Sox3 Functions in a Cell-Autonomous Manner to Regulate Spermatogonial Differentiation in Mice

Sox3 functions within germ cells to promote spermatogonial differentiation, but is not required for spermatogenesis in postpubertal animals.

Abstract

The X-linked Sox3 gene encodes a member of the Sry high-mobility group box proteins, which play a role in many developmental processes including neurogenesis and testis development. This study further examined the role of Sox3 in spermatogenesis. Males without Sox3 expression exhibited a similar number of germ cell nuclear antigen-positive germ cells at 1, 5, and 10 d postpartum (dpp) compared to their wild-type littermates, but there was significant germ cell depletion by 20 dpp. However, spermatogenesis later resumed and postmeiotic germ cells were observed by 56 dpp. The VasaCre transgene was used to generate a germ cell-specific deletion of Sox3. The phenotype of the germ cell-specific Sox3 knockout was similar to the ubiquitous knockout, indicating an intrinsic role for Sox3 in germ cells. The residual germ cells in 20 dpp Sox3^−/Y males were spermatogonia as indicated by their expression of neurogenin3 but not synaptonemal complex protein 3, which is expressed within cells undergoing meiosis. RNA expression analyses corroborated the histological analyses and revealed a gradual transition from relatively increased expression of spermatogonia genes at 20 dpp to near normal expression of genes characteristic of undifferentiated and meiotic germ cells by 84 dpp. Fluorescent-activated cell sorting of undifferentiated (ret tyrosine kinase receptor positive) and differentiated (kit receptor tyrosine kinase-positive) spermatogonia revealed depletion of differentiated spermatogonia in Sox3^−/Y tubules. These results indicate that Sox3 functions in an intrinsic manner to promote differentiation of spermatogonia in prepubertal mice but it is not required for ongoing spermatogenesis in adults. The Sox3^−/Y males provide a unique model for studying the mechanism of germ cell differentiation in prepubertal testes.

Sox3 is a member of the B1 group of protein activators that contain more than 90% homology to the high-mobility group (HMG) box of Sry and is located on the X chromosome (1, 2). Proteins in the Sox family regulate transcription and facilitate several developmental processes including neurogenesis, chondrogenesis, and testis development (3–7). Sox3 expression is localized within the proliferating neuronal precursors throughout the neural ectoderm and is maintained in undifferentiated proneural cells in adult mice (8). Some Sox3^−/Y mice show craniofacial malformations resulting from abnormal development of the pharyngeal pouch and/or pituitary defects resulting from altered formation of the Rathke's pouch (9, 10). Sox3 is also expressed within the urogenital ridge of the developing mouse embryo and within germ cells in other animals, including Xenopus and some species of fish (9, 11, 12). Sox3^−/Y mice demonstrate a spermatogenic block by 2 wk of age (13).

The proliferation of spermatogonial cells and the initiation of the first wave of spermatogenesis in prepubertal mouse testes provide the foundation for subsequent waves that occur in an organized, continuous manner in the adult. This process ensures successful sperm production throughout adult life (14, 15). Several genes have been shown to contribute to self-renewal and maintenance of the undifferentiated spermatogonia population, including zinc finger and broad-complex, tramtrack, and bric à brac domain containing 16 (Zbtb16, previously known as Plzf) and Ret (16–19). Neurogenin 3 (Neurog3) is expressed within a distinct spermatogonial population that is undergoing differentiation (20). This differentiation process also requires two spermatogenesis and oogenesis-specific basic helix-loop-helix factors, Sohlh1 and Sohlh2 (21–23). Despite these advances, the molecular pathways and major proteins involved in spermatogonia differentiation, from chains of type A spermatogonia to primary spermatocytes, are not fully understood.

The Sox3^−/Y mouse is a unique model for investigating prepubertal spermatogenesis and the mechanisms involved in spermatogonia differentiation and further maturation into primary spermatocytes. The specific objectives of this study were: 1) to determine whether the spermatogenic defect found in Sox3^−/Y mice is a result of the lack of Sox3 expression in germ cells or is a result of absent Sox3 expression in other cells that support spermatogenesis (e.g. cells within the developing pituitary); 2) to determine whether the block is progressive or reversible with aging; and 3) to determine whether the block in spermatogenesis is a result of disrupted spermatogonia proliferation or differentiation.

