ETV5 Regulates Sertoli Cell Chemokines Involved in Mouse Stem/Progenitor Spermatogonia Maintenance

Spermatogonial stem cells are the only stem cells in the body that transmit genetic information to offspring. Although growth factors responsible for self–renewal of these cells are known, the factors and mechanisms that attract and physically maintain these cells within their microenvironment are poorly understood. Mice with targeted disruption of Ets variant gene 5 (Etv5) show total loss of stem/progenitor spermatogonia following the first wave of spermatogenesis, resulting in a Sertoli cell–only phenotype and aspermia. Microarray analysis of primary Sertoli cells from Etv5 knockout (Etv5−/−) versus wild–type (WT) mice revealed significant decreases in expression of several chemokines. Chemotaxis assays demonstrated that migration of stem/progenitor spermatogonia toward Etv5−/− Sertoli cells was significantly decreased compared to migration toward WT Sertoli cells. Interestingly, differentiating spermatogonia, spermatocytes, and round spermatids were not chemoattracted by WT Sertoli cells, whereas stem/progenitor spermatogonia showed a high and significant chemotactic index. Rescue assays using recombinant chemokines indicated that C-C-motif ligand 9 (CCL9) facilitates Sertoli cell chemoattraction of stem/progenitor spermatogonia, which express C-C-receptor type 1 (CCR1). In addition, there is protein–DNA interaction between ETV5 and Ccl9, suggesting that ETV5 might be a direct regulator of Ccl9 expression. Taken together, our data show for the first time that Sertoli cells are chemoattractive for stem/progenitor spermatogonia, and that production of specific chemokines is regulated by ETV5. Therefore, changes in chemokine production and consequent decreases in chemoattraction by Etv5−/− Sertoli cells helps to explain stem/progenitor spermatogonia loss in Etv5−/− mice.


