The maintenance of spermatogonial stem cells (SSCs) provides the foundation for life-long spermatogenesis. Although glial-cell-line-derived neurotrophic factor and fibroblast growth factor 2 are crucial for self-renewal of SSCs, recent studies have suggested that other growth factors have important roles in controlling SSC fate. Because β-catenin-dependent Wnt signaling promotes self-renewal of various stem cell types, we hypothesized that this pathway contributes to SSC maintenance. Using transgenic reporter mice for β-catenin-dependent signaling, we found that this signaling was not active in SSCs in vitro and in most spermatogonia in vivo. Nonetheless, a pan-Wnt antagonist significantly reduced SSC activity in vitro, suggesting that some Wnt molecules exist in our serum-free culture system and contribute to SSC maintenance. Here, we report that Wnt5a promotes SSC activity. We found that Wnt5a-expressing fibroblasts supported SSC activity better than those not expressing Wnt5a in culture, and that recombinant Wnt5a stimulated SSC maintenance. Furthermore, Wnt5a promoted SSC survival in the absence of feeder cells, and this effect was abolished by inhibiting the Jun N-terminal kinase cascade. In addition, Wnt5a blocked β-catenin-dependent signaling. We detected the expression of Wnt5a and potential Wnt5a receptors in Sertoli cells and stem/progenitor spermatogonia, respectively. These results indicate that Wnt5a is a cell-extrinsic factor that supports SSC self-renewal through β-catenin-independent mechanisms.
Spermatogonial stem cells (SSCs) are a rare population in testes, but are responsible for life-long sperm production. These cells are present on the basement membrane of the seminiferous tubules of the testis. Recent studies suggest that SSC activity is regulated in a microenvironment that is composed of Sertoli cells, the extracellular matrix and the vasculature network (de Rooij, 2009; Shinohara et al., 1999; Yoshida et al., 2007). SSC self-renewal is known to be stimulated by glial-cell-line-derived neurotrophic factor (GDNF) and fibroblast growth factor 2 (FGF2), which are both expressed by Sertoli cells in testes (Oatley and Brinster, 2008). In vitro, GDNF and FGF2 allow for long-term SSC maintenance and expansion (Kubota et al., 2004b; Oatley and Brinster, 2008), attesting to their importance as cell-extrinsic effectors of SSC activity. However, recent studies have indicated that colony-stimulating factor 1 (CSF-1) promotes SSC self-renewal in vitro as a cell-extrinsic factor (Kokkinaki et al., 2009; Oatley et al., 2009), demonstrating that other factors are involved in controlling SSC activity.
Wnt proteins are lipid-modified, secreted morphogens that control a variety of cell processes such as embryogenesis, cell proliferation and differentiation, and stem cell regulation (Logan and Nusse, 2004; Reya and Clevers, 2005). Several Wnt ligands and signaling components are known to be expressed in the testis (Jeays-Ward et al., 2004; Kimura et al., 2006; Shackleford and Varmus, 1987), but no clear role during spermatogenesis has been identified. In this study, we examined the potential involvement of Wnt signaling in regulating SSC self-renewal.
Wnt signaling is mediated by a wide range of intracellular signaling cascades, which are roughly categorized into β-catenin-dependent and -independent pathways (Logan and Nusse, 2004). In the β-catenin-dependent Wnt pathway (termed ‘β-catenin pathway’ hereafter), Wnt ligands bind to the receptor, Frizzled (Fzd), and co-receptor, low-density lipoprotein receptor-related protein 5 or 6 (LRP5/6) (Logan and Nusse, 2004). Activation of this pathway inhibits the action of glycogen synthase kinase-3β (GSK3β) and prevents β-catenin degradation. β-Catenin then translocates to the nucleus and induces target gene expression together with transcription factors of the T cell factor/lymphoid enhancer factor (TCF/LEF) family.
β-Catenin signaling has been known to promote self-renewal of many stem cell types, including those of embryonic and postnatal origins (Kalani et al., 2008; Korinek et al., 1998; Miyabayashi et al., 2007; Reya et al., 2003). Hence, this pathway was initially regarded as a general promoter of stem cell self-renewal (Reya and Clevers, 2005). Recent studies, however, indicate that the action of β-catenin signaling is more complicated. For instance, this signaling directs embryonic and adult progenitors in the skin to differentiate (Nguyen et al., 2006) and aging muscle satellite cells and neural crest stem cells toward fibrogenesis (Brack et al., 2007) and a sensory neural fate (Lee et al., 2004), respectively. In the germ line, activation of β-catenin signaling is detrimental for development of primordial germ cells (Kimura et al., 2006). Therefore, the response of stem cells to the β-catenin pathway is cell-type specific.
In contrast to the β-catenin pathway, the mechanisms of β-catenin-independent Wnt signaling are diverse and have not been well elucidated. It is known that this signaling can be mediated by G-proteins and an intracellular Ca2+ flux, thereby activating downstream effectors, such as calmodulin-dependent kinase II (CaMKII) and protein kinase C (PKC) (van Amerongen et al., 2008). β-Catenin-independent signaling also acts through Jun N-terminal kinases (JNKs) to regulate planar cell polarity in Drosophila or convergent-extension movements during Xenopus gastrulation (van Amerongen et al., 2008). In mammalian cells, this signaling can control cell polarity and migration through the JNK cascade (Schlessinger et al., 2007). In general, the activation of β-catenin-independent pathways is known to inhibit the β-catenin pathway (Ishitani et al., 1999; Mikels and Nusse, 2006).
