Spermatogonial Stem Cell Self-Renewal Requires ETV5-Mediated Downstream Activation of Brachyury in Mice

ABSTRACT

Insight regarding mechanisms controlling gene expression in the spermatogonial stem cell (SSC) will improve our understanding of the processes regulating spermatogenesis and aid in treating problems associated with male infertility. In the present study, we explored the global gene expression profiles of the glial cell line-derived neurotrophic factor (GDNF)-regulated transcription factors Ets (E-twenty-six) variant gene 5 (Etv5); B-cell chronic lymphocytic leukemia (CLL)/lymphoma 6, member B (Bcl6b); and POU domain, class-3 transcription factor 1 (Pou3f1). We reasoned that these three factors may function as a core set of transcription factors, regulating genes responsible for maintaining the SSC population. Using transient siRNA oligonucleotides to individually target Etv5, Bcl6b, and Pou3f1 within mouse SSC cultures, we examined changes to the global gene expression profiles associated with these transcription factors. Only modest overlaps in the target genes regulated by the three factors were noted, but ETV5 was found to be a critical downstream regulator of GDNF signaling that mediated the expression of several known SSC self-renewal related genes, including Bcl6b and LIM homeobox 1 (Lhx1). Notably, ETV5 was identified as a regulator of Brachyury (T) and CXC chemokine receptor, type 4 (Cxcr4), and we showed that ETV5 binding to the Brachyury (T) gene promoter region is associated with an active state of transcription. Moreover, in vivo transplantation of SSCs following silencing of Brachyury (T) significantly reduced the number of donor cell-derived colonies formed within recipient mouse testes. These results suggest Brachyury is of biological importance and functions as part of GDNF/ETV5 signaling to promote self-renewal of mouse SSCs cultured in vitro.

Microarray profiling of ETV5-associated gene expression identified Brachyury as a novel target gene expressed in spermatogonial stem cells that is critical for self-renewal.

INTRODUCTION

Spermatogonial stem cells (SSCs) are essential to maintaining male fertility and species continuity, providing the foundation for spermatogenesis [1]. Although SSCs represent a small fraction of the testicular cell population (i.e., only approximately 1 in 3000 adult testis cells is a stem cell in the mouse testis), they are one of the most productive and biologically active adult stem cells [2]. This feature makes SSCs an excellent model to study biological processes regulating adult stem cell self-renewal and differentiation.

An understanding of the cellular and molecular mechanisms contributing to the self-renewal of SSCs has been greatly aided by a robust transplantation technique that unequivocally measures stem cell number and biological function as well as by the availability of a long-term in vitro culture system [3–7]. These assays, along with transgenic mouse models, led to the identification of glial cell line-derived neurotrophic factor (GDNF), a member of the transforming growth factor β superfamily, as a primary regulator of rodent SSC self-renewal [6–8]. The GDNF signaling pathway involves binding of GDNF ligand to GDNF receptor α1 (GFRA1), a glycophosphatidylinositol 3-kinase-anchored docking receptor, and the proto-oncogene Rearranged during Transfection (c-RET). Upon ligand-receptor binding, GDNF elicits the activation of both the phosphoinositide-kinase (PI3K) and Sarcoma (Src) family kinase (SFK) intracellular signaling mechanisms, which results in the downstream activation of the AKT signaling pathway [9–11].

Global gene profiling has identified Ets (E-twenty-six) variant gene 5 (Etv5); B-cell chronic lymphocytic leukemia (CLL)/lymphoma 6, member B (Bcl6b); and octamer binding factor 6 (POU) domain, class-3 transcription factor 1 (Pou3f1) as GDNF-regulated genes that are regarded as factors necessary for SSC survival, self-renewal, and proliferation [12, 13]. However, the mechanism by which these three GDNF-regulated factors function in controlling the fate of SSCs has not been examined. In a paradigm similar to the core regulatory group of transcription factors (i.e., POU5F1, NANOG and SOX2) that governs embryonic stem cell fate [14], it is not known if ETV5, BCL6B, or POU3F1 coregulate common gene targets essential to SSC maintenance. We reasoned that identifying genes commonly regulated by ETV5, BCL6B, and POU3f1 would provide critical information toward understanding the gene regulatory networks contributing to maintenance of the SSC population. Using siRNA gene targeting of Etv5, Bcl6b, and Pou3f1 in conjunction with gene microarray profiling, we found little overlap in the total number of genes coregulated by these three transcription factors. However, gene analysis comparing Etv5 knockdown and GDNF withdrawal gene profiles revealed that ETV5 is a critical mediator of GDNF signaling; functioning as an upstream inducer of several genes essential to regulating SSC fate, including Brachyury (T).

