Signaling through extracellular signal-regulated kinase is required for spermatogonial proliferative response to stem cell factor.

In vitro addition of stem cell factor (SCF) to c-kit-expressing A(1)-A(4) spermatogonia from prepuberal mice stimulates their progression into the mitotic cell cycle and significantly reduces apoptosis in these cells. SCF addition results in a transient activation of extracellular signal-regulated kinases (Erk)1/2 as well as of phosphatidylinositol 3-kinase (PI3K)-dependent Akt kinase. These events are followed by a rapid re-distribution of cyclin D3, which becomes predominantly nuclear, whereas its total cellular amount does not change. Nuclear accumulation of cyclin D3 is coupled to transient activation of the associated kinase activity, assayed using the retinoblastoma protein (Rb) as a substrate. These events were followed by a transient accumulation of cyclin E, stimulation of the associated histone H1-kinase activity, a delayed accumulation of cyclin A2, and Rb hyper-phosphorylation. All the events associated with SCF-induced cell cycle progression are inhibited by the addition of either a PI3K inhibitor or a mitogen-activated protein-kinase kinase (MEK) inhibitor, indicating that both MEK and PI3K are essential for c-kit-mediated proliferative response. On the contrary, the anti-apoptotic effect of SCF is not influenced by the separate addition of either MEK or PI3K inhibitors. Thus, SCF effects on mitogenesis and survival in c-kit expressing spermatogonia rely on different signal transduction pathways.

The tyrosine kinase receptor encoded by the c-kit gene and its ligand stem cell factor (SCF) 1 play a fundamental role in gametogenesis (1). Most mutations of either c-kit or SCF genes (W and Steel mutations, respectively) result in the loss of primordial germ cells in the embryonal gonad, whereas some Steel mutations affect gametogenesis after birth (2)(3).
c-kit expression is high in primordial germ cells and is downregulated in germ cells of the fetal gonad at around 13.5 days postcoitum (4). It is resumed in perinatal oocytes at the end of meiotic prophase and in proliferating spermatogonia at around 6 days postpartum (5)(6)(7). In the adult testis, c-kit expression is absent in undifferentiated spermatogonia (8), high in differentiating spermatogonia from type A 1 to B (5)(6)(7)8), and turned off in meiotic and postmeiotic cells (6 -7). A truncated form of the c-kit kinase, possibly playing a role during sperm-induced egg activation at fertilization, is expressed during spermiogenesis (9 -12).
c-kit expression in differentiating spermatogonia has led to the hypothesis that the SCF/c-kit interaction is required for the proliferation and/or survival of these cells. Several lines of evidence support this hypothesis. In vivo injection of antibodies directed against the extracellular region of c-kit selectively blocks proliferation and induces apoptosis of c-kit expressing type A spermatogonia but not of c-kit negative undifferentiated spermatogonia (7,13). Furthermore, a mutation in the c-kit docking site for the p85 subunit of phosphatidylinositol 3-kinase (PI3K), introduced by a knock-in strategy, causes a dramatic reduction of the spermatogonial population in the prepuberal testis (14 -15). A loss of spermatogonia during postnatal development is also observed in a peculiar Steel mutation, Sl 17H (3). Finally, in vitro addition of SCF, which is expressed by Sertoli cells (16 -17) under FSH control (17)(18), selectively stimulates DNA synthesis in type A but not in type B spermatogonia (17,19).
The series of molecular events leading to G 1 progression, G 1 /S transition, and mitosis have been established in several somatic cell types synchronized in G 0 through serum starvation (20 -23). Synthesis of D-type cyclins and the assembly and nuclear translocation of cyclin D/cyclin-dependent kinase 4/6 (cdk4/6) complexes is required for commitment to G 1 entry, whereas the consequent cyclin E accumulation and activation of the associated cyclin-dependent kinase 2 (cdk2) allows progression through G 1 (20 -23). Cyclin D⅐cdk4/6 complexes trigger initial phosphorylation of the retinoblastoma protein (Rb) and titrate cdk2 inhibitors (cip1/kip1 family), thus de-repressing cyclin E/cdk2 activity. Hyperphosphorylation of Rb by cyclin E/cdk2 is followed by release of the Rb-associated transcription factor E2F, which activates cyclin E transcription in a positive feedback loop, allowing the burst of cyclin E accumulation and activity in a narrow window coincident with the G 1 /S transition. E2F transcriptional activity is required to elicit timely induction of genes required for S phase progression, such as cyclin A2. Progression through the S phase coincident with the appearance of cyclin A2/cdk2 activity is followed by rapid down-regulation of cyclin E levels (20 -23).
