Follistatin and its role as an activin-binding protein

Follistatin and its role as an activin-binding protein

Hiromu Sugino^a, Kishiko Sugino^a, Osamu Hashimoto^b, Hiroki Shoji^c and Takanori Nakamura^c

^aDivision of Molecular Cytology, Institute for Enzyme Research, The University of Tokushima, Tokushima, Japan;^bDivision of Biological Chemistry and Biologicals, National Institute of Health Sciences, Setagaya-ku, Tokyo, Japan;and ^cDepartment of Endocrinology, Kagawa Medical University, Miki-cho, Kagawa, Japan

Abstract: Follistatin (FS), a specific binding protein for activin, neutralizes the diverse actions of activin by forming an inactive complex with activin. FS is a monomer derived from two polypeptide core sequences of288 (FS-288) and 315 (FS-315) amino acids originated from alternatively spliced mRNA. We purified six molecular forms of FS from porcine ovaries. Their structural differences were caused by truncation of the COOH-terminal region and/or the presence of carbohydrate chains, resulting in the formation of FS-288, FS-315and FS composed of 303 amino acids (FS-303) in various forms of glycosylation on the two potential Asn-linked glycosylation sites. All six molecular species have almost the same activin binding activity (Kd=540-680pM). By contrast, the COOH-terminal truncated form, FS-288, showed much higher affinity for heparan sulfate proteoglycans of the cell surface than FS-303,whereas the intact form of FS, FS-315, had no affinity. Furthermore, FS-288more effectively blocked the suppression of follicle-stimulating hormone (FSH) secretion from rat pituitary cells by activin. This implies that activin binds to the cell surface through FS-288which adheres to the cell surface. To clarify the physiological role of cell-associated FS, we then investigated the binding of activin to cell-associated FS and the fate of cell surface-bound activin and FS using primary cultured rat pituitary and ovarian granulosa cells. When the cells were incubated with 125I-activin A in the presence of FS-288or315, the binding of activin A to the cell surface was promoted much more markedly by FS-288than by FS-315. The amounts of radioactivity recovered in trichloroacetic acid-soluble fractions (degraded activin) from the incubation medium were greatly increased by the addition of FS-288. This increase was abolished by heparan sulfate, monensin (an endocytosis inhibitor), chloroquine (a lysosome function inhibitor) and several lysosomal enzyme inhibitors. These results suggest that cell-associated FS-288accelerates the internalization of activin into the cells, leading to its degradation by lysosomal enzymes, and that cell surface-associated FS therefore plays a role in the clearance system of activin. J. Med. Invest. 44:1-14, 1997

Key Words:follistatin, activin, heparan sulfate

INTRODUCTION
The interplay of various hormones originating from the hypothalamic-pituitary-gonadal axis regulates the development and maintenance of mammalian reproduction. The central regulators in this system are the pituitary gonadotropins, i.e., leutenizing hormone (LH) and follicle-stimulating hormone (FSH). The synthesis and secretion of both of these hormones are controlled by a combination of hypothalamic gonadotropin-releasing hormone (GnRH) and the gonadal steroids. The gonadal steroids exert feedback actions on the brain to attenuate the release of GnRH and on the anterior pituitary gland to inhibit the secretion of LH and FSH. In recent years, several hypophysiotropic protein factors that can regulate the release of FSH by pituitary cells have been identified in mammalian gonads. These include activin (1, 2), which stimulates FSH release, and inhibin (3-6) and follistatin (FS) (7), which inhibit FSH release. These protein factors have added a new dimension to the complexity of this system.
Although functionally antagonistic, activin and inhibin are structurally related; inhibin is a heterodimer composed of an α-subunit linked by disulfide bonds to one of the related β-subunits, whereas activin is a dimer composed of the inhibin β-subunits (8-13). Five β-subunits have been reported:βA, βB, βC (14), βD (15) and βE (16), whereas only a single α-subunit has been identified. There is an extensive array of possible dimers;inhibin A (αbA), inhibin B (αbB) (5, 6), activin A (βAβA)(1), activin AB (βAβB) (2) and activin B (βBβB) (17) have been isolated as dimeric proteins. DNA sequence analyses showed that all of the subunits are initially synthesized as large precursor proteins with the mature subunits residing at the COOH-terminal regions of the several clusters of multiple basic residues which can serve as potential proteolytic processing sites (Fig.1). The sizes of the mature, most abundant forms of each subunit are18kDa for the α-subunit and14kDa for the βA-and βB-subunits. Thus, depending on the combination of the gene products, protein molecules having quite different biological activities could be derived from a limited number of genes.
The finding of a homodimeric form of a β-subunit is noteworthy in view of the fact that the members of the transforming growth factor-β (TGF-β) superfamily are all homodimers. Structural studies have shown that these proteins constitute the TGF-β superfamily with diverse regulatory effects on the growth and differentiation of numerous cell types. These findings prompted us to elucidate the full spectrum of actions of activin and inhibin apart from their effects on FSH secretion. There is increasing evidence from various areas of study that activin has multiple functions in a wide variety of biological systems, e.g., the regulation of the secretion of anterior pituitary hormones such as growth hormone, prolactin and corticotropin (1,18-22), the induction of erythropoiesis(23-26), the modulation of ovarian and testicular cell functions (27-31), the promotion of survival of nerve cells(32,33), the initiation of early embryonic development of Xenopus (34-40), the stimulation of insulin secretion (41),and the enhancement of bone formation (42-44). In view of the wide spectrum of bioactivities exerted by activin, it is not surprising that the mRNA encoding β-subunits is found in a multitude of diverse tissues, including the gonads, pituitary, placenta, brain, spleen, heart, bone marrow and others (45). Activin is a growth and differentiation factor that is produced by and exert effects on a broad range of cells and tissues (Fig.2).
In contrast, inhibin is a functional antagonist to activin in some systems. This antagonism was first observed with regard to FSH secretion by pituitary cells, but has also been documented for a number of other responses. There are, however, many activin responses that do not appear to be regulated by inhibin; inhibin-independent effects have been described.
Despite the various effects of members of the TGF-β superfamily on cell phenotype and physiology, until recently very little was known about the structures of their receptors or the signaling mechanisms. An activin receptor was recently cloned from the mouse pituitary tumor-derived cell line AtT20 (46).The receptor sequence predicted a transmembrane serine/threonine-specific protein kinase. This result was unexpected, because at that time all known cytokine receptor kinases were tyrosine-specific. With the use of the activin receptor cDNA as a low-stringency hybridization probe or to direct the construction of degenerate oligonucleotides for polymerase chain reaction (PCR), a number of additional receptor serine kinases have been cloned. These are divided into two classes, called type I and type II receptors, both of which contain a serine/threonine kinase domain in their intracellular domains. Activin and other members of the TGF-β superfamily interact with both types of receptors. Activin and TGF-β can bind their individual type II receptor but fail to bind the type I receptor in the absence of the type II receptor. The ligand-bound type II receptor can associate with the type I receptor. This ligand-induced heteromerization between the type I and type II receptors results in the ligand signal transduction (Fig.3). In the current model, the function of the type II receptor is confined to ligand binding, type I receptor recruitment and transphosphorylation, whereas the type I receptor is the signaling unit in the complex (reviewed in Ref.47). Furthermore, intra-cellular mediators such as the Smad's have been identified and used for the clarification of signaling mechanisms for activin and other members of the TGF-β superfamily.
Additionally, in1987, a new class of gonadal protein factor, named FSH-suppressing protein (FSP) or FS, was identified in a side fraction derived from the purification of bovine and porcine ovarian inhibins and activins (7, 48). FS was characterized initially by its ability to suppress pituitary cellular FSH secretion in vitro. The action of FS appears to be similar to that of inhibin, but it is structurally quite different. FS is a single-chain protein that occurs in31-to39-kDa molecular mass forms, all of which have similar amino acid compositions and identical amino-terminal amino acid sequences. In a recent study, we purified an activin-binding protein which was specific to and had a high affinity for activin, from rat ovary (49) and bovine pituitary (50), and we demonstrated that it was identical to FS. Interest in the biological significance of FS had diminished, because of its weak inhibitory activity compared with that of inhibin. However, our finding that FS is an activin-binding protein sheds new light on the understanding of its physiological role;FS may participate in the regulation of the multiple actions of activin. Indeed, FS mRNA transcripts have been detected in a wide variety of tissues, which may indicate that FS, like activin, has tissue-specific effects or that FS has a universal action on different cell types. The ability of FS to bind the preiotropic growth and differentiation factor activin and thereby neutralize activin action makes this glycoprotein a potentially important regulatory factor, capable of modulating autocrine and paracrine functions and the processes of differentiation and development. In this review, we describe the biochemical properties of FS and the biological implications of FS-activin interaction in a local regulatory system.