Recently, Wnt5a has been reported to control the activity of various stem cell types in a β-catenin-independent manner. For instance, it stimulates repopulation potential of hematopoietic stem cells while inhibiting β-catenin signaling (Nemeth et al., 2007) and supports multipotentiality of mesenchymal stem cells (Bilkovski et al., 2010). Studies using vertebrates and invertebrates have indicated that β-catenin-independent Wnt5a signaling can be mediated by several receptors, such as Fzd3, Fzd5 and Fzd7 (He et al., 1997; Kawasaki et al., 2007; Medina et al., 2000). In mammalian cells, it has been shown that Ror2, a receptor tyrosine kinase, can also mediate Wnt5a signaling through JNK signaling (Mikels and Nusse, 2006).
In this study, we initially hypothesized that β-catenin signaling promotes SSC maintenance. However, results using SSC culture indicated the contrary. Further in vitro analyses showed that Wnt5a supported SSC maintenance in a β-catenin-independent manner and enhanced the survival of stem/progenitor spermatogonia. These Wnt5a effects were abolished by the inhibition of JNK signaling. Additionally, we localized Wnt5a expression to Sertoli cells in mouse testes and detected Wnt5a receptors on the cell surface of SSCs. Thus, we identify Wnt5a as an extrinsic regulator of SSC self-renewal.
β-Catenin signaling is not active in SSCs
To examine whether β-catenin signaling is involved in controlling SSC activity, we initiated SSC culture using SSC-enriched cells freshly prepared from testes of transgenic reporter mice (TCF/LEF-lacZ mice). These mice carry the lacZ reporter gene linked to β-catenin–TCF/LEF binding sites, allowing for faithful monitoring of β-catenin signaling activation (Mohamed et al., 2004a). The cultured cells formed three-dimensional aggregates of spermatogonia (Kubota et al., 2004b; Yeh et al., 2007), termed ‘clusters’ hereafter. lacZ expression was observed in a subpopulation of cluster cells (Fig. 1A), demonstrating that the cluster is a functionally heterogeneous cell community, composed of at least two cell types: β-catenin signaling-positive and signaling-negative cells.
β-Catenin-signaling cells were not observed on day 1 of culture, but appeared by day 3 and increased in number thereafter (Fig. 1B). A similar increase in signaling cells was seen when established clusters (>five passages) were used (supplementary material Fig. S1A). To assess whether SSCs were included in β-catenin-signaling or non-signaling cells, we isolated the two cell populations from established TCF/LEF-lacZ clusters using a vital fluorescent substrate of β-galctosidase and FACS. All cluster cells were found in Fraction II with a profile of side-scatterlow and relative homogeneity in size (Fig. 1D), as reported previously (Takubo et al., 2008), and which we confirmed (supplementary material Fig. S1B,C). Fraction I (side-scatterhi cells) was feeder cells. Fraction II was then separated into β-catenin-signaling and non-signaling cells (Fractions IV and III, respectively). Quantitative PCR analyses for markers of undifferentiated (Oct4, Plzf, Ngn3) and differentiating spermatogonia (Kit) indicated that both fractions similarly expressed the markers examined (supplementary material Fig. S1D). However, there was a trend showing slightly decreased expression of undifferentiated spermatogonial markers in Fraction IV, although no significance was detected. When the SSC activity of each cell population was examined by spermatogonial transplantation, nearly all SSC activity was detected in Fraction III. Virtually none was found in Fraction IV (Fig. 1F, supplementary material Fig. S1E), which was not due to cell death, because the vast majority of cells in this fraction were viable (supplementary material Fig. S1F). The low SSC activity in Fraction I was attributed to contamination. These results indicate that β-catenin signaling is not active in SSCs.
Next, we activated β-catenin signaling in cluster cells using lithium chloride, a potent inhibitor of GSK3β (Klein and Melton, 1996). We then quantified β-catenin signaling-positive cells and measured SSC activity by spermatogonial transplantation. Lithium chloride treatment increased numbers of signaling-positive cells tenfold, but significantly decreased SSC numbers (Fig. 1C). Hence, activation of β-catenin signaling reduced SSC activity in vitro.
β-Catenin signaling is activated in differentiating germ cells in vivo
To examine the activation of β-catenin signaling in vivo, we analyzed the expression of the lacZ reporter gene in TCF/LEF-lacZ mouse testes at different ages during postnatal development (Fig. 2). In mouse testes, only spermatogonia exist for the first week after birth, meiotic cells appear around 10 days post partum (d.p.p.), and haploid cells are formed around 18 d.p.p.; the first spermatozoa are found around 35 d.p.p. (McCarrey, 1993). β-Catenin-signaling cells were not found until 12 d.p.p. (Fig. 2A,B; supplementary material Fig. S2). By 1 month of age and throughout adulthood (Fig. 2C,D), signaling cells became numerous and were observed in the adluminal compartment, where meiotic and haploid cells reside. Occasionally, signaling-positive spermatogonia were found in adult testes (Fig. 2D, inset). No reporter activation was detected in the interstitial space. To verify our observations in TCF/LEF-lacZ reporter mouse testes, we examined β-catenin protein expression in adult testes using immunofluorescent staining. We detected expression in most cells along the basal layer of the seminiferous epithelium and β-catenin expression was mostly restricted to the membrane and cytoplasm in these cells (supplementary material Fig. S2E). This finding supports our result that β-catenin signaling is not activated in the cells of the basal layer and is consistent with previous results described in rat testes (Lee, N. P. et al., 2003). We also observed diffuse cytoplasmic or nuclear β-catenin expression in cells of the adluminal compartment. Occasionally, we observed nuclear β-catenin expression in the basal layer (supplementary material Fig. S2E, inset), which corresponds to rare β-catenin signaling cells observed in the basal compartment of the seminiferous epithelium in reporter mouse testes (Fig. 2D).
We next examined the pattern of reporter gene activation in TCF/LEF-lacZ mouse testes using experimental cryptorchidism and orchidopexy, an in vivo regeneration model of spermatogenesis (Nishimune and Aizawa, 1978). In cryptorchid testes, where spermatogenesis is disrupted, reporter expression was virtually abolished (Fig. 2E), even though spermatogonia and Sertoli cells were present. After orchidopexy, which induces spermatogenic regeneration, reporter gene expression was restored in the adluminal compartment (Fig. 2F). These results collectively indicate that β-catenin signaling is activated in differentiating male germ cells but not in most spermatogonia.