INTRODUCTION
In the mammalian testis, spermatogonial stem cells (SSCs) form a small subpopulation of germ cells that self-renew to maintain sperm production throughout life. SSCs reside in the basal part of the epithelium lining the seminiferous tubules and are in close association with Sertoli cells that provide germ cells with growth and differentiation factors. Despite recent progress in characterizing SSCs, their true identity has not been demonstrated, and transplantation assays have indicated that all early type A spermatogonia (A single , A paired , and A aligned ) might have stem cell potential [1,2]. A aligned sperma-togonia further develop into differentiating spermatogonia (A 1 -A 4 ), type B spermatogonia and spermatocytes that will undergo meiosis, become haploid spermatids and ultimately sperm [3]. Of particular interest is the perinatal differentiation of gonocytes and SSCs into more mature germ cells, which establishes the first and subsequent waves of spermatogenesis. This period of testicular development is associated with considerable structural changes in the seminiferous epithelium, during which the size of the testis increases over 80-fold in the mouse. This expansion is due to the fact that the somatic Sertoli cells that support germ cells proliferate extensively until day 12-16 after birth [4]. It has been postulated that this early time period, during which there is simultaneous proliferation of both Sertoli cells and spermatogonia, has regulatory requirements that are unique [5]. With each Sertoli cell division a new potential stem cell niche is created, which would be filled by the continued proliferation of stem/progenitor spermatogonia followed by lateral migration of these cells along the basement membrane [5]. Thereafter, the cyclic process of spermatogenesis initiated from SSCs nested within their niche during the first wave is maintained throughout adult life.
The niche represents the microenvironment of a stem cell within a specific tissue that regulates its fate, ensuring maintenance or homeostasis between self-renewal and differentiation. In the mammalian testis, the SSC niche is mainly defined by physical associations with Sertoli cells and the basement membrane to which SSCs adhere [6,7]. In addition, the vascular network [8] and interstitial cells found between seminiferous tubules communicate with SSCs and are critical for their spatial localization along the basement membrane. Sertoli cells produce growth factors that are essential for SSC maintenance. In particular, they secrete glial cell line-derived neurotrophic factor (GDNF), which plays a major role in promoting SSC self-renewal in vivo [9,10] and in vitro [11][12][13]. GDNF secreted by Sertoli cells binds to a receptor complex formed by association of the receptors GFRa1 (GDNF family membrane receptor alpha 1) and RET (Rearranged during transfection), which brings about the intracellular response. In the testis, GFRa1 is expressed by A single and possibly A paired and A aligned spermatogonia [11,14,15]. Hence, GFRa1 was selected to enrich stem/progenitor spermatogonia after STA-PUT isolation of type A spermatogonia in our experiments. Sertoli cells also produce factors like basic fibroblast growth factor and epidermal growth factor, which are essential for stem cell maintenance [7,12] and have stimulatory effects on Ets variant gene 5 (Etv5) expression [16].
The transcription factor ETV5 (also known as Ets-variant gene 5) is a protein that is critical for maintenance of stem/progenitor spermatogonia within their niche [17]. ETV5 is expressed by both Sertoli and germ cells, but its precise role in each cell type remains to be established [17][18][19]. Mice with a global deletion of Etv5 (Etv5 À/À ) have a first wave of spermatogenesis, but there is complete loss of stem/progenitor spermatogonia by day 36 [20], resulting in a Sertoli cell-only phenotype and ultimately aspermia in adulthood [17]. Loss of ETV5 in stem/progenitor spermatogonia induces a decrease in their selfrenewal in vivo and in vitro [19,21]. However, this does not account for the total loss of these cells. Over the years, it has become evident that the developmental fate of a stem cell is not only determined by intrinsic factors, but is also influenced by the availability and quality of stem cell niches [1]. Therefore, a critical question for understanding the role of ETV5 in spermatogenesis is to determine its function in Sertoli cells.
Chemokines are small secreted cytokines that were initially found to play a major role in leukocyte migration and communication [22,23]. However, there is growing evidence suggesting a role of chemokines in homing and maintenance of stem cells in their niche [24,25]. In the embryo, primordial germ cell (PGC) migration is mediated by the chemokine C-X-C motif ligand 12 (CXCL12), also known as stromal cellderived factor-1 (SDF-1) [26]. However, little is known about the role of chemokines in mediating communication between Sertoli cells and germ cells after birth and maintaining stem/ progenitor spermatogonia in their microenvironment.
Microarray analysis of Etv5 À/À Sertoli cells revealed a decrease in several chemokines-encoding genes such as C-Cmotif ligand 9 (Ccl9), Ccl7, Cxcl5, and Cxcl12 [17]. As a role of chemokines in stem cell homing has been proposed in other systems [24][25][26], we hypothesized that ETV5 in Sertoli cells might be regulating chemokine expression, and that these pro-teins could directly or indirectly be involved in migration and/ or retention of stem/progenitor spermatogonia in their microenvironment. Our data indicate that there is a decrease in chemoattraction of stem/progenitor spermatogonia toward Sertoli cells following loss of ETV5. This, together with a decrease in the rate of proliferation [19,21], explains the ultimate loss of these cells in Etv5 À/À mice. One of these chemokines, CCL9, is strongly expressed by Sertoli cells, and we have localized its receptor, CCR1, to undifferentiated spermatogonia. Further, we show possible protein-DNA interaction between ETV5 protein and the Ccl9 promoter. These data indicate that chemokine signaling is involved in the migration of stem/progenitor spermatogonia to newly established Sertoli cells.

Animals
C57Bl/6 dams with male pups (5-to 6-day-old) were obtained from Charles River (Boston, MA, www.criver.com). Etv5 À/À (germline knockout of exons 2-5) [17] and C57Bl/6 mice were bred and maintained in the animal facility at University of Illinois. Mice were housed at 25 C with a 12L:12D photoperiod and given water and a standard rodent chow diet. All animal experiments were approved by the IACUC at the University of Illinois and conducted in accordance with the National Institute of Health Guide for the Care and Use of Laboratory Animals. Germ cells were isolated by the STAPUT method, which utilizes gravity sedimentation on a 2%-4% bovine serum albumin (BSA) gradient [27,28]. Type A spermatogonia (Type A) were isolated from C57Bl/6 pups at days 5-6 (60-80 pups). Type B spermatogonia and leptotene/zygotene spermatocytes (Type B, L/Z) were isolated from C57Bl/6 mice at day 12 (10-12 pups), pachytene spermatocytes (P) at day 21 (5-8 mice), and round spermatids (RS) at day 45 (3-4 mice). Briefly, testes were decapsulated and subjected to a two-step enzymatic digestion protocol, to eliminate peritubular cells and Leydig cells [27][28][29]. The resulting single cell suspension, containing Sertoli cells and germ cells, was washed by centrifugation and resuspended in 0.5% BSA in Dulbecco's modified Eagle's medium (DMEM)/F12. Cells were separated based on gravity sedimentation and the different fractions were collected using a fraction collector (Bio-Rad, Hercules, CA, www.bio-rad.com). The germ cells were identified by size and morphological characteristics using a light microscope. The size of cells were $14-16 lm, 8 lm, 12-18 lm [27], and 10-11 lm [30] for Type A, Type B and L/Z, P, and RS, respectively. The cell fractions that were enriched for the respective cell types were pooled, counted, and resuspended in DMEM/F-12 supplemented with 10% synthetic Nu-Serum (BD Biosciences, San Jose, CA, www.bdbiosciences.com), penicillin-streptomycin (1 ml/100 ml; Invitrogen, Carlsbad, CA, www.Invitrogen.com) for chemotaxis assays. All enzymes were purchased from Sigma (St. Louis, MO, www.sigmaaldrich.com). For isolation of stem/progenitor spermatogonia, the type A cell suspension was plated on lectin-coated plates for up to 30 minutes (differential plating) to remove contaminating Sertoli or peritubular cells [31]. The supernatant containing Type A spermatogonia (95%-98% purity, 4 Â 10 6 cells for 60 pups, 6-day-old) was resuspended in DMEM/F-12 with 10% Nu-Serum for enrichment of stem-progenitor spermatogonia with magnetic-activated cell sorting (MACS) using GFRa1 as the selecting marker [32].