MOLECULAR HETEROGENEITY OF FOLLISTATIN
FS was originally detected as a mixture containing several bioactive polypeptides in bovine and porcine follicular fluids. Thereafter, the FS gene product processing was investigated and it was revealed that alternative splicing generated two types of FS precursor (51, 52). One is a pre-FS and the other its COOH-terminal27-amino acid truncated homologue. The corresponding mature FS isoforms comprise315 (FS-315) and288 (FS-288) amino acids, respectively.
We have established a conventional procedure for purifying FS proteins from mammalian and amphibian sources. To date, we have purified FS from rat ovary (49),porcine (17) and human (53) follicular fluids, bovine pituitary (50) and Xenopus XTC cell-culture medium (54).Every preparation demonstrated multiple bands at a32-38kDa molecular mass range when subjected to sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) under non-reducing conditions. After reduction, all these bands migrated characteristically at a higher molecular mass range (40-45kDa). In order to determine the exact chemical and biological properties of each FS isoform, we isolated each one from porcine follicular fluid and found that there are at least six (and probably more) FS molecular forms, which resulted from truncation of the COOH-terminal region and/or the presence of N-linked carbohydrate chains (Fig.4) (55). The six FS isoforms can be divided into three groups, according to the extent of truncation;i.e., full-length core proteins with 315 amino acids (FS-315), those with 303 amino acids (FS-303), and a COOH-terminal truncated protein with 288amino acids (FS-288). Characteristically, both FS-315and FS-303, but not FS-288, contain a highly acidic carboxyl-terminal extension. The FS-303molecular form is a major component of FS from mammalian sources. Interestingly, unlike mammalian FSs, the core protein of Xenopus FS may be of a single type (54), and its heterogeneity may be attributable to the number of carbohydrate chain moieties, since the N-glycanase treatment of the Xenopus preparation resulted in convergence of the multiple bands on SDS-PAGE. It has yet to be established as to which isoform group the Xenopus FS produced by XTC cells belongs. All six of the molecular species of FS we have isolated demonstrated almost identical activin-binding activity (Kd=540-680pM). The presence of various forms of FS may be physiologically significant in the regulation of the many actions of activin.

TISSUE LOCALIZATION OF FOLLISTAIN
We studied the distribution of FS in rat tissues using immunohistochemistry (56) with a polyclonal antibody directed against an FS peptide (residues123-134) as a specific immunological probe. In addition to the ovary and testis, the antiserum bound to almost all of the tissues tested, i.e., the pituitary, oviduct, uterus, kidney, adrenal gland, liver, spleen and brain cortex. In the testis, a unique immunostaining pattern was obtained with intense and specific FS immunoreactivity evident in spermatogenic cells. Interestingly, the predominant staining was in the spermatocytes and spermatid nuclei, but no immune reaction was observed in spermatogonia or spermatozoa. To our knowledge, no information has been obtained regarding how FS gets into the cell nucleus. The pituitary gland also demonstrated a characteristic staining profile. The stained spots were scattered sporadically over the anterior lobe, indicating that a specific group of anterior pituitary cells, but not all, synthesize FS. The results of our recent double immunostaining studies suggested that the cells stained with FS antibodies may be somatotrophs(57). This staining profile coincided with the distribution profile of FS mRNA determined using Northern and S1-nuclease analyses (58). Our observations clearly indicate that FS protein is ubiquitous in tissues. A similar broad tissue localization has also been demonstrated for inhibin β-subunit mRNA expression. Taken together, these findings lead us to suggest that FS, in view of its activin-binding activity, can regulate various activin actions in many tissues.

MODULATION OF ACTIVIN ACTIONS BY FOLLISTAIN
The widespread tissue distributions of activin and FS imply, but do not demonstrate conclusively, that both factors are synthesized locally and that FS regulates the physiological actions of activin in a paracrine or autocrine manner. Several lines of indirect evidence suggest that the biological activities of activin are neutralized by its binding to FS. In rat granulosa cell cultures, FS (50ng/ml) negated the stimulatory effects of activin (30ng/ml), including the expression of FSH and LH receptors and inhibin and progesterone production (Fig.5) (59). A neutralizing effect of FS on activin activity was demonstrated in the mouse Friend cell (erythroleukemia cell) bioassay. When cellular differentiation was induced by 80ng/ml activin, an inhibitory neutralizing effect of FS on erythropoiesis in vivo was observed. The continuous intraperitoneal administration of FS to normal mice resulted in a decrease of the erythroid progenitors BFU-E and CFU-E in the bone marrow and spleen (60). This supports the hypothesis that endogenous activin participates in murine erythropoiesis, a process in which FS exerts a regulatory function. In amphibian systems, activin appears to be responsible for mesodermal induction, which was also found to be inhibited by FS in a dose-dependent manner in the Xenopus animal cap assay. The coincubation of a constant amount of activin with increasing concentrations of FS prevented mesodermal tissue formation:a roughly3-fold excess of FS (on a weight basis) was needed to abolish the effect of activin (61). This also implies that there is a physiological interaction between activin and FS.

