TGF-β Superfamily Regulation of Follicle-Stimulating Hormone Synthesis by Gonadotrope Cells: Is There a Role for Bone Morphogenetic Proteins?

Abstract

Bone morphogenetic proteins (BMPs) are pleiotropic ligands in the TGF-β superfamily. In the early to mid-2000s, several BMPs, including BMP2, were shown to regulate FSH synthesis alone and in synergy with activins in immortalized gonadotrope-like cell lines and primary pituitary cultures. Activins are also TGF-β family members, which were identified and named based on their abilities to stimulate FSH production selectively. Mechanistic analyses suggested that BMP2 promoted expression of the FSHβ subunit gene (Fshb) via at least two nonmutually exclusive mechanisms. First, BMP2 stimulated the production of the inhibitor of DNA-binding proteins 1, 2, and 3 (Id1, Id2, and Id3), which potentiated the stimulatory actions of homolog of Drosophila mothers against decapentaplegic 3 (SMAD3) on the Fshb promoter. SMAD3 is an intracellular signaling protein that canonically mediates the actions of activins and is an essential regulator of Fshb production in vitro and in vivo. Second, BMP2 was shown to activate SMAD3-dependent signaling via its canonical type IA receptor, BMPR1A (also known as ALK3). This was a surprising result, as ALK3 conventionally activates distinct SMAD proteins. Although these initial results were compelling, they were challenged by contemporaneous and subsequent observations. For example, inhibitors of BMP signaling did not specifically impair FSH production in cultured pituitary cells. Of perhaps greater significance, mice lacking ALK3 in gonadotrope cells produced FSH normally. Therefore, the physiological role of BMPs in FSH synthesis in vivo is presently uncertain.

Introduction: Bone Morphogenetic Proteins, in Brief

Bone morphogenetic proteins (BMPs) are secreted ligands in the TGF-β superfamily (1). They were first extracted from demineralized bone and described as regulators of osteogenesis and chondrogenesis (2). However, in the last 30 years, it has been demonstrated that BMPs also carry out important roles in neurogenesis, organogenesis (3–5), pregnancy (6), cancer (7), and cardiovascular diseases (8). The BMP subfamily of TGF-β ligands has been classified further into at least four different subgroups based on sequence homology and phylogenetic analyses: (i) BMPs 2 and 4; (ii) BMPs 5, 6, 7, and 8; (iii) BMPs 9 and 10; and (iv) BMPs 12, 13, and 14 [also known as growth differentiation factors (GDFs) 5, 6, and 7]. BMP3 and BMP15 share greater sequence homology with the GDF ligands GDF10 and GDF9, respectively (9, 10).

Receptors for BMPs and Other TGF-β Family Members

Like other TGF-β ligands, BMPs signal through heterotetrameric type I and type II serine/threonine kinase receptor complexes (Fig. 1). The composition of these complexes determines the nature of the resulting intracellular signaling pathways. There are five type II receptors in mammals: activin receptor type IIA (ACVR2A) and IIB (ACVR2B), TGF-β receptor type II (TGFBR2), anti-Müllerian hormone receptor type II, and BMP receptor type II (BMPR2) (10, 11). BMPR2 and ACVR2A and, to a lesser extent, ACVR2B are generally considered the main type II receptors used by BMPs (12, 13).

Figure 1.

Proposed models of activin and BMP regulation of FSHβ subunit (Fshb) and inhibitor of DNA-binding protein (Id)1/2/3 transcription. Activin and BMP signaling are depicted on the left and right, respectively. ACVR2A, activin receptor type II A; ACVR2B, activin receptor type IIB; ALK, activin receptor-like kinase, type I receptor; BMPR2, BMP receptor type II; FOXL2, forkhead box L2; LDN193189, small molecule ALK2/3/6 inhibitor; SMAD, homolog of Drosophila mothers against decapentaplegic.

The nature of ligand–receptor interactions varies among TGF-β ligands, including BMPs. In some cases, the ligands first bind to type II receptors before the recruitment of type I receptors, for example, with BMP6 and BMP7. In contrast, BMP2 and BMP4 first bind type I receptors or perhaps to preassembled type I/II receptor complexes (14). Regardless of the order of receptor engagement, upon ligand binding, type I and type II receptors form higher-order tetrameric complexes in which two type II receptors trans-phosphorylate the two associated type I receptors, activating them (15).