MATERIALS AND METHODS

Cell Culture

Murine SSC cultures were established from Thy1⁺-enriched germ cell populations as previously described [6]. Although GFRA1 has been used as a marker for undifferentiated spermatogonia, transplantation studies have shown low enrichment of testis cells (∼2-fold) for SSCs with this antigen [15, 16]. In contrast, Thy1 (CD90) is reported to be a more efficient surface marker for SSC isolation, producing a 25-fold enrichment by transplantation analysis [17]. Briefly, mouse testes were harvested from Postnatal Day 6–8 C57BL/6 or ROSA26 pups (stock nos. 000664 and 002073, respectively; The Jackson Laboratory). Many cell types within ROSA26 mice, including germ cells, express β-galactosidase and can therefore be visualized following 5-bromo-4-chloro-3-indolyl-beta-d-galactoside (X-gal) staining [18]. Single-cell suspensions were prepared by digestion with 0.25% Trypsin-ethylenediaminetetra-acetic acid (EDTA; Invitrogen) and 7 mg/ml of DNase I (Sigma) as previously described [19, 20]. Viable cells in the pellet fraction were enriched by centrifugation (600 × g, 4°C, 7 min) on a 30% Percoll solution gradient. Thy1.2 microbead-conjugated antibodies (catalog no. 130-049-101; Miltenyi Biotec) were used to enrich for the Thy1⁺ germ cell population using magnetic activated cell sorting (catalog no. 130-042-201; Miltenyi Biotec). Thy1⁺ cells were seeded at a density of 0.5–1.0 × 10⁵ cells/well on 12-well plates containing mitotically inactivated SIM mouse embryo-derived thioguanine- and ouabain-resistant (STO) feeders [6]. Long-term SSC self-renewal and proliferation were supported in a chemically defined, serum-free minimal essential medium alpha (MEMα) medium (mSFM) supplemented with 20 ng/ml of GDNF (R&D Systems), 150 ng/ml of GFRA1 (R&D Systems), and 1 ng/ml of basic fibroblast growth factor (FGF2; BD Biosciences) [6, 9]. The medium was replaced every 2–3 days and cultures passaged at approximately 7-day intervals. For quantification of germ cell proliferation, 0.25% Trypsin-EDTA digestion was used to collect all cells from the culture well (STO feeders plus germ cell clumps), and the total number of cells counted was subtracted from the number of mitotically inactive STO feeders initially seeded (1.5 × 10⁵ cells). The work presented here was generated from three independent biological cultures for use in all in vitro and in vivo studies. All animal protocols were approved by the Institutional Animal Care and Use Committee of the University of Pennsylvania.

RNA Interference

Transient reduction in the expression of targeted genes within germ cells was carried out as previously described with slight modification [12]. In brief, clump-forming SSCs were removed from the feeder layer by gentle, repeated pipetting, followed by digesting cells with 0.25% Trypsin-EDTA. Digested cell clumps were resuspended in antibiotic-free mSFM containing GDNF, GFRA1, and FGF2 and plated at a concentration of 2 × 10⁵ cells/ well in a 12-well plate without a feeder layer for 2 h and then transfected. Before the transfection, 75 pmol of siRNA oligonucleotides and 2 μl of Lipofectamine RNAiMAX reagent (Invitrogen) were mixed with 200 μl of OptiMEM medium (Invitrogen). The cells were harvested for analysis after 30 h of incubation. Targeting oligonucleotides for Etv5, Bcl6b, Pou3f, Brachyury (T), and promyelocytic leukemia zinc finger protein (Zbtb16; Plzf) were purchased from Thermo Scientific/Dharmacon (catalog nos. M-040219-01, M-043010-00, L-062952-01, L-042762-01, and L-049018-01, respectively). A nontargeting siRNA was used as a control.

RNA Isolation and Quantitative RT-PCR

The RNA was harvested from transfected SSCs according to standard Trizol isolation protocols, and potential genomic DNA contamination was prevented by DNase treatment (catalog no. AM1906; Ambion, Inc.). RNA with an A₂₆₀:A₂₈₀ ratio of 1.8 or greater was applied for first-strand cDNA synthesis, and siRNA knockdown efficiencies were validated by microarray and quantitative RT-PCR (qRT-PCR) analyses. In making quantitative comparisons, relative transcript levels of specific genes within the log phase of the amplification curve were normalized to that of ribosomal protein S2 (Rps2) using the 2^−ΔΔCt method as previously described [12]. The following primers were applied for qRT-PCR (sense and antisense, respectively; 5′–3′): Zbtb16, GCTGTGTGGGAAACGCTTTC and TGCGTGGACCTCCATGTG; Bcl6b, TACTTCAAGGCTTCGCCTCTCT and CTACGTGTTCCATCTGCAAATAGG; Etv5, CCCGGATGCACTCTTCTCTATG and TCGGATTCTGCCTTCAGGAA; Pou3f1, TTCAAGCAACGACGCATCAA and TGCGAGAACACGTTACCGTAGA; CXC chemokine receptor, type 4 (Cxcr4), AACCTCTACAGCAGCGTTCTCAT and CGTGGACAATGGCGAGGTA; LIM homeobox 1 (Lhx1), CCCAGCTTTCCCGAATCCT and GCGGGACGTAAATAAATAAAATGG; and Brachyury (T), CTGGGAGCTCAGTTCTTTCG and ACCGTGTGTGTCAGTGGTGT.

Microarray Processing and Analysis

Microarray gene expression analysis was conducted by the Penn Microarray Facility and included quality-control tests of the total RNA samples [12]. In brief, total RNA was converted to cDNA using reverse transcriptase primed by a poly(T) oligomer. Fragmented and biotinylated cDNA was added to Affymetrix hybridization cocktails and hybridized for 16 h at 45°C to Affymetrix Mouse 430 2.0 GeneChips (Affymetrix, Inc.). The microarrays were then washed at low stringency (6× SSPE [1× SSPE: 150 mM NaCl, 10 mM NaH₂PO₄, and 1 mM Na-EDTA]) and high stringency (100 mM 4-morpholine-ethanesulfonic acid [Mes] and 0.1 M NaCl) and labeled with streptavidin-phycoerythrin. Fluorescence intensity was amplified by biotinylated antistreptavidin labeling with an additional aliquot of streptavidin-phycoerythrin dye. A confocal scanner was used to collect fluorescence signal after excitation at 570 nm. Affymetrix .cel files were imported into Partek Genomics Suite (Version 6.4; Partek, Inc.) where gene chip robust multichip average (GCMRA) was applied, yielding normalized, log₂-transformed intensities. Preliminary principal components analysis and pairwise correlation analysis revealed one sample from the Bcl6b knockdown group to be an outlier. GCRMA was repeated on the remaining 15 samples. The resulting probe-set list was filtered to retain those flagged as “Present” by the MAS5 algorithm (Affymetrix Expression Console, Version 1.1), yielding 29 p277 probe sets for subsequent analysis. A one-way ANOVA was performed across the four groups of samples: negative control (N), Bcl6b knockdown (B), Etv5 knockdown (E), and Pou3f1 knockdown (P), respectively. Within the ANOVA, three pairwise contrasts were also calculated (B vs. N, E vs. N, and P vs. N), each yielding a P-value and a fold-change. All resulting P-values were corrected for false-discovery rate (FDR) using the Benjamini-Hochberg step-up method as implemented in Partek. For the arrays depicted in the heat-map array, .cel files from both of GDNF withdrawal and Etv5 knockdown microarray data were imported to Partek Genomics Suite (Version 6.4), performing GCRMA normalization yielding log₂-transformed expression values. For each probe set, the median expression was calculated across all 14 samples, and the probe-set median was subtracted from each individual probe-set value. Data were obtained from the Etv5 list (E vs. N, 10% FDR, >50% significant, P < 0.05, change in gene expression from control, 125 probe sets) and GDNF withdrawal list (>2-fold significant, P < 0.05, change in gene expression, 278 probe sets), and intersection of the two lists (15 probe sets) was calculated. Colors in the heat map reflect deviation from the median expression for each probe set.

