Activin was discovered in the 1980s as a gonadal protein that stimulated FSH release from pituitary gonadotropes and was thought of as a reproductive hormone. In the ensuing decades, many additional activities of activin were described and it was found to be produced in a wide variety of cell types at nearly all stages of development. Its signaling and actions are regulated intracellularly and by extracellular antagonists. Over the past 5 years, a number of important advances have been made that clarify our understanding of the structural basis for signaling and regulation, as well as the biological roles of activin in stem cells, embryonic development and in adults. These include the crystallization of activin in complex with the activin type II receptor ActRIIB, or with the binding proteins follistatin and follistatin-like 3, as well as identification of activin's roles in gonadal sex development, follicle development, luteolysis, β-cell proliferation and function in the islet, stem cell pluripotency and differentiation into different cell types and in immune cells. These advances are reviewed to provide perspective for future studies.
Inhibin and activin were purified from gonadal fluids based on their ability to inhibit or stimulate (respectively) release of FSH from cultured pituitary gonadotropes. Following purification and molecular characterization, these hormones were identified as members of the TGF-β superfamily, which now includes more than 40 ligands including TGF-β, inhibin, myostatin, and bone morphogenetic proteins (BMPs; Vale et al. 1988). Activin receptors were discovered later by functional cloning, revealing a type II receptor with serine/threonine kinase activity as the binding moiety (Mathews & Vale 1993), which was later found to phosphorylate a type I receptor after ligand binding (Tsuchida et al. 2008). These discoveries along with the identification and characterization of extracellular activin antagonists such as follistatin (FST) and follistatin-like 3 (FSTL3; Welt et al. 2002), spurred a torrent of research activity into activin's biological roles in a wide spectrum of tissues and systems ranging from fate determination in embryos to homeostatic mechanisms in adults (reviewed in Mather et al. (1992), (1997), DePaolo (1997), Welt et al. (2002)). One of the continuing challenges in this field has been to decipher the in vivo activities of ligands that act primarily through autocrine and paracrine mechanisms where the value of the more classical endocrinological ablation and replacement paradigms is limited. Moreover, the requirement for activin during development has restricted the utilization of global knockout technology to address this issue. The more recent advent of conditional knockout technology, regulatable expression systems, and siRNA suppression technology is opening the door for more detailed investigations of activin's tissue-specific roles in adults.
The basics of activin signaling have been known for some time, although emerging details of cross-talk with other signaling pathways and an increasingly complex web of intracellular regulators adds new wrinkles at a regular pace (reviewed in Tsuchida et al. (2008)). This review will first focus on recent advances in our understanding of the structural basis for activin's interactions with its receptors and antagonists along with new information about how activin signaling and biosynthesis are regulated, after which we will summarize advances in our understanding of activin's biological roles, primarily in adults, published since our last comprehensive review (Welt et al. 2002). The role of activin in endocrine cancers, while important, was too vast to be included in this review.
Advances in activin structure–function relationships
Activin-receptor crystal structures
The canonical activin signaling pathway involves an activin dimer binding to one of two type II receptors (ActRIIA or ActRIIB) in complex with activin's type I receptor ActRIB (a. k. a, activin receptor-like kinase 4, Alk4), which ultimately leads to phosphorylation of the type I receptor. This receptor then phosphorylates the second messenger molecules Smad2 and Smad3 that, once activated, complex with a common Smad, Smad4. This complex then translocates to the nucleus where it activates gene transcription. Smad7 was identified as an inhibitory Smad that blocks phosphorylation of Smads 2 and 3, thereby providing a short-loop negative feedback regulation for this signaling pathway (reviewed in Tsuchida et al. (2008)). Another level of regulation occurs extracellularly through soluble antagonists like FST and FSTL3 (Schneyer et al. 2001) or membrane bound modifiers like BMP and activin receptor membrane bound inhibitor (BAMBI) and Cripto (a GPI-anchored membrane protein, which inhibits activin signaling by forming an ‘inert’ complex with activin and ActRIIA; Onichtchouk et al. 1999, Kelber et al. 2008). Despite focused efforts; crystallization of activin alone was unsuccessful, suggesting that there may be considerable flexibility in how the two disulfide-linked subunits of activin are aligned. This was confirmed when activin A was crystallized in complex with ActRIIB extracellular domain (Thompson et al. 2003). In this structure, ActRIIB binds to the outer edges of the so-called finger regions on the activin dimer. Interestingly, the activin dimer in this model was tightly folded into a compact conformation unlike the extended configuration of other TGF-β family members previously crystallized (Griffith et al. 1996, Hinck et al. 1996). Activin's flexibility and compact conformation was confirmed in a more highly resolved crystal structure with ActRIIB and suggested that this movement allowed binding sites for the type I receptor to be exposed once the type II receptor sites were occupied, thereby suggesting a mechanism for how the hexameric complex consisting of the activin dimer with two type II and two type I receptors is organized and activated (Greenwald et al. 2004). Support for this model comes from the remarkable achievement of crystallization of the ternary signaling complex comprised of a related ligand, BMP-2 and both receptor subtypes (Allendorph et al. 2006). These results showed that the ligand is the central coordinator of the complex with an absence of direct contacts between the receptors themselves (Allendorph et al. 2006). Moreover, they show that receptor specificity depends on specific unique residues arranged around a conserved hydrophobic core on type I receptors (Allendorph et al. 2006).