COMPLEX FORMATION BETWEEN ACTIVIN AND FOLLISTAIN
The ability of FS to bioneutralize the activities of activin in various assay systems suggests that there is a stoichiometric relationship between FS and activin. We therefore investigated whether activin binds to FS to form a biologically inactive complex in vivo (17). Porcine follicular fluid was collected on ice by aspiration from ovaries of various sizes and subjected directly to gel filtration. The immunoblotting analysis of the resulting fractions indicated that more than enough FS exists in follicular fluid to generate an activin-FS complex. The absence of free activin was confirmed by the finding that no FSH-release stimulating activity was detected in any fraction. Next, the activin-FS complex was purified by successive steps of affinity chromatography on dextran sulfate-Sepharose and fast protein liquid chromatography (FPLC) gel permeation on a Superose12HR column. This procedure exploits the high affinity of FS and its complex for sulfated polysaccharides. The complex thus obtained was subjected to reverse-phase high-performance liquid chromatography (HPLC), which caused it to dissociate and yield its components, activin and FS. Interestingly, three isoforms of activins, A, AB and B, were found to be present as complexes with FS in the follicular fluid. It is also noteworthy that the biological activity of activin B was significantly lower than those of activins A and AB in various assay systems, such as the stimulation of FSH secretion, the induction of erythrodifferentiation and the potentiation of gonadotropin receptor expression on ovarian cells, whereas activin B was a potent inducer of Xenopus mesoderm. Activin B may therefore play a specific role in early embryonic development. These results indicate that mammalian ovarian cells, probably granulosa cells, synthesize and secrete three activin isoforms and FS, which easily generate an inert complex in follicular fluid. Accordingly, to establish the precise physiological significance of the interplay between FS and activin in the ovary, it is important to identify the changes that the components of the complex undergo in relation to the development of the ovary or during the reproductive cycle.

ACTIVIN BINDING CAPACITY OF FOLLISTAIN
Although the relationship between activin and FS has not been characterized fully, there appears to be a stoichiometric interaction between them. We examined their interaction using gel permeation FPLC with two Superose 12HR columns connected in tandem. The amounts of the proteins produced were calculated from the amino acid composition analysis results. When activin and FS were premixed in a1:2molar ratio and subjected to FPLC, a single activin-FS complex peak was observed. Neither free activin nor free FS peaks were detected. Even when increasing amounts of FS were added, the complex peak remained unchanged and a free FS peak emerged (Fig.6), which indicates that one mole of activin bound to two moles of FS to form an inactive large molecular weight complex, and that activin therefore has two binding sites for FS. A1:1molar mixture of activin and FS showed significant FSH-release stimulatory activity (about 40% of the initial activin activity remained), whereas no such activity was detected in a 1:2 molar mixture (containing a2-fold molar excess of FS). This stoichiometric relationship between FS and activin closely reflects the antagonistic effect of FS on the activin actions described above. In studies of the inhibitory effect of FS on activin action, at least two moles of FS were needed to neutralize the bioactivity of one mole of activin. This raised the question of whether inhibin can also bind to FS, because the inhibin molecule contains a β-subunit derived from activin. Therefore, we used the gel permeation FPLC procedure to determine the capacity of inhibin binding to FS. When a1:1molar mixture of inhibin and FS was gel-filtered, the elution peak shifted forward and the formation of a large molecular complex was observed. A2-fold molar excess of FS to inhibin caused no change in the height of the complex peak, which indicates that inhibin bound to FS to form a1:1molar complex. These results demonstrate that activin has two binding sites (whereas inhibin has only one) for FS, and imply that FS binds to both through the common β-subunit.
This explanation of the activin and inhibin FS-binding capacities is consistent with the results obtained by Shimonaka et al. (62). However, these studies give no information about the affinity of inhibition for FS compared with that for activin. Therefore, we tested the abilities of inhibin and activin to displace 125I-activin bound to FS;activin competed for binding with an ED50 of20ng/ml, whereas inhibin was at least1,000-fold less potent in this respect (ED50 of24μg/ml). This association of inhibin with FS may not lead to the neutralization of the activity of either inhibin or FS. Further studies are needed to elucidate the physiological implications of the interaction between inhibin and FS.
Another binding protein for inhibin and activin, α2-macroglobulin (α2M), found in human serum and follicular fluids was recently identified (63, 64). To test its ability to bind activin in human serum, SDS-PAGE under non-reducing conditions of human serum was performed using 3.6% gels, and the proteins were transferred to Immunobilon membranes, which were incubated with125I-activin and subjected to autoradiography. The major α2(α2M) M band was observed at370kDa, which indicated that α2M may be the primary activin-binding protein in serum. However, the affinity of activin for α2M is much lower than that for FS, since typical stoichiometric binding was not observed in the gel filtration experiments. In addition, several other bands, including some at around500kDa and180kDa, were also detected on the blots. Although these serum proteins that are capable of binding activin and probably inhibin have not been identified, they may serve as binding proteins for activin, inhibin and possibly other factors as well as α2M to protect these molecules from proteolytic degradation and/or affect their clearance from the bloodstream.

