|Home | About | Journals | Submit | Contact Us | Français|
GnRH and activin independently and synergistically activate transcription of the FSH β-subunit gene, the subunit that provides specificity and is the limiting factor in the synthesis of the mature hormone. This synergistic interaction, as determined by two-way ANOVA, is specific for FSHβ and may, therefore, contribute to differential expression of the two gonadotropin hormones, which is critical for the reproductive cycle. We find that the cross-talk between the GnRH and activin signaling pathways occurs at the level of p38 MAPK, because the synergy is dependent on p38 MAPK activity, which is activated by GnRH, and activin cotreatment augments p38 activation by GnRH. Both the Smad and activator protein-1 binding sites on the FSHβ promoter are necessary and sufficient for synergy. After cotreatment, Smad 3 proteins are more highly phosphorylated on the activin-receptor signaling-dependent residues on the C terminus than with activin treatment alone, and c-Fos is more highly expressed than with GnRH treatment alone. Inhibition of p38 by either of two different inhibitors or a dominant-negative p38 kinase abrogates synergy on FSHβ expression, reduces c-Fos induction by GnRH, and prevents the further increase in c-Fos levels that occurs with cotreatment. Additionally, p38 is necessary for maximal Smad 3 C-terminal phosphorylation by activin treatment alone and for the further increase caused by cotreatment. Thus, p38 is the pivotal signaling molecule that integrates GnRH and activin interaction on the FSHβ promoter through higher induction of c-Fos and elevated Smad phosphorylation.
The p38 subfamily of mapk consists of four isoforms, α, β, γ, and δ (1). Although p38 was first identified as important for inflammatory and stress responses, subsequently it has been shown that p38 also plays a role in apoptosis, differentiation, and other cellular processes (2). In particular, p38 is activated after GnRH treatment of pituitary gonadotrope cells, together with the ERK1/2 and c-Jun N-terminal kinase (JNK) branches of MAPK (3–5). GnRH is secreted into the hypophyseal portal system by a small population of hypothalamic neurons. It binds its G-protein-coupled receptor, which is expressed specifically by anterior pituitary gonadotrope cells, to induce expression and secretion of the gonadotropin hormones LH and FSH (6).
FSH is a heterodimer of α- and β-subunits and is required for fertility of female mice, because FSH β-subunit null mice are infertile due to a block in folliculogenesis before the antral stage (7). The unique FSH β-subunit confers specific biological activity, whereas the α-subunit is in common with LH, TSH, and chorionic gonadotropin (8). In addition to providing biological specificity, β-subunit gene expression is the limiting factor in FSH synthesis. FSH β-subunit gene transcription is induced primarily by GnRH and activin (6, 9, 10). We previously demonstrated that GnRH-induced activator protein-1 (AP-1) binds a novel AP-1 site in the mouse FSHβ proximal promoter and that overexpression of AP-1 proteins induces FSHβ transcription (11). The AP-1 transcription factor is a heterodimer of c-Fos and c-Jun immediate-early genes. In the gonadotrope cell line αT3-1, GnRH induction of c-Fos promoter activity was partially dependent on p38 (12).
Interestingly, p38 has also been implicated in TGFβ and activin signaling. Activin, a member of the TGFβ family, was originally identified as a regulator of FSH synthesis that was secreted by the gonads. Activin increases the release of FSH from the pituitary (13) and induces FSHβ expression in gonadotrope cells (10). Follistatin is a structurally unrelated protein that binds activin, making it biologically inactive (14). Activin and follistatin are also expressed within the pituitary and by the gonadotrope cell itself and can function in an autocrine or paracrine manner (15, 16). Activin, upon binding its receptors, activates receptor-associated Smads, Smad 2 and 3, which then associate with Smad 4 and translocate to the nucleus (17). Smad 3 and 4 bind DNA with low affinity at the Smad-binding element (SBE) to induce target genes (18), whereas Smad 2 does not bind DNA directly. For activin induction of FSHβ, Smad 3 seems to be a limiting factor, because its overexpression can induce the FSHβ reporter (19, 20). Additionally, receptors for the TGFβ family members can activate other intracellular kinases, such as TGFβ receptor-associated kinase (TAK1) (21, 22). TAK1, in turn, activates MAPK kinase (MAPKK), which is responsible for p38 activation. Furthermore, p38 is more highly activated after combined GnRH and activin treatment of the gonadotrope-derived cell line LβT2 than with GnRH treatment alone (23).
LβT2 cells, created in our laboratory, represent mature gonadotropes (24) and mirror in vivo gonadotrope responses. Therefore, LβT2 cells are an excellent homologous cell model for the study of regulation of gonadotropin synthesis. Activin and GnRH are the most potent regulators of FSH synthesis. There are reports in the literature that these two hormones can interact in vivo to induce the FSHβ gene (25). Furthermore, in transgenic animals, GnRH alone did not increase transgene expression, but in combination with activin, GnRH doubled the expression compared with activin alone (26). The interaction of these two hormones was also observed in isolated pituitary cells in culture (27). Synergistic induction of a luciferase reporter driven by the FSHβ promoter was observed in LβT2 cells (19). This interaction may be an important mechanism that regulates FSHβ expression independently of LHβ (also induced by GnRH), particularly at the time after the surge and ovulation, when LH levels precipitously drop, but FSH is still present to maintain folliculogenesis for the following cycle (28).
In this report, we determine the molecular mechanism of interaction and synergistic induction of FSHβ gene expression by GnRH and activin. Synergism was determined by the statistical method described in the most detail by Slinker (29) for identification of interactions between two treatments. Both the Smad DNA binding site and an AP-1 binding site in the FSHβ promoter are necessary and sufficient for synergistic induction by GnRH and activin. Furthermore, c-Fos is induced to a more elevated level with cotreatment than with GnRH treatment alone, and Smad 3 is activated more highly by cotreatment than with activin treatment alone. Both of these effects of the cotreatment are abrogated with the inhibition of p38 MAPK activity. Thus, synergistic induction is dependent on the p38 branch of the MAPK pathway.