Seven type I receptors have been identified in mammals. These receptors are often referred to as activin receptor-like kinases (ALKs) (16). Activins are also TGF-β ligands, and ACVR2A was the first receptor in the family identified (17). As most of the remaining receptors were subsequently cloned via homology, the ALK terminology was adopted in the course of the naming and characterization of the type I receptors. The canonical type I receptors for BMPs are ALK1 (also known as ACVR-like 1), ALK2 [ACVR type I (ACVR1)], ALK3 [BMP receptor type IA (BMPR1A)], and ALK6 [BMP receptor type IB (BMPR1B)]. The remaining type I receptors classically mediate the actions of activins, nodal, TGF-βs, and several GDFs: ALK4 [ACVR type IB (ACVR1B)], ALK5 [TGFBR type I (TGFBR1)], and ALK7 [ACVR type IC (ACVR1C)] (10, 11, 18). Activated type I receptors initiate intracellular signaling via phosphorylation of intracellular proteins in the homolog of Drosophila mothers against decapentaplegic (SMAD) family (10, 19).

Intracellular Signaling via SMAD Proteins

SMADs are classified into three subtypes: the receptor-regulated SMADs (R-SMADs), SMADs 1, 2, 3, 5, and 8; the common partner SMAD, SMAD4; and inhibitory SMADs, SMADs 6 and 7 (19). TGF-β ligands typically signal through one of two broadly defined pathways: the SMAD2/3 pathway, which is activated by the type I receptors ALK4/5/7, and the SMAD1/5/8 pathway, which is activated by ALK1/2/3/6 (11). Receptor-substrate recognition is governed by a solvent-exposed loop between subdomains IV and V of the type I receptor kinase (the L45 loop) (20) and by the so-called L3 loop region in the C-terminal MAD homology 2 domain of the R-SMADs. As BMPs typically bind ALK1, ALK2, ALK3, and/or ALK6, they activate the SMAD1/5/8, or BMP, pathway (10, 11, 19, 21). It should be noted that ALK1 expression is restricted to endothelial cells and principally mediates the actions of BMP9/10 (18). Therefore, for the remainder of this mini review, we will focus on ALK2/3/6 when referring to BMP type I receptors.

Type I receptors phosphorylate R-SMADs on C-terminal serine residues. Once phosphorylated, these SMADs dissociate from the receptors and form oligomeric complexes with SMAD4 in the cytosol (19, 22). The R-SMAD/SMAD4 complexes then accumulate in the nucleus, where they act as DNA-binding transcription factors (Fig. 1). All R-SMADs (except full-length SMAD2) and SMAD4 possess a β-hairpin in their N-terminal MAD homology 1 domains that enables their binding to the major groove of DNA (23, 24).

BMP Regulation of Reproductive Physiology

As indicated above, BMPs play pleiotropic roles throughout the body and during various stages of life. In recent years, these ligands have been implicated in reproductive function (25). Endocrine regulation of reproductive physiology is controlled by hormones and growth factors from the gonads, hypothalamus, and anterior pituitary gland (26). BMPs are highly expressed in all of these tissues and play important roles therein (27–32). For example, several BMPs regulate ovarian follicle recruitment, maturation, and ovulation in females, as well as spermatogenesis in males (28–30). At the level of the hypothalamus, BMP7 modulates dendrite formation in gonadotropin-releasing hormone neurons (33). In the pituitary gland, BMPs play critical roles in organogenesis and cell-lineage specification (34, 35). BMP2 and BMP4 regulate the commitment of precursor cells to the adrenocorticotropin-producing corticotrope and TSH-producing thyrotrope cell lineages (36). Pituitary gonadotrope cells produce two glycoprotein hormones: FSH and LH, which regulate critical aspects of gonadal function (26). In recent years, BMPs have been implicated in the selective regulation of FSH synthesis and secretion by gonadotropes (37, 38).