AFFINITY OF FOLLISTAIN FOR HEPARAN SULFATE OF PROTEOGLYCANS
In the course of an FS purification study, we noticed that it possessed a unique and strong affinity for sulfated polysaccharides, such as dextran sulfate, heparin and sulfated cellulose. It was recently proposed that the binding of basic fibroblast growth factor (FGF) to its receptor requires prior binding to the heparan sulfate side chains of proteoglycans or heparin(65). We therefore attempted to ascertain the physiological significance of FS-sulfated polysaccharide interactions in regulating activin signal transduction. Rat granulosa cells cultured in serum-free medium were incubated with various concentrations of FS, and the amounts of cell-bound FS were determined by adding 125I-activin and counting the specifically bound radioactivity;this yielded a typical ligand saturation curve with an apparent Kd of5nM. Heparin and heparan sulfate, but not chondroitin, keratan or dermatan sulfate competed strongly for this binding. When granulosa cells were treated with various glycosaminoglycan-degrading enzymes before or after adding FS to the cultures, heparinase and heparitinase treatment resulted in significant suppression of the binding. These results show that FS has a high affinity for the heparan sulfate side chains of granulosa cell surface proteoglycans (66).
As discussed above, our FS preparation contains at least six isoforms, which result from truncation at the COOH-terminal region and/or the presence of carbohydrate chains. Interestingly, the affinity for the heparan sulfate chains was found to differ quite markedly depending on the sequence of core proteins(55). The COOH-terminal truncated form (FS-288) showed a very high affinity for the granulosa cells, with a Kd of2nM. In contrast, the full-length form (FS-315) showed no affinity, whereas the midsized isoforms (FS-303) showed moderate affinity (Fig.7). The carbohydrate chains did not appear to affect their affinity for the cell surface. The association of the short and midsized FS isoforms with the cell was abolished by adding heparan sulfate or heparin. High concentrations of the sulfated polysaccharides (>10μg/ml) had no effect on the interaction between activin and FS. To obtain direct evidence of FS binding to heparan sulfate, we carried out heparan sulfate-Sepharose affinity chromatography. Almost all of the short isoform (FS-288) applied bound to a heparan sulfate-Sepharose column, whereas the full-length isoform (FS-315) was recovered quantitatively in the breakthrough fraction, indicating that it had no affinity for this column. These results clearly indicate that the amino acid sequence of the COOH-terminal is important for FS binding to the heparan sulfate chains of cell surface proteoglycans.
The sequence dependence of FS isoforms on its cellular association was also confirmed by determining the transient expression of FS cDNA in COS cells (55). Cells transfected with the FS-288 cDNA secreted FS-288, which bound to the COS cells themselves, whereas cells transfected with the FS-315construct produced the FS-315protein, which did not bind to the cells. This result shows that the short form of FS (FS-288) expressed and secreted by COS cells can bind to the cell surface with a much higher affinity than that of the long form (FS-315).
The biological response first associated with FS was inhibition of activin-induced FSH secretion from cultured rat pituitary cells. To determine whether this inhibitory activity is affected differentially by the various FS isoforms, the potencies of the short, midsized and long FS forms were compared in this system in vitro. The COOH-terminal truncated FS (FS-288) was more potent than FS-303and FS-315. There was a5-fold difference between the inhibitory effects of FS-288 (ED50=2ng/ml) and FS-303 (ED50=10ng/ml), and a2-fold difference between those of FS-303 and FS-315 (ED50=20ng/ml), results which are consistent with the differences in the binding affinities of these forms to cell surfaces (55). Moreover, a recent study demonstrated that the in vivo FSH suppressant activity of recombinant FS-288 in ovariectomized rats was greater and longer-lasting than that of inhibin A, which suggests that the COOH-terminal region of FS is important for inhibiting the FSH-releasing activity of activin and probably for exerting other bioneutralizing effects on the various actions of activin (67). Proteoglycans similar to those on ovarian granulosa cells may also be present in the pituitary, and various FS isoforms found locally therein may be bound by them.

INTERFERENCE BY FOLLISTATIN WITH ACTIVIN BINDING TO ACTIVIN RECEPTORS
Activin receptor mRNAs have been reported to be distributed in the rat pituitary (68), but the receptor protein in the pituitary has yet to be identified. We attempted to analyze the pituitary activin receptor by affinity cross-linking 125I-activin A to rat pituitary cells using the bifunctional chemical cross-linker DSS (69). Very faint, but definite, cross-linked bands of 80 and 100kDa were observed (Fig.8), which corresponded well with those of types I and II activin receptors coexpressed transiently in COS-7cells (Fig.8). The binding of labeled activin to both types I and II activin receptors was abolished in the presence of excess FS-288or FS-315 (Fig.8). These findings indicate that the activin-FS complex cannot bind to activin receptors and would account for the inhibitory effect of FS on the activin-induced stimulation of FSH secretion by pituitary cells. De Winter et al. recently demonstrated that the preincubation of radioiodinated activin A with FS abolished the activin binding to type II activin receptors and consequently, its binding to type I activin receptors, and they proposed that FS can neutralize activin bioactivity by interfering with activin binding to type II activin receptors (70). Our affinity cross-linking experiments also showed inhibition by FS of activin binding to its receptors on rat pituitary cells. This result can be explained by the hypothesis that FS masks the as yet unidentified receptor binding domain of the activin molecule, resulting in the failure of activin to transduce its signal in responsive cells.

ASSOCIATION OF FOLLISTAIN WITH RAT PITUITARY CELL SURFACES
Although the formation of activin-receptor complexes was prevented by adding FS, a broad band with a molecular mass ranging from45to65kDa was visible after treatment with either FS-288or FS-315 (Fig.8). This band with a low molecular mass was not related to activin receptors, because it was also yielded by COS-7cells that were not transfected with activin receptor DNAs (Fig.8)and the labeling of this band was completely inhibited by incubation with heparan sulfate (Fig.8) and with excess unlabeled activin, indicating that labeled activin stays on the cell surfaces via FS bound to the heparan sulfate side chains of proteoglycans. It should be noted that FS-288yielded a much more intense band than that yielded by FS-315 (Fig.8), which was consistent with our above-mentioned finding that FS-288showed much higher affinity than FS-315for heparan sulfate proteoglycans on rat ovarian granulosa cells (55). To test this finding applied to pituitary cells, rat pituitary cells were incubated with various concentrations of radioiodinated FS. As expected, FS-288showed a quite high affinity for the pituitary cells, whereas the affinity of FS-315was low. The association of FS with the cells was completely suppressed by excess heparan sulfate or heparin, indicating that FS-288binds to heparan sulfate on pituitary cell surfaces.
It is well documented that the interaction between FGF and heparin-like molecules in the extracellular matrix is important for various biological functions, such as the protection of this factor against proteolytic degradation and its concentration on cell surfaces (65). The role of heparin-like molecules in signal transduction of FGF is noteworthy:the binding of basic FGF (bFGF) to its receptor requires prior binding either to the heparan sulfate side chains of cell-surface glycosaminoglycan or to free heparin to present the ligand to the receptor. De Winter et al. attempted to determine whether cell surface-bound FS-288presents activin A to the activin receptors on human erythroleukemic K562cells, and they found that FS-288and the activin A-type IIA activin receptor complex were not coprecipitated by an anti-type IIA activin receptor antibody (70), suggesting that, unlike bFGF, cell surface-associated FS cannot present ligands to signaling receptors. Judging from these results, FS appears to be nothing more than a negative regulator for activin, and its function is to form an inactive complex with activin to neutralize the activity of activin.