Activin antagonist crystal structures
FST, originally identified and isolated from both bovine and porcine follicular fluids on the basis of its inhibition of pituitary FSH secretion, is a monomeric glycosylated polypeptide chain that binds activin with high affinity and neutralizes most of its biological actions. FSTL3 was cloned from a B-cell leukemia line and it was originally named FST-related gene (FLRG) based upon primary sequence homology to FST. Subsequently, FSTL3 was also found to be a high-affinity activin binding protein (reviewed in Welt et al. (2002)).
As detailed previously, point mutagenesis and domain swapping studies were used to determine portions of the FST molecule that were critical for its neutralization of activin activity (Welt et al. 2002). FST is comprised largely of three successive 73–77 residue, 10-cysteine FST domains (FSDs), preceded by a 63-residue N-terminal domain (Welt et al. 2002). Residues on FSD2 and on the N-domain were found to be critical for activin binding (Sidis et al. 2001, 2005, Keutmann et al. 2004). Moreover, the N-domain, followed by at least two other FSDs were required for activin binding and neutralization (Keutmann et al. 2004). Rearrangement or substitution among the FSDs impaired or abolished activin binding, indicating that the number and order of FSDs were important for full activity (Keutmann et al. 2004). While these observations were instructive and consistent with binding to proposed receptor binding surfaces on activin, it was not until FST was crystallized with activin A that the significance of these observations were fully appreciated (Thompson et al. 2005). Two FST molecules envelop the activin dimer, covering a large percentage of the surface and completely blocking both type I and type II receptor binding sites (Thompson et al. 2005). Moreover, it appeared that the C-terminus of one FST molecule contacted the N-terminus of the other FST molecule as if to lock them in place (Thompson et al. 2005), suggesting a possible mechanism to account for the nearly irreversible kinetics identified much earlier (Schneyer et al. 1994, 1996). The FST-activin structure also confirmed the importance of FSD2 in contacting a large portion of the activin surface, thereby contributing substantially to the overall affinity of the interaction (Keutmann et al. 2004, Thompson et al. 2005).
More recently, FSTL3, which has one fewer FSD than FST, has been crystallized in complex with activin A revealing a similar overall structure with two FSTL3 molecules surrounding the activin dimer (Stamler et al. 2008). However, in this case, the N-domain of FSTL3 contacts activin more extensively than in FST, enhancing its importance to activin binding and perhaps even compensating for the absence of a third FSD in FSTL3 (Stamler et al. 2008). The N-domain contacts may also influence ligand binding specificity (Stamler et al. 2008), thereby accounting for the previously determined difference in BMP binding between FST and FSTL3 (Sidis et al. 2006).
Myostatin is also a member of the TGF-β superfamily closely related to activin, and FST can bind myostatin with relatively high-affinity, although not as high as activin itself (Sidis et al. 2006). Through site directed mutagenesis, it was found that mutations in FSD2, or replacement of FSD2 with FSD1 reduced activin binding but maintained myostatin binding activity, whereas mutations in FSD1 or replacement of FSD1 with another copy of FSD2 diminished myostatin binding but not activin binding (Schneyer et al. 2008). Thus, it was possible to create a myostatin-selective analog of FST that may be useful for designing new treatments for muscle-wasting disorders (Schneyer et al. 2008).
Regulation of activin activity by antagonists
Although it has been known for years that FST is produced as three isoforms resulting from alternative splicing and post-translational proteolytic processing (Sugino et al. 1997), the relative activities of these different isoforms remained to be fully elucidated. As reviewed previously (Welt et al. 2002), the shorter FST288 isoform has a higher affinity for heparin-sulfated proteoglycans located on the cell surface allowing FST288 to be concentrated at the outer surface of the plasma membrane and potentially prevent action of autocrine acting activin in addition to activin arriving from other cells or tissues (i.e. paracrine or endocrine acting activin). On the other hand, the longest isoform, FST315, was localized primarily in the circulation (Schneyer et al. 2004), consistent with its reduced affinity for heparin in the unliganded state. The intermediate isoform, FST303, was found primarily in gonadal extracts and fluids and thought to be derived by proteolytic processing of the C-terminal acidic tail (Sugino et al. 1997). FSTL3, which shares a number of structural features with FST, does not contain a heparin binding sequence and thus, does not bind cell-surface proteoglycans under normal conditions (Sidis et al. 2002). These biochemical distinctions suggested that the different FST isoforms and FSTL3 had distinct biological actions in vivo (Schneyer et al. 2001, Tortoriello et al. 2001, Sidis et al. 2002).