Sertoli Cell Isolation and Culture
Sertoli cells were isolated from 5-day-old Etv5 À/À or wild-type (WT) pups according to standard methods widely used by others and our laboratories [4,[33][34][35][36]. Briefly, testes were decapsulated and digested using a 2-step enzymatic digestion protocol, to eliminate peritubular and Leydig cells. The resulting cell suspension, containing Sertoli cells and spermatogonia, was trypsinized to eliminate clumps, resuspended in serum-free DMEM/F-12 media (minimal media), and plated on matrigel-coated plates. The media was changed after 4 hours to remove spermatogonia, and the Sertoli cells were cultured for 48 hours. This method of isolation yields a highly homogenous Sertoli cell population with a purity of >95%. Lack of serum averts the proliferation of any residual peritubular cells. Although contamination by peritubular cells is unavoidable, no germ cells and only <1% Leydig cells can be detected. This method also preserves the functionality of Sertoli cells for at least 7 days, such as response to hormones and growth factors [33,34,[36][37][38][39].

Chemotaxis Assays
Chemotaxis assays were performed using 8-lm-pore Cytoselect 24-well assay plates (colorimetric format; Cell Biolabs, San Diego, CA, www.CellBiolabs.com). Sertoli cells isolated from 5day-old Etv5 À/À and WT testes were seeded at a concentration of 2.5 Â 10 5 cells in the lower chambers of the transwell units and cultured in minimal media for 48 hours prior to migration assays. WT stem/progenitor spermatogonia or other germ cells were seeded in minimal media into the upper chamber at a concentration of 5 Â 10 5 cells/insert. The cells were allowed to migrate for 24 hours at 34 C, 5% CO 2 , and maximum humidity. DMEM/F12 with 10% fetal bovine serum (FBS) was used as a positive control in the lower chamber in all experiments, and minimal media alone, without any added cells, chemokines or serum, was used as a negative control. After 24 hours of incubation, the cells that migrated through pores in the membrane were fixed, stained, and lysed. The optical density of the lysed solution was measured at 560 nm and the chemotactic index was calculated. The chemotactic index is the optical density measurement of stem/progenitor spermatogonia that migrated toward the chemoattractant in the lower chamber compared with the negative control (minimal media or Etv5 À/À Sertoli cells).
Initially a dose-response of recombinant proteins CCL9, CCL12, and CXCL5 to attract C18-4 cells was performed. C18-4 is an immortalized SSC cell line that expresses many SSC markers such as GFRa1, OCT-4, and deleted in azoospermia-like (DAZL) [40,41]. Chemokines were added to DMEM/F-12 in the lower chamber of the transwell unit and incubated with C18-4 cells in the upper chamber for 24 hours. The optimum concentration was standardized as 50 ng/ml for all proteins individually (not shown).

RNA Extraction and Real-Time Polymerase Chain Reaction
RNA was extracted from isolated and cultured Sertoli and germ cells using the RNeasy mini or micro kit according to manufacturer's instructions (Qiagen, Valencia, CA, www.qiagen.com). The samples were treated with DNase to prevent DNA contamination. First-strand cDNA was synthesized from total RNA (1 or 0.5 lg) using Superscript reverse transcriptase III and random primers (Invitrogen, Carlsbad, CA, www.Invitrogen.com). The expression value of each gene was normalized to the amount of an internal control gene (18S ribosomal RNA) to calculate a relative amount of RNA in each sample. The expression value of each gene in WT or mock siRNA controls was arbitrarily defined as 1 unit. Real-time polymerase chain reaction (PCR) assays were carried out in triplicate (three technical Â three biological repeats) and the normalized expression values for all WT, Etv5 À/À Sertoli samples, and germ cells averaged. A relative quantitative fold change was determined using the DD Ct method. ABI Taqman gene expression assays used for specific transcripts were: Mm 01617100_m1 (Ccl12), Mm 00443113_m1 (Ccl7), Mm 00441260_m1 (Ccl9), Mm 00436451_g1 (Cxcl5), Mm 01216147_m1 (Ccr1), Mm00436304_m1 (Ret) and Hs 9999901_s1 (18S) (Applied Biosystems, Carlsbad, CA, www.appliedbiosystems.com).