FATE OF ACTIVIN ASSOCIATED WITH FOLLISTATIN ON CELL SURFACES
To investigate the behavior of cell surface-associated activin via FS, we examined the binding of activin to FS associated with pituitary cell surfaces(69). In the presence of various con-centrations of FS-288 or FS-315, rat pituitary cells were incubated with increasing amounts of 125I-activin A (0-100ng/ml), and the cell-bound radioactivities (activin-FS complex) were then determined (Fig.9). The binding activity of activin A alone was difficult to detect, probably due to the very small number of activin receptors on pituitary cells. However, as expected, FS-288 markedly increased the affinity of activin A for cell surfaces, in a dose-dependent manner, whereas FS-315did not enhance the activin A binding to cell surfaces. These results suggest that activin A can adhere strongly to cells by forming a complex with FS-288on the cell surfaces. Next, we followed the behavior of the cell-associated activin-FS complex and found that it was degraded endocytotically. Rat pituitary cells were incubated with radioiodinated activin A (40ng/ml) in the presence of increasing concentrations of FS-288 or FS-315 for various incubation periods, after which the radioactivities recovered from the trichloroacetic acid (TCA)-soluble fractions (degraded activin) of the incubation media were determined using a γ-spectrometer. As shown in Fig.10, FS-288stimulated the activin A degradation significantly time- and dose-dependently to a greater extent than did FS-315. This stimulatory effect of FS-288 was abolished by adding heparin or heparan sulfate to the culture medium. This result reflects the cell-surface adhesiveness of the complex:the more strongly the activin A-FS-288complex binds to cell-surface heparan sulfate, the more easily it is degraded. Moreover, this degradation was dependent on the number of pituitary cells:increasing their number stimulated the degradation of activin A bound to the cell surfaces via FS. These degradation data were obtained by monitoring the degradation of 125I-activin. SDS-PAGE and gel filtration of the complex samples demonstrated that the FS component of the activin-FS complex was also degraded.

ENDOCYTOTIC DEGRADATION OF ACTIVIN A IN LYSOSOMES
The degradation of cell-bound activin and/or FS by pituitary cells led us to hypothesize that endocytotic internalization occurs in the cells and the resulting endocytotic vesicles ultimately fuse with lysosomes, after which most of the vesicle contents are rapidly broken down. To explore this idea, we examined the effects of various types of lysosomal enzyme inhibitor on activin A degradation in rat pituitary cells (69). The results are summarized in Table1. The lysosomal enzyme inhibitors reduced the degradation significantly, but the serine protease inhibitor aprotinin had no effect. The addition of chloroquine, which increases the pH inside lysosomes, to intact cultured cells inhibited the activin A breakdown markedly. Both heparin and heparan sulfate significantly suppressed the activin degradation, strongly suggesting that the degradation does not occur until FS binds to the pituitary cells. The almost complete inhibition of the degradation by monensin indicated that the endocytotic process is essential for the degradative process.
Autoradiographic experiments using radioiodinated activin A were also performed (Fig.11). Pituitary cells were incubated with 125I-activin A at 37Ž in the presence or absence of FS, heparan sulfate and chloroquine for12h, and the cells were washed with acid/salt buffer to strip 125I-activin A from their surfaces and then autoradiographed. FS-288accelerated the uptake of activin A by pituitary cells markedly, and to a greater extent than did FS-315. Heparan sulfate significantly suppressed the uptake, a result which agreed well with the degradation data described above. The coincubation with chloroquine inhibited the degradation of activin A taken up by the cells, probably in the lysosomes, resulting in activin A accumulation within the cells. Microscopic observations supported our hypothesis that activin A bound to pituitary cell surfaces via FS-288becomes internalized and packaged into endocytic vesicles, which fuse with lysosomes, followed by proteolytic degradation of their contents.
The number of growth factors and cytokines found to bind to heparin and heparan sulfate is steadily increasing; these include FGF, granulocyte-macrophage colony stimulating factor, interleukin-3, pleiotrophin, hepatocyte growth factor, vascular endothelial growth factor and midkine, among others. On the basis of our results described above, we speculated that, like the activin A-FS-288complex, these growth factors bind to cell-surface heparan sulfate proteoglycans, then become internalized and eventually are degraded in lysosomes. Indeed, we found that 125I-bFGF bound to rat cultured pituitary cells in a dose-dependent manner and that its intracellular degradation was time-dependent. It is therefore conceivable that such endocytotic degradation of growth factors is a common mechanism for eliminating signaling molecules from cell surfaces.

SUPPRESSION OF FSH SECRETION-INHIBITING ACTIVITY OF FOLLISTAIN BY HEPARAN SULFATE
FS was originally identified as an inhibitor of FSH secretion by cultured pituitary cells, but its potency was only10-30%of that of inhibin (7,48). The mechanism by which FS acts is unclear, but it has been suggested that it binds to endogenous activin and neutralizes activin-stimulated FSH secretion. In an attempt to elucidate how the interaction between FS and proteoglycans participates in the FSH-suppressing action of FS, we examined the effect of heparan sulfate on the inhibitory activity of FS on rat pituitary cells(69). Dose-response curves for the inhibition of basal FSH secretion into the culture medium by FS-288and FS-315in the presence and absence of heparan sulfate (10μg/ml) were plotted, and revealed that FS-288(ED50=5ng/ml) was about7times more potent than FS-315(ED50=35ng/ml), and that heparan sulfate reduced the inhibitory activity of FS-288 by about50%,whereas it had no effect on that of FS-315. These results suggest that cell-associated FS-288plays a more positive role than FS-315in controlling the action(s) of activin at cell surfaces, and show the importance of the cell-surface adhesiveness of FS to its role in controlling activin activity.

DEGRADATION OF ACTIVIN IN RAT CULTURED GRANULOSA CELLS
We reported that activin stimulated the FSH-induced differentiation of rat granulosa cells and enhanced the induction by FSH of LH receptor expression in and progesterone production by rat cultured granulosa cells(27). As previously observed with cultured pituitary cells, the presence of FS-288markedly increased the affinity of activin for heparan sulfate on granulosa cell surfaces, whereas FS-315did not enhance the activin binding to the surfaces of these cells.
We demonstrated that rat cultured granulosa cells produced FS protein and secreted it into the medium and that FSH, but not LH, stimulated these events (71). Rat granulosa cells were incubated with 125I-activin A (40ng/ml) in the absence or presence of FSH (100ng/ml) or/and heparin (10mg/ml), and FSH stimulated activin A degradation by about2-fold compared with that in the absence of FSH. This stimulatory effect was abolished by the addition of heparin. Indeed, ligand blotting of the culture medium demonstrated that both FS-315and FS-288were produced and secreted into the medium in the presence of FSH. Moreover, both FS-315and FS-288mRNA expressions were confirmed by reverse transcription (RT)-PCR. These results indicate that, in response to FSH, granulosa cells produce and secrete FS-288, which itself adheres to the cells, and that the cell-associated FS-288 can capture activin, leading to its endocytotic degradation.