Western Blot Analysis

Cells were harvested in modified RIPA buffer (1% Triton X-100, 50 mM Tris-HCL, 135 mM NaCl, 0.1% sodium deoxycholate, 2 mM EDTA, 50 mM NaF, 2 mM sodium orthovanadate, 10 μg/ml of aprotinin, 10 μg/ml of leupeptin, and 1 mM PMSF). Protein lysates were separated by SDS-PAGE gels and transferred to nitrocellulose membranes. Blots were blocked in 5% nonfat dry milk in 1× PBS plus 0.1% Tween-20 for 2 h and incubated with specific antibodies overnight. Antibodies used in the present study were anti-Zbtb16 (1:500; catalog no. OP128, 2A9; Calbiochem Co.), anti-POU3F1 (1:500; SC11661, C-20; Santa Cruz Biotechnology), anti-Brachyury (1:500; catalog no. SC17743, N-19; Santa Cruz Biotechnology), anti-Bcl6b (1:500; polyclonal rabbit anti-BCL6B; generous gift from D. Fearon at Welcome Trust Institute), anti-ETV5 (1:500; catalog no. SC22807, H-100; Santa Cruz Biotechnology), and anti-actin (1:5000; Sigma-Aldrich, Inc.). Blots were incubated with horseradish peroxidase (HRP)-conjugated secondary antibodies (1:2000) and proteins detected using enhanced chemiluminescent reagent (Pierce).

Immunohistochemistry

Testes from adult (age, 2 mo) mice were isolated and fixed in 4% paraformaldehyde. The testes were processed for histology by the Histology and Gene Expression Core at University of Pennsylvania. Tissue sections were deparaffinized, hydrated, and subject to antigen retrieval as previously described [9]. Tissue section were blocked using 10% normal goat serum followed by 1-h incubation at room temperature with goat anti-Brachyury antibody (catalog AF2085; R&D Systems) or a goat immunoglobulin (Ig) G antibody. Samples were washed in PBS and incubated for 20 min with biotin-labeled secondary antibody at room temperature. Sections were then washed and incubated with streptavidin-conjugated HRP and developed using the HistoStain SP substrate kit (Invitrogen). Nuclei were labeled with 4′,6-diamidino-2-phenylindole (DAPI).

SSC Transplantation

The number of donor colonies in a recipient testis following transplantation of a cell population reflects the number of SSCs present [21]. To assess the impact of transient gene knockdown on SSC self-renewal in the in vitro germ cell cultures, we used an approach previously reported [12]. In brief, 1 × 10⁵ cells/ml were seeded on individual wells of a 12-well tissue-culture plate and incubated for 30 h at 37°C in 5% CO₂ in air with gene-specific siRNA or negative-control oligonucleotides. Following incubation, cells were collected and replated on inactivated STO feeder cells in mSFM with the three growth factors. The number of cells in culture was determined at the end of 1 wk of culture. Clump-forming germ cells, enriched for SSCs, were then transplanted into the testis (10⁴ cells/testis) of adult, busulfan-treated (60 mg/kg) 129/SvCP × C57BL/6 recipient mice (stock no. 101043; The Jackson Laboratory). Two months after transplantation, testes were harvested and analyzed for donor cell-derived colonies of spermatogenesis by X-gal staining (Fig. 1A). Each set of treated cells was transplanted into four recipient mice, and each gene knockdown experiment was replicated at least twice to include 16–24 recipient testis. The final assessment of the affect of siRNA targeting is based on the following formula [17]: SSC number/10⁵ Thy1⁺ cells cultured = (number of donor-derived colonies of spermatogenesis) × (10⁵ total cells harvested/10⁵ cells transplanted) × (1/10⁵ Thy1⁺ cells originally cultured).

FIG. 1.

The effect of siRNA knockdown of Etv5, Bcl6b, Pou3f1, and Zbtb16 on in vitro-cultured SSCs. A) Experimental procedure to determine the effect of Etv5, Bcl6b, Pou3f1, and Zbtb16 gene silencing on in vitro-cultured SSCs. B) Quantitative RT-PCR validation of siRNA-knockdown efficiency. The expression level of Etv5, Bcl6b, Pou3f1, and Zbtb16 was significantly decreased to 28%, 29%, 38%, and 17%, respectively (P < 0.05). C) Western blot time-course analyses of Etv5, Bcl6b, Pou3f1, and Zbtb16 siRNA treatment in mouse germ cells (N represents 18-h nontargeting siRNA treatment). D) Average number of colonies formed in recipient testes from 10⁵ cells that were initially treated with siRNA and then further cultured for 7 days before transplantation. When Etv5, Bcl6b, and Pou3f1 were silenced, the number of donor-derived colonies was significantly decreased to 51% (43.6 ± 3.6 colonies), 38% (32.4 ± 2.6 colonies), and 72% (61.9 ± 4 colonies), respectively, as compared to controls (86.2 ± 5.1 colonies). The number of donor-derived colonies when Zbtb16 was silenced (86.5 ± 6.2 colonies) showed no significant change compared to negative controls (85.6 ± 10.6 colonies). All data are presented as the mean ± SEM for two (Etv5, Bcl6b, and Pou3f1) or three (Zbtb16) independent replicate cultures, resulting in 16 or 24 testes, respectively. An asterisk (*) denotes a significant difference between treatment means (P < 0.05).