We report evidence that SCF acts as a mitogenic factor in cultured c-kit-expressing spermatogonia and that both mitogen-activated protein kinase kinase (MEK)-and PI3K-dependent pathways are required for the proliferative response. The * This work was supported by grants from Ministero dell'Universita' e della Ricerca Scientifica e Tecnologica and from Agenzia Spaziale Italiana. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
‡ To whom correspondence should be addressed. Tel.: 39-06-72596272; Fax: 39-06-72596268; E-mail: pellegrino.rossi@med. uniroma2.it. 1 The abbreviations used are: SCF, stem cell factor; aa, amino acids; Erk, extracellular signal-regulated kinase; PI3K, phosphatidylinositol 3-kinase; Rb, retinoblastoma protein; MEK, mitogen-activated proteinkinase kinase; PBS, phosphate-buffered saline; GST, glutathione Stransferase; TUNEL, terminal dUTP nick-end labeling; cdk, cyclin-dependent kinase. mitogenic effect is not accompanied by an increase in total cellular amount of cyclin D3 (24), but it is associated with a rapid change in its subcellular localization. We also show that SCF is an anti-apoptotic factor for spermatogonia, but the MEK-or the PI3K-dependent pathways are not sufficient on their own to promote the survival response.

EXPERIMENTAL PROCEDURES
Isolation and Culture of Mouse Spermatogonia-Spermatogonia were obtained from either 5/6-or 8-day-old Swiss CD-1 mice, as reported previously (17). Spermatogonial stem cells and proliferating but undifferentiated spermatogonia are the prevalent germ cell types at 5-6 days of age, whereas differentiating (type A 1 -A 4 , intermediate, and type B) spermatogonia predominate at 8 days of age (25)(26). Briefly, germ cell suspensions were obtained by sequential collagenase-hyaluronidase-trypsin digestions of freshly withdrawn testes from 20 animals. To release cells completely, after the trypsin treatment, the pellet was resuspended in 1 ml of culture medium and pipetted at least 30 times and then brought to 20 ml with culture medium adding 2 mg/ml DNase and 10% fetal calf serum. Cell suspension was plated in Petri dishes (5 ml/dish) for 3 h in a humidified incubator at 32°C to promote adhesion of somatic cells. At the end of this pre-plating treatment, enriched germ cell suspensions were washed from fetal calf serum, and spermatogonia were then cultured in Eagle's minimal essential medium supplemented with 1 mM DL-lactic acid, 2 mM sodium pyruvate, non-essential amino acids (Life Technologies, Inc.). For time course experiments, spermatogonia were either left untreated or stimulated with SCF (100 ng/ml, Genzyme) at different time points and then they were processed as described below. Where indicated cells were also incubated 1 h before SCF addition with 10 M U0126 (catalog number V1121, Promega), with 10 M LY294002 (catalog number 270-038-M005, Alexis), or with 1 M tyrphostin AG490 (catalog number 658401, Calbiochem), all dissolved in Me 2 SO. In these experiments, an equal volume of the Me 2 SO solvent was also added in control and SCF-treated cultures. Nuclear morphology of spermatogonia after the pre-plating time and after 24 h of culture in the absence or constant presence of SCF and/or the signaling inhibitors was assessed after hypotonic shock of 10 5 cells (75 mM KCl) followed by fixation in methanol:acetic acid solution (3:1). Cells were then dropped onto glass slides to allow spread-ing of the nuclei and stained with Giemsa solution. Spermatogonia nuclei were judged as in interphase, metaphase, or apoptotic and counted from quadruplicate experiments. Somatic nuclei were excluded from the counts, and purity of spermatogonia was assessed as about 85% after the pre-plating treatment and almost 100% after 24 h of culture.
DNA synthesis was studied by [ 3 H]thymidine incorporation followed by autoradiography as previously described (17). In these experiments, incubation with [ 3 H]thymidine was performed during the last 4 h of the 24 h culture period.