This hypothesis was tested by producing recombinant FST isoforms and assessing their binding kinetics for different TGF-β ligands as well as their ability to inhibit biological activity when added as proteins (simulating paracrine/endocrine actions) or when transfected into cells (simulating autocrine actions). Activin binding kinetics were comparable between the FST isoforms and FSTL3 but cell-surface binding differed markedly (Sidis et al. 2006) as expected. Interestingly, inhibition of endogenous (e.g. autocrine) activin activity correlated closely with surface binding activity, whereas inhibition of exogenous (paracrine or endocrine) activin was not significantly different among the FST isoforms, consistent with their similar binding affinities (Sidis et al. 2006). Expression of FST315 as a membrane-anchored protein transformed it from a soluble protein with weak antagonism to autocrine activin into an isoform that inhibited autocrine activin activity equal to that of the FST288 isoform (Sidis et al. 2006), supporting the concept that membrane localization via the heparin-binding sequence significantly increases inhibition of autocrine-acting activin. In addition, myostatin and BMP-6 and -7 were inhibited by the FST isoforms, albeit to a lesser degree than that of activin, and this inhibition also correlated with their cell-surface binding (Sidis et al. 2006). These results support the hypothesis that differential cell-surface binding is the primary determinant of FST isoform biological activity and that these isoforms may have distinct bioactivities in vivo.
In support of the hypothesis, that FST isoforms have different in vivo actions, FST315 or FST288 cDNAs were expressed transgenically under control of the human promoter and crossed into the FST null mouse line (Lin et al. 2008), which had been previously demonstrated to be neonatally lethal (Matzuk et al. 1995c). While the FST288 expressing TG mice did not rescue this lethality, the FST315 expressing construct allowed mice to survive to adulthood with partial reversal of the skeletal abnormalities seen in FST null mice (Lin et al. 2008). However, these mice exhibited neonatal growth retardation, impaired tail growth and female infertility that may have resulted from failure of corpus luteum (CL) formation and a decline in ovarian follicle population (Lin et al. 2008). These results suggest that the FST288 isoform may be required for normal vascularization and ovarian function, whereas the FST315 isoform is sufficient for other regulatory actions that when absent, result in neonatal death (Lin et al. 2008), consistent with isoform-specific actions of FST in vivo. More detailed analysis awaits production of mice capable of producing individual FST isoforms from the endogenous FST locus.
Biochemical differences among the FST isoforms and FSTL3 also govern their biosynthesis and secretion. Using 35S pulse-chase labeling, FST315 was found to be synthesized and secreted the fastest while FST288 was secreted more slowly, with some FST288 remaining within the cell after extended incubation (Saito et al. 2005). FSTL3 was the slowest to be secreted with newly synthesized proteins being both secreted and transported to the nucleus (Saito et al. 2005). Interestingly, this nuclear FSTL3 was glycosylated but not to the same degree as secreted FSTL3, suggesting it may be derived from an endoplasmic reticulum (ER)-resident pool that is transported directly to the nucleus without first being secreted (Saito et al. 2005). These results indicate that both FST and FSTL3 may have both intracellular and extracellular roles that are governed to some degree by heparin-binding affinity (Saito et al. 2005).
Regulation of activin synthesis
Inhibin and activin share a common β-subunit with activin constituting a dimer of two β-subunits, whereas inhibin contains a related α-subunit linked to the β-subunit (Vale et al. 1988). Thus, regulation of α-subunit biosynthesis and dimerization with the β-subunit can alter activin biosynthesis in cells that synthesize both subunits such as ovarian granulosa cells. The mature portion of the α-subunit is glycosylated and elimination of the carbohydrate chain at asp 268 inhibited inhibin secretion without altering activin secretion, while mutations of the β-subunit that induced its glycosylation also favored inhibin production (Antenos et al. 2007). These observations indicate that the degree of glycosylation determines the relative proportion of inhibin and activin produced in such cells (Antenos et al. 2007). Activin also stimulated production of both α- and β-subunit mRNA and protein, as well as the required proteolytic processing that removed mature hormone from the precursor portions of the subunit (Antenos et al. 2008). Interestingly, activin also stimulates furin biosynthesis through a smad2/3-dependent process, suggesting the presence of a short feedback loop to enhance secretion of mature activin and inhibin (Antenos et al. 2008).
Advances in biological actions of activin in adults and embryonic stem cells
Activin and reproduction
Biological roles for activin have been proposed in a number of reproductive organs including the gonads, uterus, and pituitary, where it regulates processes such as folliculogenesis, spermatogenesis, and pregnancy (reviewed in Welt et al. (2002)). Over the past several years, significant progress has been made in delineating the roles and mechanisms of activins in embryonic ovary development, follicle development, luteolysis, and testis development, as well as in regulation of FSH-β gene transcription.