ROLE OF FOLLISTATIN IN NEURAL INDUCTION IN XENOPUS EMBRYOS
Hemmati-Brivanlou et al. recently constructed a truncated type II activin receptor and demonstrated that the inhibition of the signal transduced by this receptor led to neuralization in developing Xenopus embryos (72). They proposed that FS RNA neuralized Xenopus ectodermal explants directly in the absence of detectable mesoderm(73). It has been argued that heparan sulfate proteogly-cans are involved in mesoderm development and neural induction in Xenopus embryos. Furuya et al. found that heparan sulfate proteoglycans were expressed exclusively in the animal hemisphere at the early gastrula stage of Xenopus embryos and appeared predominantly on the sheath of the neural tube, the notochord and epithelium(74). In light of these recent findings, we speculated that FS bound to heparan sulfate proteoglycans removes activin signals from cell surfaces, as observed with rat pituitary and ovarian granulosa cells, and we therefore examined the affinity of FS for Xenopus heparan sulfate as follows. Xenopus heparan sulfate was purified from Xenopus embryos at the tadpole stage, labeled with tritium acetate, and incubated with the required FS. The resulting incubation mixture was filtered through nitrocellulose membranes, which trapped heparan sulfate bound to FS. FS-288 was found to bind to Xenopus heparan sulfate in a dose-dependent manner, whereas FS-315 showed only slight affinity for it. When a PCR was carried with a pair of suitable primers to amplify the Xenopus short form of FS, the short-form FS band did not appear until the gastrula stage. Since there is no information about the entire cDNA sequence of Xenopus long-form FS, its expression could not be detected. Taken together, our preliminary results strongly suggest that the short form of Xenopus FS binds to embryonic cell surfaces, and eliminates the activin signal from the activin target cells, thereby making them ready for neural induction.
Two organizer factors, chordin and noggin, were recently found to be bone morphogenetic factor (BMP)-binding proteins (75, 76). These two factors bind to BMP with high affinity and abolish its activity by blocking binding to all cognate surface receptors. Interestingly, both chordin and noggin were observed to have affinity for heparin and heparan sulfate, like FS. Therefore, BMP may bind to its target cells via cell surface-bound chordin or noggin, followed by endocytotic degradation, whereby BMP signaling may be removed from its target cell surfaces.

CONCLUSION
FS binds stoichiometrically to activin to form an inactive complex, which results in the blockade of various activin bioactivities. However, the physiological significance of this complex formation is not fully understood. In addition, little is known about the importance of the cell-surface adhesiveness of FS regarding its role in controlling activin bioavailability.
Heparan sulfate reduced the FSH-suppressing activity of FS-288more effectively than it did that of FS-315,lending support to the hypothesis that FS isoforms have different affinities for cell surfaces as well as differing roles in the local modulation of activin function. Significant binding of radioiodinated activin A to pituitary cell surfaces was observed only in the presence of FS. As expected, FS-288promoted this binding markedly and to a greater extent than FS-315. When incubated with pituitary cells in the presence of FS-288, activin A in the medium appeared to be trapped by cell-associated FS-288. Therefore, after being captured on the cell surface, activin A may, together with FS-288and proteoglycans, be ingested by endocytotic vesicles, which ultimately fuse with primary lysosomes and are degraded. Most of the vesicle contents were found to be hydrolyzed into small breakdown products and secreted to the exterior. There is little doubt that activin A is broken down by such an endocytotic degradation process, because various types of inhibitor of each stage of this process significantly blocked the activin A degradation:the inhibitors tested included monensin, a proton ionophore and an endosome-lysosome fusion inhibitor;chloroquine, a lysosome function inhibitor, and several lysosomal protease inhibitors. As was the case with radioiodinated activin A, the proteolytic degradation of 125I-labeled FS-288in pituitary cells was observed. Taking these findings together, we hypothesized that the endocytotic degradation of growth factors via cell-surface heparan sulfate is necessary to erase the growth factor signals from the surrounding cell surfaces when they become excessive and thus useless. It has been established that the binding of a signaling ligand to its receptor stimulates a biological response and triggers a sequence of events leading to cellular desensitization to the ligand in order to regulate the responsiveness of the target cell to the ligand. We propose that, in addition to such receptor-mediated endocytosis, there must be a scavenger mechanism for clearing signaling molecules away from their target cell surfaces (Fig.12). Our results demonstrated that cell-associated FS-288 (carboxy-terminal truncated FS) accelerates the endocytotic internalization of activin into rat pituitary cells, rat granulosa cells and probably Xenopus animal hemisphere cells, leading to its degradation by lysosomal enzymes. Cell-associated FS therefore plays a role in the system responsible for clearing the activin signal from cell surfaces.

ACKNOWLEDGEMENTS
We thank Dr. Y. Eto for providing recombinant human activin A, Dr. S. Shimasaki for the gift of recombinant human follistatins, and Dr. Y. Hasegawa for donating bovine inhibin A. We are grateful to the NIDDK Pituitary Hormone Program for supplying rat FSH RIA kit. This work was supported in part by grants from the Ministry of Education, Science, and Culture of Japan, from the Naito Foundation, the Uehara Memorial Foundation and the Sankyo Foundation of Life Science.