Immunofluorescence Analysis and TUNEL Assays-Control and 1-h SCF-treated spermatogonia, preincubated or not with U0126 or LY294002, were spotted onto poly-L-lysine glass slides and fixed for 10 min at room temperature in 2% paraformaldehyde. Cells were washed in PBS, permeabilized 10 min with PBS, 0.1% Triton X-100 and incubated for 30 min at room temperature with PBS, 0.5% bovine serum albumin. Cells were incubated overnight at 4°C in a humidified chamber with mouse monoclonal anti-cyclin D3 antibody at a final concentration of 2 g/ml and then 1 h at room temperature with cyanin 3-conjugated anti-mouse IgG (Calbiochem). Slides were washed and mounted in 50% glycerol in PBS and immediately examined by fluorescence microscopy. Nuclei were counterstained with 1 g/ml Hoechst (catalog number 33342, Sigma). Control experiments were performed using mouse non-immune IgGs instead of the specific antibody.
For in situ detection of apoptotic cell death, control and SCF-treated spermatogonia, preincubated or not with U0126, LY294002, or AG490, after a 24-h period of culture were spotted onto poly-L-lysine glass slides, fixed for 10 min at room temperature in 2% paraformaldehyde, and subjected to TUNEL assay with an in situ cell death detection kit (catalog number 1684817, Roche Molecular Biochemicals) by following the manufacturer's instructions. Nuclei were counterstained with 1 g/ml Hoechst (catalog number 33342, Sigma). Slides where then examined by fluorescence microscopy. A, the time course of SCF-induced Erk1/2 activation in spermatogonia from 8-day-old mice was monitored through Western blot analysis using specific anti-phospho-Erk antibodies. Blots were stripped and reprobed with anti-Erk2 antibodies for loading control. This experiment was repeated three times with similar results. B, Erk1/2 activation after 15 min of SCF treatment is blocked by a selective MEK inhibitor but not by a selective PI3K inhibitor, as shown by Western blot analysis using specific anti-phospho-Erk antibodies in cells that had been preincubated with either U0126 or LY294002. Blots were stripped and reprobed with anti-Erk2 antibodies for loading control. C, the time course of Akt activation in spermatogonia from 8-day-old mice was monitored through Western blot analysis using specific anti-phospho-Akt antibodies. Blots were stripped and reprobed with anti-Akt antibodies for loading control. D, Akt activation after 15 min of SCF treatment is blocked by a selective PI3K inhibitor but not by a selective MEK inhibitor, as shown by Western blot analysis using specific anti-phospho-Akt antibodies in cells that had been preincubated with either U0126 or LY294002. Blots were stripped and reprobed with anti-Akt antibodies for loading control. E, SCF does not induce Erk1/2 activation in undifferentiated spermatogonia, as shown by Western blot analysis of germ cells from 5-to 6-day-old mice using specific anti-phospho-Erk antibodies. Blots were stripped and reprobed with anti-Erk2 antibodies for loading control. larly enriched in differentiating spermatogonia (25,26), which express high levels of c-kit (5)(6)(7)(8)14). Fig. 1 shows that, after 24 h of culture, several nuclei with characteristic features of apoptosis, such as reduced size and intense chromatin staining, can be observed in untreated cells. In SCF-treated cultures, the frequency of such cells is clearly reduced (see the last paragraph of this section), and a clear increase in the number of mitotic figures (nuclei showing condensed metaphase chromosomes) can be appreciated. These data confirm that SCF is required to maintain the proliferative state of differentiating spermatogonia cultured in vitro (17,24). We studied DNA synthesis and cell cycle progression in these cultures by using [ 3 H]thymidine incorporation and metaphase counting. SCF induces a 2-fold increase in the number of [ 3 H]thymidine incorporating cells and a 3-fold increase of metaphase counts with respect to the control after 24 h of culture (Table I). We also analyzed the effects of SCF addition in germ cell populations from 5-to 6-day-old mice, when undifferentiated spermatogonia are the predominant cell types (25,26), and c-kit expression is not detectable (5,7,8,27,28). c-kit signaling pathways activated in cell cycle progression have been shown to involve PI3K, MEK, and Janus-activated kinase 2 (JAK2) in different cell types (14,15,29,30). In mouse spermatogonia, PI3K activation has been shown to be involved in SCF-dependent proliferation (14,15,24); however, the possible involvement of MEK-and JAK2-dependent pathways has not been studied. To investigate whether these c-kit-activated signaling pathways mediate the mitogenic activation of spermatogonia observed in vitro, the proliferation assays were performed in the presence of inhibitors selective for each of the three different pathways: the MEK inhibitor U0126, the JAK2 inhibitor tyrphostin AG490, and the PI3K inhibitor LY294002 (Table I). The inhibition of the MEK pathway abolished the SCF-induced increase in both [ 3 H]thymidine incorporation and metaphase counts, demonstrating that the integrity of this pathway is required for SCF induction of mitogenesis. Inhibition of PI3K pathway also abolished SCF mitogenic effect, indicating that both MEK and PI3K pathways are required. On the contrary, inhibition of JAK2 signaling had no effect on SCF-stimulated [ 3 H]thymidine incorporation.