Electromobility Shift Assays
Electromobility shift assays (EMSAs) were carried out using 3 0 biotinylated probes (Invitrogen, Carlsbad, CA, www.invitrogen.com) designed to include the three Ets binding sites (EBSs) found in the Ccl9 genomic sequence (Accession number AL596122) and a chemiluminescent kit (Cat. No. 20148, Thermo Scientific, Pittsburgh, PA, www.thermo.com). Sense and antisense primers to the following sequences (EBS in bold) were annealed and used as probes; EBS1-AATTTGAACTTTGGAAAGCCCCAAGTCACC, EBS2-GGAGAAATCCTTCTGGAAATATTTACTCTC and EBS3-CTCTAGTACCAGGGAATTTAACACCCTCTTC. EBS1 and EBS2 did not produce a shift (data not shown). Mutant sequence to EBS3 used as probes was: CTCTAGTAC-CACTCGAGTTAACACCCTCTTC. Nuclear extracts were prepared from TM4, a Sertoli cell line, transfected 48 hours with an Etv5 expression vector [42] (pErmFLAG, a generous gift from Dr. John Shannon, University of Cincinnati) or with control vector. Nuclear extracts were incubated in binding reaction buffer (Thermo Scientific, Pittsburgh, PA, www.thermo.com) with 2.5% glycerol, 50 mM KCl, 2 mM MgCl 2, NP-40 (0.04%), 50 ng/ll polydeoxyinosinepolydeoxycytidine acid (poly dI/dC), 1 mM dithiothreitol, 2 mM EDTA, and 20 fM biotinylated (B) probe at room temperature for 20 minutes. Competitor reactions contained nonbiotinylated (NB) probes at 200-fold excess. For supershift assays, the reaction mixture was incubated with 0.5 lg of ETV5 antibody (sc-22807, Santa Cruz Biotechnology, Santa Cruz, CA, www.scbt.com) for an additional 20 minutes. Screening of different concentrations of ETV5 antibody and NB probe was not performed in this study. Each reaction was fractionated on a 4% nondenaturing acrylamide gel.
For producing purified ETV5 protein, a cDNA fragment containing the ETV5 binding domain was generated using PCR and inserted into the pET-41a(þ) vector (Novagen/EMD Biosciences, Gibbstown, NJ, www.emdchemicals.com) according to manufacturer's instructions, and the sequence verified. The pET41a-ETV5 expression construct was transfected into E. Coli BL21 (DE3) cells. The cells were grown with 30 lg/ml kanamycin, and optimal expression was induced with 1 mM isopropyl beta-D-1-thiogalactopyranoside (IPTG). The ETV5 protein was purified using B-PER 6XHis purification kit (Thermo Scientific, Pittsburgh, PA, www.thermo.com) according to manufacturer's instructions. The ETV5 protein was incubated with binding buffer, B or NB probe to perform EMSAs.

RNA Interference
Ccr1 downregulation: Stem/progenitor spermatogonia from WT testes were seeded into 24-well plates at a concentration of 2 Â 10 4 in 200 ll of DMEM/F-12 with 10% Nu-Serum (BD Biosciences, Franklin Lakes, NJ, www.bdbiosciences.com). The cells were then transfected with Ccr1 or mock siRNAs (Invitrogen, Carlsbard, CA, www.invitrogen.com) using RNAiFect kit according to manufacturer's instructions (Qiagen, Valencia, CA, www.qiagen.com). RNA was isolated after 24 hours to assess the efficiency of gene silencing. For chemotaxis assays, stem/progenitor spermatogonia that were transfected with Ccr1 siRNA for 24 hours were then allowed to migrate toward WT Sertoli cells for 24 hours and the chemotactic index was assessed.

Statistical Analysis
For all experiments, including RNA interference studies, data were collected from three independent experiments, each done in triplicate. For the chemotaxis assays, data were obtained from three independent experiments, each done in duplicates or triplicates. Data were analyzed by one-way analysis of variance followed by Tukey-Kramer's multiple comparison tests. Statistical analyses were performed using Graph Pad Prism 4.0 (Graph Pad Software, Inc., San Diego, CA). Data were expressed as mean 6 SEM, and differences between groups were considered significant at p < .05.