Wnt5a promotes SSC maintenance as a cell-extrinsic factor in vitro
The activation of β-catenin signaling in some cluster cells (Fig. 1) suggests that Wnt ligands are present in our SSC culture system. This is notable, because our culture medium is serum free and does not contain Wnt ligands (Kubota et al., 2004b). If Wnt ligands exist in the culture, therefore, they must be expressed by feeder cells and/or cluster cells. To test this, we first cultured clusters derived from β-actin–GFP (B6GFP) mouse cells (Okabe et al., 1997) and separated them from feeder cells using FACS. We examined the expression of Wnt1, Wnt2b and Wnt3a, all of which are classically categorized into a β-catenin-dependent class of Wnt ligands, as well as those of Wnt5a, Wnt6 and Wnt11, which are often recognized as the β-catenin-independent class (van Amerongen et al., 2008). Using RT-PCR, we detected transcripts of Wnt2b, Wnt5a and Wnt11 in feeder cells, but only Wnt1 transcripts in cluster cells (Fig. 3A).
We next asked whether Wnt ligands affect SSC maintenance by using two soluble antagonists of Wnt signaling, Dickkopf-1 (Dkk1) and secreted frizzled-related protein 1 (sFRP1). We used these antagonists because Dkk1 specifically blocks β-catenin signaling by binding to LRP5/6, whereas sFRP1 inhibits both β-catenin-dependent and -independent signaling by binding Wnt ligands (Logan and Nusse, 2004; Schlessinger et al., 2007). When Dkk1 was applied to TCF/LEF-lacZ cluster cultures, cluster numbers did not change, but numbers of β-catenin signaling cells markedly declined in a dose-dependent manner (Fig. 3B). Because cluster numbers correlate with SSC numbers (Yeh et al., 2007), these results suggest that inhibition of β-catenin signaling did not affect SSC activity. However, sFRP1 dose-dependently reduced cluster numbers (Fig. 3C). Collectively, these results imply that β-catenin-independent Wnt signaling is involved in SSC maintenance.
Wnt5a is most often associated with β-catenin-independent signaling and can also block β-catenin signaling (Mikels and Nusse, 2006). Our data showed that β-catenin signaling was not active in SSCs and that Wnt5a was expressed by feeder cells in our SSC culture. These observations led us to hypothesize that Wnt5a contributes to SSC maintenance in vitro. To test this, we competitively added recombinant Wnt5a at increasing doses in the presence of sFRP1 and quantified clusters in a 6-day culture (Fig. 3D). We found that Wnt5a dose-dependently restored cluster numbers that had been affected by sFRP1. Next, we cultured clusters on a feeder layer of L-fibroblasts stably transfected with Wnt5a (L-Wnt5a) or parental L-cells, which lack Wnt5a transcripts (Mohamed et al., 2005) (our results). SSC activity was determined by cluster numbers and spermatogonial transplantation. We detected a 2.8-fold increase in cluster numbers with L-Wnt5a feeders (Fig. 3E). Similarly, transplantation results showed a 2.2-fold increase in SSC activity with L-Wnt5a feeders (Fig. 3E), indicating that Wnt5a promotes SSC maintenance as a cell-extrinsic factor in vitro.
Because we used feeder cells in all the above experiments, Wnt5a could have affected SSC maintenance indirectly through feeder cells. Thus, clusters were removed from feeder cells by gentle pipetting, by which cluster cells are recovered with a >90% purity (Oatley et al., 2006) (our data), and were cultured on Matrigel with or without Wnt5a for 4 days. No clusters emerged under these conditions. SSC activity was then measured by transferring these cells to our standard SSC culture conditions (cluster-formation assay). We found that SSC activity was significantly higher in the presence of Wnt5a (Fig. 3F). Likewise, spermatogonial transplantation showed that Wnt5a increased SSC numbers 3.8-fold, compared with levels in cultures without Wnt5a (Fig. 3G), demonstrating that SSC maintenance by Wnt5a is not indirect through feeder cells.
β-Catenin-independent signaling mediates Wnt5a action
When TCF/LEF-lacZ clusters were cultured on feeder cells with added Wnt5a for 6 days, the number of β-catenin-signaling cells significantly declined (Fig. 4A), indicating that Wnt5a suppresses β-catenin signaling in clusters and suggesting that Wnt5a affects SSC maintenance in a β-catenin-independent manner. We thus blocked β-catenin-independent pathways using inhibitors of the CaMKII, PKC, G-protein and JNK cascades and assessed the effects on cluster formation. Cluster cells were exposed to each inhibitor for the first 3 days of culture, and clusters were quantified on day 6. The results showed that although cluster numbers did not change with inhibition of CaMKII, PKC and G-protein signaling, they declined significantly with inhibition of JNK signaling using a cell-permeable competitive peptide (JNK Inhibitor III) and a small molecule inhibitor (SP600125) in a dose-dependent manner (Fig. 4B,C, filled bars). Furthermore, when clusters treated with SP600125 were passaged and cultured under our standard SSC culture condition, numbers of secondary clusters declined dose-dependently (Fig. 4C, open bars). To eliminate the possibility of indirect effects through feeder cells, we cultured cluster cells on Matrigel with Wnt5a alone or Wnt5a plus a JNK inhibitor and quantified SSCs using spermatogonial transplantation. The data showed that the inhibition of JNK signaling abolished the Wnt5a-induced increase in SSC numbers to control levels (Fig. 4D, open bars; supplementary material Fig. S3A). The cluster-formation assay generated similar results (Fig. 4D, filled bars), suggesting that Wnt5a promotes SSC maintenance in vitro through the JNK signaling pathway.