SCF Activates Both Extracellular Signal-regulated Kinases (Erk)1/2 and Akt Kinases in c-kit-expressing Spermatogonia-
Since the MEK and PI3K inhibitors were effective in the inhibition of SCF-induced proliferation of spermatogonia, we studied the Erk1/2 and PI3K activation pathways induced by SCF in these cells in a time course experiment. Fig. 2A shows that MEK was activated as early as 5 min from the addition of SCF, since an increase of phospho-Erk1/2 could be detected with respect to the control. The activation of both Erks was maximal at 15 min and then decreased to the control levels within 1 h, showing that SCF induces a transient Erk1/2 activation. SCFinduced increase of phospho-Erks is specifically regulated by MEK activation, since in the presence of U0126 the phospho-Erk1/2 bands were no longer detectable (Fig. 2B). To study the activation of PI3K, we monitored the phosphorylation state of its substrate, the Akt kinase. SCF stimulation induces a rapid and persistent Akt phosphorylation (Fig. 2C). Inhibition of PI3K with LY294002 completely blocked SCF-induced Akt phosphorylation (Fig. 2D). The two signaling pathways were independently regulated by SCF, since the presence of LY294002 or U0126 did not interfere with Erk1/2 or Akt activation, respectively (Fig. 2, B and D). This result indicates that, even though both MEK and PI3K activities are required for SCF-induced mitogenic effect in spermatogonia, no cross-activation occurs between these two signaling pathways in response to SCF stimulation.
As expected from the observation that SCF did not induce an increase of proliferation in germ cells from 5-to 6-day-old mice, SCF addition did not modify the phosphorylation state of Erk1/2 in these cells at any time point studied (Fig. 2E).
SCF Induces a Very Rapid G 1 /S Transition through Sequential Induction of Cyclin E and Cyclin A2-To study the effect of SCF addition on the spermatogonial cell cycle, we analyzed the expression of cyclins specifically expressed during the G 1 /S phase by Western blot. SCF addition did not modify the levels of cyclin D3, a D-type cyclin that is predominantly expressed in proliferating spermatogonia (31,32), at any time point studied (Fig. 3A), nor c-Myc levels (Fig. 3B), which are often up-regulated during mitogenic stimulation in other cell types (20 -23). However, the levels of cyclin E were up-regulated after 1 h from SCF addition and decreased after 3 h (Fig. 3C). Cyclin A2, which is expressed in proliferating spermatogonia (33), was up-regulated between 10 and 16 h after SCF addition, and it returned to the control levels after 24 h (Fig. 3D).
To exclude the possibility that in vivo exposure of spermatogonia to endogenous SCF had already caused sustained expression of cyclin D3 prior to their isolation for the in vitro culture, we stimulated spermatogonia with SCF following an overnight incubation in the absence of the growth factor. Even under these conditions, cyclin D3 did not increase upon SCF stimulation at any time point checked (Fig. 3E). The levels of cyclin A2, however, reached a maximum about 12 h after SCF addition (Fig. 3F).
SCF Treatment Causes a Transient Increase in the Activity of Cyclin D3-and Cyclin E-associated Kinase Activities-The increase of cyclin E and subsequently of cyclin A2 levels in the absence of detectable cyclin D3 quantitative modifications prompted us to investigate whether the activity of cyclin D3⅐cdk4 complex was affected by SCF addition. It is known that D-type cyclins induced by growth factors activate cyclin-dependent kinases (cdk4 and cdk6) to initiate Rb phosphorylation, which is then completed by cyclin E/cdk2 and cyclin A/cdk2 (20 -23). Western blot analysis of spermatogonial extracts with a mouse cross-reactive monoclonal anti-Rb antibody directed against aa 332-344 of human Rb showed that hyper-

FIG. 5. A rapid and transient activation of cyclin D3-associated Rb kinase activity and cyclin E-associated H1 kinase activity is induced by SCF treatment.