Expression of Specific Chemokines Is Decreased in Sertoli Cells from Etv5 2/2 Mice
Our initial microarray analysis demonstrated a decrease in expression of chemokine-encoding genes [17]. Hence, we performed a chemokine profile using PCR array gene analysis of WT and Etv5 À/À Sertoli cells. The results established significant decreases in expression of 14 of the 74 chemokinerelated genes analyzed (data not shown). We confirmed by real-time PCR that expression of Ccl7, Ccl9, Ccl12, and Cxcl5 was decreased $50% (Fig. 1A). Expression of five housekeeping genes and other Sertoli genes (Fig. 1B) was unchanged in freshly isolated Etv5 À/À compared with WT Sertoli cells, indicating that decreases observed in chemokinerelated mRNA transcripts in Etv5 À/À Sertoli cells were specific, and not related to in vitro manipulation.

Etv5 2/2 Sertoli Cells Have Decreased Chemoattractive Ability
Stem/progenitor spermatogonia were isolated from 5-to 6day-old pups using the STAPUT method and MACS purification with GFRa1 antibody as described earlier. About 80% of the isolated cells expressed GFRa1 and all expressed GCNA1, indicating that the germ cell population was enriched for GFRa1 stem/progenitor spermatogonia ( Fig. 2A-2C). Type B spermatogonia, L/Z, and P spermatocytes were characterized based on the size and expression of mVH, a germ cell marker (Fig. 2D, 2E). RS were characterized based on the size.
The chemotactic index was defined as the number of germ cells migrating toward Sertoli cells. This index was significantly decreased (over 50%, p < .001) when GFRa1 þ germ cells migrated toward Etv5 À/À Sertoli cells compared with WT Sertoli cells (Fig. 3A-3C). The chemotactic index toward Etv5 À/À Sertoli cells was not statistically different from that toward the negative control, which was minimal media alone. This indicated that Etv5 À/À Sertoli cells had no demonstrable chemoattractive ability for stem/progenitor sper-matogonia. The chemotactic index toward the FBS positive control was approximately threefold greater than that toward the negative control. The GFRa1 þ cells that had migrated  . Chemoattractive ability of WT Sertoli cells is specific for stem/progenitor spermatogonia. Chemotaxis assays were performed to assess if Sertoli cells had a specific effect on GFRa1 þ cells compared with cKIT þ , type B spermatogonia, leptotene/zygotene spermatocytes, pachytene spermatocytes, or round spermatids. A mesoangioblast stem cell line (M25.2) was also used as a control. The chemotactic index of stem/progenitor spermatogonia (GFRa1 þ ) toward WT Sertoli cells was significantly greater than all other cell types. The chemotactic index of germ cell fractions other than GFRa1 þ cell fraction migrating toward WT Sertoli cells was not significantly different from that migrating toward minimal media alone. The results are representative of three independent experiments and are expressed as mean 6 SEM. Values with different letters were significantly different (p < .05). Abbreviations: GFRa1, GDNF familiy membrane receptor alpha 1; L/Z, leptotene/ zygotene spermatocytes; type B, type B spermatogonia; WT, wild type. also expressed mVH, confirming that these cells were germ cells (Fig. 3D).

Sertoli Cells Chemoattract Only Stem/Progenitor Spermatogonia and Not Other Germ Cells
The number of GFRa1 þ germ cells (stem/progenitor spermatogonia) migrating toward WT Sertoli cells was significantly higher (p < .01) compared with other germ cell fractions (Fig. 4). The migration of other germ cell fractions, including cKIT þ cells (differentiating spermatogonia) toward WT Sertoli cells was not significantly different from migration toward minimal media alone. A mesoangioblast cell line (M25.2, a generous gift from Dr. Suzanne Berry, University of Illinois) was also used as a control and there was no increase in the chemotactic index of these cells toward Sertoli cells (Fig. 4). These results indicate that the chemoattractive ability of Sertoli cells is a specific effect on stem/progenitor spermatogonia only.