Materials and Methods

Animals

Sox3^−/Y mice were bred on a C57BL/6 background as described previously (13). The germ cell-specific Sox3-null animals were generated by crossing transgenic males with Cre recombinase under the control of the germ cell-specific Vasa gene to females with a floxed Sox3 allele (9, 24). Because the resulting animals have a mixed FVB/C57BL6 background, Sox3^+/Y and Sox3^−/Y males were generated on the same mixed background for appropriate comparisons between the genotypes. All animal use was in accordance with procedures approved by the Northwestern University Animal Use and Care Committee.

Testis histology and immunohistochemistry

The weight of both testes was measured immediately after removal and was calculated as a ratio to the body weight. A one-way ANOVA was used to determine whether any of the groups differed from wild type. Testes from male littermates were fixed in 10% neutral buffered formalin, processed, and embedded in paraffin. Testis cross-sections were cut at 5–7 μm. Sections were stained with hematoxylin and eosin or with periodic acid Schiff's reagent and hematoxylin (PAS-H). Five or more nonadjacent sections from at least five different animals per genotype at ages 10, 20, 56, and 84 d postpartum (dpp) were examined. Morphological similarities or differences were noted, and representative sections were photographed (Figs. 1 and 2). Immunostaining of the desired proteins was performed using a standard protocol and anti-NEUROG3 (Santa Cruz Biotechnology, Inc., Santa Cruz, CA; sc-25655, 1:100) or anti- synaptonemal complex protein 3 (SYCP3) (Novus Biologicals, Littleton, CO; NB300-230SS, 1:250) and antirabbit AlexaFluor488 (Molecular Probes, Inc., Eugene, OR; A11034, 1:500) (13). Micrographs were taken using a Zeiss Axioskop microscope (Carl Zeiss, Thornwood, NY) and an Optronics MicroFire digital camera (Optronics Engineering, Goleta, CA).

Fig. 1.

Histological analysis of images of Sox3^+/Y and Sox3^−/Y testes of prepubertal and adult mice. Sections from 10 dpp and 20 dpp mice were stained with hematoxylin and eosin. Sections from 56 dpp and 84 dpp mice were stained with PAS-H. Bar, 50 μm.

Fig. 2.

Germ cell-specific ablation of Sox3 on the FVB/C57BL6 mouse strain. A, Comparison of testis to body weight of Sox3^+/Y, Sox3^−/Y, and VasaCre⁺;Sox3^fl/Y FVB/C57BL6 males. Each data point represents the mean of three or more animals. Error bars represent the sem. One-way ANOVAs determined the three genotypes within each age group to be significantly different. *, P ≤ 9.96 × 10⁻⁴. B, Histological comparison of Sox3^+/Y, Sox3^−/Y, and VasaCre⁺;Sox3^fl/Y testis cross-sections stained with PAS-H. Tubules that contain elongated spermatids are shown (*). Bar, 50 μm.

Analysis of germ cell nuclear antigen (GCNA)-positive germ cells

Testes from three to four C57BL/6 animals per genotype at 1, 5, 10, and 20 dpp were fixed and sectioned as above. A GCNA antibody was used to stain germ cells as described previously (13). GCNA-positive germ cells, within 100 or more tubules, were counted. The average number of GCNA-positive germ cells per tubule was determined per animal in each genotype. Student's two-tailed t test was used to assess statistical significance, defined as P ≤ 0.05. Representative photomicrographs from Sox3^+/Y and Sox3^−/Y cross-sections were taken at each age.

Analysis of differentiated germ cells

Sections from three to five C57BL/6 Sox3^+/Y and Sox3^−/Y males age 56 and 84 dpp were stained with PAS-H. At least 50 tubules from two nonadjacent sections per animal were analyzed. In all of the Sox3^+/Y and most of the Sox3^−/Y tubules, the most differentiated germ cells present were determined based on the acrosomal staining and epithelial stage of round spermatids (15). Some Sox3^−/Y tubules did not contain spermatids and therefore, the most differentiated germ cells were determined to be spermatocytes (sc), spermatogonia (sg), or lacked germ cells (only Sertoli cells present, 0). The most differentiated germ cells present ranged from no germ cells to step 16 elongated spermatids. The percent of tubules containing a particular type of most differentiated germ cells was determined for each animal, and the results were presented as the mean for each germ cell type for each genotype. Representative photomicrographs from Sox3^+/Y and Sox3^−/Y cross-sections at 56 and 84 dpp were taken. An ANOVA was used to determine whether the groups differed over all of the categories and to compare the genotypes among abnormal tubules categories, 0 through steps 6–7, and tubules containing normal tubule differentiation, steps 9–16.