Because both GnRH and activin are critical regulators of FSHβ gene expression, we sought to ascertain whether these hormones can interact and cross-talk to transcriptionally regulate the endogenous FSHβ gene. LβT2 cells, after overnight starvation, were treated with vehicle, GnRH, activin, or both hormones in combination. RNA was extracted, reverse transcribed, and subjected to quantitative real-time PCR. The endogenous mouse FSHβ gene is induced 9-fold by 5 h GnRH treatment, 54-fold by 5 h activin, and 108-fold by cotreatment (Fig. 1A). This is far greater than simply additive and indicates synergistic interaction between these two hormones, determined by two-way ANOVA (P < 0.0001) and a method by Slinker (29). To better illustrate this synergistic interaction, we also represent the same results using the Slinker method (Fig. 1B). Nonparallel lines, without activin (circles) and with activin (squares), connecting vehicle and GnRH treatment indicate that GnRH and activin interact and that the combined effect is not an independent induction by each hormone. Although both GnRH and activin induce LHβ, as we have determined before (30), when both hormones were added together, the induction did not significantly differ from GnRH or activin alone (Fig. 1C). Thus, synergistic induction of the FSHβ gene by GnRH and activin is not replicated on the LHβ gene, indicating that this effect is not due to an induction of GnRH receptor and illustrating the specificity of the regulation of FSHβ induction.
To analyze the molecular mechanism for this synergy, we turned to transient transfection assays and transfected 398 bp of the 5′ proximal regulatory region of the mouse FSHβ gene linked to a luciferase reporter into LβT2 cells. Luciferase was induced 6-fold with 5 h GnRH treatment, 3.5-fold with activin, and 20-fold with cotreatment (Fig. 2A). The same results are represented in Fig. 2B, and again statistical analysis indicates that GnRH and activin synergistically interact (P = 0.0058). Because both endogenous mRNA levels (Fig. 1A) and the reporter gene under the control of the FSHβ promoter (Fig. 2A) show synergism, these results also indicate that the synergistic interaction between GnRH and activin is not primarily due to changes in FSHβ mRNA stability, but rather to regulation of transcriptional activity. In the following experiment (Fig. 2C), we treated LβT2 cells with combinations of different concentrations of GnRH and activin. At every concentration of activin and GnRH, cotreatment showed synergistic interaction between these two hormones as analyzed by two-way ANOVA (P = 0.0078 for 1 nm GnRH and 1 ng/ml activin; P = 0.0108 for 1 nm GnRH and 10 ng/ml activin; P < 0.0001 for 10 nm GnRH and 1 ng/ml activin; P = 0.0011 for 10 nm GnRH and 10 ng/ml activin; Fig. 2C). Thus, even using physiological levels of hormones, at which it is difficult to study molecular mechanisms due to our limited detection levels, the interaction between GnRH and activin can be synergistic and may contribute to the specific secondary surge of FSH.
Follistatin is an activin binding protein that prevents activin from interacting with its receptor. To analyze its role, we treated LβT2 cells with follistatin. Follistatin alone reduced the level of FSHβ promoter activity to 58% of control, whereas follistatin cotreatment with GnRH reduced the fold induction by GnRH by 30% (Fig. 3A). This is probably due to inhibition of the actions of autocrine activin, secreted by the LβT2 cells, which may contribute to the expression of the FSHβ promoter even in untreated cells (31). Analyzed by two-way ANOVA, results indicate that follistatin and GnRH negatively interact (P < 0.0001). Smad 7 serves as a dominant-negative Smad, because it prevents phosphorylation of receptor Smads and thus prevents formation of an active Smad heterodimer of Smad 2/3 and Smad 4 (32, 33). To ascertain the importance of the signaling downstream of activin receptor, we cotransfected an expression vector containing the Smad 7 cDNA with the FSHβ-luciferase reporter into LβT2 cells. Overexpression of Smad 7, similar to follistatin treatment, reduced the level of FSHβ expression to 68% compared with the vehicle-treated cells transfected with control vector and reduced the fold induction by GnRH. More significantly, Smad 7 overexpression reduced the induction by cotreatment to the level of induction by GnRH alone (Fig. 3B). These results, together with the results obtained with follistatin treatment, indicate that synergy is occurring through the activin receptor-driven Smad signaling pathway. They also indicate that autocrine, endogenously secreted activin synergizes with GnRH to achieve maximal GnRH induction.
To identify which promoter elements convey GnRH, activin, and their synergistic response, we mapped the responsive regions of the mouse FSHβ promoter using truncation analysis. LβT2 cells were transiently transfected with a series of truncations of the FSHβ gene 5′-flanking region, ranging in length from 1500 to 95 bp upstream of the transcriptional start site. Figure 4 displays the expression of the reporter gene as fold induction over vehicle-treated control for each of the truncated promoter regions. Activin responsiveness maps to a region distinct from GnRH responsiveness (Fig. 4). GnRH response maps to two distinct regions of the proximal FSHβ promoter, between −95 and the transcriptional start site and between −304 and −230 bp from the start site of transcription (Fig. 4A). This result is in agreement with our previous report (11), although in the current study, we have used a larger region of the promoter, −1500 bp from the transcriptional start site, whereas previously we had used only the first 398 bp. However, the additional distal sequences do not seem to contain GnRH-responsive elements as the fold induction of the −1500 truncation does not differ from the fold induction of the −304 truncation. On the other hand, activin responsiveness decreases with each truncation of the promoter (Fig. 4B). This gradual decrease in responsiveness to activin indicates that there are likely multiple activin-responsive elements in this promoter. The biggest decrease in the responsiveness to activin occurs when the promoter is truncated from −304 to −230. This region contains the full SBE located at −267, as reported previously (19, 20, 34, 35). The most proximal response to activin is found between −230 and −194 from the transcriptional start site. The most proximal region, which retains the most responsiveness to GnRH, does not respond to activin at all. Thus, most of the activin responsiveness maps to a region distinct from GnRH responsiveness, again indicating the interaction between these two hormones is not at the level of either of the receptors but more downstream in signaling pathway. There are two significant drops in the level of induction by cotreatment (Fig. 4C), between −1500 and −1000 and between −304 and −230. Thus, the region between −304 and −230 is the region that is involved in GnRH and activin induction and in synergy between the two hormones.