REFERENCES
1. Vale W, Rivier J, Vaughan J, McClintock R, Corrigan A, Woo W, Karr D, Spiess J:Purification and characterization of an FSH releasing protein from porcine ovarian follicular fluid. Nature321:776-779, 1986
2. Ling N, Ying SY, Ueno N, Shimasaki S, Esch F, Hotta M, Guillemin R:Pituitary FSH is released by a heterodimer of the β-subunits from the two forms of inhibin. Nature321:779-782, 1986
3. Robertson DM, Foulds LM, Leversha L, Morgan FJ, Hearn MTW, Burger HG, Wettenhall REH, de Kretser DM:Isolation of inhibin from bovine follicular fluid. Biochem Biophys Res Commun126:220-226, 1985
4. Miyamoto K, Hasegawa Y, Fukuda M, Nomura M, Igarashi M, Kangawa K, Matuo H:Isolation of porcine follicular fluid inhibin of 32K daltons. Biochem Biophys Res Commun129:396-403,1985
5. Ling N, Ying SY, Ueno N, Esch F, Denoroy L, Guillemin R:Isolation and partial characterization of a Mr 32,000 protein with inhibin activity from porcine follicular fluid. Proc Natl Acad Sci USA82:7217-7221, 1985
6. Rivier J, Spiess J, McClintock R, Vaughan J, Vale W: Purification and partial characterization of inhibin from porcine follicular fluid. Biochem Biophys Res Commun133:120-127, 1985
7. Ueno N, Ling N, Ying SY, Esch F, Shimasaki S, Guillemin R:Isolation and partial characterization of follistatin:A signal-chain Mr 35,000 monomeric protein that inhibits the release of follicle-stimulating hormone. Proc Natl Acad Sci USA 84:8282-8286, 1987
8. Mason AJ, Hayflick JS, Ling N, Esch F, Ueno N, Ying SY, Guillemin R, Niall H, Seeburg PH:Complementary DNA sequences of ovarian follicular fluid inhibin show precursor structure and homology with transforming growth factor-β. Nature 318:659-663, 1985
9. Forage RG, Ring JM, Brown RW, McInerney BV, Cobon GS, Gregson RP, Robertson DM, Morgan FJ, Hearn MTW, Findlay JK, Wettenhall REH, Burger HG, de Kretser DM:Cloning and sequence analysis of cDNA species coding for the two subunits of inhibin from bovine follicular fluid. Proc Natl Acad Sci USA83:3091-3095, 1986
10. Mayo KE, Cerelli GM, Spiess J, Rivier J, Rosenfeld MG, Evans RM, Vale W:Inhibin A-subunit cDNAs from porcine ovary and human placenta. Proc Natl Acad Sci USA83:5849-5853, 1986
11. Mason AJ, Niall HD, Seeburg PH:Structure of two human ovarian inhibins. Biochem Biophys Res Commun135:957-964, 1986
12. Esch FS, Shimasaki S, Cooksey K, Mercado M, Mason AJ, Ying SY, Ueno N, Ling N:Complementary deoxyribonucleic acid (cDNA) cloning and DNA sequence analysis of rat ovarian inhibins. Molec Endocrinol1:388-396, 1987
13. Mason AJ, Berkemeier LM, Schmelzer CH, Schwall RH:Activin B:Precursor sequences, genomic structure and in vitro activities. Molec Endocrinol3:1352-1358, 1989
14. Hoten G, Neidhart H, Shneider C, Pohl J: Cloning of a new member of the TGF-β family:A putative new activin βC chain. Biochem Biophys Res Commun206:608-613, 1995
15. Oda S, Nishimatsu S, Murakami K, Ueno N:Molec-ular cloning and functional analysis of a new activin β subunit:A dorsal mesoderm-inducing activity in Xenopus. Biochem Biophys Res Commun210:581-588, 1995
16. Fang J, Yin W, Smiley E, Wang SQ, Bonadio J: Molecular cloning of the mouse activin βE subunit gene. Biochem Biophys Res Commun228:669-674, 1996
17. Nakamura T, Asashima M, Eto Y, Takio K, Uchiyama H, Moriya N, Ariizumi T, Yashiro T, Sugino K, Titani K, Sugino H:Isolation and characterization of native activin B. J Biol Chem267:16385-16389, 1992
18. Murata T, Ying YS:Transforming growth factor-β and activin inhibit basal secretion of prolactin in a pituitary monolayer culture system. Proc Soc Exp Biol Med198:599-605, 1991
19. Kitaoka M, Kojima I, Ogata E:Activin A:A modu-lator of multiple types of anterior pituitary cells. Biochem Biophys Res Commun157:48-54, 1988
20. Billestrup N, Gonzalez-Manchon C, Potter E, Vale W: Inhibition of somatotroph growth and GH bios-ynthesis by activin in vitro. Mol Endocrinol4:356-362, 1990
21. Bilezikjian LM, Corrigan A, Vale W:Activin A modulates growth hormone secretion from cultures of rat anterior pituitary cells. Endocrnology126:2369-2376, 1990
22. Bilezikjian LM, Blount AL, Campen CA, Gonzalez-Manchon C, Vale W:Activin A inhibits POMC mRNA accumulation and ACTH secretion of AtT20cells. Mol Endocrinol5:1389-1395, 1991
23. Eto Y, Tsuji T, Takezawa M, Takano S, Yokozawa Y, Shibai H:Purification and characterization of erythroid differentiation factor (EDF) isolatied from human leukemia cell line THP-1.Biochem Biophys Res Commun142:1095-1103, 1987
24. Yu J, Shao L, Lemas V, Yu AL, Vaughan J, Rivier J, Vale W:Importance of FSH-releasing protein and inhibin in erythrodifferentiation. Nature330:765-767, 1987
25. Broxmeyer HE, Lu L, Cooper S, Schwall RH, Mason AJ, Nikolics K:Selective and indirect modulation of human multipotential and erythroid hematopoietic progenitor cell proliferation by recombinant human activin and inhibin. Proc Natl Acad Sci USA85:9052-9056, 1988
26. Lebrun JJ, Vale WW:Activin and inhibin have antagonistic effects on ligand-dependent heteromer-ization of the type I and type II activin receptors and human erythroid differentiation. Mol Cell Biol17:1682-1691, 1997
27. Sugino H, Nakamura T, Hasegawa Y, Miyamoto K, Abe Y, Igarashi M, Eto Y, Shibai H, Titani K: Erythroid differentiation factor can modulate follicular granulosa cell functions. Biochem Biophys Res Commun153:281-288, 1988
28. Hutchinson LA, Findlay JK, de Vos FL, Robertson DM:Effects of bovine inhibin, transforming growth factor-β and bovine activin-A on granulosa cell differentiation. Biochem Biophys Res Commun146:1405-1412, 1987
29. Hasegawa Y, Miyamoto K, Abe Y, Nakamura T, Sugino H, Eto Y, Shibai H, Igarashi M:Induction of follicle stimulating hormone receptor by erythroid differentiation factor on rat granulosa cell. Biochem Biophys Res Commun156:668-674, 1988
30. Hsueh AJW, Dahl KD, Vaughan J, Tucker E, Rivier J, Bardin CW, Vale W:Heterodimers and homodimers of inhibin subunits have different paracrine action in the modulation of luteinizing hormone-stimulated androgen biosynthesis. Proc Natl Acad Sci USA84:5082-5086, 1987