Chromatin Immunoprecipitation Assays

The DNA binding of ETV5 to the Brachyury (T) target gene was determined using a commercial chromatin immunoprecipitation (ChIP) assay as per manufacturer recommendations (EZ CHIP; catalog no. 17–371RF; Upstate Co.). In brief, 1 × 10⁷ clump-forming SSCs were fixed with formaldehyde to covalently cross-link proteins and DNA. Fixation was terminated using 1% glycine, and cells were rapidly harvested into lysis buffer. Lysates were sonicated to shear DNA to optimal lengths of 200–500 bp using a Sonic Dismembrator (model 500; Fisher Scientific) at a setting of 10 cycles for 20 sec, interspersed with a rest period of 30 sec. Sheared protein/DNA samples were incubated overnight at 4°C with an anti-ETV5 antibody (2 mg/ml; catalog no. SC22807, H-100; Santa Cruz Biotechnology). Antibody-bound DNA fragments were collected using Protein G-Sepharose beads, and DNA fragments were purified and frozen for subsequent PCR analysis. PCR amplification was carried out using the following primer pairs: primer set A (−225 to −46): sense, 5′-AAAGGAATGTCCTTAGCGCG-3′; antisense, 5′-TCCCCTCCCCATAAATAC-3′; primer set B (+1397 to +1638): sense, 5′-CAGAGGTTCTCACCGAGAGG-3′; antisense, 5′-GTGGAGAGGGAAGCTCAATG-3′; primer set C (−864 to −625): sense, 5′-CAGAGGTTCTCACCGAGAGG-3′; antisense, 5′-GTGGAGAGGGAAGCTCAATG-3′, which were designed according to the online software OligoPerfect Designer (http://tools.invitrogen.com; Invitrogen). Control ChIP using RNA polymerase II and amplification of the glyceraldehyde phosphate dehydrogenase promoter was provided within the EZ CHIP kit.

Statistical Analyses

Independently established cell cultures of SSCs were run in at least duplicate or triplicate for transfection and transplantation experiments. The differences between cell numbers and colony numbers were determined by one-way ANOVA using SPSS 15 statistical software (SPSS, Inc.).

RESULTS

Etv5, Bcl6b, and Pou3f1, but Not Zbtb16, Are Required for GDNF-Dependent SSC Self-Renewal and Survival In Vitro

Previously, ETV5, BCL6B, and POU3F1 were shown to be transcription factors required for maintenance of the SSC [9, 12, 13]. However, the exact function of these individual transcription factors and their subsequent transcriptional gene targets in regulating the SSC population remains largely unknown. To expand our current understanding of the expression of transcription factors, Etv5, Bcl6b, Pou3f1, and Zbtb16 (a frequently used marker of mouse SSC), were transiently silenced within in vitro germ cell cultures enriched for SSCs. After 30 h of incubation, transient siRNA-mediated gene silencing of Etv5, Bcl6b, Pou3f1, and Zbtb16 decreased mRNA expression to 28%, 29%, 38%, and 17%, respectively, as compared to nontargeting controls (Fig. 1B). Western blot analyses further validated that 30-h siRNA treatment dramatically reduced protein expression (Fig. 1C). Following exposure to siRNA, germ cells were plated onto new feeders and cultured for 1 wk, a time period that reflects slightly more than the doubling time of 5.6 days for mouse SSCs (Fig. 1A [6]). Transient silencing in the expression of Etv5, Bcl6b, and Pou3f1, but not Zbtb16, significantly decreased the number of germ cells in vitro (Supplemental Fig. S1; all Supplemental Data are available online at www.biolreprod.org). To determine the acute effects that gene silencing has on the SSC population undergoing self-renewal, siRNA-treated germ cells and appropriate control germ cells were transplanted into the testes of busulfan-treated recipient mice. As previously reported [12], this approach enables quantification of SSCs in vivo and can determine the reduction in the number of SSCs that occurs in the germ cell culture following transient silencing of Etv5, Bcl6b, Pou3f1, or Zbtb16. Approximately 2 mo following transplantation, the mice were killed and the testes stained with X-gal to determine the number of donor cell-derived colonies. Transient silencing of Etv5, Bcl6b, and Pou3f1 within germ cell cultures significantly decreased the formation of donor cell-derived colonies to 50%, 37%, and 70%, respectively (Fig. 1D). In contrast to these findings, a transient decrease in Zbtb16 mRNA and protein did not significantly reduce donor cell colony formation (Fig. 1D). The findings presented here further validate the importance of these factors in maintaining the SSC population and clearly demonstrate that a transient loss of these GDNF-regulated transcription factors reduces the number of SSCs in in vitro germ cell cultures.

