes floxed genes. Oral administration of tamoxifen (5 mg) to the mice resulted in conditional deletion of exon 9 and 10 of Hsp90aa 1^flox, thus producing Hsp90α^−/−/CreER^T2 and Hsp90α^−/+/CreER^T2 mice (Fig. 1B).

Hsp90αKO mice grew normally, nearly comparably to wild type (WT) controls (Fig. 1C). However, we found that male 8-week-old Hsp90αKO mice were infertile and had small scrotums compared to the WT mice. Indeed, their testes were significantly smaller, close to one third the normal size, although the seminal vesicles were not significantly altered (Fig. 1C). Microscopic sperm counts revealed that no spermatozoa could be recovered from the caudal epididymis of 8 to 10 week Hsp90αKO mice. Notably, the size of the prepubescent testes (day 17 after birth) was nearly comparable between Hsp90αKO and WT mice (data not shown).

To determine which cell type in the testes is affected in Hsp90αKO mice, histological analysis was performed. By H&E staining, there was only a marginal reduction of germ cell development in Hsp90αKO day 17 testes; however, developmental arrest became apparent with progressive age (8 weeks) (Fig. 1D, left panel). TUNEL analysis revealed apoptosis occurring in germ cells after the pachytene stage in day 17 testes and there was increased severity at 8 weeks (Fig. 1D, right panel). The results clearly indicate that reduced numbers of developmental germ cells is caused by apoptosis induction in Hsp90αKO testes with increasing age. The summary of statistics for numbers of TUNEL positive cells is shown in Table 1. Overall, our results obtained from the analysis of Hsp90αKO testes are consistent with those of Grad et al., who used Hsp90aa1 mutant mice produced by the gene trap method (Grad et al., 2010).

To determine the effect of conditional deletion of the Hsp90aa1 gene in adult mice, Hsp90α floxed mice were treated with oral tamoxifen. Genomic DNA was extracted from freshly isolated tissues (Fig. 2A) and the region between Exon 8 and Exon 11 of the Hsp90aa1gene was amplified by PCR, as indicated in Fig. 1A. Administration of tamoxifen resulted in deletion of the floxed Exons 9 and 10, which was demonstrated by identification of a 1.2 kb band (Fig. 2B). The Hsp90α⁺ and Hsp90α^flox alleles could be identified as 2.0 kb and 2.1 kb bands, respectively (Fig. 2B). Although Exon 9 and Exon 10 were deleted, levels of Hsp90α were not reduced at day 6 except in liver (Fig. 2C, day 6). Accordingly, both Hsp90α^flox/flox/CreER^T2 and Hsp90α^flox/+/CreER^T2 mice had normal sized testes (Fig. 2D, day 6).

We next recovered tissues at day 30 after administration of tamoxifen (Fig. 2A). Unlike the case on day 6, Hsp90α was markedly downregulated in all tissues tested (Fig. 2C, day 30). On day 30, testes of Hsp90α^flox/flox/CreER^T2 but not Hsp90α^flox/+/CreER^T2 mice showed atrophy, diminishing to a size similar to that of Hsp90α KO testes (Fig. 2D, day 30). Indeed, the weight of the testes in Hsp90α^flox/flox/CreER^T2 mice were around one-third that of Hsp90α^flox/+/CreER^T2 and WT mice (Fig. 2E).

Histological analysis revealed pronounced, severe apoptosis and deletion of spermatocytes in testes with conditionally deleted Hsp90α, similar to what was observed in the conventional Hsp90α KO mice. No germ cell development, including the pachytene stage, remained (Fig. 3, upper panel). A TUNEL assay demonstrated that germ cells after pachytene stage underwent apoptosis, leading to a complete arrest of spermatogenesis and cell death of meiotic spermatocytes (Fig. 3, lower left panel). By contrast, Leydig and Sertoli cells seem to be not much impaired, rather closely resembling normal cells. Hsp90β does not compensate for Hsp90α during germ cell development in the testes of the conditional knockout mice. Relevant to this point, Hsp90β appears to be mainly expressed in Sertoli cells, whereas Hsp90α is expressed specifically in primordial and mature germ cells (Lee, 1990; Vanmuylder et al., 2002). Therefore, the limited expression of Hsp90β in germ cells might explain why Hsp90β could not rescue germ cell development in Hsp90α KO testes.

Androgen-AR interaction plays an important role in spermatogenesis since AR functions as testosterone-dependent transcription factor, initiating expression of an array of androgen-responsive genes (Chang et al., 1988a; Chang et al., 1988b; Lubahn et al., 1988; Heinlein and Chang, 2002). Based on the fundamental importance of AR, we wondered whether AR expression was normal in individual cells (germ cells) of the testis in the absence of Hsp90α. We used specific antibodies and performed immunohistochemical analysis for estrogen receptor (ER)α, ERβ and AR of both WT and Hsp90α KO testes (both 8-week-old). On germ cells, we found dominant expression of ERα and AR in spermatogonia and to a lesser extent in spermatocytes of WT testis, whereas the pattern was disrupted and/or even reversed in Hsp90α KO testis (Fig. 4). In order to discriminate between spermatogonia and spermatocytes, we further performed immunohistochemistry in mirror sections of 8 week and day 17 testes with antibodies specific to synaptonemal complex protein 3 (SCP3) that is expressed in spermatocyte but not in spermatogonia. Consistent with the results in Fig. 4, AR was expressed in germ cells containing both spermatocytes (SCP3 positive) and spermatogonia (SCP3 negative) in WT testes, whereas it was positive in spermatocytes but not in spermatogonia in Hsp90α KO testis (Fig. 5A). In prepuberal testes, on the other hand, AR was positive in only spermatocytes in both WT and Hsp90α KO testes (Fig. 5B). Thus, AR is weakly or not expressed in spermatogonia even in WT testis before puberty, suggesting that AR in spermatoginia is critical in initiation of germ cell development. In other words, most seminiferous tubules showed strong expression of AR (and/or ERβ) in spermatocytes, but little or none in spermatogonia in Hsp90α KO mice, implying that Hsp90α KO testis lacks the androgen-AR signaling in spermatogonia required for subsequent germ cell development.