Other proteins in the TGF-β superfamily—the activins and inhibins—are better known for their roles in FSH regulation (39, 40). According to current dogma, activins are produced within the pituitary and stimulate FSH production (41, 42). Inhibins are synthesized by the gonads and feedback to the pituitary to suppress FSH (43–45). FSH and LH are dimeric proteins that share an α subunit and have hormone-specific β subunits. Activins and inhibins selectively regulate FSH by increasing and decreasing, respectively, the transcription of the FSHβ subunit gene (Fshb) (46, 47). Several lines of evidence suggest that BMPs may similarly regulate Fshb expression (48, 49). To place these BMP effects in context, we first briefly review mechanisms of activin and inhibin action in gonadotrope cells.

Activin and Inhibin Regulation of FSH Production

Activins come in three forms: activin A, activin B, and activin AB (50). In vitro, these ligands can bind to three different type II receptors: ACVR2A, ACVR2B, and BMPR2 (13, 51, 52). All three activins bind and signal via the type I receptor ALK4, whereas activin B and AB can also use ALK7 (53, 54). ALK4 and ALK7 phosphorylate SMAD2 and SMAD3 (20, 55). As SMAD2 is likely to be dispensable for Fshb expression (56, 57), we focus on SMAD3 hereafter. In gonadotrope cells, phospho-SMAD3 forms complexes with SMAD4 and the forkhead transcription factor forkhead box L2 (FOXL2) (58). SMAD3/4-FOXL2 bind to the Fshb promoter to drive its transcription (57–60) (Fig. 1). Mice lacking SMAD3, SMAD4, and/or FOXL2 in gonadotropes are FSH deficient and either subfertile or sterile (56, 61, 62). Inhibins suppress Fshb expression (39, 44, 45) by competing with activins (and/or related ligands) for binding to type II receptors (63). There are two forms of inhibin: inhibin A and inhibin B. Inhibin A binds to ACVR2A with high affinity but only in the presence of a coreceptor, betaglycan [also known as TGFBR type III (TGFBR3)]. Inhibin B antagonism of activin signaling is betaglycan independent in mice in vivo but may similarly rely on a coreceptor for high-affinity binding to ACVR2A (64).

The specific type I and type II receptors mediating activin actions in gonadotropes in vivo have not yet been resolved. FSH levels are significantly reduced in Acvr2a global knockout mice (17). Residual FSH production in these animals may reflect compensation by one or more type II receptors. Global Bmpr2 or Acvr2b knockout mice die during embryonic development or shortly after birth, respectively, precluding an assessment of the role of these receptors in FSH synthesis in vivo (65, 66). Based on in vitro evidence, ACVR2A and BMPR2, but not ACVR2B, mediate activin A- and activin B-induced Fshb transcription (13). Gonadotrope-specific (conditional) knockout animals of the different type II receptors, alone or in combination, will elucidate the relative role(s) of these receptors in Fshb expression in vivo.

With respect to type I receptors, cultured pituitaries of ALK7 (product of the Acvr1c gene) knockout mice produce FSH normally, both basally and in response to activin A or B (67). As such, activins are thought to signal principally through ALK4 to stimulate Fshb expression in gonadotropes. Acvr1b mice die before gastrulation, precluding their use to assess the receptor’s role in FSH production in vivo (68). Here, too, a conditional knockout approach will be necessary. Of note, Fshb expression and/or promoter activity from several species (e.g., mouse, rat, sheep) are nearly abolished (27, 69, 70) by a small molecule ALK4/5/7 inhibitor, SB431542 (71) (e.g., Fig. 2). These data indicate that autocrine/paracrine activins (or related ligands) signal via one or more of these type I receptors to stimulate FSH production. The precise identity of the ligand(s) is not yet clear. In rat pituitary cultures, an activin B bioneutralizing antibody suppresses FSH synthesis and secretion (72). In murine pituitary cultures, neither activin A nor activin B antibodies suppress FSH (38). Moreover, activin B-deficient mice appear to have elevated, rather than reduced, FSH levels (73). Thus, it will be critical to determine both the relevant TGF-β ligand(s) and its/their type I receptors to appreciate fully mechanisms underlying FSH regulation in vivo. Data collected over the past two decades suggested a potential role for BMPs in FSH synthesis.

Figure 2.

Noggin does not regulate Fshb expression in murine pituitary cultures. Pituitary cells were isolated from male wild-type C57BL/6 mice and treated in culture with noggin (30 or 300 ng/mL) or SB431542 (384 ng/mL, 1 µM) for 24 hours. Fshb mRNA expression was determined by quantitative RT-PCR. Data represent the means (+SEM) of three to four independent experiments. ****P < 0.0001, one-way ANOVA, followed by Sidak multiple comparisons.