RT-PCR analysis

RNA from one testis per animal was isolated with Trizol (Invitrogen, San Diego, CA; 15596) using the manufacturer's protocol. Each sample was treated with deoxyribonuclease I (Promega Corp., Madison, WI; M6101) and reextracted with phenol-chloroform-isoamyl alcohol (Ambion, Austin, TX; 9720). A 2-μg aliquot was transcribed using SuperScript II Reverse Transcriptase (Invitrogen; 18064-014) and Oligo dTs (Promega; C110A). PCRs included 1 μm total primer pair, 100 ng of cDNA and IQ SYBR Green Supermix (Bio-Rad Laboratories, Inc., Hercules, CA; 170-8880). The samples were amplified using a Bio-Rad iCycler iQ. The C_T of the gene of interest was subtracted from the C_T of Rpl19, and relative expression was calculated with the mean Sox3^+/Y expression set equal to 1. Primer sequences are listed as followed in the 5′ to 3′ direction: Rpl19 forward, GCCAATGCCAACTCCCGTCAG; Rpl19 reverse, CGCTTTCGTGCTTCCTTGGTCTTA; Ret forward, CTGAGTTCAACCTTCTGAAACAAGT, Ret reverse, TATAGATGTGATCGAAAAGGGACTC; Zbtb16 forward, GAGACACACAGACAGACCCATACT; Zbtb16 reverse, CACACATAACACAGGTAGAGGTACG; Stra8 forward, AACGGTATCTCAACTTTTACAAGCA; Stra8 reverse, ATTTCTCCTCTGGATTTTCTGAGTT, Dmc1 forward, CGTGCCTATACTAGTGAGCATCAG; Dmc1 reverse, CCTGAAAGGTCATAGTTGCTCCTG; Sycp3 forward, GACTGTATTTACTCCTGCCCAAGG; Sycp3 reverse, GCTTCCCAGATTTCCCAGAATG; Acr forward, TGGACCCTGCTGTCTACCTCATTT; Acr reverse, TAGTCGATGTGACACGCCCATTGT. Each primer was used to amplify adult wild-type testis cDNA. These bands were sequenced to verify amplification of the desired PCR product. RNA from at least five animals was compared for each analysis. Student's two-tailed t test was used to assess statistical significance, defined as P ≤ 0.05.

Fluorescent-activated cell sorting (FACS) analysis

Seminferous tubules from a single testis were washed in 1 mg/ml collagenase at 37 C for 15 min to remove interstitial cells. A single-cell suspension was made by digesting the tubules as described by Ogawa et al. (25). Cells were filtered using a MACS preseparation filter (Miltenyi Biotech, Auburn, CA; 130-041-407). The suspension was incubated with two fluorescein-conjugated antibodies, anti-CD117-APC (eBioscience, San Diego, CA; 17-1171) and anti-Ret-PE (Santa Cruz Biotechnology, Inc.; sc-1290) in MACS Separation Buffer (Miltenyi Biotech; 130-091-221) at 4 C for 20 min. The samples were washed and resuspended in the separation buffer (500 μl total) containing 50 μl AccuCount fluorescent particles 10 μm (SpheroTec, GbmH, Martinsried, Germany; ACFP-100–3) and 5 μl of 100 μg/ml 4,6-diamidino-2-phenylindole (DAPI) solution. The samples were run through the CyAn flow cytometer (Dako Cytomation, Carpinteria, CA) for FACS analysis. The first 10,000 counts were analyzed for each sample. DAPI-positive cells were excluded from the analysis, and gates for ret tyrosine kinase receptor (RET)- and KIT-positive cells remained constant for all samples. The cell populations were determined according to the manufacturer's instructions (SpheroTec AccuCount).