Activin and gonadal sex determination
Emerging evidence suggests that FST and activins influence differentiation of the indifferent gonad as well as the early stages of folliculogenesis. Embryonic XX, but not XY gonads express FST, with mRNA becoming detectable in the E11.5 XX mouse gonad (Menke & Page 2002, Yao et al. 2004). In support of a role for FST in regulating development of female gonads, FST null XX gonads develop a coelomic vessel, the rudiment of the testis-specific vasculature structure that organizes and patterns the developing testis, suggesting that FST possesses an anti-testis function (Yao et al. 2004). In the mouse XX gonad on E12.5, β-B but not β-A subunit mRNA is expressed, which is in contrast to the expression of both β-B and β-A subunit mRNA in XY gonads. Interestingly, in FST and β-B subunit double knockout embryos, vasculature development in XX gonads was reversed from the FST null phenotype to wild type (i.e. female) while local application of activin B induced ectopic formation of coelomic vessel in cultured XX gonads, and β-B knockout XY gonads exhibited defects in coelomic vessel formation (Yao et al. 2004). These results suggest that activin B induces the formation of the coelomic vessel in both XY and XX gonads, but this action is normally inhibited by FST in the XX gonad (Yao et al. 2004, 2006a). Although the role of activin and FST in gonadal differentiation has not been established in other species, Xia et al. (2008b), found that FST concentrations were higher in fetal blood and allantoic fluids from female fetuses compared with male fetuses during early pregnancy in sheep, suggesting that FST may also have important roles in sexual development in non-rodent species.
Activin and follicle development
Initial follicle pool formation is critical for normal fertility in adult-female mammals. Studies on developing mouse and human ovaries suggest that activins are involved in regulation of germ cell proliferation during the developmental period leading up to primordial follicle formation, which occurs at 24 h after birth in the mouse and 21 weeks of gestation in the human (Martins da Silva et al. 2004, Bristol-Gould et al. 2006). In mouse ovaries from postnatal days 0 and 4, activin β-A and β-B subunits, and activin receptors ActRIIA, ActRIIB, and ALK4 are expressed in both somatic cells and oocytes, suggesting that neonatal mouse ovaries produce activin and are activin-responsive (Bristol-Gould et al. 2006). Administration of activin A to female pups from birth to day 4 increased germ cell and granulosa-cell proliferation, as well as the number of primordial follicles formed from the germ cell nests, without any effect on apoptosis in these cells (Bristol-Gould et al. 2006). Activin subunits, β-A and β-B, the receptors ActRIIA and ActRIIB, and downstream effectors Smad2/3 are also expressed in human fetal ovaries at 14–19 weeks of gestation (Martins da Silva et al. 2004, Coutts et al. 2008). Interestingly, β-A subunit mRNA expression increased twofold from 14 to 19 weeks of gestation, and treatment of human ovaries between 14 and 17 weeks of gestation with activin A dramatically increased oogonial proliferation (Martins da Silva et al. 2004). These results suggest that activin signaling may play an important role in determining the size of the primordial follicle pool in humans as well as mice. Since this is the total endowment of follicles for the female's lifetime, these results also suggest that activin signaling may be important in premature ovarian failure where functional follicles are prematurely exhausted (Anasti 1998).
Activins and FSTs were originally isolated from follicular fluids and these proteins are predominantly expressed in granulosa cells. Multiple roles have been proposed for granulosa cell activin including regulation of proliferation, differentiation, and steroidogenesis (reviewed in Welt et al. (2002)). More recently discovered actions for activin include synergistic actions of activin with FSH to promote G1/S transition and cell proliferation in rat primary granulosa cells, a process accompanied by induction of CycD2 expression and phosphorylation of retinoblastoma protein (Ogawa et al. 2003, Park et al. 2005). In addition, induction of CycD2 by activin/FSH involved activation of phosphatidylinositol 3-kinase and Akt pathways (Park et al. 2005). Activin was also shown to increase the ERα and β in mouse granulosa cells both in vitro and in vivo suggesting that the activin and estrogen signaling pathways intersect to regulate granulosa cell activity and follicle development (Kipp et al. 2007).
Genetic elimination of activin β-A resulted in mouse pups that died within 24 h of birth with craniofacial and other defects (Matzuk et al. 1995b) while activin β-B deficient mice survived to become fertile adults but offspring of βB-deficient females died perinatally, perhaps from insufficient lactation (Vassalli et al. 1994). To explore the role of activin A in ovarian follicle development, granulosa cell-specific conditional β-A subunit knockout mice were created, resulting in subfertility with enhanced CL survival and accumulation within the ovary (Pangas et al. 2007). These results, in combination with the observation that replacement of mature β-A with mature β-B results in subfertility (Brown et al. 2000), suggest that both activin A and B have important roles within the ovary with activin A being functionally dominant over activin B (Pangas et al. 2007).
Transgenic overexpression of FST in ovaries led to a blockade in follicular development before antral follicle formation suggesting FST had important roles in ovarian physiology (Guo et al. 1998), perhaps through regulation of activin action. Consistent with this observation, granulosa cell-specific overexpression of FSTL3, the related activin antagonist, resulted in fewer antral follicles in adult females (Xia et al. 2004), while deletion of FSTL3 resulted in mild subfertility in females and increased testis size in males (Mukherjee et al. 2007). Since global FST null mice died just after birth and are therefore not useful for analyzing adult FST roles, Jorgez et al. (2003) created a conditional fst allele and used these mice to delete FST production from ovarian follicles. This led to decreased antral follicle number in 3-month old mice, as well as decreased litter number and size (Jorgez et al. 2003), suggesting that optimized signaling of activins, and/or other TGF-β family members bound by FST and FSTL3 (e.g. BMP-6 and -7, see (Sidis et al. 2006, Schneyer et al. 2008)) are required to maintain normal development from secondary to antral follicles (Jorgez et al. 2003). An additional phenotype developed as the females matured with accelerated loss of follicles, early ovarian senescence and elevated gonadotropins, (Jorgez et al. 2003) similar to the human infertility condition known as premature ovarian failure (Anasti 1998).