To confirm that Wnt5a activates the JNK cascade, we performed western blot analysis following short-term feeder-free culture of cluster cells. The data showed that Wnt5a-treatment led to a significant increase in levels of the activated (phosphorylated) form of JNK (JNK-P), despite seemingly high basal levels of JNK-P (supplementary material Fig. S3B). This apparent basal level could be due to insulin in our culture medium (Lee, Y. H. et al., 2003). Addition of the inhibitor diminished these JNK-P levels below basal levels and by ~90% of activated levels (supplementary material Fig. S3B). Total JNK levels were unaffected by these treatments. We further examined whether JNK activity is found in spermatogonia in vitro and in vivo by assessing the expression of JNK-P in clusters and adult testis. Immunofluorescence for JNK-P showed diffuse expression throughout all cluster cells with increased intensity restricted to a small subset of cells within clusters (supplementary material Fig. S3C). A similar pattern of expression was also observed in testis as diffuse JNK-P expression throughout the cells of adult seminiferous tubules and increased expression restricted to the cells along the basal compartment (supplementary material Fig. S3D). Therefore, these results collectively demonstrate that Wnt5a signaling activates the JNK cascade in spermatogonia.
Wnt5a inhibits apoptosis through β-catenin-independent signaling
When cluster cells were cultured feeder-free with Matrigel for 4 days, we observed a trend that more cells were detected in the presence of Wnt5a, compared with untreated controls, although the differences were not statistically significant (Fig. 5A). This suggested that Wnt5a might have affected SSC maintenance by altering cell proliferation and/or survival. Hence, we first analyzed cell cycle profiles of cluster cells cultured on Matrigel with or without Wnt5a for 4 days, using flow cytometry. The results showed that the majority of cluster cells were in the G0–G1 phase and that Wnt5a did not alter the percentage of cycling cells (Fig. 5B). Similar results were obtained when analyzed at 2 days in culture (supplementary material Fig. S3E). Thus, Wnt5a did not affect the cell cycle profile of cluster cells under these conditions. Because the induction of quiescence is known to support hematopoietic stem cell activity (Nemeth et al., 2007), we analyzed cluster cells at G0 by detecting Ki-67-negative cells in the G0–G1 fraction. Flow cytometric analyses indicated that ~13% of G0–G1 cells on average were negative for Ki-67, which was not affected by Wnt5a (Fig. 5C). Hence, Wnt5a apparently does not induce quiescence in cluster cells.
To examine the effect of Wnt5a on cell death, cluster cells were cultured on Matrigel with Wnt5a, and 48 hours later, subjected to TUNEL analysis. Flow cytometric analyses showed that 52.6±9.2% of cluster cells were apoptotic in the absence of Wnt5a in contrast to 21.9±2.3% in its presence (Fig. 5D), demonstrating that Wnt5a inhibits apoptosis of cluster cells.
Finally, we asked whether Wnt5a inhibits apoptosis through JNK signaling. Clusters were cultured on Matrigel and treated with Wnt5a alone, a JNK inhibitor alone, or both. The proportion of apoptotic cells was determined using TUNEL staining and flow cytometry. Data showed that Wnt5a reduced the number of apoptotic cells whereas the inhibition of JNK signaling negated this effect (Fig. 5E). Importantly, the inhibitor alone did not affect apoptosis, compared with control levels. Because the inhibition of JNK signaling in feeder-free cultures abolished the effect of Wnt5a on both SSC activity (Fig. 4D) and cluster cell survival (Fig. 5E), these results suggest that Wnt5a promotes SSC maintenance by inhibiting apoptosis through JNK signaling.
Wnt5a is detected in Sertoli cells in testes
To gain insight into the potential involvement of Wnt5a in regulating SSCs and spermatogonia in vivo, we examined its expression in testes during postnatal development using RT-PCR (Table 1). Wnt5a transcripts were detectable in all developmental stages examined as well as in adult cryptorchid testes (Table 1 and Fig. 6B). Because neonatal and cryptorchid testes contain no meiotic and haploid germ cells, these results suggest that Wnt5a is expressed in somatic cells and/or spermatogonia in mouse testes. To identify the cell types expressing Wnt5a, we used in situ hybridization in neonatal testes. Wnt5a-expressing cells were observed only in the seminiferous epithelium and clearly in Sertoli cells (Fig. 6A), in good agreement with a previous report that suggested Sertoli cells express Wnt5a in mouse testes (O'Shaughnessy et al., 2007).Wnt5a transcripts were also detected in two Sertoli cell lines (Fig. 6B). The intimate contact of spermatogonia with Sertoli cells made it difficult to determine the staining in spermatogonia. Hence, we FACS-purified spermatogonia from testes of transgenic Oct4–GFP mice, which express GFP specifically in spermatogonia at least up to 7 d.p.p. (Yoshimizu et al., 1999) (and our results). The RT-PCR analyses of GFP+ spermatogonia at 6–7 d.p.p. showed that Wnt5a transcripts were undetectable in spermatogonia (Fig. 6B). These results indicate that Sertoli cells express Wnt5a in mouse testes.
SSCs express Wnt5a receptors
If Wnt5a acts on SSCs, Wnt5a receptors must be expressed by SSCs. We thus examined the expression of various Wnt receptors in cluster cells. RT-PCR analyses using FACS-purified B6GFP clusters detected transcripts of Fzd3, Fzd5, Fzd7 and Ror2 (supplementary material Fig. S4A). Interestingly, all these receptors are known for their ability to transduce β-catenin-independent Wnt5a signaling (He et al., 1997; Kawasaki et al., 2007; Medina et al., 2000; Mikels and Nusse, 2006). We also detected transcripts encoding LRP5 and LRP6 (supplementary material Fig. S4A), which mediate β-catenin signaling (Pinson et al., 2000) (see Discussion). To examine the protein expression of Fzd3, Fzd5, Fzd7 and Ror2 in clusters, we used immunofluorescent staining and flow cytometry. Immunostaining showed that all cells in clusters expressed Fzd5, Fzd7 and Ror2 (Fig. 7A). Flow cytometric analyses supported these results; histograms of staining intensity showed all cluster cells were positively stained, compared with the negative controls (Fig. 7A). A similar profile was observed when cluster cells were stained for integrin α6, an established SSC marker (Fig. 7B). Thus, Fzd5, Fzd7 and Ror2 are expressed by all cluster cells, including SSCs.