Results

Histological analysis of Sox3^−/Y prepubertal and postpubertal testes

Proliferating spermatogonia are situated along the basement membrane, and primary spermatocytes are first visible within normal tubules from 10 dpp males (14). There were no apparent morphological differences between the Sox3^+/Y and Sox3^−/Y testis sections at this time point (Fig. 1). However, by 20 dpp, when postmeiotic germ cells are visible in Sox3^+/Y testes, there was a substantial decrease in germ cells within the Sox3^−/Y testes. Many Sox3^−/Y testis cross-sections contained only Sertoli cells. The Sox3^−/Y postpubertal testes (56 and 84 dpp) also displayed fewer germ cells, although many Sox3^−/Y tubules contained postmeiotic germ cells and all appropriate germ cell associations expected for their stage. This apparent recovery of spermatogenesis in the Sox3^−/Y testes appeared more complete by 84 dpp than at 56 dpp.

Analysis of germ cell-specific Sox3 knockout mice

Because Sox3 is expressed in other tissues, it is possible that the spermatogenic defect in the Sox3^−/Y males was due to a secondary effect of Sox3 deficiency in the brain or other organs. A germ cell-specific Cre recombinase transgenic mouse (VasaCre) was used previously to delete a targeted gene within germ cells (26). VasaCre mice were crossed with Sox3-floxed mice to yield offspring with a germ cell-specific excision of Sox3 (VasaCre⁺;Sox3^fl/Y). Animals with Sox3 deleted only in the germ cell population developed normally. No craniofacial malformations were observed, unlike those occasionally observed in mice with a ubiquitous deletion of Sox3 (9, 10). Testis weight and histology of VasaCre⁺;Sox3^fl/Y males were compared with Sox3^+/Y and ubiquitous Sox3^−/Y testes of the same mouse strains (Fig. 2A). The weight of VasaCre⁺;Sox3^fl/Y testes, like the testes weights from Sox3^−/Y males, was significantly less than Sox3^+/Y testes at 20, 56, and 84 dpp. Additionally, gross histology of the testes was similar in VasaCre⁺;Sox3^fl/Y and Sox3^−/Y animals (Fig. 2B). Both genotypes displayed significant germ cell loss by 20 dpp with many tubules containing only Sertoli cells. Like the ubiquitous knockout, there were postmeiotic germ cells and appropriate germ cell associations within many tubules of the germ cell-specific knockout, indicative of ongoing spermatogenesis by 56 and 84 dpp (Fig. 2B). Examples of tubules that contain elongated spermatids are shown in Fig. 2B “*”.

Quantitation of germ cells in Sox3^+/Y and Sox3^−/Y testes at 1, 5, 10, and 20 dpp

Germ cells within testis cross-sections were counted to determine whether there was a difference between the number present in Sox3^+/Y and Sox3^−/Y testes at ages before the spermatogenic block is first apparent in the Sox3^−/Y animals, and once the Sox3^−/Y testes are phenotypically altered (Fig. 3). The GCNA antibody was used to stain the germ cells (brown cells), and hematoxylin was used as a nuclear counterstain (blue cells). Gonocytes cease to proliferate by 16.5 d postcoitum and resume proliferation at 1.5 dpp (27). Therefore, testes from 1, 5, and 10 dpp animals were investigated to determine whether the initial gonocyte populations were different at these early time points, before and after gonocyte proliferation (Fig. 3, A–F). The number of GCNA-positive germ cells within the tubules was similar at each of these time points (Fig. 3, C, F, and I). However, by 20 dpp, less than 12% of the germ cells present in the Sox3^+/Y were present within the Sox3^−/Y animals (Fig. 3L).

Fig. 3.

Quantification of germ cells per tubule in Sox3^+/Y and Sox3^−/Y testes of prepubertal mice. A, Representative histological cross-section of GCNA-positive cells (brown) of a Sox3^+/Y 1 dpp; D, 5 dpp; G, 10 dpp; and J, 20 dpp testes. B, Representative cross-sections of GCNA-positive cells of Sox3^−/Y 1 dpp; E, 5 dpp; H, 10 dpp; and K, 20 dpp testes. All sections were counterstained with hematoxylin (blue). Bar, 50 μm. C, Each bar represents the mean ± sem for the number of GCNA-positive germ cells per tubule for three animals at 1 dpp; F, 5 dpp; I, 10 dpp; and L, 20dpp. *, P ≤ 0.05 by Student's t test.