Regulation of FSH Synthesis by BMPs in Vitro

Like activins, BMP7 (and BMP6) can bind ACVR2A (74). Therefore, reduced FSH levels in Acvr2a global knockout mice cannot unequivocally be linked to impaired activin action (17). Notably, a BMP7-bioneutralizing antibody significantly impaired luciferase activity in pituitary cultures derived from transgenic mice in which the ovine Fshb promoter drives a luciferase reporter (oFshb-luc) (38). The same antibody suppressed endogenous FSH secretion from murine, rat, and ovine pituitary cultures, suggesting that BMP7 may be produced in the pituitary and stimulate FSH production across species, at least in vitro. Bmp7 mRNA is expressed in primary murine gonadotropes and in an immortalized murine gonadotrope-like cell line, LβT2 cells (27, 31). These observations motivated further investigations of BMP signaling in gonadotropes, specifically in the context of FSH regulation.

Like BMP7, BMP15 stimulated Fshb expression in LβT2 cells and in rat pituitary cultures (75). However, BMP15 is not expressed in the pituitary in vivo (31). BMP3 did not regulate luciferase activity in primary cultures from the ovine Fshb luciferase reporter (oFshb-luc) transgenic mice (38). BMP2 (and BMP4) stimulated murine, porcine, and ovine Fshb promoter-reporter activities alone and synergistically with activin A in LβT2 cells (27). To our knowledge, BMP-dependent regulation of the rat Fshb and human FSHB promoters has not been investigated. BMP2 and BMP4 also synergistically stimulated Fshb mRNA expression and/or FSH release with activin A in LβT2 cells (27, 37, 76). Bmp2/4 mRNA levels are low in these cells and primary gonadotropes, but both genes are expressed in other pituitary cell lineages (34, 77), indicating that these ligands are more likely to act as paracrine rather than autocrine factors in gonadotropes.

Mechanisms of BMP2 Action in Vitro

BMP2 was 10-fold more potent than BMP6 or BMP7 in inducing murine Fshb transcription (27). For this reason and because the concentrations of BMP6/7 needed to stimulate FSH were often supraphysiological (38, 74, 75), most subsequent studies used BMP2 or BMP4. Importantly, exogenous BMP4 inhibited Fshb expression and FSH secretion in cultured ewe pituitaries (78, 79). However, the activin type I receptor inhibitor SB431542, but not the BMP2/4 inhibitor noggin (80), suppressed FSH secretion from these cultures, suggesting that endogenous activins (or related ligands), rather than BMP2/4, are the primary regulators of FSH (79). Collectively, these data indicate that BMP effects on FSH may be both ligand and context/species specific.

Initial mechanistic studies suggested that BMP2 might regulate murine Fshb transcription via the receptors BMPR2, ACVR2A, and ALK3 (48) and via the receptor-regulated SMAD, SMAD8 (27). However, the data for SMAD8 derived exclusively from overexpression analyses in LβT2 cells. Moreover, Smad8 knockout mice are viable and fertile (81), although it should be noted that FSH levels were not reported in these animals. Subsequently, two nonmutually exclusive mechanisms of BMP2 regulation of Fshb were articulated (Fig. 1). First, BMP2 was shown to stimulate the expression of inhibitor of DNA-binding proteins 1, 2, and 3 (Id1, -2, and -3) in LβT2 cells. BMP2 signaled via SMAD1 and SMAD5, but not SMAD8, to regulate Id3 transcription (82). BMPs similarly regulate Id1 transcription via SMAD1/5 in other cells (83–85). Importantly, all three Id proteins interacted with SMAD3, the R-SMAD through which activins regulate Fshb transcription (86). Synergistic actions of BMP2 and activin A were Id2 and Id3 dependent, and the stimulatory effects of overexpressed SMAD3 on murine Fshb promoter activity were potentiated by coexpressed Id1, -2, or -3 (86). These data suggested that the combined actions of SMAD3 and Id proteins might explain, at least in part, the synergistic induction of Fshb transcription by activins and BMPs (Fig. 1). How Id proteins modify SMAD3 function has not yet been resolved.