By contrast, Fzd3 staining was apparently heterogeneous in clusters, because Fzd3-positive and Fzd3-negative cells were observed (Fig. 7C). Double staining for TCF/LEF-lacZ activity and Fzd3 expression indicated that Fzd3 expression on cluster cells did not correlate with TCF/LEF-lacZ expression (supplementary material Fig. S4B). With flow cytometry, we could not clearly resolve these two populations (Fzd3-negative and Fzd3-positive) (Fig. 7C). Therefore, to determine Fzd3 expression on SSCs, we isolated Fzd3+ cells from adult mouse testes and measured their SSC activity using spermatogonial transplantation. To this end, SSC-enriched testis cells were prepared from B6ROSA adult mice using Percoll-based cell separation (Kubota et al., 2004a) (Fig. 7D; supplementary material Fig. S4C), followed by immunomagnetic cell sorting for β2-microglobulin-negative cells (Fig. 7D). Resulting cells were analyzed for the expression of Fzd3 and Thy1 with flow cytometry. The data showed that on average, 17% of sorted cells were Thy1−Fzd3+ and 2.1% were Thy1+Fzd3+; the remainder expressed neither molecule (Fig. 7E). We did not detect a Thy1+Fzd3− population, indicating that all Thy1+ cells express Fzd3 in the testis. Spermatogonial transplantation demonstrated that approximately 80% of SSCs were found in the Thy1+Fzd3+ population, whereas no SSC activity was detected with Fzd3− cells (Fig. 7F). Minor SSC activity in Thy1−Fzd3+ cells probably resulted from contaminating SSCs from the Thy1+Fzd3+ population. Collectively, these results indicate that SSCs express receptors that are known for their ability to transduce β-catenin-independent Wnt5a signaling.
Wnt signaling is involved in several developmental processes during embryogenesis, where tight cell–cell communication is crucial to generate a properly patterned embryo (Logan and Nusse, 2004). This necessity for cell–cell communication is applicable to the biology of stem cells, which intimately interact with their surrounding environment. It might not be surprising therefore that Wnt signaling has been implicated in the maintenance of various stem cell types. One of the first examples described was in the mouse intestine where β-catenin-dependent Wnt signaling is essential for homeostasis and tumorigenesis (Korinek et al., 1998). Furthermore, this pathway is known to promote the in vitro expansion of neural stem cells and hematopoietic stem cells (Kalani et al., 2008; Reya et al., 2003). Therefore, classically, β-catenin signaling has been deemed a general regulator of stem cell self-renewal. However, recent evidence has indicated that the pathway can also support stem cell differentiation, and the effects of Wnt signaling on stem cell maintenance appear to be cell-type specific. Our study identifies for the first time that β-catenin-independent signaling mediated by Wnt5a, but not the β-catenin-dependent pathway, can promote SSC activity.
It has been reported that heterogeneity exists in germ-cell clusters (Dann et al., 2008; Oatley et al., 2009) and that SSCs represent a small subpopulation of cluster cells (Kanatsu-Shinohara et al., 2005; Yeh et al., 2007). In this study, we were able to visualize the heterogeneity of cluster cells in functional terms by activation of β-catenin signaling (Fig. 1A). Because the signaling-positive cells comprised a minority population, we hypothesized that these cells were SSCs. However, our results revealed that cluster cells with activated β-catenin signaling had lost SSC function (Fig. 1). This finding is in line with our in vivo observations (Fig. 2) that most, if not all, spermatogonia did not show active β-catenin signaling, whereas more differentiated cells did. These results imply that β-catenin signaling is associated with differentiation during spermatogenesis and that there might be mechanisms that suppress activation of β-catenin signaling in spermatogonia.
Our data indicate that Wnt5a is involved in such a mechanism and promotes SSC maintenance as a cell-extrinsic factor in vitro (Figs 3, 4). The following evidence supports this conclusion. First, Wnt5a was expressed in feeder cells in vitro. Second, feeder cells expressing Wnt5a supported SSC maintenance better than those that do not express Wnt5a. Third, a pan-Wnt inhibitor, sFRP1, diminished cluster formation, and this effect was competitively overcome by Wnt5a. Fourth, Wnt5a supported significantly more SSCs in feeder-free culture conditions, suggesting that Wnt5a can act directly on SSCs. In addition, we found that Wnt5a is expressed in Sertoli cells in the seminiferous epithelium, suggesting that Wnt5a acts as an environmental factor for SSC regulation in vivo. Further studies are necessary to address this notion.
We have also found that Wnt5a promotes SSC maintenance by supporting cell survival, and as a possible mechanism of action, we have shown that inhibition of a β-catenin-independent mechanism (i.e. JNK signaling) blocks Wnt5a-mediated SSC maintenance (Fig. 4). In a feeder-free culture condition, Wnt5a did not affect the cell cycle but suppressed cell death, and inhibition of JNK signaling abolished the pro-survival effect of Wnt5a (Fig. 5). However, it is possible that Wnt5a also blocks differentiation of SSCs and spermatogonia. Our data showing that Wnt5a reduces the number of β-catenin-signaling-positive cells, which do not have SSC activity, supports this possibility. Nonetheless, we note that these two populations are composed of undifferentiated spermatogonia and did not differ drastically in expression of the SSC markers examined (supplementary material Fig. S1D). Therefore, it will be important to further characterize β-catenin-positive and -negative cluster cells, and gene expression profiling might reveal the difference between these two cell populations at the molecular level.