Recombinant CCL9 Restores the Chemoattractive Ability of Etv5 2/2 Sertoli Cells
As Ccl9, Ccl12, and Cxcl5 were the chemokines showing the greatest (more than 60%) downregulation in Etv5 À/À Sertoli cells, we sought to rescue the decreased ability of Etv5 À/À Sertoli cells to chemoattract stem/progenitor spermatogonia using these chemokines. Recombinant proteins were added at a final concentration of 50 ng/ml, individually or in combination, to Etv5 À/À Sertoli cells in the lower chamber of the transwell units and chemotaxis assays performed. Only recombinant CCL9 was able to rescue the chemoattractive ability of Etv5 À/À Sertoli cells to WT levels as there was no statistically significant difference between the chemotactic index of stem/progenitor spermatogonia toward untreated WT and Etv5 À/À Sertoli cells that were treated with recombinant CCL9 (Fig. 5). Addition of CXCL5 to Etv5 À/À Sertoli cell cultures significantly increased the chemoattractive ability above the Etv5 À/À levels, but did not fully restore it to WT levels. Recombinant CCL12 did not show any effect (Fig. 5). The combination of all three chemokines also significantly increased the chemoattractive ability above Etv5 À/À levels, but did not restore it to that seen with WT Sertoli cells (Fig. 5).

Expression of CCL9 and CCR1 in the Testis
As CCL9 was the most active chemokine tested, we confirmed its expression in WT Sertoli cells by immunohistochemistry (Fig. 6A) and its downregulation in Etv5 À/À mice (Fig. 6B). The receptor for CCL9, CCR1, was expressed in gonocytes, and later in undifferentiated spermatogonia, but not in Sertoli cells (Fig. 6C, 6D). This pattern of expression was confirmed by real-time PCR (data not shown). Western blot analysis of WT Sertoli cell extracts indicated that these cells expressed CCL9 and that there was a downregulation of CCL9 protein expression starting 10 days after birth with no expression of the protein after 35 days (Fig. 6E). Therefore, a functional interaction between CCL9 and CCR1 in the testis is likely, but might be restricted to the proliferative period of Sertoli cells.

ETV5 Interaction with the Ccl9 Promoter
To assess if there was interaction between ETV5 and the Ccl9 promoter, EMSAs were performed. The Sertoli cell line TM4 was transfected with pErmFLAG [42], which resulted in a 10,000-fold increase of mRNA expression of Etv5 compared with nontransfected controls. Nuclear extracts prepared from these cells were incubated with three biotinylated putative ETV5-binding site sequences (EBS1-3) in the Ccl9 promoter region. Gel shifts were observed when nuclear extracts were combined with biotinylated EBS3 sequences (B EBS3), indicating protein-DNA interaction (Fig. 6F). A 200-fold excess of nonbiotinylated EBS3 probes (NB EBS3) substantially, but not completely, outcompeted protein-DNA binding, suggesting that other factors might be present in the complex. Incubation with ETV5 antibody seemed to alter protein-DNA binding, rather than producing a consistent supershift, which is in agreement with the data of Lin et al. [42]. Shifts of protein-DNA complexes were not observed when nuclear extracts from TM4 cells transfected with the control vector were incubated with B EBS3 or when mutant EBS3 sequences were incubated with cells transfected with the Etv5 construct (Fig.  6F). Purified ETV5 protein also produced a gel shift that was almost completely outcompeted with the NB EBS3 probe, confirming the data obtained with nuclear extracts. Taken together, these results indicate that ETV5 is a good candidate regulator of CCL9 expression.

Downregulation of Ccr1 Induces a Reduction in the Migration of Stem/Progenitor Spermatogonia
Ccr1 was downregulated in primary stem/progenitor spermatogonia $70% using RNA interference (Fig. 7A). Stem/progenitor spermatogonia downregulated for Ccr1 were subjected to chemotaxis assays. There was a 14% reduction in the chemotactic index of these cells toward WT Sertoli cells, which was statistically significant (Fig. 7B). Although there was a reduction in migration of these cells, compensatory mechanisms by other chemokines or receptors probably limited this effect.