Etv5, Bcl6b, and Pou3f1 Share Only Modestly Overlapping Gene Expression Profiles

The impact of GDNF-regulated transcription factors ETV5, BCL6B, and POU3F1 on global gene expression as it relates to the SSC self-renewal process is poorly understood. Importantly, it remains unknown whether a disparity or a redundancy exists in the transcriptional targets regulated by ETV5, BCL6B, and POU3F1. To explore the overlapping genes potentially coregulated by the above factors, the global gene expression profiles of mouse germ cells cultures, in which Etv5, Bcl6b, or Pou3f1 had been transiently silenced, was assessed using microarray analyses. For each of the siRNA treatments, microarray gene probe values, in which expression levels between the siRNA-treated groups and nontargeting control were significantly different (P < 0.05) and at least 50% up- or down-regulated, were compared. Internal validation of siRNA treatments showed that the expression of Etv5, Bcl6b, and Pou3f1 within the individually treated germ cell cultures decreased to 31%, 48%, and 49 %, respectively, as compared to nontargeting siRNA controls (Fig. 2A). In the Bcl6b siRNA treatment, 54 gene probes were down-regulated, and 507 gene probes were up-regulated (Fig. 2, B and C, and Supplemental Table S1-1). Following Etv5 siRNA treatment, 43 gene probes were down-regulated, and 82 gene probes were up-regulated (Fig. 2, B and C, and Supplemental Table S1-2). Lastly, 64 gene probes were down-regulated, and 53 gene probes were up-regulated, in SSCs transfected with the Pou3f1 siRNA oligonucleotide (Fig. 2, B and C, and Supplemental Table S1-3). Comparative analyses of gene profiles suggests little similarity in the genes commonly down-regulated by silencing of Etv5, Bcl6b, and Pou3f1, respectively, because lamin B2 (Lmnb2) (Supplemental Table S2) was the only gene to be commonly down-regulated among the three siRNA treatments (Fig. 2B). The down-regulated gene profile of Etv5 shared three genes in common with the Bcl6b down-regulated gene profile (Bcl6b, Dhcr24, and Lmnb2) and shared two genes in common with the Pou3f1 gene profile (Lmnb2 and Sema7a). Affymetrix Expression Console (Version 1.1) was used to ascertain the biological processes and molecular function of the genes regulated by each of the three siRNA treatments. Genes down-regulated following Etv5 silencing were largely associated with cellular and molecular functions (e.g., Brachyury (T), Cxcr4, Eomes, Foxo2, and Tec). Transient silencing of Bcl6b resulted in the down-regulation of genes largely associated with cell-to-cell signaling and interaction (e.g., Itga5 and Pvr) and cellular growth and proliferation (Fosl1, Lmna, and Prmt). Pou3f1 silencing resulted in the down-regulation of genes associated with posttranslational modification and protein folding (e.g., Isg15 and Usp18) as well as chemotaxis (e.g., Ccl5 and Cxcl10).

FIG. 2.

Overlapping gene expression profiles following transient siRNA silencing of Etv5, Bcl6b, and Pou3f1. A) Percentage of gene expression for Etv5, Bcl6b, and Pou3f1 within gene microarray following siRNA knockdown. Compared to negative controls, the expression of Etv5, Bcl6b, and Pou3f1 was decreased to 31%, 48%, and 49%, respectively. All data are presented as the mean ± SEM for the expression values of gene probes from three gene microarrays of three independent replicate cultures treated with siRNA to Etv5, Bcl6b, or Pou3f1. An asterisk (*) denotes a significant difference between treatment means (P < 0.05). B) The overlapping gene profile of significantly down-regulated genes common among SSCs treated with siRNA targeting Etv5, Bcl6b, or Pou3f1. C) The overlapping gene profile of up-regulated genes common among SSCs treated with siRNA targeting Etv5, Bcl6b, or Pou3f1.

In all, 33 gene probes were commonly up-regulated among the three siRNA treatments (e.g., Alcam, Casp3, Exoc4, and Fbn1) (Fig. 2C and Supplemental Table S2), with 38 gene probes shared in common between Bcl6b and Pou3f1 (e.g., Arf6 and Col11a1). The up-regulated gene profile of Etv5 shared 52 gene probes in common with the Bcl6b up-regulated gene profile (e.g., Camk1d and Zeb1) and shared 38 gene probes in common with the Pou3f1 gene profile (e.g., Kctd14 and Pi3kip1). A common biological function of genes up-regulated following knockdown of Etv5, Bcl6b, and Pou3f1 included cell death, as observed by the increase in apoptotic-related genes, such as Bcl2, Casp3, Atxn2, and Fbn1. This latter observation provides a partial mechanism for previous reports demonstrating that abrogation of Bcl6b or Pou3f1 increases the number of germ cells undergoing apoptosis in vitro [12, 13]. Overall, results suggest modest similarity in the gene profiles regulated by ETV5, BCL6B, and POU3F1, and the differences in their respective gene profiles suggests these transcription factors have different roles in regulation SSC fate.

ETV5 Is an Upstream Transcription Factor that Regulates Multiple Known SSC Self-Renewal Related Genes

Although only modest overlap was found in the genes coregulated by ETV5, BCL6B, and POU3F1, it was observed that siRNA-targeted inhibition of Etv5 decreased the expression of Bcl6b, but not Pou3f1. This finding suggests that ETV5 regulates Bcl6b as part of the signaling events elicited by GDNF. To better evaluate the relationship between the genes regulated by GDNF signaling and those controlled by ETV5, we compared the gene expression profiles of a previous GDNF withdrawal study [12] to that the Etv5 siRNA knockdown gene profiles. Using stringent cutoff values to reduces the inclusion of false-positive or false-negative genes in the data analyses, the GDNF withdrawal gene array produced 278 probe sets (>2-fold change in gene expression), and the Etv5 gene list produced 125 probe sets (10% FDR, 50% change in gene expression from control). A Venn diagram in Figure 3A demonstrates the intersection of the two array studies, with only 15 genes probes found to be in common between each study. These 15 genes were further profiled and displayed as a heat map, and based on the gene overlap, we observed that both withdrawal of GDNF and silencing of Etv5 within SSC-enriched germ cell cultures resulted in the significant decrease of several genes, including self-renewal related genes Bcl6b and Lhx1 [12] as well as Brachyury (T) and Cxcr4, the function and importance of which within the maintenance of SSCs has not yet been described (Fig. 3A). The expression of these target genes in Etv5 siRNA-treated SSCs was further validated by qRT-PCR, demonstrating that loss of Etv5 significantly decreased the expression of Bcl6b, Lhx1, Brachyury (T), and Cxcr4 to 41%, 56%, 40%, and 71%, respectively (Fig. 3B).

FIG. 3.