Second, according to in vitro data, BMP2 also acted directly via SMAD3 to regulate Fshb (49). As reviewed above, BMPs typically signal via the type I receptors ALK2/3/6 and the associated SMAD1/5/8. However, in LβT2 cells, BMP2 stimulated both SMAD1/5/8- and SMAD2/3-dependent signaling (49). BMP2 induced the expression of the subunit that comprises activin B [inhibin βB (Inhbb)] (27), providing a means for indirect activation of the SMAD2/3 pathway. BMP2 also stimulates Inhbb expression in ovarian granulosa cells (87). It is important to note that an increase in Inhbb expression could affect activin B or inhibin B production (or both). However, neither an activin B antibody nor the activin type I receptor SB431542 completely blocked BMP2 activation of SMAD2/3 signaling in these cells (49). Instead, BMP2 acted via its canonical type I receptor ALK3 to directly stimulate SMAD3 phosphorylation and action.

Interestingly, whereas a constitutively active form of ALK4 (and ALK5 or ALK7) (54, 88) significantly increased Fshb promoter activity, constitutively active ALK3 (or ALK2 or ALK6) could not when expressed on its own (27, 48, 49). However, when coexpressed with BMPR2 (short isoform only) or activin type II receptors, constitutively active ALK3 robustly stimulated Fshb promoter activity (27, 48, 49). Likewise, constitutively active ALK3 only stimulated SMAD2 and SMAD3 phosphorylation in the presence of coexpressed type II receptors (49). As shown with in vitro kinase assays, ALK3 can recognize SMAD3 as a substrate, particularly in the presence of BMPR2. It should be noted that constitutively active ALK3 stimulated the SMAD1/5/8 pathway and Id expression independently of type II receptor expression, suggesting that the primary actions of BMP2 on Fshb expression may be via SMAD3 rather than Id proteins (27, 49) (Fig. 1).

In Vitro Evidence Challenging a Role for BMPs in FSH Regulation

Although the above data indicate that BMPs can regulate Fshb expression/transcription and highlight potential mechanisms underlying the actions of BMPs, many of the studies relied on exogenous ligands, overexpression analyses, and immortalized cells (in particular, LβT2 cells). Therefore, the physiological relevance of BMPs was not demonstrated with the majority of these approaches. Of notable exception, and as reviewed above, a BMP7 antibody reduced FSH secretion from primary pituitary cultures of three different species (38). However, the effects were modest, particularly in rodents, and the antibody crossreacted with BMP6. Although it did not crossreact with activin A, it is not clear whether the antibody inhibited the actions of other TGF-β family ligands. Furthermore, as reviewed above, the BMP inhibitor noggin did not impact FSH secretion from ovine pituitary cultures (79). Noggin, in contrast to the ALK4/5/7 inhibitor, also failed to impair Fshb mRNA expression in murine pituitary cultures (Fig. 2). Noggin inhibits BMP ligands with different potencies (80, 89, 90). Therefore, it is possible that BMPs with lower affinity for noggin remain capable of regulating FSH production in culture.

As all BMPs signal via ALK2/3/6, we treated cultured murine pituitaries with the small molecule ALK2/3/6 inhibitor LDN-193189 (91). LDN-193189 significantly reduced Fshb mRNA levels without affecting expression of the LHβ subunit (Lhb) (92). These results were similar to those with the ALK4/5/7 inhibitor (SB431542) and suggested that an endogenous ligand that signals via ALK2/3/6 stimulates Fshb expression in these cultures. However, the effects of LDN-193189 on Fshb were also observed in primary cells in which ALK2 and ALK3 were ablated using a Cre-lox recombination strategy (92). Moreover, deletion of ALK2/3 did not affect basal Fshb expression but blocked the effect of exogenous BMP2 on Id1 expression (92). These data suggest that the effects of LDN-193189 were nonspecific. ALK6 is expressed at very low levels in murine pituitary and was not upregulated in the ALK2/3 knockdown cultures (31, 77, 92). Therefore, it is unlikely that compensation by ALK6 occurred. The inhibition of BMP2 action on Id1 expression by ALK2/3 ablation was also consistent with this idea. Collectively, these data cast doubt on vital roles for endogenous BMPs in FSH synthesis, at least in cultured pituitaries.