Several FST isoforms were recently identified in ovine (Xia et al. 2008a) and bovine gonads (Glister et al. 2006) by western blotting. The relative amounts of these FST isoforms varied during follicle maturation although the total FST concentration in both ovine and bovine follicular fluids declined with follicle development, suggesting that regulation of activin action by intrafollicular FST changes as follicles mature (Glister et al. 2006, Xia et al. 2008a). In fact, FST isoform distribution and biological role appears to depend on biochemical features distinct within the FST isoforms themselves such as the ability to bind to heparin sulfated proteoglycans on the surface of cells (Sidis et al. 2006). Thus, the longest isoform that has little cell-surface binding ability is largely found in serum (Schneyer et al. 2004) while follicular fluid appears to contain predominantly the intermediate sized isoform, FST303 (Sugino et al. 1993). This model is further supported by a mouse model where a human FST315 minigene was expressed transgenically under the direction of the human FST promoter and then crossed with the FST null mouse model described above (Lin et al. 2008). These FST315/303-expressing mice survived to adulthood but females were subfertile with reduced numbers of primordial, primary, and secondary follicles, and no antral follicles and CL (Lin et al. 2008).
Activin and luteolysis
Using cocultures of cells isolated from human CL, Myers et al. (2007, 2008) demonstrated that activin derived from luteinized granulosa cells increased expression of matrix metalloproteinase-2 (MMP-2) by fibroblast-like cells, and that this activity was inhibited by FST. MMPs are key proteolytic enzymes involved in the degradation of the extracellular matrix, and MMP-2 is an important luteolytic agent during the demise of CL (Myers et al. 2007), thereby suggesting that activin may play an active role in regression of CL survival. This notion is supported by analyses of ovary-specific deletion of activins in mice in which the ovaries displayed progressive accumulation of CLs and the absence of CL regression (Pangas et al. 2007). Additional support for this concept comes from the observation that FST mRNA expression is dramatically reduced in regressing CLs compared with newly formed CLs in rats which created a low FST/high-activin environment (Erickson & Shimasaki 2003). On the other hand, FST release from the ovary increased at luteolysis in sheep, which may be another mechanism to create a low FST environment through FST secretion rather than keeping the FST in follicular fluid (Xia et al. 2008b).
Activin and testicular function
Smaller testes and reduced fertility were observed in mice when activin signaling was down regulated by inactivation of ActRIIA (Matzuk et al. 1995a), or when FST (Guo et al. 1998) or FSTL3 (Xia et al. 2004) were transgenically overexpressed. Moreover, in the FSTL3 transgenic mice, reduced testis weight was observed as early as postnatal day 7 (Xia et al. 2004). The mechanism whereby activin might control testis size was suggested in early studies showing that activin can directly promote Sertoli cell proliferation in culture (Mather et al. 1990), and supported by more recent observations that the primary source for activin-A in the testis was peritubular myoid cells which acts in a paracrine fashion to promote Sertoli cell proliferation only when the Sertoli cells were obtained between neonatal days 3 and 9 (Buzzard et al. 2003). Taken together, these results support the concept that disrupted activin signaling may limit Sertoli cell proliferation in a critical developmental window which persists as reduced testis weight in adults because the number of Sertoli cells established during this proliferative phase in neonates constitutes an upper limit to spermatogenic output in adults (Sharpe et al. 2003). Therefore, activin signaling during early-testicular development appears to determine ultimate testis size and function in adults, and may be regulated by activin antagonists including FSTL3 (Xia et al. 2004) and FST (Guo et al. 1998).
Activin actions in the pituitary
Activin was originally purified on the basis of its ability to stimulate FSH release from pituitary gonadotropes (Vale et al. 1988) and it was subsequently shown that activin is a critical regulator of FSH-β biosynthesis (Weiss et al. 1992). More recent studies have shown that Smad3 is the principal regulator of activin-induced FSH-β subunit transcription, while Smad2 may be more involved in regulation of basal FSH-β transcription in murine immortalized LβT2 gonadotrope cells (Suszko et al. 2003, 2005, 2008, Bernard 2004, Gregory & Kaiser 2004, Lamba et al. 2006). This functional difference between Smad2 and Smad3 appears to be attributed to the second of the two loops in the Smad2 MH1 domain, which is thought to bury a β-hairpin loop structure thereby preventing Smad2 from directly binding its cognate DNA (Suszko et al. 2005). In addition, activin stimulation of FSH-β gene transcription involves Smad3 partners, including Pitx2c (a pituitary-specific transcription factor; Suszko et al. 2008) and TALE homeodomain proteins Pbx1 and Prep1 (Bailey et al. 2004).