Therefore, we concentrated on this region to analyze potential sites by mutations, internal deletions, and EMSA. First, EMSA were used to determine whether complexes from the LβT2 cell nuclear extracts can be detected that change after GnRH, activin, or cotreatment. For that purpose, a radiolabeled 33-bp sequence from −275 to −243 was incubated with nuclear extracts from LβT2 cells treated with GnRH and/or activin as indicated above each lane (Fig. 5). A full Smad consensus site is found at −267, and Smad proteins bind to a 33-bp probe encompassing this site, as we will address below, consistent with previous reports (19, 20). Using the same probe, we identified a complex present in the extracts after GnRH treatment (marked with arrow 1) and a complex that intensifies after cotreatment (lane 5; arrow 2). The GnRH-induced complex was identified as AP-1, because it supershifts with antibodies that recognize c-Fos protein (marked with asterisk and an arrow, lane 8).
To determine which nucleotides are needed for AP-1 binding, competitions with 200-fold excess of unlabeled mutant oligonucleotides were used (Fig. 6, A–D). Three-base-pair scanning mutations (A–K) were introduced into the wild-type oligonucleotide (Fig. 6A). First, to confirm Smad binding to its consensus site, we used Cos-1 cells to overexpress Smad 3 and Smad 4 protein (Fig. 6B). An arrow indicates a complex that forms with overexpression of Smad 3 and Smad 4 in Cos-1 nuclear extracts (lane 2), compared with the nuclear extracts from Cos-1 cells transfected with empty vector control (lane 1). Oligonucleotides with mutations in the SBE (mutations C–F) cannot bind Smad proteins and do not compete with the labeled wild-type probe. Figure 6C represents extracts from activin-treated LβT2 cells. In the lane marked SS for supershift, Smad 4 antibody was included to induce a supershift marked with an asterisk. Due to autocrine activin, activin treatment (lane 1) caused only a minor increase in the band intensity (marked with an arrow) relative to control (lane 2). However, competitions indicate that the complex is dependent on the SBE site for binding (mutations C–F), the same residues bound with overexpressed Smad proteins in Fig. 6B. Supershift and dependence on the SBE site for binding indicate that activin, both endogenous and exogenous, causes Smad proteins to bind to the SBE. In a similar competition EMSA, using nuclear proteins from cells treated with GnRH for 2 h (Fig. 6D, lanes 2–14, whereas lane 1 is extracts from control cells), mutants H and I cannot compete successfully for the complex induced by GnRH identified in Fig. 5 as AP-1. There is a 1-bp mismatch between the nucleotides identified as required for AP-1 binding (AGAGTCA) and a consensus AP-1 site (TGAGTCA). The putative −256 AP-1 site is juxtaposed to the −267 Smad site (Fig. 6A). Thus, Smad proteins, both overexpressed in Cos-1 cells, which serve as control, and from LβT2 cells, bind the SBE that is adjacent to the newly identified putative AP-1 element, which is bound by GnRH-induced AP-1.
To assess the contribution of these sites, the SBE at −267 and the AP-1 at −256, and of our previously identified AP-1 site at −76 (11) to FSHβ induction, we introduced selective mutations into the FSHβ-luciferase reporter (Fig. 7). Mutation of the putative −256 AP-1 element does not reduce induction by GnRH alone, indicating that this is not a functional element for GnRH induction, although it can be bound in vitro, under conditions of probe excess in EMSA. This is consistent with our previous results that the only functional AP-1 site in the FSHβ promoter is the −76 site (11), because the mutation of the site −76 AP-1 completely abrogated induction by c-Fos and c-Jun overexpression. Every mutant reporter that makes a change in the −76 site, alone or in combination with the Smad site mutations, reduced the fold induction by GnRH by 30% (significant reduction in fold induction is marked with a dagger). Every mutant that changed bases in the Smad site, also alone or in combination with either AP-1 element, reduced the fold induction by activin by 20%. Significant reduction in synergistic induction of the FSHβ promoter was achieved by double mutations changing both the Smad and −76 AP-1 sites. Interestingly, there was no further effect on GnRH or activin induction of a double mutation, but the effect on synergy was significant. Although the −256 AP-1 site was not functional for GnRH induction of the FSHβ gene, it may contribute to synergy, because mutation in this site coupled with mutations in the Smad and −76 AP-1 sites completely abrogated the synergy, making the induction with cotreatment only additive, as analyzed by two-way ANOVA (P < 0.0001 for wild-type; P = 0.0340 for the Smad and −76 AP-1 double mutant vs. P = 0.4377 for the triple mutant). The result that GnRH induction significantly decreased when the promoter was truncated from −304 to −230 prompted a search for other putative GnRH-responsive elements. To further analyze that region, we created 10-bp internal deletions throughout the whole region; however, none of these deletions significantly changed fold induction by GnRH except the deletion that eliminated the Smad site at −267 (data not shown). The lack of a GnRH-only responsive element in this region, together with the result that both follistatin treatment and Smad 7 overexpression reduce fold induction by GnRH, implies that GnRH can synergize with the autocrine, endogenous activin secreted by the LβT2 cells.