Comparison of gene profiles following 30-h transient knockdown of Etv5 and 18-h GDNF withdrawal. A) The 278 gene probes found to be significantly regulated in a previous GDNF withdrawal gene array analysis (2-fold change in gene expression [12]) were compared to the 125 gene probes significantly altered by Etv5 siRNA treatment. Top) The Venn diagram shows the intersection of 15 gene probes common between each profile analysis. Bottom) Heat map analysis of the 15 genes overlapping between two sets of microarray data; from left to right: Etv5 control (four samples, orange), Etv5 knockdown (four samples, blue), GDNF withdrawal (three samples, purple) and no GDNF withdrawal control (three samples, red). The fold-expression levels of down-regulated (green) and up-regulated (red) genes range from −3.5- to 3.5-fold. B) From the above gene list, the expression of Bcl6b, Lhx1, Brachyury (T), and Cxcr4 were dramatically down-regulated following either Etv5 siRNA treatment or GDNF-withdrawal treatments, and the expression of these genes was further validated by qRT-PCR. Following Etv5 siRNA treatment of SSCs, the relative expression values for Bcl6b, Lhx1, Brachyury (T), and Cxcr4 decreased to 41%, 56%, 40%, and 71%, respectively, as compared to negative controls. All data are presented as the mean ± SEM for three independent replicate cultures treated with control or Etv5 siRNA and profiled by qRT-PCR. An asterisk (*) denotes a significant difference between treatment means (P < 0.05).

Brachyury (T) Gene Is Directly Regulated by ETV5 and Is Required for Proliferation and Self-Renewal of Mouse SSCs

The above findings suggest that Brachyury (T) is the gene most influenced by GDNF withdrawal or alteration in Etv5 expression and represents a potentially novel regulator of SSC self-renewal. To verify the expression of Brachyury in SSC-enriched germ cell cultures, Western blot analysis was carried out comparing the protein lysates of two independent SSC-enriched germ cell cultures, gently aspirated from STO feeders, with protein lysates from STO feeders alone as well as from whole testis. The expression of Brachyury was detected in two SSC-enriched germ cell cultures, but not in the STO feeder cells (Fig. 4A). Brachyury expression is also detected in the whole-testis lysate. To further evaluate Brachyury expression, immunohistochemistry was performed on the testis of adult (age, 2 mo) mice using an anti-Brachyury antibody or IgG negative control (Fig. 4B and Supplemental Fig. S2). Brachyury expression was observed in spermatogonia near the basement membrane of the mouse testis, but its expression increased in preleptotene, leptotene, and some pachytene spermatocytes (Fig. 4B).

FIG. 4.

Detection of Brachyury expression. A) Western blot analysis of Brachyury expression in SSC-enriched germ cell cultures. SSC-enriched germ cells were gently aspirated and collected, leaving behind the STO feeders. This approach yields 96% ± 1% pure population of germ cells [12]. Mitotically active and inactive STO cells, along with whole-testis lysates, were also assessed. The expression of Brachyury is observed in SSC-enriched germ cell cultures but not STO feeder cells. B, top) Immunohistochemical analysis of Brachyury expression on adult (2-mo-old) testis. Detection of Brachyury is seen in spermatogonia on the basement membrane, but its expression is increased in preleptotene, leptotene, and some pachytene spermatocytes. B, bottom) DAPI nuclear staining was used to aid identification of spermatogenic stages. Bar = 50 μm.

A previous report indicated that PEA3 (polyoma virus enhancer activator 3; E1AF or ETV4), another ETS family member, has a putative binding site (5′-ACAGGAAG-3′) in the upstream promoter region of the Brachyury (T) gene [22], and a similar binding sequence may also be utilized by ETV5 in the direct regulation of Brachyury (T) in self-renewing SSCs. Using ChIP assays, we cross-linked all protein bound to DNA and specifically immunoprecipitated all ETV5-bound DNA fragments. Subsequent PCR amplification using a primer set designed to flank the ETV5-binding site within the promoter region of the Brachyury (T) gene loci (−225 to −46 relative to the transcription start site) produced a 179-bp DNA product (Fig. 5, A and B). No amplification product of ETV5-bound DNA was observed using control primer sets targeting regions not known to possess ETS-binding sites (primer sets B and C) (Fig. 5B, bottom). Along with IgG negative controls, these results suggest that ETV5 targets the ETS-binding site contained within the Brachyury (T) promoter region. Moreover, ETV5/ETS binding within the Brachyury (T) promoter region is associated with an active state of transcription, because amplification of RNA polymerase II-bound DNA using the same primers designed to flank the ETS-binding site in the Brachyury (T) promoter region produced a similar amplicon (Fig. 5B, top). This finding strongly suggests that as a GDNF-regulated gene, ETV5 binding to the Brachyury (T) promoter has a role in regulating Brachyury (T) expression. This is further supported by previous gene array experiments that also demonstrated Brachyury (T) as one of the most tightly regulated genes in response to GDNF withdrawal (∼5-fold decrease after GDNF withdrawal) [12]. However, the importance of Brachyury (T) in maintaining the SSC population has not been addressed. Therefore, transient siRNA-mediated silencing was employed to repress endogenous Brachyury (T) expression to approximately 20% compared to negative siRNA control treatments (Fig. 5C). Following 7 days of culture after transfection, this transient loss of Brachyury (T) resulted in a decrease in cell number as compared to approximately 55% of negative controls (1.08 ± 0.08 × 10⁵ cells vs. 1.98 ± 0.02 × 10⁵ cells) (Fig. 5D). Importantly, when these cells were transplanted into recipient testis (after 7 days in culture posttransfection), the transient silencing of Brachyury (T) decreased the number of donor colonies formed from 99.3 ± 4.1 per 10⁵ treated germ cells in nontargeting control groups to 63.0 ± 4.0 per 10⁵ treated germ cells in Brachyury (T) siRNA-treated SSCs (Fig. 5E). This result demonstrates the requirement for Brachyury expression for the in vitro maintenance of the mouse SSC population.

FIG. 5.