AR is chaperoned by Hsp90 before its association with testosterone and Hsp90 inhibitors caused AR-Hsp90 complex to dissociate, which might cause degradation of AR by the proteasome (Whitesell and Cook, 1996; Zhang et al., 2006). Although the precise reason why AR is missing in spermatogonia of the Hsp90α KO testis is unknown, it is plausible that AR undergoes degradation following deletion of Hsp90α gene since it is usually chaperoned by Hsp90α and not by Hsp90β. Although absent in spermatogonia, AR seemed to reappear in germ cells of Hsp90α KO testis. This might be because the proteasome activity in germ cells (beyond the pachytene stage) already undergoing apoptosis is downregulated, which eventually causes accumulation of AR within the cells. The balance between Hsp90α-mediated chaperone activity and proteasomal activity determines the expression level of AR in germ cells. Hsp90α deficiency might influence this balance, giving rise to the observed aberrant expression pattern of AR in germ cells including spermatogonia. It is also possible that Hsp90α simply regulates (facilitates) transcriptional activation of AR gene in spermatogonia.

Germ cell-specific AR KO mice did not show testicular atrophy, nor did they show spermatogenesis arrest, indicating that the androgen-AR interaction in germ cells is dispensable for their development (Tsai et al., 2006). These observations conflict with our hypothesis that Hsp90α-dependent AR expression in spermatogonia is essential in subsequent germ cell development. However, the germ cell-specific AR KO mice were generated by mating the floxed AR mice with synaptonemal complex protein 1 promoter (Sycp1-cre) transgenic mice (Tsai et al., 2006). Sycp1 is expressed at an early stage of male meiosis, leptene to zygotene, therefore, the floxed AR allele might not be deleted in spermatogonia of these mice.

Sertoli cell- and Leydig cell-specific AR KO mice have a testicular phenotype very similar to the Hsp90α KO, i.e., no sperm formation, disrupted germ cell development and testicular atrophy (Chang et al., 2004; De Gendt et al., 2004; Holdcraft and Braun, 2004; Xu et al., 2007). However, in our Hsp90α KO mice, both Sertoli cells and Leydig cells appeared normal. It is possible that the AR-mediated transcriptional cascade in Sertoli- and Leydig-cells of our mice is not impaired due to the compensatory effect of Hsp90β, since this isotype rather than Hsp90α is mainly expressed in these cell types (Lee, 1990; Vanmuylder et al., 2002).

In summary, we have established Hsp90α conditional deletion mutant mice and found that spermatogenesis in adult testes was completely abrogated through apoptosis caused by inducible ablation of the Hsp90α gene. In addition, AR expression in spermatogonia was downregulated in Hsp90αKO testis. Although our results raised the possibility that AR-dependent signaling is essential in the subsequent germ cell development, further studies will be required to test this hypothesis.

Conditional HSP90α-null mice were generated by deleting exons 9 and 10, which encode the C-terminal region of the protein, in a cre-recombinase-dependent manner (Imai et al., 2011). Tamoxifen-dependent, inducible deletion of the floxed exons was performed according to a previous report (Hikida et al., 2009). Briefly, mice were orally administrated 100 µl 4-hydroxy tamoxifen (Sigma T5648) dissolved in corn oil (Sigma C8267) at a dose of 50 mg/kg, twice in a two-day interval. Mice were analyzed 4 d and 28 d after the last injection. Deletion efficiency of exons 9 and 10 was assessed by genomic PCR. The samples were subjected to Western blotting analysis using antibodies specific for Hsp90α (mouse mAb EMD-17D7, Calbiochem.), Hsp90β (rabbit polyclonal Abs, Lab Vision Corporation, Fremont, CA) and actin (Sigma).

The primers used for genotyping of mice after oral administration of tamoxifen were forward (within exon 8): ttctctgaaggactactgtaccagaatgaa, and reverse (downstream of exon 11): gtgtgggggtgcgcaaagaaaagacacatt. The primers for detection of the cre-recombinase gene were forward: caccctgttacgtatagcc, and reverse: ctctgaccagagtcatcct.

Enzyme immunohistochemistry was performed on paraffin sections (5–6 µm) of the mouse testis, as described previously (Koji et al., 2001; Song et al., 2011). The antibody dilutions were as follows: rabbit anti-AR antiserum (Millipore AB561) was at 1:800, mouse monoclonal anti-ERα (BioGenex ER88) was at 1:100, mouse monoclonal anti-ERβ (BioGenex ERβ88) was at 1:200, and rabbit anti-SCP3 antibody (Novus Biologicals NB300-231) was at 1:750 (1.33 µg/ml). After the reaction with horseradish peroxidase (HRP)-conjugated second antibody, the sites of HRP deposition were visualized with 3,3′-diaminobenzidine-4HCl (DAB) and H₂O₂ in the presence or absence of nickel and cobalt ions (Shukuwa et al., 2006). In every experimental run, negative control samples were prepared by reacting the sections with normal rabbit serum (for AR) or normal mouse IgG (ERα and ERβ) instead of the specific first antibody. To identify apoptotic cells, Terminal deoxynucleotidyl transferase (TdT)-ated dUTP nick end-labeling (TUNEL) staining was performed according to the method (Gavrieli et al., 1992) with a slight modification (Koji et al., 2001).