To assess whether any of these sites are sufficient for synergy, we created multimers containing four copies of the sites indicated in Fig. 8 and ligated them to the heterologous minimal thymidine kinase (TK) promoter and luciferase reporter. Induction by GnRH of the −256 AP-1 site multimer did not reach statistical significance, indicating that this site is not sufficient for GnRH induction. Consistent with our previous results, the multimer of the −76 AP-1 site is sufficient for induction by GnRH (11) but not for activin induction. The Smad site is sufficient for induction by activin, as expected, but none of these sites alone are sufficient for synergy. However, when the Smad and −256 AP-1 sites are multimerized, which means that the four copies of the region between −269 and −248 of the mouse FSHβ promoter are linked to the TK promoter driving luciferase expression, they are shown to be sufficient to convey GnRH and activin synergy to this heterologous promoter (P = 0.0117). Furthermore, when the Smad site is added to the −256 AP-1 site, GnRH induction reached statistical significance (P = 0.0170), as opposed to the result with the −256 AP-1 site alone (P = 0.0590). The multimer of the −76 AP-1 and Smad sites together was also sufficient for synergy (P < 0.0001) and reached a 220-fold induction by cotreatment. To assess whether any AP-1 site linked to the Smad site can achieve synergy, a multimer of a consensus AP-1 site (cons AP-1) and the Smad site was tested and found to be induced by GnRH by 230-fold, activin by 34-fold, and cotreatment by 615-fold. Thus, the results presented in Figs. Figs.77 and and88 demonstrate that both the Smad site and the AP-1 sites are necessary and sufficient for the synergistic interaction between activin and GnRH to induce the FSHβ gene.
To determine the reason that Smad and AP-1 sites are necessary and sufficient for synergistic activation of FSHβ gene expression, we analyzed the proteins that bind these sites. Smad 3, as indicated previously, is the primary transcription factor phosphorylated by activin receptor binding by activin in LβT2 cells. Upon phosphorylation, Smad 3 dimerizes with Smad 4, translocates to the nucleus, and binds the Smad binding site within the DNA. We tested the levels of Smad 3 phosphorylation after activin, GnRH, and cotreatment of LβT2 cells. Whole-cell lysates were analyzed using Western blots probed with antibodies specific for the phosphorylated form of Smad 3 at the C-terminal serine residues. As expected, control quiescent cells do not have detectable levels of phosphorylated Smad proteins. Upon 1 h activin treatment, Smad proteins are phosphorylated, and the level of phosphorylation increases with cotreatment (Fig. 9A). The increase with cotreatment was small, but consistent, and, in every experiment, varied between 1.4- and 1.5-fold over activin treatment alone, which was a statistically significant change. These results were observed by carefully titrating the protein amounts loaded onto the gels and quantifying the Western blots with a chemiluminescence reader (to avoid redundancy, quantification was presented for the blots in Fig. 12, and the results presented in Fig. 9 had the same quantification profile). The concentration of Smad proteins in the cells does not change as shown by Western blots detecting total Smad 3 protein (Fig. 9A). Thus, cotreatment of cells with GnRH and activin together causes higher levels of Smad activation than activin treatment alone. This is the first report that shows that G protein-coupled receptor activation can affect phosphorylation of Smad molecules at the C-terminal residues, a region normally phosphorylated by activin receptor.
We then transfected Smad 3 expression vectors into LβT2 cells. A recent report has shown that Smad 2 cannot be overexpressed in LβT2 cells and that Smad 3 alone is sufficient for FSHβ induction, presumably because Smad 4 levels are not the limiting factor for activin action in these cells (36). Smad 3 overexpression is also sufficient for synergy with GnRH (Fig. 9B), which is in agreement with the previously published report by Gregory et al. (19). We also overexpressed Smad 3 proteins that contain mutations in amino acids that can be phosphorylated either in the linker region or in the C-terminal region. It has been reported that the linker region of Smad 3 can be phosphorylated by the MAPK pathway, but the functional significance of this remains controversial. Activin receptor activation leads to phosphorylation of serines located in the C-terminal region. Mutation of the residues that can be phosphorylated in the linker region to alanine increases transcriptional activity of Smad 3, indicating that phosphorylation of the linker region negatively affects Smad 3 transcriptional activity. Phosphorylation of the C-terminal region is important for the role of Smad 3 in FSHβ expression, because mutation of serines in the C terminus to alanines decreases Smad 3 transcriptional activity compared with the wild-type protein. Both increased C-terminal phosphorylation by cotreatment (Fig. 9A) and decreased transcriptional activation by Smad 3 with the C-terminal region mutation (Fig. 9B) indicate the importance of serine phosphorylation caused by activin-receptor signaling.
We have demonstrated previously that GnRH treatment induces c-Fos, FosB, c-Jun, and JunB (but not JunD) in LβT2 cells (11). Although activin treatment alone does not induce any of the AP-1 isoforms in LβT2 cells, c-Fos and FosB are induced more strongly with 1 h GnRH with activin cotreatment than with GnRH alone, but c-Jun and JunB are not (Fig. 10). Although GnRH alone causes not only induction of c-Jun but also c-Jun phosphorylation, there was no additional increase in c-Jun phosphorylation after cotreatment. The increase in c-Fos levels with cotreatment was consistently 1.6- to 1.8-fold over GnRH treatment alone, as quantified by chemiluminescence, and after c-Fos levels were normalized to β-tubulin levels, which serves as a loading control. Therefore, the cotreatment causes accumulation of a higher concentration of Fos isoforms in the cells than GnRH treatment alone.
Because GnRH activates three MAPK pathways, ERK1/2, p38, and JNK (3–5), and activin can use p38 pathways in other cell types (37, 38), it was important to address the contribution of MAPK pathways to synergistic induction of the FSHβ gene. Using specific inhibitors for these pathways (Fig. 11A), we determined that the inhibition of ERK1/2 reduced GnRH induction and the fold induction by cotreatment, but the synergy, although reduced (P = 0.043 vs. P = 0.008 for control), still remains. This is in agreement with Gregory et al. (19). Inhibition of the JNK pathway does not have any effect on induction of FSHβ by GnRH, activin, or cotreatment. Inhibition of p38, however, completely abolished induction by GnRH, activin, and synergy, indicating that p38 activity is critical. To alleviate potential concerns regarding the specificity of this pharmacological approach, we next used a different p38 inhibitor, PD169316, for comparison with the one used in Fig. 11A (SB202190). Lower concentrations of both inhibitors significantly reduced fold induction by GnRH, activin, and cotreatment (for synergy, P = 0.01 for lower concentration of inhibitors, whereas P = 0.0058 for control), whereas higher concentrations completely abolished induction by either hormone and by their combination (Fig. 11B). There was no change in basal expression with any of the inhibitors (data not shown). Furthermore, to confirm the role of p38, the expression vector for the dominant-negative p38 kinase that has mutations in the phosphorylation/activation Thr-180/Tyr-182 residues was cotransfected into LβT2 cells with the FSHβ-luciferase reporter. Expression of the dominant-negative p38 also reduced the fold induction by GnRH and synergy. It did not, however, completely abolish synergy, either due to an insufficient level of expression to inhibit all of the endogenous p38 kinase or due to the variety of the p38 isoforms present in the cell, of which some may not be inhibited with this dominant-negative protein. Nevertheless, the results with the specific inhibitors and dominant-negative p38 kinase illustrate the importance of this branch of MAPK for the synergistic interaction.