A, representative kinase assays on equal amounts of cyclin D3 immunoprecipitates using GST-Rb as a substrate (see "Experimental Procedures" for details). Immunoprecipitates (IP) were also subjected to Western blot analysis showing that equal amounts of cyclin D3 were present in all samples (lower panel). This experiment was repeated four times with similar results. B, kinase assays using GST-Rb as a substrate on cyclin D3 immunoprecipitates from equal amounts of extracts obtained from cells that had been preincubated with selective MEK or PI3K inhibitors. C, Western blot analysis using anti-cdk4 and anti-cyclin D3 antibodies of equal amounts of the same cell extracts utilized in B. D, kinase assays using histone H1 as a substrate on cyclin E immunoprecipitates from equal amounts of cell extracts. E, Western blot analysis using anti-cyclin E antibodies in cells that had been preincubated with selective MEK or PI3K inhibitors. phosphorylation of Rb (the slower migrating bands) started to be detectable as early as 2 h after SCF addition and reached a plateau after 16 h (Fig. 4A). Similar results were obtained using a polyclonal anti-Rb antibody directed against a peptide corresponding to 15 aa at the carboxyl terminus of human Rb (data not shown). This indicates that the different mobility of Rb in SCF-treated and control samples is actually due to changes in the phosphorylation state of Rb and not to a change in its molecular size due to proteolytic cleavage at the carboxyl terminus (a phenomenon which is often associated to apoptotic death in some cell types). Rb was timely hyperphosphorylated after SCF treatment also in spermatogonial cultures that had been subjected to overnight growth factor deprivation (Fig. 4B).
In order to test whether the increase of Rb phosphorylation was due to cyclin D3/cdk4 activation by SCF, cell extracts from spermatogonia were immunoprecipitated with anti-cyclin D3 antibodies, and then a kinase assay using GST-Rb as a substrate was performed. Fig. 5A shows a significant increase of cyclin D3-associated Rb-kinase activity after 1 h of stimulation with SCF, which became less evident after the 2nd h of culture. As a control, we performed similar kinase assays on cyclin D3 immunoprecipitates using histone H1 as a substrate, and no stimulation of H1 phosphorylation was observed after SCF addition (data not shown).
U0126 or LY294002 abolished SCF-induced stimulation of cyclin D3-associated Rb kinase activity (Fig. 5B), but they did not modify total cellular levels of cyclin D3 and cdk4 (Fig. 5C). Thus, both the MEK and the PI3K pathways converge at the level of regulation of cyclin D3-dependent kinase activity, rather than at the level of cyclin D3 or cdk4 synthesis or stabilization.
As expected from the time course of cyclin E accumulation in SCF-treated cells (Fig. 3C), cyclin E/cdk2 kinase activity, monitored using histone H1 as a specific substrate, was strongly induced after 1 h in the presence of SCF and decreased to the control levels after 4 h of stimulation (Fig. 5D). SCF-induced cyclin E accumulation was also abolished after pretreatment with either U0126 or LY294002 (Fig. 5E).
SCF Induces Nuclear Localization of Cyclin D3 through MEK-and PI3K-dependent Pathways-The subcellular localization of cyclin D3 was studied by immunofluorescence experiments. As shown in Fig. 6, the immunofluorescence staining of spermatogonia using an anti-cyclin D3 antibody revealed that SCF induced a marked increase in nuclear localization of the FIG. 6. A rapid change in the subcellular localization of cyclin D3 is induced by SCF treatment, and it is abolished by pretreatment with either MEK or PI3K inhibitors. Representative immunofluorescence study using specific anti-cyclin D3 antibodies. Hoechst counter-staining was used to control the subcellular localization of cyclin D3. This experiment was repeated five times with identical results. cyclin in the majority of the cell population after 1 h, while in the control cultures the fluorescence was restricted to the narrow ring of cytoplasm typical of this cell population. Pretreatment with U0126 or LY294002 completely abolished the nuclear accumulation of cyclin D3-induced by SCF in the 1st h of culture (Fig. 6) suggesting that both MEK and PI3K pathways are involved in promoting cyclin D3 nuclear relocation upon activation of the SCF receptor. No change in the subcellular localization of cyclin E, which is predominantly nuclear, was observed after SCF addition (data not shown).