FSH biosynthesis is also regulated at the transcriptional level by gonadal steroids, glucocorticoids and GnRH, and these actions are modulated by activin signaling. For example, androgen and activin synergistically stimulate FSH-β transcription, and the stimulation by androgen or activin alone is dependent on the presence of an intact Smad binding element or androgen response element (respectively) within the FSH-β promoter (Spady et al. 2004, Thackray & Mellon 2008). Glucocorticoids and activin also synergistically activate FSH-β transcription and this synergy occurs at the level of the FSH-β promoter (McGillivray et al. 2007). The role of GnRH in regulation of activin induced FSH-β transcription is more complex, because chronic administration (20–24 h) inhibits activin action (Shafiee-Kermani et al. 2007), whereas acute (4–6 h) GnRH synergistically enhances it (Gregory et al. 2004) and the interaction can also be species specific, making interpretation of heterologous reporter experiments challenging (Lamba et al. 2008, Wang et al. 2008b).
Activin also modulates transcription of LH-β subunit and GnRH receptor mRNA (Cherrington et al. 2005, Coss et al. 2005) as well as a number of other genes in LβT2 cells including inhbb, inha, gdf9, and the 17β-HSD gene (Zhang et al. 2006). Activin also regulates the number of GnRH receptors and hence, response to GnRH (Gregory & Kaiser 2004). Thus, in addition to direct regulation of FSH-β biosynthesis, activin has important actions in modulating LH-β biosynthesis and release through regulation of GnRH action, together supporting a major role for activin in modulating neuroendocrine reproductive control.
Activin and glucose metabolism
Roles for activins and related members of the TGF-β family in regulating pancreatic β-cell number and/or function have been proposed for some time, and more recent results have refined those concepts, although the precise ligands and their actions in the pancreas remain to be definitively demonstrated. For example, activin can enhance glucose stimulated insulin secretion in β-cells from rats and humans in culture (Totsuka et al. 1988, Verspohl et al. 1993, Furukawa et al. 1995, Florio et al. 2000), and β-cell proliferation is enhanced with activin treatment (Brun et al. 2004, Li et al. 2004). In addition, all components of the activin/TGF-β signaling pathway have been identified in β-cells, including the ligands themselves, receptors, second messengers, and antagonists (Ogawa et al. 1993, Wada et al. 1996, Yamaoka et al. 1998). However, from these studies alone it was not possible to determine if activin or related ligands had physiological roles regulating β-cell proliferation or function in vivo.
Yamaoka et al. (1998) demonstrated that transgenic overexpression of a dominant negative type II activin receptor in islets resulted in islet hypoplasia, reduced insulin content and impaired glucose tolerance, although the ratios of α- and β-cells was not altered. Although this study implicated activin in regulating β-cell proliferation, maturation, and possibly functions it was not possible in this model to differentiate between developmental effects and regulation of β-cells in adults. To circumvent this limitation, Smart et al. (2006) used a conditional transgenic strategy to overexpress Smad7 in β-cells. Smad7 is known as an inhibitory Smad which, unlike the signaling Smads, inhibits signaling in the activin/TGF-β pathway (Bhushan et al. 1998). Expression of Smad7 was inhibited in embryonic and neonatal mice by treating with doxycycline (Dox) and then induced in adult mice by removal of Dox, resulting in inhibition of the Smad2/Smad3 signaling pathway of activin and TGF-β (Smart et al. 2006). In these mice, β-cell gene expression, insulin content, serum insulin, and glucose tolerance were all reduced compared with controls, and these effects were reversible with Dox administration (Smart et al. 2006). Thus, activin and/or related ligands have important roles in regulating β-cell number and function in rodents.
Further support for a role for activin in regulating β-cell proliferation and function was reported by Mukherjee et al. (2007) when they created a global FSTL3 null mouse. Among other metabolic alterations, islets were nearly twice as large and contained mostly β-cells, insulin levels were slightly increased, and both glucose tolerance and insulin sensitivity, as assessed by glucose and insulin tolerance tests respectively, were enhanced. These observations suggest that removal of a natural activin antagonist resulted in increased activin bioactivity and increased β-cell mass and function (Mukherjee et al. 2007).
Although in vitro biological studies usually find nearly identical actions of activin A and B (Welt et al. 2002), activin B was shown to also use Alk7 as an alternative type I receptor (Tsuchida et al. 2004), suggesting activins A and B might have distinct actions in vivo. This concept is supported by the different phenotypes observed with activin A and B knockout mice (Matzuk et al. 1995b). In addition, Bertolino et al. (2008) found that Alk7 (and presumably, its ligand activin B) is a negative regulator of β-cell function, a finding in opposition to the reported stimulatory actions of activin A described above. Alk7 KO mice developed progressive hyperinsulinemia, reduced insulin sensitivity and impaired glucose tolerance. It appears that activin B acts by reducing glucose-stimulated calcium influx (Bertolino et al. 2008). Moreover, activin B KO mice have hyperinsulinemia consistent with activin B negatively regulating insulin production and secretion in β-cells (Bertolino et al. 2008).