Recently, we and our collaborators demonstrated that cotreatment of LβT2 cells with GnRH and activin increases p38 activation but not ERK1/2 MAPK activation when compared with GnRH treatment alone (23). We next tested whether p38 can affect c-Fos induction in LβT2, as has been shown in another gonadotrope cell line, αT3-1 (12). c-Fos induction by GnRH is dependent on both p38 and ERK1/2, because inhibitors of p38 and ERK (but not of the JNK pathway), reduced the level of c-Fos induction by GnRH (data not shown).
Because c-Fos induction is dependent on p38 and p38 activity is augmented after cotreatment, we tested the role of p38 in increasing expression of c-Fos by cotreatment, as observed in Fig. 10. The cells were first treated with the inhibitor of p38 and the hormones, and then whole-cell lysates were analyzed by Western blot. As demonstrated in Fig. 10, c-Fos protein levels were higher after cotreatment than with GnRH treatment alone. Inhibition of p38 activity reduced the level of induction by GnRH by 35% and completely abrogated the difference in the induction by GnRH vs. cotreatment (Fig. 12A).
To test whether p38 contributes to increased Smad 3 phosphorylation after cotreatment over activin treatment alone, Fig. 12B shows that the p38 inhibitor reduced Smad phosphorylation by activin to 62%, and, additionally, eliminated the difference in Smad 3 phosphorylation after cotreatment vs. activin treatment alone. Thus, p38 activity is important both for higher c-Fos protein levels induced by GnRH and activin cotreatment as well as GnRH induction of c-Fos. Furthermore, p38 is critical for the increase in Smad 3 phosphorylation obtained with activin cotreatment with GnRH over activin treatment alone. Moreover, p38 is involved in Smad phosphorylation by activin in the C-terminal region.
In this report, p38 MAPK is shown to be a point of convergence between the GnRH and activin signaling pathways. Inhibition of p38 reduces fold induction of FSHβ by GnRH and activin and abrogates synergistic interaction between these two hormones. Synergy is dependent on both Smad and AP-1 binding sites in the FSHβ promoter, and the proteins that bind them, Smad 3 and c-Fos, respectively, are more highly activated by cotreatment in a p38-dependent manner. p38 is involved in increased c-Fos induction by cotreatment and augmented Smad phosphorylation on the activin receptor-dependent C-terminal phosphorylation sites. Thus, p38 is a point of cross-talk between these two pathways and a key signaling molecule that integrates the synergistic interaction of GnRH and activin that is specific for the FSHβ gene and likely contributes to the differential expression of the two gonadotropin hormones, which is critical for the reproductive cycle.
We previously reported that p38 is more highly activated after activin and GnRH cotreatment than with GnRH treatment alone, although we did not detect activation of p38 by activin treatment alone (23). The lack of detection of activin-induced changes in p38 phosphorylation could be due to insufficient sensitivity of Western blotting for detection of slight changes in activation or because the antibodies recognize only phosphorylation at the classical dual Thr-180/Tyr-182 site, whereas activin causes phosphorylation at a different site. For example, it has been reported that T cell receptor can cause phosphorylation of Tyr-323 of p38, which can cause autophosphorylation of Thr-180/Tyr-182 (39). Alternatively, it is possible that in LβT2 cells, activin alone is not sufficient to activate p38 despite its activation of TAK1 (22) and, through it, possible activation of MAPKK3/6, which can, in turn, activate p38. On the other hand, in some cell types, TGFβ (but less so activin) is sufficient to activate p38 (21, 38, 40, 41). It is probable that in LβT2 cells, due to autocrine, endogenously secreted activin, p38 is already active at a higher basal level, and exogenous activin cannot augment that, whereas other signaling pathways such as GnRH can. The mechanisms by which activin treatment augments GnRH activation of p38 are not known and are the subject of further study.
Elevated expression of c-Fos and augmented Smad phosphorylation are both downstream of p38 and are both critical for synergy between GnRH and activin on FSHβ gene expression. We determined that p38 is critical for c-Fos induction in the LβT2 gonadotrope cell line that endogenously expresses FSHβ, consistent with a report using another gonadotrope-derived cell line, αT3-1 (12). c-Fos protein levels are even higher after cotreatment. Higher activation of p38 with cotreatment, coupled with its role in c-Fos expression, is a probable cause of augmented c-Fos expression with cotreatment compared with GnRH treatment alone. We show that inhibition of p38 not only reduces the level of c-Fos protein induction by GnRH by 35% but also abrogates the difference between cotreatment and GnRH treatment alone. TGFβ can induce AP-1 isoforms, particularly c-Jun and Jun B in other cell types (42); however, we could not detect activin induction of any of the AP-1 proteins, either by Western blot or by reporter assay using the c-Fos or c-Jun promoters (data not shown). Activin, therefore, is not sufficient to induce AP-1 isoforms in LβT2 cells. Because p38 can contribute to their expression by other hormones, this may also imply that, in LβT2 cells, activin is not sufficient to activate p38 and that signals from both activin and GnRH are required.