Categorization and quantitation of differentiation in postpubertal tubules

To better understand the extent of spermatogenesis recovery in postpubertal animals, individual tubules from 56 and 84 dpp were categorized based on the differentiated germ cells present (Fig. 4). Categories were assigned according to the spermatid step number (steps 1–16), spermatocyte (sc), or spermatogonia (sg) that represented the most differentiated germ cells present within a given tubule. Acrosomes were stained with periodic acid Schiff's reagent (PAS) to identify round and elongated spermatids (pink), and the cross-sections were counterstained with hematoxylin (blue, Fig. 4, A, B, D and E). The types of germ cells were identified based on their morphological characteristics and associations with other germ cells (15). Normal tubules contained elongated spermatids (steps 9–16).

Fig. 4.

Quantification of tubules containing specified types of differentiated germ cells. A, Representative histological cross-section of a Sox3^+/Y 56 dpp and B, Sox3^−/Y 56 dpp testes stained with PAS-H. Acrosomes are visible within round spermatids or along elongated spermatids (bright pink). C, Each bar represents the mean ± sem for the type of the most differentiated germ cells within greater than or equal to 100 tubule cross-sections for four to five animals at 56 dpp. Student's t tests were used to determine significant differences between genotypes for each category of germ cells. Significance is defined as P ≤ 0.05. The following category was significantly different between the two genotypes: for sc, P ≤ 0.005. D, Representative histological cross-section of a Sox3^+/Y 84 dpp and E, Sox3^−/Y 84 dpp testis stained with PAS-H. Acrosomes are visible within round spermatids or elongated spermatids (bright pink). F, Each bar represents the mean ± sem for the type of most differentiated germ cells within greater than or equal to 100 tubule cross-sections for three to four animals at 84 dpp. Student's t tests were used as described for 56 dpp. The following categories were significantly different: for sc, 6–7 and 14; P ≤ 0.05. Bar, 50 μm. 0, tubules without germ cells; sp, spermatogonia; sc, spermatocytes.

The Sox3^−/Y animals display tubules that represent each stage of the epithelial cycle. Almost all Sox3^−/Y tubules contain spermatogonia to the most differentiated germ cells that are appropriate for the stage of that tubule. Therefore, it is unlikely that there are additional spermatogenic blocks that occur once spermatogenesis begins to recover in the Sox3^−/Y adults. Approximately 75% of tubules within the 56 dpp Sox3^−/Y animals contain postmeiotic germ cells, including elongated spermatids (Fig. 4C, steps 1–16). This percentage increases to almost 80% by 84 dpp (Fig. 4F). These data provide further support for the recovery and continuation of spermatogenesis within the Sox3^−/Y animals.

Histological analysis of germ cells within 20 dpp Sox3^−/Y testes

To further understand the spermatogenic block and recovery in Sox3^−/Y animals, germ cells were analyzed at 20 dpp, when the spermatogenic block was most severe. Markers for spermatogonia (NEUROG3 and ZBTB16) and spermatocytes (SYCP3) were used to identify the type of germ cells present. As shown in Fig. 5, A and B, antibodies against NEUROG3 identified clusters of spermatogonia in both Sox3^−/Y and Sox3^+/Y tubules. Similar results were seen using a ZBTB16 antibody (data not shown). In contrast, perinuclear staining of SYCP3 was present in many germ cells within the Sox3^+/Y tubules but not in Sox3^−/Y tubules (Fig. 5, C and D). These findings suggested that, although Sox3^−/Y testes contained spermatogonia at 20 dpp, the progression into meiosis was hindered.

Fig. 5.

Immunohistological analysis of NEUROG3 and SYCP3 in Sox3^−/Y 20 dpp testis cross-sections. A, NEUROG3 immunostaining in Sox3^+/Y and B, Sox3^−/Y testis cross-sections (green). C, SYCP3 immunostaining Sox3^+/Y and D, Sox3^−/Y testis cross-sections (green). All sections were counterstained with DAPI nuclear stain (blue). Bar, 50 μm.