DISCUSSION
It has been recently demonstrated that the absence of ETV5 in SSCs decreases their proliferation [19,21] and contributes to the decline in their numbers [20]. However, decreased Figure 5. Chemokines increase the chemoattractive ability of Etv5 À/À Sertoli cells. Recombinant proteins CCL9, CXCL5, and CCL12 were added individually or in combination along with Etv5 À/À Sertoli cells. CCL9 and CXCL5 significantly increased the chemoattractive ability of Etv5 À/À Sertoli cells. Addition of CCL9 was able to restore the chemoattractive ability of Etv5 À/À Sertoli cells to WT levels. Adding CCL12 did not increase the chemoattraction of these cells. The combination of all three chemokines significantly increased the chemoattractive ability of Etv5 À/À Sertoli cells, but did not restore it to WT levels. Results are representative of three independent experiments and are expressed as mean 6 SEM. Values with different letters were significantly different (p < .05). Abbreviations: CCL9, C-C-motif ligand 9; CXCL5, C-X-C motif ligand 5; CCL12, C-C-motif ligand 12; Etv5, Ets variant gene; WT, wild type. proliferation would reduce, but not eliminate stem/progenitor spermatogonia in the seminiferous tubules, and the critical question is how the loss of ETV5 ultimately results in a total loss of stem/progenitor spermatogonia.
The present study sought to identify and determine the functional significance of chemokines produced by Sertoli cells that are regulated by ETV5. Our results indicate that downregulation of specific chemokines such as CCL9 and CXCL5 in Etv5 À/À Sertoli cells leads to impairments in the ability of these Sertoli cells to chemoattract stem/progenitor spermatogonia. This may be a key aspect of the mechanism by which loss of ETV5 protein leads to a progressive loss of germ cells in the Etv5 À/À testis. Downregulation of chemokines is specific, because the expression of other genes is not affected in freshly isolated Etv5 À/À Sertoli cells. If a loss of chemokine expression had occurred as a result of subsequent culture conditions, and not of gene ablation, then both WT and Etv5 À/À Sertoli cells would have been affected, without changes in differential expression.
The stem cell niche plays an essential role in maintaining the balance between self-renewal and differentiation of stem cells. As stem/progenitor spermatogonia have direct cell-tocell contact only with Sertoli cells in the seminiferous tubules, and Sertoli cells are the major source of ETV5 in the postnatal and adult testis [5,17], changes in Etv5 À/À Sertoli cells are likely involved in the loss of germ cells in these knockout mice. The loss of stem/progenitor spermatogonia in the Etv5 À/À testes starts between day 4 and day 8 after birth [20], and by day 16 the testis weight is already 35% smaller than WT [20]. In the WT testis, CCL9 is expressed at least until day 10, if not longer (but not after day 35; Fig. 6E). Therefore, it is plausible that lack of CCL9 in Etv5 À/À mice might impair stem/progenitor spermatogonia maintenance and migration shortly after birth. In mice, proliferation of Sertoli cells (creating new niches) occurs until day 16. As CCL9 is present during this period, and that we demonstrate migration using spermatogonia and Sertoli cells aged 5-8 days, it is possible that chemotaxis ensures stem/progenitor spermatogonia filling the niches [5]. A reduction in the migration of spermatogonia to fill new niches formed by proliferating Sertoli cells might result in Sertoli cell only regions, as observed in the Etv5 À/À testis, and could contribute to the overall decline in testis weight. Similarly, the lack of colonization of Etv5 À/À seminiferous tubules by WT stem/progenitor spermatogonia [18] could be attributed, at least in part, to a lack of proper chemokine expression.
Previous literature indicates that chemokine signaling may be involved in maintaining stem cells and progenitors in other niches. For example, chemokines secreted by the hematopoietic stem cell (HSC) niche are putative regulatory factors of HSCs [24,43]. The aorta-gonad-mesonephros region generates HSCs and serves as the HSC microenvironment. In particular, in this system chemokines such as CCL9 and CCL3 are produced by the cells comprising the niche, and these chemokines regulate and enhance the repopulating ability of HSCs in vitro [24,44]. Consistent with a role of niche cells signaling through chemokines to stem cells, neural stem cells (NSCs) express CXCR4, the cognate receptor for the chemokine CXCL12. CXCL12 treatment of quiescent NSCs enhances proliferation, promotes chain migration, and activates intracellular molecular pathways mediating recruitment of these cells toward an injury [25]. Therefore, chemokines and their receptors might play a similar role in spermatogenesis as they do in hematopoiesis and neurogenesis, where they regulate homing/retention of stem cells in the surrounding microenvironment and provide signals for cell growth and differentiation.
Although Sertoli cell chemoattraction or a role of Sertoli cell chemokines in the postnatal testis has not been previously demonstrated, the chemokine CXCL12 is essential for migration of PGCs to the gonadal ridges, and its receptor, CXCR4, is expressed by these cells [26]. PGC and gonocyte migration is not affected in Etv5 À/À mice as stem/progenitor numbers were comparable in neonatal Etv5 À/À and WT mice [20]. This is consistent with our present results that CXCL12 was not among the chemokines decreased in Etv5 À/À testes.