We observed that Wnt5a significantly increased JNK-P levels (supplementary material Fig. S3B), supporting the results of our inhibitor screening experiments (Fig. 4) and apoptosis analyses (Figs 4, 5). Although this increase appears modest (~1.4-fold), it might be sufficient to mediate the effects of Wnt5a. However, it is possible that some other mechanisms also act downstream of Wnt5a signaling, because β-catenin-independent Wnt signaling remains elusive. Nonetheless, our data suggest JNK signaling to be a potential mediator of Wnt5a signaling in SSCs and undifferentiated spermatogonia.
Wnt5a has unique characteristics that are potentially important as a cell-extrinsic regulator of SSC function. Wnt5a acts over short distances and can regulate cell polarity. For instance, it contributes to the redistribution of cytoskeletal proteins and surface receptors, including Fzd3, and centrosome reorientation (Schlessinger et al., 2007; Witze et al., 2008). Wnt5a has also been shown to enhance polarization of melanoma cells toward a chemokine gradient, thus facilitating the response of a cell to directional cues (Witze et al., 2008). Therefore, it is tempting to speculate that Wnt5a is involved in the regulation of cell polarity in SSCs, thereby contributing to SSC fate control and their directional localization in the seminiferous epithelium. Because Wnt5a-knockout mice die perinatally (Yamaguchi et al., 1999), in vivo analyses of such Wnt5a actions will require a conditional-knockout mouse model, and our study provides the foundation to explore such a research direction.
For Wnt signaling mechanisms in general, it is known that distinct receptor–ligand pairings, but not properties intrinsic to Wnt ligands, determine which signaling pathway is activated (Mikels and Nusse, 2006; van Amerongen et al., 2008). Although we detected on SSCs the expression of all receptors that have been reported to mediate β-catenin-independent Wnt5a signaling (Fig. 7), it is not known which receptor was responsible for Wnt5a action. In this regard, Ror2 is an interesting target, because the Wnt5a–Ror2 interaction has been demonstrated to lead to the activation of JNK signaling (Mikels and Nusse, 2006; Oishi et al., 2003).
A recent study has reported that Wnt3a, which activates β-catenin signaling, stimulates proliferation of a spermatogonial cell line in vitro (Golestaneh et al., 2009). We detected transcripts encoding LRP5/6 in clusters (supplementary material Fig. S4A). Because these receptors act as essential receptor subunits that uniquely transduce β-catenin signaling (Logan and Nusse, 2004), our data suggest that these cells have the machinery to respond to Wnt3a. However, we also found that β-catenin-signaling-positive cells did not possess SSC activity and blocking the function of LRP5/6 by Dkk1 did not alter cluster numbers. We, and others, localized β-catenin expression on the plasma membrane and cytoplasm of most spermatogonia (Fig. S2E) (Lee, N. P. et al., 2003) rather than in the nucleus, the latter of which is a typical consequence of β-catenin signaling activation. Therefore, we speculate that Wnt3a might affect the activity of progenitor spermatogonia, rather than SSCs.
Finally, which molecules activate β-catenin signaling in clusters (Fig. 1A)? We detected transcripts of Wnt2b in feeder cells and those of Wnt1 in cluster cells. A recent DNA microarray analysis has identified the expression of R-spondin in clusters (Oatley et al., 2006). All these molecules are known to stimulate β-catenin signaling (Hendrickx and Leyns, 2008) and are candidate molecules that activate the signaling in clusters. In this regard, Wnt1 is known to be expressed by spermatids in testes (Shackleford and Varmus, 1987). It is thus not surprising that we observed activation of β-catenin signaling in germ cells in the adluminal compartment. Interestingly, however, β-catenin signaling activation was seen approximately 1 week before spermatids emerge during postnatal development (Fig. 2). Although the cause of this earlier-than-expected activation is unclear, we note the expression of Wnt2b in testes from the time of birth (Table 1). Combined with the expression of mRNA encoding Wnt1 and R-spondin in cluster cells, these factors could be responsible for the earlier activation of β-catenin signaling in vivo. These observations lead us to suspect that β-catenin-dependent and -independent pathways might crosstalk and contribute to coordinated regulation of SSC activity and spermatogenesis.
In summary, our study demonstrates that Wnt5a supports SSC self-renewal in part by promoting their survival. Our results suggest that JNK signaling is at least one of the mediators of this pro-survival effect of Wnt5a in a β-catenin-independent manner. We also found that other Wnt ligands are expressed in feeder cells and in the testis. These observations suggest that although Wnt5a contributes to SSC maintenance, it might represent only one aspect of the complex mechanisms driven by Wnt molecules that control SSC activities and spermatogenesis.
Materials and Methods
TCF/LEF-LacZ mice (from Daniel Dufort, McGill University, Montreal, Canada) have CD-1 genetic background and carry the lacZ reporter gene driven by the β-catenin–TCF/LEF responsive elements (Mohamed et al., 2004a; Mohamed et al., 2005). B6ROSA mice are F1 hybrids of C57BL/6 (B6) and ROSA26 (B6;129S-Gt(ROSA)26Sor/J) mice, which express lacZ ubiquitously (Zambrowicz et al., 1997). B6GFP mice (C57BL/6-Tg(CAG-EGFP)1Osb/J) express GFP ubiquitously. Oct4GFP mice (GOF-18ΔPEOct4/GFP) were generated as Yoshimizu et al. (Yoshimizu et al., 1999) by Jacquetta Trasler (McGill University) and express GFP specifically in spermatogonia up to ~7 days of age. Experimental cryptorchidism and orchidopexy were induced as described previously (Nishimune et al., 1978). Animal procedures were approved by the Animal Care and Use Committee of McGill University.