SCF-mediated Prevention of Spontaneous Apoptosis in Cultured Spermatogonia Is Not Blocked by the Separate Inhibition of MEK-or PI3K-dependent Pathways-It has been proposed that SCF acts as a survival factor that prevents apoptosis in differentiating spermatogonia (13,19,34). In line with this we observed that during the 24 h of culture, concomitantly with the increase of the metaphase counts, SCF decreased the frequency of cells showing morphologies typical of apoptosis (Fig.  1). Such effect was quantified by TUNEL staining, which specifically detects DNA fragmentation associated with apoptotic cell death ( Fig. 7 and Table II). In order to verify whether the mitogenic and antiapoptotic effects of SCF on spermatogonia share common activation pathways, we explored the possible involvement of PI3K and/or MEK/Erk in SCF-mediated prevention of apoptosis. Contrary to what we observed about the mitogenic effect, neither the MEK-nor the PI3K-dependent pathway was essential for the activation of the survival response, since preincubation with either U0126 or LY294002 of SCF-treated spermatogonial cultures does not interfere with the antiapoptotic effect of SCF ( Fig. 7 and Table II). Similarly, addition of the JAK2 inhibitor tyrphostin AG490 has no detectable effect. Only the simultaneous addition of MEK and PI3K inhibitors partially reverts the SCF-activated survival response (Table II). DISCUSSION In this paper we show that two signal transduction pathways are involved in c-kit-induced proliferation of cultured spermatogonia. SCF addition results in a transient activation of Erk1/2 kinases and in a parallel activation of the PI3K-dependent Akt kinase. Inhibition of either the MEK or the PI3K signaling completely abolished SCF-induced DNA synthesis and cell cycle progression, whereas inhibition of JAK2-dependent pathways had no effect.
SCF stimulation of germ cell proliferation is not followed by phenomena that are commonly observed in G 0 -G 1 -arrested somatic cells after addition of different mitogenic stimuli (20 -23). SCF does not increase cyclin D3 or c-Myc cellular levels, and it does not affect levels of other positive or negative regulators of the G 1 /S transition, such as the cdc25a phosphatase or the cdk inhibitors of both the Cip1/Kip1 and the Ink family (data not shown). However, SCF induces a marked change in the subcellular localization of cyclin D3; within the 1st hour of SCF treatment, cyclin D3, which is predominantly cytoplasmic in control cells, becomes highly concentrated in the nucleus. Nuclear translocation of D-type cyclins has been also reported in the case of cAMP-dependent proliferation of primary thyrocytes (35) and 17␤-estradiol-dependent proliferation of the uterine epithelium (36). Our data show that nuclear translocation of a D-type cyclin can be stimulated also by the activation of a tyrosine kinase receptor.
SCF-induced nuclear accumulation of cyclin D3 in spermatogonia is coincident with a transient induction of its associated Rb kinase activity. These events are followed by a very rapid induction of the G 1 /S transition, monitored through transient accumulation of cyclin E at very early times of culture and activation of its associated histone H1 kinase activity, followed by induction of cyclin A2 (a marker of the S phase) at later times, and progressive hyperphosphorylation of Rb.
The observation that SCF mitogenic stimulus provokes such a rapid activation of the G 1 /S transition in differentiating spermatogonia is in agreement with pioneering autoradiographic studies by Monesi (37) on DNA synthesis in these cells, showing that duration of the "resting phase preceding DNA synthesis" (i.e. the G 1 phase) is very short, ranging between 2 and 3 h.
Inhibition of either MEK or PI3K signaling completely abolished SCF-induced increase in nuclear localization of cyclin D3, cyclin D3-associated Rb-kinase activation, cyclin E induction, and cell cycle progression in c-kit-expressing spermatogonia. Thus, the contemporaneous activation of both these pathways by SCF is essential to trigger G 1 /S transition in these cells.