Spermatogonial and meiotic gene expression in prepubertal and postpubertal testes

Whole-testis mRNA from 10, 20, 56, and 84 dpp Sox3^+/Y and Sox3^−/Y males was used to further evaluate the role of Sox3 in the spermatogenic block and subsequent recovery. Genes expressed in undifferentiated spermatogonia, such as Ret and Zbtb16, were significantly increased in 20, 56, and 84 dpp Sox3^−/Y testes when compared with Sox3^+/Y testes (Fig. 6A). These results coincide with the finding of fewer differentiated germ cells and a greater overall percent of undifferentiated germ cells within Sox3^−/Y tubules at these ages. Stimulated by retinoic acid protein 8 (Stra8), an early marker for germ cell differentiation, was significantly reduced in 10 dpp Sox3^−/Y testes, when primary spermatocytes are normally first seen. Stra8 expression in Sox3^−/Y mice was similar to that in Sox3^+/Y mice at 56 and 84 dpp, when spermatogenesis begins to recover in Sox3^−/Y mice (Fig. 6B). The meiosis-specific dosage suppressor of mck1 homolog (Dmc1) is expressed in germ cells that are undergoing meiosis. Like Stra8 expression at 56 and 84 dpp, there was no difference in Dmc1 expression in Sox3^−/Y and Sox3^+/Y testes (Fig. 6B). Genes expressed in meiotic and postmeiotic germ cells, such as Sycp3 and Acrosin (Acr), were significantly decreased in animals at 56 dpp but were more similar to Sox3^+/Y levels by 84 dpp (Fig. 6C). These results are consistent with the fewer number of spermatids observed in Sox3^−/Y tubules at 56 dpp and continued recovery at 84 dpp. Acr was expressed in very low levels in both Sox3^+/Y and Sox3^−/Y testes at 20 dpp, when the first postmeiotic germ cells that began to express this gene were first detectable in wild types. The levels appeared similar between the two genotypes because this method of analysis is not able to distinguish differences between minimal mRNA values. Overall the results support the phenotype that was observed in the micrograph analyses.

Fig. 6.

RT-PCR analysis of Sox3^+/Y and Sox3^−/Y gene expression in testes. Each bar represents the mean ± sem relative expression of four or more animals at 10, 20, 56, and 84 dpp. *, P ≤ 0.05 for Student's t test. ND represents no signal detected for that gene. A, Expression of Ret and Zbtb16 relative to the Rpl19 housekeeping gene. B, Stra8 and Dmc1 relative to Rpl19. C, Sycp3 and Acr relative to Rpl19 expression.

FACS analysis of undifferentiated and differentiated spermatogonia in Sox3^+/Y and Sox3^−/Y tubules

Undifferentiated and differentiated spermatogonia were counted using a technique similar to what has previously been used to identify spermatogonial stem cells (28). Glial cell line derived neutrophic factor signals through Glial cell line derived neutrophic factor family receptor α 1 and RET (29). These receptors are localized at the cell surface of undifferentiated spermatogonia, whereas the KIT receptor tyrosine kinase that binds the stem cell factor ligand, or steel, is expressed on the surface of differentiated spermatogonia (19, 30, 31). FACS, with fluorescent-conjugated antibodies to RET and KIT, was used to count the number of undifferentiated or differentiated spermatogonia, respectively, in Sox3^+/Y and Sox3^−/Y animals. The number of RET ⁺,KIT⁻ cells was similar in Sox3^+/Y and Sox3^−/Y tubules at 10, 20, 56, and 84 dpp (Fig. 7A). However, there were fewer RET⁻,KIT⁺ spermatogonia within Sox3^−/Y tubules in comparison to Sox3^+/Y tubules at 10 and 20 dpp and significantly fewer in 56 and 84 dpp animals (Fig. 7B). These data show that fewer spermatogonia differentiate in Sox3^−/Y mice in comparison to Sox3^+/Y mice, indicating that a spermatogenic block in Sox3^−/Y mice occurs at the point of spermatogonial differentiation.

Fig. 7.