Brachyury is essential for the in vitro self-renewal of mouse SSCs and is a bona fide gene target of ETV5. A) The presence of an Ets-family transcriptional binding site, 5′-ACAGGAAG-3′ within the Brachyury (T) promoter region has been previously described [20]. A pair of primers (primer set A) was designed to span the binding position −225 to −46, relative to the transcription start site, producing a 179-bp PCR product. B) PCR enrichment from ChIP assays using antibodies against ETV5 or RNA-Polymerase II. First gel, from left to right: 100-bp DNA marker, ETV5 pull down 1, ETV5 pull down 2, Input DNA 1, Input DNA 2, IgG Control 1, IgG Control 2, RNA Polymerase II 1, and RNA Polymerase II 2. Amplification of DNA pull-down products using primer set A, which includes the ETS-binding motif and spans positions −225 to −46 relative to the transcription start site, produced a 179-bp band in ETV5 pull-down lanes, signifying ETV5 binding to ETS-binding site in the Brachyury (T) promoter. A similar DNA amplicon, using primer set A, was produced in samples immunoprecipitated with anti-RNA polymerase II antibody, suggesting the Brachyury (T) promoter region containing the ETS-binding site is transcriptionally active. Second control gel, from left to right: 100-bp DNA marker, amplification of ETV5-bound DNA using primer set A as well as negative-control primer sets B and C. Primer set B spans the Brachyury intron (+1397 to +1638), whereas primer set C was designed to flank the region −864 to −625 upstream of the Brachyury transcriptional start site. Neither primer set B nor primer set C contains known ETS-binding motifs. Input DNA serves as the control. C) Quantitative RT-PCR showed that Brachyury (T) gene expression was reduced in germ cells to nearly 20% to that of control after 30-h of incubation with Brachyury (T) siRNA compared to negative control. Three independent replicate cultures were treated with control or Brachyury (T) siRNA and profiled by qRT-PCR (values are presented as the mean ± SEM). D) Following 7 days in culture, cells that had been transfected with siRNA targeting Brachyury (T) yielded a significantly lower cell number (1.08 ± 0.08 × 10⁵ cells) compared to negative control (1.98 ± 0.02 × 10⁵ cells). E) After 1 wk of cell culture following transfection with Brachyury (T) siRNA, the treated cells were transplanted into the testes of recipient mice. The number of donor-derived colonies formed from Brachyury (T) siRNA-treated germ cell cultures was decreased from 99.3 ± 4.1 to 63.0 ± 4.0 colonies per 10⁵ cells initially treated and cultured. For D and E, two independent replicate cultures were treated with control or Brachyury (T) siRNA, resulting in 16 testes per treatment (n = 8 mice), respectively (values are presented as the mean ± SEM). An asterisk (*) denotes a significant difference between treatment means (P < 0.05).

DISCUSSION

The maintenance of SSCs relies on multiple intrinsic and extrinsic signaling mechanisms to regulate self-renewal, differentiation, and apoptosis. The sum total of SSC self-renewal is representative of SSCs undergoing proliferation, differentiation, and apoptosis, and any impairment to this delicate balance can result in a decline in the number of SSCs both in vitro and in vivo. GDNF is an important molecule regulating SSC activity, because perturbation of its signaling disrupts spermatogenesis [8]. Extrinsic GDNF signaling activates the SFK pathway, which in turn is associated with regulating the expression of Etv5, Bcl6b, and Lhx1. The Ets families of transcription factors are involved in the regulation of numerous cellular process, including growth, apoptosis, development, and differentiation [23]. ETV5 is essential for both the in vitro and in vivo self-renewal of SSCs, and the present results further support a role for ETV5 in the maintenance of the SSC pool [12]. The gene overlap between Etv5 knockdown and GDNF withdrawal studies via heat-map analyses revealed that the expression of Bcl6b, Lhx1, Brachyury (T), and Cxcr4 were all dramatically reduced in both treatments, implicating ETV5 as an upstream effector of all four genes following activation of the GDNF-signaling cascade.

The transient silencing of Zbtb16 did not show a significant effect on the ability of SSCs to undergo self-renewal in vitro (Fig. 1D and Supplemental Fig. S1). This is further supported by the culture of Zbtb16^−/− spermatogonial progenitor cells (SPCs), which can be maintained for long term [24]. However, the expression of Zbtb16 is required, because Zbtb16^−/− mice undergo a progressive loss of spermatogenesis and become infertile [25, 26]. Recently, Hobbs et al. [24] demonstrated that the abrogation of Zbtb16 expression increased the activity of mammalian target of rapamycin complex 1 (mTORC1) in SPCs, which in turn inhibited the response of SPCs to GDNF. Subsequent inhibition of mTORC1 relieves the SPCs of the defects caused from loss of Zbtb16^−/−. This suggests that Zbtb16, although not dependent upon GDNF for its expression, is capable of augmenting GDNF signaling via reduced activity of mTORC1 [24].

Whereas Pou5f1 (also known as Oct-4) is an essential factor for the maintenance of the embryonic stem cell population [27, 28], it is Pou3f1 (also known as Oct-6) that is involved in regulating SSC fate [13]. This is further validated by the present results demonstrating that siRNA-mediated gene silencing of Pou3f1 impairs in vitro maintenance of SSCs and dramatically impacts the number of colonies formed within the testes of recipient mice. However, results show that the expression of Pou3f1 is not affected by the silencing of Etv5, indicating that unlike Bcl6b, the GDNF regulation of Pou3f1 is via another signaling mechanism. Until now, the profile of Pou3f1-regulated genes within SSCs has remained unknown; however, results from our microarray analyses demonstrate that transient silencing of Pou3f1 causes the down-regulation of genes associated with processes regulating chemotaxis, posttranslational modifications, and protein folding. Additionally, we find that Pou3f1 silencing results in the increased expression of several apoptosis-related genes (Arf6, Casp3, and Casp7) (see Supplemental Table S1). This is consistent with recent findings showing that silencing of Pou3f1 significantly increases the number of SSCs undergoing apoptosis in culture [13].