MEK and PI3K cooperation in promoting cell proliferation has been explained by the observation that MEK-dependent Erk stimulation often promotes the synthesis whereas PI3Kdependent Akt activation leads to the stabilization of D-type cyclins (20 -23). Here we show a novel effect of the cooperation between these two pathways, culminating in modulation of the subcellular localization, rather than of total cellular levels, of a D-type cyclin. It has been shown that Erk activation can trigger a transient induction of p21 Cip1/Waf1 (38), which in turn can play a positive role in the assembly, in the nuclear translocation, and in the activation of cyclin D⅐cdk4/6 complexes (39). However, we found that p21 Cip1/Waf1 is barely detectable in spermatogonia at early times of culture, and SCF treatment does not cause any increase in its cellular levels (data not shown). Alternatively, Erk-dependent pathways might regulate phosphorylation of cyclin D3 residues homologous to Thr-156 of cyclin D1, whose mutation is known to inhibit nuclear import of the cyclin D1⅐cdk4 complexes (40).
The nuclear localization of D-type cyclins is also regulated by the PI3K pathway through the inhibition of glycogen synthase kinase 3␤ exerted by Akt. Indeed, inhibition of glycogen synthase kinase 3␤-dependent phosphorylation of cyclin D1 at the Thr-286 residue is coupled to the maintenance of nuclear localization of this cyclin during the G 1 /S transition (41)(42). We suggest that a similar mechanism may regulate cyclin D3 localization in response to SCF.
It has been reported that mouse spermatogonia isolated from 5-day-old mice and propagated on a feeder layer for an undefined period express higher levels of cyclin D3 when stimulated with SCF, and this would correlate with stimulation of DNA synthesis (24). In the present study we demonstrate that primary cultures of spermatogonia freshly isolated from 8-day-old mice are fully responsive to SCF, but no increase in cyclin D3 levels can be observed, even when SCF treatment is performed after overnight growth factor deprivation. Moreover, we found that freshly isolated spermatogonia from 5-to 6-day-old mice are not stimulated by the growth factor. The possibility exists that, even though the cell population used by Feng et al. (24) should not express c-kit at the beginning of culture (5, 7), it could eventually acquire SCF responsiveness during the coculture period.
Our data are in agreement with two recent reports (14,15) in which mutant mice were generated in which the c-kit codon tyrosine 719 (the docking site for the p85 subunit of PI3K) was converted to phenylalanine. The Y719F mutation induced a sex-and tissue-specific defect in postnatal gametogenesis, since males are completely sterile (14,15). A complete block of DNA synthesis was observed in germ cells at 8 days of age, when c-kit expressing differentiating type A 1 -A 4 spermatogonia are present and predominant (14). Our in vitro results show that, in addition to activating the PI3K pathway, SCF must also induce a transient Erk activation in order to elicit proliferation of spermatogonia.
It has been proposed that SCF acts merely as a survival factor that prevents apoptosis in differentiating spermatogonia, which are assumed to be intrinsically committed to proliferate (13,19,34). We actually found that SCF addition also partially inhibits apoptosis occurring in germ cells from 8-dayold mice after 24 h of culture. However, the anti-apoptotic effect observed in vitro was not inhibited by the separate addition of either the MEK-or of the PI3K inhibitor, whereas both inhib-itors on their own can impair the mitogenic response. Thus, the Erk1/2 activation and the PI3K-mediated Akt activation that we observed in cultured spermatogonia are not essential for SCF inhibition of apoptosis. Distinct SCF-activated signal transduction pathways must be involved in the pro-survival response, since even the simultaneous addition of both MEKand PI3K inhibitors does not completely suppress SCF antiapoptotic effect. In agreement with our in vitro observations, abolishment of c-kit-mediated PI3K signaling in c-kit Y719F knock-in mice was not associated to increased apoptosis in spermatogonia at 8 days of age (14).
In conclusion, our data indicate that soluble SCF stimulates proliferation of c-kit expressing and differentiating type A 1 -A 4 spermatogonia in vitro through both MEK-and PI3K-dependent pathways, by triggering nuclear relocation of cyclin D3 and a rapid G 1 /S transition. Moreover, they show that the SCFmediated proliferative and survival effects on spermatogonia depend on the activation of different combinations of intracellular signal transduction pathways.  7).
b Statistical analysis (analysis of variance test) was performed using a program PSI-plot 3.0 from Polysoftware International. p values versus control: p Ͻ 0.005. c Statistical analysis (analysis of variance test) was performed using a program PSI-plot 3.0 from Polysoftware International. p values versus control: p Ͻ 0.05.