This is the first report showing that p38 activity and GnRH treatment contribute to Smad 3 phosphorylation at the Ser-423/Ser-425 dual site, which is normally phosphorylated by recruitment to the activated TGFβ/activin receptors (43). In a few recent reports in other systems, the p38 pathway was shown to be critical for activin action. Activin can activate p38 and regulate erythroid gene expression (37). The p38 pathway is also required for growth inhibition of human breast cancer cells by activin (38). However, in those reports, Smad and TAK1-MAPKK3/6-p38 were considered to be separate pathways activated by activin. Herein, we show that p38 is important for Smad 3 phosphorylation at the Ser-423/Ser-425 site. Phosphorylation of Smad proteins by the MAPK pathway has been reported to occur in the linker region of the Smad structure (44, 45). In this study, we used antibodies that recognize Smad 3 proteins phosphorylated in the TGFβ/activin-regulated region, i.e. phosphorylated serines 423 and 425 in Smad 3. Interestingly, that region seems to be more highly phosphorylated by GnRH and activin cotreatment in a p38-dependent manner. There are several potential explanations. First, p38 could inhibit the phosphatase that dephosphorylates Smad 3. The only Smad phosphatase identified to date is PPM1A (46), which is present in the cell at a steady-state level and constitutively active. Thus, we tested whether the levels of PPM1A protein in the cell are regulated by GnRH, activin, or p38. With 1-h treatments, a time frame in which we observed changes in Smad phosphorylation, it is unlikely that p38 affects expression level of the phosphatase, but it can affect degradation. However, we did not detect a change in the levels of the phosphatase under any treatment, with or without p38 inhibition (data not shown). Also, we did not detect a shift in molecular weight using a high-percentage gel, which can indicate a protein modification that might affect the activity (data not shown). A second potential mechanism for how p38 could affect Smad phosphorylation is that p38 activity could provide positive feedback at the level of receptor activation, which phosphorylates Smad proteins. TAK1 activity can be limited by p38 through feedback control (47), whereas a TAK1 binding protein, TAB1, which can also bind p38α, can be phosphorylated by p38 (48), but the potential for phosphorylation of activin receptor by p38 or by TAK1 has not been addressed. Finally, because it has been shown that p38 phosphorylates the linker region of Smad proteins, it is possible that linker region phosphorylation extends the half-life of phosphorylation at Ser-423 and Ser-425 by causing different folding of the protein and protection from the phosphatase. TGFβ growth inhibition of breast carcinoma cells requires both the Smad and p38 pathways, which can be independently activated, but also p38 can be upstream from Smads (41). Inhibition of p38 caused a loss of Smad transcriptional activity in these cells and lowered Smad 2/3 phosphorylation in the linker, although not in the C-terminal region. It is likely that a similar scenario is involved in LβT2 cells, although we were not able to test the phosphorylation of the linker, because the antibodies are not commercially available. The aforementioned report, however, did not detect a role for p38 in C-terminal phosphorylation of Smads, as we did. Maybe this difference is due to the cell lines used (transformed carcinoma cells vs. more differentiated gonadotrope-derived cells), different hormones (TGFβ vs. activin), or the cellular process that was assayed (cell growth inhibition vs. activation of cell-specific genes). Alternatively, we may have been able to detect the role for p38 in Smad C-terminal phosphorylation because we very carefully titrated the amount of protein used in the Western blots. We did not detect changes in the total Smad protein content, which also served as loading control, but the change in Smad phosphorylation by cotreatment, although small, was consistent. We were unable to perform chromatin immunoprecipitation assays to assess whether this higher C-terminal phosphorylation results in increased recruitment to the promoter, due to the lack of available antibodies. To our knowledge, there is no report in the literature in which Smad chromatin immunoprecipitation has been performed after activin treatment. Further examination of all of these possible mechanisms by which p38 could affect Smad C-terminal phosphorylation will be a particularly interesting future direction.
LH and FSH are differentially synthesized during the estrous and menstrual cycle. This differential synthesis is crucial for follicle growth and selection. Differential synthesis of the mature hormones is mirrored by the differential transcription of the corresponding β-subunits. LHβ is synthesized at a higher level before the surge to provide high LH levels for ovulation (49). FSHβ exhibits a secondary transcriptional peak on the morning of estrus, and this secondary surge is required for follicle selection and maturation (28, 50, 51). Several hypotheses can explain differential regulation of LH and FSH within the same cell. Changes in GnRH pulse frequency throughout the cycle are proposed to cause differential regulation of LH and FSH (52, 53). Progesterone feedback from the ovary has an inhibitory effect on LH synthesis, whereas it is stimulatory for FSH (Thackray, V. G., and P. L. Mellon, submitted). Our recent report determined that glucocorticoids, similarly to progesterone, also inhibit LH and stimulate FSH and synergize with activin to induce FSHβ (34). Finally, the actions of activin increase during estrus due to the fall in circulating inhibin levels, which preferentially stimulates FSH to much higher levels than LH. Herein, we demonstrate that synergistic interaction of GnRH and activin is specific for FSHβ induction and does not occur for LHβ. This interaction may occur specifically in estrus due to the unopposed actions of activin and may contribute to the secondary peak of FSH.
We determined that synergy between activin and GnRH on the FSHβ gene is not due to activin induction of the GnRH receptor (54). In a previous paper from this laboratory regarding the regulation of ovine FSHβ gene expression (31) and in the current paper concerning the mouse FSHβ gene, follistatin significantly reduced GnRH induction of the FSHβ gene, further indicating that activin and GnRH pathways interact. Moreover, the same report showed (31) that removal of activin by follistatin does not abolish GnRH induction of the LHβ promoter. Herein, synergy was not observed in LHβ induction, indicating that the activin effect cannot be solely through the up-regulation of the GnRH receptor. Furthermore, GnRH and activin responsiveness map to distinct regions of the FSHβ promoter, as demonstrated by truncation analysis. The most proximal region, which retains the most responsiveness to GnRH, does not respond to activin at all, again indicating that the synergy is not due to the induction of the GnRH receptor.