Chemotaxis assays were performed to study the functional role of chemokines on stem/progenitor spermatogonia. These studies clearly indicated that Sertoli cells were chemoattractive for GFRa1 þ stem/progenitor spermatogonia, the first demonstration that Sertoli cells have a chemotactic effect on these cells and that other germ cell populations such as type B spermatogonia, L/Z, P spermatocytes, and RS were not chemoattracted by Sertoli cells. In addition, the chemoattractive ability of Sertoli cells was significantly decreased in Etv5 À/À mice, indicated by a reduced number of stem/progenitor spermatogonia migrating toward Etv5 À/À Sertoli cells compared with their age-matched WT controls. Our results demonstrate that two of the three chemokines identified as being reduced in Etv5 À/À Sertoli cells were chemoattractive for stem/progenitor spermatogonia (CCL9 and CXCL5). However, our results also indicated that adding a cocktail of the three recombinant chemokines (CCL9, CCL12, and CXCL5) to Etv5 À/À Sertoli cells did not restore their chemoattractive ability to WT levels. Therefore, the cell migratory behavior might vary depending on the concentration or the combination of the chemoattractants used and the combination of chemokines may synergistically increase or decrease the migratory behavior of stem cells [45]. In the present study, CCL12 did not increase the chemoattractive ability of Etv5 À/À Sertoli cells, suggesting that this chemokine does not have a chemotactic role postnatally. The inability of the combination of all chemokines to fully restore values of Etv5 À/À Sertoli cells back to WT levels, indicates that there might be changes in Etv5 À/À Sertoli cells other than reduction in the chemokines examined that affected chemoattraction. These are likely other growth factors, adhesion molecules, and/or proteolytic enzymes produced by Sertoli cells that have a role in the complex interactions between Sertoli cells and stem/ progenitor spermatogonia. The behavior and activity of chemokines is also controlled in a complex manner by several mechanisms that include synergistic interactions, glycosaminoglycans binding properties, and interactions with accessory molecules [46]. These factors, absent in our culture conditions, might influence their mobilization, transport, and rate of degradation, thereby determining the sites and duration of their action.
As CCL9 was sharply reduced in Etv5 À/À Sertoli cells and attracted stem/progenitor spermatogonia at the highest level in vitro, we further studied the expression patterns of CCL9 and its receptor CCR1 in the testis. Immunohistochemical localization of CCL9 confirmed its expression in Sertoli cells, consistent with our PCR-array and real-time PCR data. Similarly, CCL7 and its receptor, which are also expressed in Sertoli cells and spermatogonia, might also function in the maintenance of stem/progenitor spermatogonia within the niche [1]. Though there was strong CCL9 protein expression in seminiferous tubules neonatally, it decreased to undetectable levels between day 10 and day 35, supporting the hypothesis that proliferating Sertoli cells create new niches and stem/progenitor spermatogonia migrate along the basement membrane to occupy them [5]. Furthermore, our EMSA data indicated that ETV5 is a good candidate regulator of Ccl9 expression. EMSAs performed with different concentrations of NB probes or different ETV5 antibody concentrations will clarify in the future whether other factors are present in the protein-DNA complex, and the degree of specificity of the interaction.
As expected, the receptor for CCL9, CCR1, was expressed by undifferentiated spermatogonia. Microarray analysis of Thy1 þ testis cell populations also showed that expression of Ccr1 and many other chemokine receptors was increased more than 10-fold in these cells [47] suggesting the importance of chemokine signaling in maintenance of stem/progenitor spermatogonia in their microenvironment. In the chemotaxis assays performed with Ccr1 siRNA spermatogonia, there was only a 14% reduction in the chemotactic index suggesting a redundancy of chemokine signaling by either the receptors or the ligands. Although chemokines and their receptors have been shown to be crucial for stem cell homing and maintenance in the hematopoietic and neural systems, targeted deletions of their genes usually only led to mild phenotypic defects [48][49][50]. Similarly, Ccr1 À/À mice are fertile and have very minor phenotypes in other organs [49] suggesting a compensatory mechanism in chemokine signaling. Hence, transplantation studies of germ cells that are knocked out for a single chemokine receptor will not give an insight into the complex signaling mechanism of chemokines produced by Sertoli cells in maintaining stem/progenitor spermatogonia in their microenvironment.

CONCLUSION
Our data show for the first time that Sertoli cells are chemoattractive for stem/progenitor spermatogonia in the prepubertal testis. We demonstrated that ETV5 is responsible for this process, possibly by regulating the production of chemokines such as CCL9 by Sertoli cells (supporting information Figure  S1). However, our data also indicate that more than one chemokine is involved. A decrease in chemokine production by Etv5 À/À Sertoli cells, and the resulting decline of their chemotactic activity reduce the migration of stem/progenitor spermatogonia to newly formed Sertoli cells/niches. This helps to explain the loss of germ cells in Etv5 À/À mice.