FACS analysis of RET⁺,KIT⁻ and RET⁻,KIT⁺ cells within Sox3^+/Y and Sox3^−/Y tubules. Each bar represents the mean ± sem relative expression of four or more animals at 10, 20, 56, and 84 dpp. *, P ≤ 0.05. A, Number of RET⁺,KIT⁻ cells represented as a percent of Sox3^+/Y cell numbers. B, Number of RET⁻,KIT⁺ cells represented as a percent of Sox3^+/Y cell numbers.

Discussion

Studies of genetic pathways that control spermatogenesis are of interest for defining potential causes of male infertility as well as potential targets for contraception. Previous studies by our laboratory demonstrated that deletion of Sox3 causes a spermatogenic block in mice, characterized by seminiferous tubules that are largely devoid of germ cells at 14 dpp (13). Of note, testes from 56 dpp were reported to contain tubules with relatively normal spermatogenesis alongside others with only Sertoli cells (13). Tubules with expected germ cell associations, based on seminiferous epithelial staging, were identified in the current study, in postpubertal (56 and 84 dpp) Sox3^−/Y animals (15). Only six Sox3^−/Y tubules, of more than 1500 tubules analyzed, were missing the type of undifferentiated germ cells normally associated with the elongated spermatids observed in that tubule. All other tubule cross-sections that contained elongated spermatids also contained the expected spermatogonia, spermatocytes and round spermatids, based on the stage in which the elongated spermatid is normally associated. However, only approximately 75% of the tubules in Sox3^−/Y mice contained postmeiotic germ cells at 56 dpp. This percentage increased by 84 dpp, suggesting a recovery in spermatogenesis over time. Consistent with these results, mRNA analysis from whole testes at these time points revealed an expression pattern of meiotic and postmeiotic genes in Sox3^−/Y testes that was more similar to Sox3^+/Y expression by 84 dpp. The relatively normal distribution of differentiated spermatids, and therefore normal distribution of seminiferous epithelial stages, supports the hypothesis that germ cells within Sox3^−/Y tubules initiated a spermatogenic wave that began to repopulate the testis. Thus, whereas many genetic models associated with impaired spermatogenesis progress with age (e.g. Zbtb16^−/−, Nanos2^−/−, Etv5^−/−), the Sox3 mutation is associated with partial escape and recovery from the observed prepubertal spermatogenic block (17, 18, 32, 33). Overcoming the block indicates that Sox3 is not necessary for spermatogonial stem cell self-renewal.

The progeny from VasaCre animals bred to Sox3^fl/Y animals demonstrated a defect similar to the ubiquitous Sox3-knockout animals. Although some Sox3-null animals show abnormal development of the ventral diencephelon and altered gonadotropin levels, the phenotype of VasaCre⁺;Sox3^fl/Y males underscores the importance of intrinsic Sox3 expression within germ cells for spermatogenesis in prepubertal mice and demonstrates the direct effects of Sox3 ablation on spermatogenesis (34).

Although testis histology appears normal at 10 dpp, analyses of Sox3^−/Y whole-testis mRNA revealed low Stra8 expression in comparison with Sox3^+/Y. Stra8 is expressed in differentiated spermatogonia and preleptotene spermatocytes at the initiation of meiosis (35). Therefore, decreased expression of Stra8 may indicate fewer spermatogonia undergoing differentiation. This is corroborated by the fewer RET⁻,KIT⁺ differentiated spermatogonia found in 10 dpp Sox3^−/Y tubules. Therefore, the spermatogenesis defect in Sox3^−/Y males may have occurred in the differentiated spermatogonial population before visible phenotypic differences. This block in spermatogonial differentiation resulted in a significant decline in differentiated germ cells by 20 dpp. The germ cells in 20 dpp tubules maintained expression of NEUROG3, a marker of undifferentiated spermatogonia, but not SYCP3, a marker of primary spermatocytes. These tubules also maintained normal numbers of RET⁺,KIT⁻ undifferentiated spermatogonia, suggesting that germ cells in the Sox3^−/Y testes have not entered or have not progressed normally through meiosis at 20 dpp, even though the undifferentiated spermatogonia population was established in the prepubertal testis.