Recipient mice and spermatogonial transplantation
Recipient mice were prepared and spermatogonial transplantation was performed as described (Yeh et al., 2007). Recipients for B6ROSA donor cells were 129/SvEv×B6 F1 hybrids, and those for TCF/LEF–lacZ cells, Ncr nu/nu mice (Taconic). Recipient testes were analyzed for SSC quantification following staining with 5-bromo-4-chloro-3-indolyl β-D-galactoside (X-gal) 2 months after transplantation.
SSC cultures were generated using Thy1-positive testis cells from mice at 6–8 days post partum (d.p.p.), as described previously (Yeh et al., 2007). Cultures were maintained with a STO feeder layer (Kubota et al., 2004b; Yeh et al., 2007) and ‘growth factors’ GDNF (20 ng/ml), GFRα1 (75 ng/ml) and FGF2 (1 ng/ml). All in vitro experiments, except those where indicated, were conducted using established cluster cultures (i.e. >5 passages) on STO feeders with 40 ng/ml GDNF, 300 ng/ml GFRα1 and 1 ng/ml FGF2. The day of cell seeding was designated as day 0. In one series of experiments, L cells or L cells stably transfected with Wnt5a (Mohamed et al., 2005) (from Kazuki Kuroda, University of Ottawa, Ottawa, Canada) were used as feeders. For short-term feeder-free cultures, culture plates were coated with Matrigel (BD Biosciences), diluted 1:2, and incubated overnight at 4°C.
To activate β-catenin signaling, 5 mM lithium chloride was added to B6ROSA or TCF/LEF–lacZ cluster cultures on day 3 (three replicates). On day 6, TCF/LEF–lacZ clusters were trypsinized and reacted with X-gal to quantify signaling-positive cells using a hemocytometer. B6ROSA clusters were transplanted into recipient testes to quantify SSCs.
To inhibit Wnt signaling, Dickkopf-1 (Dkk1; R&D Systems) or secreted frizzled-related protein-1 (sFRP1; R&D Systems) was added on day 0 to TCF/LEF–lacZ or B6ROSA cluster cultures and replenished on day 3. On day 6, B6ROSA clusters were quantified visually, and signaling-positive cells in TCF/LEF–lacZ clusters, using a hemocytometer. At least three experiments were done. The same experimental schedule was used for competition assays of recombinant Wnt5a (R&D Systems) against sFRP1. Clusters were visually quantified on day 6. Because cluster numbers semi-quantitatively indicate SSC numbers, SSC activity can be measured by this ‘cluster-formation assay’ (Yeh et al., 2007).
To examine inhibition of β-catenin signaling by Wnt5a, TCF/LEF–lacZ clusters were cultured for 6 days with 400 ng/ml Wnt5a. β-Catenin-signaling cells were then quantified using a hemocytometer (four experiments). To examine effects of Wnt5a in the absence of feeders, B6ROSA clusters were removed from STO feeder cells by gentle pipetting (Oatley et al., 2006) and trypsinized into single cells. Resulting cells were placed onto Matrigel with Wnt5a, and 4 days later, they were trypsinized and subjected to the cluster-formation assay (passaging onto STO feeder cells with 1:1 split) and to spermatogonial transplantation. Three experiments were performed for the cluster-formation assay and six for spermatogonial transplantation.
To inhibit Wnt5a intracellular signaling, inhibitors against CaMKII (KN-93; Calbiochem), PKC (GF109203X; Biomol), G-proteins (Pertussis Toxin; Sigma), and JNK (JNK Inhibitor III and SP600125; Calbiochem) were added on day 0 to B6ROSA clusters. On day 3, the inhibitors were withdrawn, and on day 6, clusters were quantified. Inhibitor doses were determined according to previous reports (Bennett et al., 2001; Holzberg et al., 2003; Katada and Ui, 1981; Marley and Thomson, 1996; Toullec et al., 1991). At least three experiments were performed. In one series of experiments, clusters were treated with SP600125 as above; on day 6, they were trypsinized and subjected to the cluster-formation assay in three experiments. In feeder-free cultures on Matrigel, Wnt5a and/or SP600125 was incubated for 4 days, followed by the cluster formation assay in four experiments.
To analyze Wnt5a effects on cell cycle and apoptosis, TCF/LEF–lacZ or B6ROSA clusters were cultured feeder-free on Matrigel for 2 or 4 days and subjected to flow cytometry.
Whole-mount testis staining
TCF/LEF–lacZ mouse testes were fixed in 4% paraformaldehyde, whereas adult testes were fixed in Bouin's solution, and reacted with X-gal overnight. Paraffin-sections (5 μm) were counterstained with nuclear Fast Red.
Flow cytometric analysis and sorting
Flow cytometric analyses and fluorescent-activated cell sorting (FACS) were done using FACScan and FACSAira, respectively (Beckton Dickinson). All reactions with antibodies (supplementary material Table S1) were at 4°C for 30 minutes with gentle agitation. To isolate β-catenin-signaling cells, TCF/LEF–lacZ clusters were digested into single cells with trypsin and 1 μg/ml DNase. Cells resuspended at 8–9×106 cells/ml in PBS were reacted with 500 μM fluorescein di-β-D-galactopyranoside (FDG, Marker Gene Technologies) in double-distilled H2O for 1 minute at 37°C, following the manufacturer's protocol. Reaction was arrested with 10 mM HEPES and 4% FBS in PBS at 4°C. Cells were resuspended in Modified Eagle's Medium (MEM; Invitrogen) with 1% FBS before FACS. Experimental gates were established using control cells: B6ROSA (positive) and B6 (negative) cluster cells. B6ROSA and B6 cells showed >96% positive and <1% positive, respectively. Data were collected from two experiments.