Experimental reduction of β-cell mass has been shown to induce proliferation of ductal tissues within remaining pancreas and differentiation of insulin-producing cells to compensate for lost islets in a process known as neogenesis (Bonner-Weir 1994). More recently, it was found that activin expression is increased in ductal tissues and co-localized with the cytokeratin maker for ductal cells after reduction of β-cell mass with streptozotocin or partial pancreatectomy (Zhang et al. 2002). Activin was also found to increase proliferation of rat ductal cells in vitro, as well as to induce their differentiation into insulin producing cells expressing PDX-1, Ngn-3, and other genetic markers of β-cells (Park et al. 2007). Glucose-stimulated insulin secretion was also enhanced in these cells, and when transplanted, were able to reverse streptozotocin-induced diabetes in rats (Park et al. 2007). These observations suggest that activin may also influence replication and differentiation of islet-cell progenitors located in ductal epithelium.
While activin's role in regulating β-cell proliferation and function appears to be important for maintaining normal glucose metabolism in adults, its role with other islet cell types remains relatively unexplored. However, activin was shown to decrease glucagon gene expression in α-cell lines and in human islets in culture and inhibit proliferation of α-cell lines in culture by inhibiting synthesis of Arx (Mamin & Philippe 2006). These results suggest that activin A may have opposite roles in regulating α- and β-cells in adults in ways that enhance glucose uptake and decrease its synthesis, actions favorable to relieving symptoms of diabetes.
Activin and stem cell differentiation
Embryonic stem cells, both mouse and human, were first cultured on mouse feeder cell layers to maintain pluripotency during expansion. The need for completely defined culture conditions for use in human embryonic stem cell (hESC) therapies drove efforts to identify culture medium components that facilitated hESC proliferation while maintaining pluripotency. Activin and other TGF-β family members signaling through Smads 2 and 3 were found to maintain hESC pluripotency and self-renewal (Beattie et al. 2005, James et al. 2005, Vallier et al. 2005) and that blocking activin signaling resulted in hESC differentiation (Vallier et al. 2005, Wu et al. 2008). Moreover, activin often interacts with other growth factors like fibroblast growth factor (FGF) to maintain pluripotency (Vallier et al. 2005, Xiao et al. 2006). The mechanism for these activities appears to involve inhibition of BMP signaling by the interaction of activin and FGF pathways (Willems & Leyns 2008), thereby sustaining expression of such stem cell genes as Nanog, Oct4 and Sox2 through Smad binding to the Nanog promoter (Xu et al. 2008). It was additionally reported that this activin and nodal mechanism seems to be universal among mammals since it maintains pluripotency of epiblast stem cells from mice and rats, demonstrating that previously observed differences between mouse and hESCs could be overcome and that both could be maintained in the pluripotent state with activin or nodal treatment (Brons et al. 2007, Ogawa et al. 2007).
Upon removal or inhibition of activin signaling, hESC differentiation ensues, either in embryoid bodies or as monolayers. Another goal of ESC research is to direct hESC differentiation into distinct pathways. Recapitulating normal embryonic development, where activin is important in endoderm/mesoderm formation and patterning (Tremblay et al. 2000, Kumar et al. 2003), the reapplication of activin induces hESCs to form endoderm (Yao et al. 2006b, Funa et al. 2008, Sumi et al. 2008) and even differentiate into distinct cell types such as insulin producing β-cells (Shi et al. 2005, Frandsen et al. 2007, Jiang et al. 2007, Phillips et al. 2007, Jafary et al. 2008), hepatocytes (Hay et al. 2008), parathyroid-like tissue (Bingham et al. 2008), neuroectoderm (Smith et al. 2008), blood cells (Nostro et al. 2008, Pearson et al. 2008), and cardiomyocytes (Kitamura et al. 2007), perhaps through modulation of other important developmental signaling systems such as sonic hedgehog (Mfopou et al. 2007, Mfopou & Bouwens 2008). Thus, activin has critical activities in regulating hESC pluripotency, and then differentiation into endoderm and its tissue derivatives that appear to be useful for directly ESC differentiation in vitro.
Activin and immune response
The importance of TGF-β family members in immune system function was first demonstrated by the targeted disruption of TGF-β1 in mice, which resulted in a severe multifocal inflammation (Shull et al. 1992). Subsequent studies revealed that activins (reviewed in Jones et al. (2004b)) and BMPs (Gould et al. 2002, Lee et al. 2003, Hagen et al. 2007) are also important regulators of inflammatory responses in various cell types and organs.
Activin expression has been detected in many immune cells including monocytes (Eramaa et al. 1992, Abe et al. 2002), macrophages (Ebert et al. 2007), dendritic cells (Robson et al. 2008), T and B lymphocytes (Ogawa et al. 2006, 2008), and mast cells (Cho et al. 2003, Funaba et al. 2003), and its expression increases when these cells are activated by immune stimuli and toll-like receptor agonists. For example, microglial cells (quiescent central nervous system resident macrophage-like cells) and peritoneal macrophages released activin A and its binding protein FST after treatment with TLR2 agonist Pam3Cys, TLR4 agonist LPS, or TLR9 ligand CpG (cytosin–guanosin oligodesoxynucleotide 1668; Michel et al. 2003b, Ebert et al. 2007).