Equivocal Role for BMP Signaling in FSH Synthesis in Mice in Vivo

Whereas BMP regulation of Fshb expression has been actively investigated in vitro, the roles of these proteins and their receptors in vivo have received comparatively little attention. As in vitro studies indicated that BMP2 preferentially signaled via the type I receptor, ALK3 (encoded by Bmpr1a), we used a conditional (Cre-lox) knockout approach to interrogate the receptor’s function in gonadotropes in mice (92). FSH synthesis and fertility were fully intact in gonadotrope-specific Bmpr1a knockout mice of both sexes. BMP2 induction of Id3 mRNA expression was blocked in gonadotropes purified from conditional knockouts, indicating that ALK3 had been functionally ablated (92). These data showed that ligands that signal exclusively via ALK3 are not required for quantitatively normal FSH production in mice. However, it was possible that a BMP ligand, such as BMP7, which preferentially signals via alternative type I receptors, such as ALK2 (encoded by Acvr1) or ALK6 (encoded by Bmpr1b), or via multiple (redundant) receptors, could compensate for the loss of ALK3. To address this possibility, we generated mice lacking ALK2 and ALK3 specifically in gonadotropes. Litter size was also normal in these animals (92).

Whereas these data suggest that BMP signaling via ALK2 and ALK3 in gonadotropes is dispensable for FSH production and/or fertility in mice, there are a few important caveats to consider. First, BMPs can also signal via ALK6. However, as noted above, ALK6 is expressed at low levels in the pituitary (77, 93), making compensation unlikely. Second, FSH production was not measured in gonadotrope-specific ALK2 or ALK2/3 knockout mice. Nonetheless, normal litter sizes in the double knockouts suggest that FSH production was likely intact. Indeed, ablation of the two receptors in cultured pituitaries impaired BMP2 stimulation of Id1 expression without affecting Fshb levels. Finally, the Cre-driver used becomes active during embryogenesis (embryonic day 12.75) (94). Gonadotrope plasticity and developmental compensatory mechanisms could therefore mask a role for BMP signaling via ALK2/3 during adult life. Ablation of the receptors in gonadotropes of adult mice would permit a direct test of this possibility (95).

Conclusions

Despite the excitement generated by initial findings in the early 2000s, the available in vitro and in vivo data now indicate that BMPs may play only a minor role in the regulation of FSH synthesis by pituitary gonadotropes. That said, in vivo analyses have, thus far, been limited to mice and to BMP type I receptors therein. As the identities of the TGF-β ligands regulating FSH production in vivo have not yet been fully cataloged, we cannot definitely rule out roles for all BMPs. Nevertheless, the TGF-β ligands that stimulate FSH production appear to act via the type I receptors ALK4, ALK5, and/or ALK7, none of which are conventionally considered BMP receptors. Indeed, we are unaware of any BMPs that signal via these type I receptors. ALK4/5/7 classically activate the SMAD2/3 pathway, and it is now clear that FSH production, at least in mice, depends on SMAD3 and SMAD4 (but not SMAD2) (56). Roles for SMADs 1, 5, and 8 (the BMP SMADs) in FSH synthesis have not been systematically investigated in vivo, but it is clear that none can compensate for the loss of SMAD3. As described above, BMP2 can activate the SMAD2/3 pathway in vitro, but this activity is driven via ALK3, which appears dispensable for FSH production in vivo. The characterization of the specific type I and type II receptors required for FSH production by gonadotropes in vivo will greatly aid efforts to identify the relevant endogenous TGF-β ligands, whether they are BMPs and/or other members of the family.

Glossary

Abbreviations:

ACVR

activin receptor

ALK

activin receptor-like kinase

BMP

bone morphogenetic protein

BMPR

bone morphogenetic receptor

FOXL2

forkhead box L2

Fshb

FSHβ subunit gene

GDF

growth differentiation factor

Id

inhibitor of DNA-binding protein

Inhbb

inhibin βB

LDN-193189

small molecule activin receptor-like kinase, type I receptor 2/3/6 inhibitor

R-SMAD

receptor-regulated homolog of Drosophila mothers against decapentaplegic

SMAD

homolog of Drosophila mothers against decapentaplegic

TGFBR

TGF-β receptor