The siRNA-mediated knockdown of Bcl6b in SSCs has been associated with promoting apoptosis [12], and the present results demonstrate that a loss of Bcl6b up-regulates genes associated with apoptosis, which supports Bcl6b as a critical factor required for in vitro growth of germ cells. BCL6B shares very similar biochemical characteristics to its homologue, BCL6, including a comparable affinity for a common consensus binding site, in which it acts as a transcriptional repressor [29]. Consistent with this function, a large disparity between up-regulated versus down-regulated genes was observed in SSCs treated with Bcl6b siRNA, with 507 gene probes up-regulated and only 54 gene probes down-regulated following Bcl6b silencing (nearly a 10-fold difference in the number of responding genes) (Fig. 2), highlighting the predominant nature of Bcl6b as a transcriptional repressor in this cell population. Therefore, the proposed regulation of Bcl6b by ETV5 might be expected to share a similar gene profile in Etv5-silenced cells. Rather, little similarity was found between the two profiles, and this difference may be due to the transient nature of the siRNA approach in causing Etv5 inhibition, in which the 30-h transfection time in vitro is not sufficient to allow a complete loss of Bcl6b expression. Interestingly, the transcriptional repressor activity of BCL6B is dependent upon BCL6 [30], and although previous gene profiling suggests that Bcl6 gene expression is not GDNF-dependent [12], it is possible that BCL6, either constitutively expressed or under the regulation of a separate growth factor, may function as a cofactor required by BCL6B for its transcriptional repressor activity and subsequent maintenance of the SSC population.

Whereas both Bcl6b and Lhx1 were previously shown to be important for SSC maintenance [12], both Cxcr4 and Brachyury (T) are newly identified here as Etv5-regulated genes. Comparative microarray results demonstrated that Cxcr4 is another downstream target of both GDNF signaling and ETV5 in cultured SSCs. A similar cellular regulation of Cxcr4 by GDNF activation of ETV4 and ETV5 is also observed in ureteric budding during kidney branching morphogenesis, where abrogation of ETV5 is associated with a dramatic reduction in Cxcr4 expression [31]; this relationship is similar to our current findings and suggests that Cxcr4 is a target of ETV5. The interaction between CXCR4 and its ligand, CXCL12, is involved in many cellular functions, including the homing of hematopoietic stem cells within the bone marrow and guidance of migratory primordial germ cells to the gonad [32]. In spermatogenesis, a proposed role for the CXCR4-CXCL12 axis is in the maintenance of undifferentiated spermatogonia within the postnatal testes [33, 34]. Targeted disruption of Etv5 within the Sertoli cells of mouse testes results in the decreased expression of Cxcr4 and its cognate ligand, CXCL12, resulting in a gradual depletion of the germ cell population [34]. Although ETV5 is present in both germ cells and Sertoli cells, the regulation of ETV5 in Sertoli cells appears to be mediated by FGF signaling [34], whereas in germ cells, ETV5 is regulated by GDNF [12]. The significance in the differential regulation of ETV5 in these two cell types requires further investigation, but it is interesting that in Sertoli cells, ETV5 regulates the expression of additional chemokines, including CCL9 and CXCL5 [35]. Taken together, these findings suggest that in addition to CXCL12/CXCR4 signaling, ETV5 coordinates the expression of other chemokines in Sertoli cells that may have an important role in maintaining these and other SSC niche factors. Similarly, ETV5 in germ cells regulates the expression of CXCR4 and, perhaps, other chemokine receptors as part of a coordinated response to the chemokine factors produced by the Sertoli cells.

The expression of Brachyury (T) was identified as a new gene central for SSC maintenance and self-renewal. As a member of the T-box family of transcription factors, Brachyury (T) is required in the early stages of mesoderm determination and differentiation [36, 37]. Brachyury (T) is also expressed in multipotent germline stem cells derived from the testis of neonatal mice [38], and aberrant expression of Brachyury (T) is found in a number of tumors, including human testis tumors [39]. Yamaguchi et al. [22] previously identified a PEA3-binding site within the Brachyury (T) promoter site; PEA3 belongs to an ETS subfamily that includes three highly homologous factors, Ets-related molecule (ERM; Etv5), Ets-related-81 (ER81; Etv1), and PEA3 (Etv4). Although individual ETS transcription factors display sequence specificity, different ETS factors are capable of binding to the same ETS-binding site [23, 40]. ChIP analyses suggests that direct binding of ETV5 to the Brachyury (T) promoter region is associated with its transcriptional regulation (Fig. 4B), and this is further supported by the reduced expression of Brachyury (T) observed following the transient silencing of ETV5 or withdrawal of GDNF [12]. Importantly, transient silencing of Brachyury (T) significantly reduced the in vitro self-renewal potential of SSCs as demonstrated by decreased in vivo colony formation in recipient mice, strongly suggesting that Brachyury (T) is a GDNF-ETV5 target gene involved in the maintenance of SSC self-renewal. Although the function of Brachyury at the molecular level remains to be determined, in vitro silencing of Brachyury (T) in human lung carcinoma (H460) cells increases cellular proliferation as compared to negative-control transfected cells. This increase in cell growth is associated with decreased expression of Cdk inhibitor p21 (Cdkn1a) and cullin 1 (Cul1), both negative regulators of the cell cycle [41]. It is possible that in mouse SSCs, Brachyury, through control of cell-cycle progression, may regulate proliferation and self-renewal.

Rather than acting as a core group of transcription factors, the gene profiles of ETV5, BCL6B, and POU3F1 are each mostly independent of one another, with the repertoire of genes in each profile differing in regards to their molecular function(s), and this suggests that GDNF maintenance of the SSC population functions through several regulatory pathways that are in part controlled by ETV5, BCL6B, and POU3F1. Notably, activation of ETV5 appears to be a “master” regulator of the GDNF signaling pathway, and according to our findings, Etv5, although it does not control Pou3f1, acts as an upstream regulator of Bcl6b, Lhx1, Brachyury (T), and Cxcr4 gene expression in the mouse SSC. Moreover, Brachyury (T) is a novel factor required for SSC self-renewal.