Blood dendritic cells and monocyte-derived dendritic cells express activin mRNA, and secretion of activin A protein was dramatically increased in the presence of Escherichia coli, or CD40 ligand (CD40L), or various TLR ligands (Robson et al. 2008). Secreted activin A then negatively regulated production of cytokines (IL-6, IL-12p70, TNF-α, and IL-10) and chemokines (IL-8, IP-10 RANTES, and MCP-1) induced by the CD40L/CD40 pathway in an autocrine/paracrine manner (Robson et al. 2008). These inhibitory functions of activin released by dendritic cells may play an important role in preventing uncontrolled release of cytokine/chemokine and recruitment of immune effectors within the local microenvironment (Robson et al. 2008). Activin was also expressed in activated CD4+CD25− T cells in the spleen and promoted differentiation of macrophages into alternatively (M-2 phenotype) activated macrophages (Ogawa et al. 2006). In addition, activin treatment inhibited IFN-γ- or LPS-induced NO2 production, and suppressed LPS-induced pinocytosis and expression of TLR4 in mouse macrophages (Zhang et al. 2005, Ogawa et al. 2006, Wang et al. 2008a). Through direct action on resting B cells, activin A enhanced IgG and IgE production (Ogawa et al. 2008). These results indicate that activin is produced by a variety of cell types in the immune system and has regulatory actions on many of them, suggesting an intricate autocrine/paracrine system whereby activin modulates certain immune responses.
In addition to upregulation of activin expression in activated immune cells, activin expression is also stimulated by inflammatory mediators including LPS in endothelial cells (Wilson et al. 2006) and Sertoli cells (Okuma et al. 2005, Riccioli et al. 2006, Jones et al. 2007). Furthermore, increased activin A expression was also found in many other cell types in inflamed or wounded tissues (Hubner et al. 1996, Ota et al. 2003, Fumagalli et al. 2007, Kaitu'u-Lino et al. 2008). Thus, activin appears to play an important role both in immune response and immune reactions during wound repair, as well as perhaps influencing wound repair directly (Hubner et al. 1999).
As a result of these immune system actions, elevated systemic activin A concentrations were found in the serum of patients with septicemia or angina (Michel et al. 2003a, Smith et al. 2004). In fact, a rapid increase in circulating activin A, peaking within 1 h, was observed after LPS administration to sheep or mice, a response which preceded the systemic release of TNF-α, IL-6, IL-1β, and FST (Jones et al. 2004a, 2007). Blockade of activin action with FST treatment altered the profiles of TNF-α, IL-6, and IL-1β after LPS stimulation and enhanced survival from a lethal LPS dose, suggesting that activin modulates the release of other key proinflammatory cytokines (Jones et al. 2007). These results demonstrate that activin plays a crucial role in the inflammatory response and that blocking its actions may have therapeutic potential in treatment modalities for acute inflammatory disorders (Jones et al. 2007).
Over the past 5 years, substantial progress was made in elucidating the biological actions of activin in adult animals, taking advantage of novel techniques that allow regulation of the timing of genetic alterations in mice as models for human disease. The important actions so far enumerated indicate that activin has many important homeostatic functions in adult humans and is thus likely also to contribute to disease in those individuals where activin expression or bioactivity is altered. While no associations with human disease have yet been identified for activins A or B in genome wide association studies, FST and/or FSTL3 levels are altered, for example, in heart disease (Lara-Pezzi et al. 2008), suggesting that such associations might be uncovered in the near future as more individuals are genotyped. It is therefore likely that new actions of activin, as well as its involvement in disease, will be identified with additional research and that pharmaceuticals based on modulation of activin signaling having therapeutic value may be discovered in the next few years.
Declaration of interest
Authors have no conflicts to declare.
This work was supported in part by the NIH (R01DK075058) to A L S.
JafaryHLarijaniBFarrokhiAPirouzMMollamohammadiSBaharvandH2008Differential effect of activin on mouse embryonic stem cell differentiation in insulin-secreting cells under nestin-positive selection and spontaneous differentiation protocols. Cell Biology International32278–286.
Martins da SilvaSJBayneRACambrayNHartleyPSMcNeillyASAndersonRA2004Expression of activin subunits and receptors in the developing human ovary: activin A promotes germ cell survival and proliferation before primordial follicle formation. Developmental Biology266334–345.
ParkYMaizelsETFeigerZJAlamHPetersCAWoodruffTKUntermanTGLeeEJJamesonJLHunzicker-DunnM2005Induction of cyclin D2 in rat granulosa cells requires FSH-dependent relief from FOXO1 repression coupled with positive signals from Smad. Journal of Biological Chemistry2809135–9148.
Shafiee-KermaniFHanSOMillerWL2007Chronic gonadotropin-releasing hormone inhibits activin induction of the ovine follicle-stimulating hormone beta-subunit: involvement of 3′,5′-cyclic adenosine monophosphate response element binding protein and nitric oxide synthase type I. Endocrinology1483346–3355.