Germ cell loss, due to irradiation or chemical treatment, and subsequent recovery through the initiation of proliferation and differentiation of the residual progenitors have been studied in both mice and rats (36–39). Additionally, spermatogonia have the ability to repopulate testes in which busulfan is used to deplete germ cell numbers (20, 40). The spermatogonia that remain within the Sox3^−/Y prepubertal animals may act in a similar way to repopulate the depleted testes over time. These data, along with the analyses of differentiated germ cells in 56 and 84 dpp Sox3^−/Y tubules, support the hypothesis that a mechanism exists to detect a reduced population of differentiated germ cells within the testis (20, 40, 41). Zbtb16^−/− mice contained spermatogonia that underwent differentiation but not self-renewal, causing a progressive loss of germ cells over time (17, 18). Loss of Etv5 in Sertoli cells also resulted in an initial wave of spermatogenesis followed by a loss of germ cells over time as the stem cell pool was depleted (33). Our data suggest that Sox3 is not required for maintenance or repopulation of spermatogenesis but does affect the differentiation of spermatogonia in prepubertal mice. Sox3^−/Y mice are therefore a unique model to study spermatogonial differentiation and recovery mechanisms.

The mechanism by which the Sox3^−/Y spermatogenesis resumes is unknown. However, several plausible explanations might pertain. First, a redundant or compensatory Sox or other HMG box containing protein could function in place of Sox3. Because Sox3 is a member of a protein family that includes more than 20 HMG-box proteins, it is possible that one or more redundant or compensatory proteins could support spermatogenesis in animals after the first wave, potentially explaining the recovery in adult Sox3^−/Y animals. Previous studies have shown that Sox proteins are redundant in some developmental processes and that other family members can compensate for a closely related Sox protein (42–44). For example, the SoxB1 genes, Sox1/2/3, are redundant in lens development and function together during neurogenesis (45, 46). Sox1 was not detected in testes from Sox3^+/Y or Sox3^−/Y at 10, 20, 56, or 84 dpp. Sox2 was expressed at similar levels in both Sox3^+/Y and Sox3^−/Y testes at these ages (data not shown). Therefore, these genes are unlikely to compensate for the absence of Sox3 and thereby account for the recovery of spermatogenesis in the older Sox3^−/Y animals. Although Sox30 is not as closely related to Sox3 as the other SoxB1 genes, it is expressed in pachytene spermatocytes and could act as a redundant protein in the testis (47). Sox30 was reduced by approximately 45–66% in Sox3^−/Y testes at 20, 56, and 84 dpp, also making it less plausible as a candidate to compensate for Sox3 (data not shown). Other genes outside of the Sox family also contain an HMG box. The HMG box chromosomal protein 1 (HMGB1) and HMG box-containing protein 1 (HBP1) are both expressed in testes and could play a role in the recovery seen in the Sox3^−/Y adult animals (48, 49). Second, redundant signaling pathways could function to activate downstream targets of Sox3 and can therefore bypass Sox3. For example, Sox3 has been described as directly controlling Wnt signaling, by inhibiting β-catenin expression, in developing zebrafish embryos (50). Third, the progenitor population in young animals and adults may be different and/or contain unique differentiation pathways. If this is the case, germ cell repopulation in the adult may be able to circumvent Sox3 deficiency. Future investigations will test these hypotheses.

In conclusion, these results demonstrate that Sox3 is an intrinsic germ cell factor that is important for spermatogonial differentiation in prepubertal males. This was most apparent in Sox3^−/Y and VasaCre⁺;Sox3^fl/Y testis cross-sections at 20 dpp. Partial recovery of spermatogenesis is visible in adult animals. Analyses of Sox3^−/Y testes demonstrated a normal number of germ cells at 1, 5, and 10 dpp, and the majority of germ cells that remained in 20 dpp testes were spermatogonia and were not undergoing meiosis. These results, in addition to the decreased number of differentiated spermatogonia, suggest that there is a block in spermatogonial differentiation. There are no reports, to date, that describe a germ cell-autonomous spermatogenic block within juvenile or prepubertal mice that was overcome in the adult. Sox3^−/Y animals are therefore a unique model that can provide insight into the pathways that control the progression of spermatogenesis in the prepubertal animals and persistent sperm production in the adult.