Cell cycle profiles were examined with B6ROSA cluster cells. Cells were trypsinized and fixed in 70% ethanol, followed by incubation with 40 μg/ml propidium iodide (PI) and 100 μg/ml RNase at 37°C for 30 minutes. Data were collected from four experiments with 5000–30,000 events collected per sample. To determine G0 phase cells, B6ROSA cells were fixed in 1% paraformaldehyde for 30 minutes and 70% ethanol overnight and stained for Ki-67 expression, followed by incubation with PI and RNase. Experimental gates were established from unstained controls. Data were collected from three experiments with 5000–10,000 events collected per sample.
Apoptosis profiles were analyzed with B6ROSA cluster cells by staining with the APO-Direct Apoptosis Detection Kit (BD Biosciences). Three experiments were performed with 5000–10,000 events collected per experimental group.
To determine the expression of Wnt5a receptors on cluster cells, clusters from B6ROSA or B6-GFP mice were removed from STO feeder cells using gentle pipetting and dispersed using a micropipette into single cells for flow cytometric analyses. At least 10,000 events were collected. Data were from three to five experiments per particular receptor, except for integrin-α6, from two experiments.
Fzd3 expression on SSCs was analyzed using testis cells of 3-month-old B6ROSA mice (six testes/experiment). A single cell suspension was separated on a Percoll gradient as in (Kubota et al., 2004a). The SSC-containing fraction was further enriched for SSCs using immunomagnetic cell sorting against β2-microglobulin. Six million cells in 1 ml DMEM were used per sorting. β2-microglobulin-negative cells, representing 5–7.8% of total testis cells, were subjected to flow cytometric analyses for Fzd3 and Thy1 expression. Gates were established from primary antibody-omitted negative controls. Data were collected from two experiments; six recipient testes per group per experiment.
RNA isolation and RT-PCR analysis
Total RNA was extracted using TRIzol (Invitrogen). Complementary DNA was synthesized using Superscript III Reverse Transcriptase (Invitrogen) with random hexamers. Primer sequences are shown in supplementary material Table S2. Primers for other mRNAs were as reported previously: Wnt (Mohamed et al., 2004b), Fzds (Chen et al., 2004; Torday and Rehan, 2006), Ror2 (Mikels and Nusse, 2006) and LRP5/6 (Stump et al., 2003). PCR was performed using the program: 95°C for 3 minutes followed by 25–30 cycles of 95°C for 30 seconds, 51.1–61.5°C for 30 seconds, 72°C for 30 seconds with a final extension at 72°C for 5 minutes. Quantitative PCR was performed with QuantiTect SYBR Green PCR Kit (Qiagen) on a Rotogene 6000 (Corbett Research) with the program: 94°C for 15 minutes followed by 40 cycles of 94°C for 15 seconds, 60°C for 30 seconds, 72°C for 35 seconds. Annealing temperatures were as published or determined through Primer3 software (Rozen and Skaletsky, 2000).
In situ hybridization
B6ROSA testes were fixed in 4% paraformaldehyde and then in increasing concentrations of sucrose before cryosection at 10 μm. Sections were treated with 20 μg/ml proteinase K (Invitrogen) at 65°C for 10 minutes and post-fixed in 4% paraformaldehyde. Samples were acetylated in 0.25% acetic anhydride (Fisher) in 0.1 M triethanolamine pH 8.0 for 20 minutes. Hybridization was carried out using a digoxigenin-labeled riboprobe generated from a pGEM32F (Promega) vector carrying a 360 bp PCR fragment containing the Wnt5a coding sequence (Parr et al., 1993) (from Daniel Dufort, McGill University). Counterstaining was with YOYO-1 (Invitrogen). Images were captured with LSM 510 Meta laser-scanning confocal microscope (Zeiss).
To detect β-catenin expression, adult ROSA26 mouse testes were cryosectioned and fixed in ice-cold methanol before immunodetection. To identify putative Wnt5a receptors, B6ROSA clusters were fixed with 4% paraformaldehyde for 20 min. TCF/LEF–lacZ clusters were fixed with 0.5% glutaraldehyde for 5 minutes and subsequently reacted with X-gal for 6 hours before chromogenic immunodetection for Fzd3. Antibodies and their concentrations are listed on supplementary material Table S1. Hoechst 33342 (Invitrogen) was used for nuclear staining. Primary antibodies were omitted in negative controls. In some cases, image contrast was adjusted using Photoshop CS2 to better reflect our visual observations.
Western blot analysis
B6GFP cluster cells were cultured on Matrigel overnight in the absence of growth factors prior to addition of Wnt5a, or Wnt5a and SP600125, or vehicle alone for 2 hours, as described previously (Farias et al., 2009). Equal amounts of whole-cell lysate from each group were resolved using SDS-PAGE, proteins were transferred to polyvinylidene difluoride membranes, and immunoblotting for JNK-P or total JNK levels was performed. The blots were stripped after data acquisition, α-tubulin was immunoblotted, and visualized as a loading control. Visualization was obtained using ECL Plus Western Blotting Detection Kit (GE Healthcare). Quantification of blot intensities was performed using ImageJ software (NIH, Bethesda, MD) as per the developer's protocol. Data were collected from three separate experiments. Antibodies and their concentrations are listed on supplementary material Table S1.
Data presentation and Statistics
Data were expressed as mean ± s.e.m. Numbers of clusters in vitro and SSCs detected with spermatogonial transplantation were indicated as those per cm2 of growth surface in culture, unless specified otherwise. Significance was determined using Student's t-test or ANOVA followed by Fisher's Test for Least Significant Difference. P<0.05 determined significance.
The authors thank Daniel Dufort, Jacquetta Trasler, Kazuki Kuroda, Riaz Farookhi, and Hugh Clarke for transgenic mice and reagents. The authors also thank K. Ebata, C. Park and J.-C. Neel for technical support and K. Orwig and R. Farookhi for comments on the manuscript.
Supplementary material available online at http://jcs.biologists.org/lookup/suppl/doi:10.1242/jcs.080903/-/DC1
- Accepted March 6, 2011.
- © 2011.