Thus, there is an activin-independent GnRH effect and, at least in part, it involves the −76 AP-1 site. The attempt to identify other GnRH-only responsive sites was unsuccessful and awaits further investigation. Activin also has GnRH receptor-independent effects because the response to activin maps partially to the region upstream of −340, which does not show additional responsiveness to GnRH. That prompts postulation that the interaction between these two hormones is at the level of the promoter and the transcription factors they activate or at the level of the signaling pathways.
The only transcription factor known to date to be involved in GnRH regulation of the FSHβ promoter is AP-1. We previously determined that GnRH induces AP-1, and AP-1 binds the FSHβ promoter. Activin, on the other hand, activates Smad transcription factors, and we and others have determined that Smads bind the FSHβ promoter. Interaction between Smads and AP-1 has been demonstrated (55). Furthermore, Smad and AP-1 can interact and induce the AP-1-responsive element of the human collagenase promoter in response to TGFβ (56). In that report, Smad 3 binding to the DNA was important, but the DNA sequence of the SBE was not critical, making the AP-1 site sufficient. Thus, functional interaction between Smad 3 and c-Jun/AP-1 was dependent on interaction of both proteins with DNA and each other, although the nucleotide sequence requirement for Smad 3 binding was very loose. In contrast, in the FSHβ promoter, both the canonical SBE and the AP-1 sites are important. Multimers of either AP-1 site, −76 or −256, were not induced by either Smad 3 overexpression (data not shown) or with activin treatment. It may be that these particular AP-1 sites don’t contain even the minimal sequence requirement for Smad 3 binding. TGFβ regulation of the c-Jun promoter also required both Smad and AP-1 sites, which Smads and AP-1 proteins bind independently (42). Mutation of the AP-1 site alone, however, abrogated the TGFβ response. In contrast, in our study, the AP-1 site was necessary for the response to GnRH, but not to activin, whereas the Smad site was necessary for the response to activin but not to GnRH. Only with cotreatment were both sites necessary for the observed synergy. In the FSHβ promoter, it is likely that neither hormone alone is sufficient for these sites on the DNA to functionally interact and that Smad and AP-1 proteins are activated by activin and GnRH, respectively, although more highly with cotreatment due to cross-talk at the point of increased p38 activity. Subsequent to activation, Smad and AP-1 proteins bind their corresponding sites on the DNA. Therefore, we postulate that after DNA binding, physical interaction between AP-1 and Smads can stabilize and augment recruitment to low-affinity binding sites or functional cooperation can facilitate the recruitment of coactivators and provide higher affinity for the polymerase complex.
In this report, we demonstrate that both Smad and AP-1 sites are critical and sufficient for synergy. This implies that the factors binding these sites interact. We also showed that cotreatment of GnRH and activin causes both an increased induction of c-Fos protein compared with GnRH treatment alone and a higher level of Smad phosphorylation compared with the activin treatment alone. More importantly, synergy, augmented Smad phosphorylation, and elevated c-Fos expression are all dependent upon p38 activity.
RNA was extracted from LβT2 cells with Trizol reagent (Invitrogen/GIBCO, Carlsbad, CA) according to the manufacturer’s instructions. Contaminating DNA was removed with DNA-free reagent (Ambion, Austin, TX), and 2 μg RNA was reverse transcribed using Superscript III First-Strand Synthesis System (Invitrogen, Carlsbad, CA). Quantitative real-time PCR was performed in an iCycler from Bio-Rad (Hercules, CA), using the iQ SYBR Green Mastermix PCR Kit also from Bio-Rad and the primers FSHβ forward, GCCGTTTCTGCATAAGC; FSHβ reverse, CAATCTTACGGTCTCGTATACC; LHβ forward, CTGTCAACGCAACTCTGG; LHβ reverse, ACAGGAGGCAAAGCAGC; GAPDH forward, TGCACCACCAACTGCTTAG; and GAPDH reverse, GGATGCAGGGATGATGTTC under the following conditions: 95 C for 15 min, followed by 40 cycles at 95 C for 15 sec, 54 C for 30 sec, and 72 C for 30 sec. For LHβ measurements, the equivalent of 50 ng starting RNA (as quantified before reverse transcription) was used in each reaction, and for FSHβ measurements, 200 ng was used, due to low basal expression, whereas for GAPDH, 10 ng was sufficient. Each sample was assayed in triplicate, and the experiment was repeated four times. A standard curve with dilutions of 10 pg/well, 1 pg/well, 100 fg/well, and 10 fg/well of a plasmid containing LHβ cDNA, FSHβ cDNA, or GAPDH cDNA was generated in each run with the samples. In each experiment, the amount of FSHβ or LHβ was calculated by comparing a threshold cycle obtained for each sample with the standard curve generated in the same run. Replicates were averaged and divided by the mean value of GAPDH in the same sample. After each run, a melting curve analysis was performed to confirm that a single amplicon was generated in each reaction.
LβT2 cells were plated in 12-well plates 1 d before transfection. Transfection was performed in DMEM with 10% fetal bovine serum using Fugene 6 reagent (Roche Molecular Biochemicals, Indianapolis, IN) following the manufacturer’s instructions. Each well was transfected with 0.5 μg mouse FSHβ-luciferase plasmid. Plasmid construction and preparation was described previously (11). The sequences of all of the promoter fragments were confirmed by dideoxynucleotide sequencing. An expression plasmid containing β-galactosidase driven by the herpes virus TK promoter was cotransfected with FSHβ-luciferase and used as an internal control. Expression vectors (0.2 μg/well) for Smad 3 (provided by Joan Massague) or Smad 7 (provided by Rik Derynck) or their empty vector controls were cotransfected, as indicated in specific experiments. The expression vector for the dominant-negative p38 kinase was obtained from Jiahuai Han and cotransfected as indicated. Sixteen hours after transfection, the cells were switched to serum-free DMEM supplemented with 0.1% BSA, 5 mg/liter transferrin, and 50 nm sodium selenite. The following day, the cells were treated with 10 nm GnRH (Sigma Chemical Co., St. Louis, MO), 10 ng/ml activin (Calbiochem, La Jolla, CA), or with the combination of the two hormones for 5 h unless otherwise indicated. In Fig. 2, follistatin (R&D Systems, Minneapolis, MN) was added to the cells for the same amount of time. The cells were then lysed with 0.1 m K-phosphate buffer (pH 7.8) with 0.2% Triton X-100. Equal volumes of each lysate were placed in 96-well plates, and luciferase activity was measured on a luminometer (Berthold Technologies, Oak Ridge, TN) by injecting 100 μl of a buffer containing 100 mm Tris-HCl (pH 7.8), 15 mm MgSO4, 10mm ATP, and 65 μm luciferin per well. Galactosidase activity was measured using the Galacto-light assay (Tropix, Bedford, MA) following the manufacturer’s instructions. All transfection experiments were performed in triplicate and repeated at least three times. Luciferase values from reporter gene-transfected cells were consistently at least 100 times higher than values from mock-transfected cells. Results represent the mean ± SEM of three repeated experiments. P < 0.05 was considered statistically significant, as determined by ANOVA and by a two-way approach and the method of Slinker (29) to determine synergy after cotreatment, followed by post hoc comparisons with Tukey-Kramer honestly significant difference test, using the statistical package JMP 5.0 (SAS, Cary, NC). Asterisks mark statistically significant differences between single-hormone treatment and the control-treated cells and pound signs mark synergistic interaction, unless otherwise indicated in the figure legends.
Cos 1 cells, transfected for 24 h with expression vectors containing Smad 3 and Smad 4 cDNAs or vector control and kindly provided by Rik Derynck, or LβT2 cells, after treatment with GnRH or activin, were scraped in hypotonic buffer [20 mm Tris-HCl (pH 7.4), 10 mm NaCl, 1 mm MgCl2 with protease inhibitors (aprotinin, pepstatin, and leupeptin at 10 μg/ml each), and 1 mm phenylmethylsulfonyl fluoride] and allowed to swell on ice. Cells were broken open by passing through a 25-gauge needle and the nuclei pelleted by centrifugation. Nuclear proteins were extracted in hypertonic buffer [20 mm HEPES (pH 7.8), 420 mm KCl, 1.5 mm MgCl2 with protease inhibitors, and 20% glycerol]. Two micrograms of nuclear protein per sample were used in the binding reaction [10 mm HEPES (pH 7.8), 50 mm KCl, 5 mm MgCl2,5mm dithiothreitol, 0.1% BSA, 0.1% Nonidet P-40 with 0.5 μg/ml poly-dIdC, and 2 fmol per reaction of end-labeled probe]. Oligonucleotides were labeled with T4 kinase using [γ-32P]ATP. In the competition experiments, competitor oligonucleotides were added 10 min before addition of the probe, as were the antibodies in the supershift assays. The Smad 4 and c-Fos antibodies and nonspecific IgG were obtained from Santa Cruz Biotechnology (Santa Cruz, CA). The reaction was loaded on a 5% acrylamide gel in 0.25× Tris-borate-EDTA and run on 0.5 V/cm2 constant voltage. After drying, gels were exposed to autoradiography.
Mutagenesis of the FSHβ-luciferase plasmid was performed using the QuikChange Site-Directed Mutagenesis Kit (Stratagene, La Jolla, CA) according to the manufacturer’s protocol. From the wild-type sequence, the −267 SBE site GTCTAGAC was mutated to GTAGCTAC to create a Smad site mutant. Mutation of the downstream −76 AP-1 site was previously described (11), whereas the −256 AP-1 site AGAGTCA was mutated to AGAAACA. Mutations were confirmed by dideoxyribonucleotide sequencing performed by the DNA Sequencing Shared Resource, University of California, San Diego, Cancer Center, which is funded in part by National Cancer Institute Cancer Center Support Grant P30 CA23100.
After overnight starvation and hormone treatment, LβT2 cells were rinsed with PBS and lysed with lysis buffer [20 mm Tris-HCl (pH 7.4), 140 mm NaCl, 0.5% Nonidet P-40, 0.5 mm EDTA, with protease inhibitors (aprotinin, pepstatin, and leupeptin at 10 μg/ml each), and 1 mm phenylmethylsulfonyl fluoride]. Protein concentrations were determined with Bradford reagent (Bio-Rad), and an equal amount of protein per sample was loaded on SDS-PAGE gels. After proteins had been resolved by electrophoresis and transferred to a polyvinylidene fluoride membrane, they were probed with specific antibodies for Fos, Jun, phospho-c-Jun, phospho-Smad 2/3 (sc-11769, made in rabbit), or Smad 2/3 (sc-6032) (Santa Cruz Biotechnology). The bands were detected with secondary antibodies linked to horseradish peroxidase and enhanced chemiluminescence reagent (Amersham Pharmacia, Piscataway, NJ). Western blots were quantified by Gene-Gnome Bio Imaging chemiluminescence reader (Syngene, Frederick, MD).
We thank Rik Derynck who kindly provided the Smad 3, 4, and 7 expression vectors, Joan Massague who provided Smad 3 wild-type and phosphorylation site mutant expression vectors, and Jiahuai Han for the plasmid containing the dominant-negative p38 cDNA. We appreciate the time and insight of many Mellon laboratory members who participated in helpful discussions and the technical assistance of Susan Mayo.
This research was supported by National Institute of Child Health and Human Development, National Institutes of Health (NIH), through a cooperative agreement (U54 HD012303) as part of the Specialized Cooperative Centers Program in Reproduction Research (P.L.M.). This work was also supported by NIH Grant R01 HD020377 (to P.L.M.). D.C. was supported by NIH NRSA F32 HD41301 and the Lalor Foundation. H.A.E. was supported by the Howell Foundation. P.L.M. is a member of the Biomedical Sciences Graduate Program.
Disclosure Summary: The authors have nothing to disclose.