31. Mather JP, Attie KM, Woodruff TK, Rice GC, Philiips DM:Activin stimulates spermatogonial proliferation in germ-Sertoli cell cocultures from immature rat testis. Endocrinology127:3206-3214, 1990
32. Schubert D, Kimura H, LaCorbiere M, Vaughan J, Karr D, Fischer WH:Activin is a nerve cell survival molecule. Nature344:868-870, 1990
33. Hashimoto M, Kondo S, Sakurai T, Etoh Y, Shibai H, Muramatsu M:Activin/EDF as an inhibitor of neural differentiation. Biochem Biophys Res Commun173:193-200, 1990
34. Asashima M, Nakano H, Shimada K, Kinoshita K, Ishii K, Shibai H, Ueno N:Mesodermal induction in early amphibian embryos by activin A (erthroid differentiation factor). Roux's Arch Dev Biol198:330-335, 1990
35. Asashima M, Nakano H, Uchiyama H, Davids M, Plessow S, Loppnow-Blinde B, Hoppe P, Dan H, Tiedemann H:The vegitalizing factor belongs to a family of mesoderm-inducing proteins related to erythroid differentiation factor. Naturwissenschaften77:389-391, 1990
36. Smith JC, Price BMJ, Van Nimmen K, Huylebroeck D:Identification of a potent Xenopus mesoderm-inducing factor as a homologue of activin A. Nature345:729-731, 1990
37. van den Eijnden-Van Raaij AJM, van Zoelent EJJ, van Nimmen K, Koster CH, Snoek GT, Durston AJ, Huylebroeck D:Activin-like factor from a Xenopus laevis cell line responsible for mesoderm induction. Nature345:732-734, 1990
38. Thomsen G, Woolf T, Whitman M, Sokol S, Vaughan J, Vale W, Melton DA:Activins are expressed early in Xenopus embryogenesis and can induce axial mesoderm and anterior structures. Cell63:485-493, 1990
39. Mitrani E, Ziv T, Thomsen G, Shimoni Y, Melton DA, Bril A:Activin can induce the formation of axial structures and is expressed in the hypoblast of the chick. Cell63:495-501, 1990
40. Asashima M, Nakano H, Uchiyama H, Sugino H, Nakamura T, Eto Y, Ejima D, Nishimatsu S, Ueno N, Kinoshita K:Presence of activin (erythroid differentiation factor) in unfertilized eggs and blastulae of Xenopus laevis. Proc Natl Acad Sci USA88:6511-6514, 1991
41. Totsuka Y, Tabuchi M, Kojima I, Shibai H, Ogata E: A novel action of activin A:Stimulation of insulin secretion in rat pancreatic islets. Biochem Biophys Res Commun156:335-339, 1988
42. Centrella M, McCarthy TL, Canalis E:Activin-A binding and biochemical effects in osteoblast-enriched cultures from fetal-rat parietal bone. Mol Cell Biol11:250-258, 1991
43. Hashimoto M, Shoda A, Inoue S, Yamada R, Kondo T, Sakurai T, Ueno N, Muramatsu M:Functional regulation of osteoblastic cells by the interaction of activin A with follistatin. J Biol Chem267:4999-5004, 1992
44. Ogawa Y, Schmidt DK, Nathan RM, Armstrong RM, Miller KL, Sawamura SJ, Ziman JM, Erickson KL, de Leon ER, Rosen DM, Seyedin SM, Glaser CB, Chang RJ, Corrigan AZ, Vale W:Bovine bone activin enhances bone morphogenetic protein-induced ectopic bone formation. J Biol Chem267:14233-14237, 1992
45. Meunier H, Rivier C, Evans RM, Vale W:Gonadal and extragonadal expression of inhibin α, βA and βB subunits in various tissues predicts diverse functions. Proc Natl Acad Sci USA85:247-251, 1988
46. Mathews LS, Vale WW:Expression cloning of an activin receptor, a predicted transmembrane serine kinase. Cell65:973-982, 1991
47. Mathews LS:Activin receptors and cellular signaling by the receptor serine kinase family. Endocr Rev15:310-325, 1994
48. Robertson DM, Klein R, de Vos FL, McLachlan RI, Wettenhall RE, Hearn MT, Burger HG, de Kretser DM:The isolation of polypeptides with FSH suppressing activity from bovine follicular fluid which are structually different to inhibin. Biochem Biophys Res Commun149:744-749, 1987
49. Nakamura T, Takio K, Eto Y, Shibai H, Titani K, Sugino H:Activin-binding protein from rat ovary is follistatin. Science247:836-838, 1990
50. Kogawa T, Nakamura T, Sugino K, Takio K, Titani K, Sugino H:Activin-binding protein is present in pituitary. Endocrinology128:1434-1440, 1991
51. Shimasaki S, Koga M, Esch F, Mercado M, Cooksey K, Koba A, Ling N:Porcine follistatin gene structure supports two forms of mature follistatin produced by alternative splicing. Biochem Biophys Rec Commun152:717-723, 1988
52. Shimasaki S, Koga M, Esch F, Cooksey K, Mercado M, Koba A, Ueno N, Ying SY, Ling N, Guilemin R: Primary structure of the human follistatin precursor and its genomic organization. Proc Natl Acad Sci USA85:4218-4222, 1988
53. Yokoyama Y, Nakamura T, Nakamura R, Ihara M, Aono T, Sugino H:Identification of activins and follistatin proteins in human follicular fluid and placenta. J Clin Endocrinol Metab80:915-921, 1995
54. Fukui A, Nakamura T, Sugino K, Takio K, Uchiyama H, Asashima M, Sugino H:Isolation and characterization of Xenopus follistatin and activins. Dev Biol159:131-139, 1993
55. Sugino K, Kurosawa N, Nakamura T, Takio K, Shimasaki S, Ling N, Titani K, Sugino H:Molecular heterogeneity of follistatin, an activin-binding protein:Higher affinity of the carboxyl-terminal truncated forms for heparan sulfate proteoglycans on the ovarian granulosa cell. J Biol Chem 268:15579-15587, 1993
56. Kogawa K, Ogawa K, Hayashi Y, Nakamura T, Titani K, Sugino H:Immunohistochemical localization of follistatin in rat tissues. Endocrinnol Jpn38:383-391, 1991
57. Kogawa K:Function and localization of follistatin, a binding protein for activin. Ph. D. Thesis Acta, University of Tokyo, 1992
58. Shimasaki S, Koga M, Buscaglia ML, Simmons DM, Bicsak TA, Ling N:Follistatin gene expression in the ovary and extragonadal tissues. Mol Endocrinol3:651-659, 1989
59. Nakamura T, Hasegawa Y, Sugino K, Kogawa K, Titani K, Sugino H:Follistatin inhibits activin-induced differentiation of rat follicular granulosa cells in vitro. Biochem Biophys Acta1135:103-109, 1992
60. Shiozaki M, Sakai R, Tabuchi M, Nakamura T, Sugino K, Sugino H, Eto Y:Evidence for the participation of endogenous activin A/erythroid differentiation factor in the regulation of erythropoiesis. Proc Natl Acad Sci USA89:1553-1556, 1992
61. Asashima M, Nakano H, Uchiyama H, Sugino H, Nakamura T, Eto Y, Ejima D, Davids M, Plessow S, Cinchocka I, Kinoshita K:Follistatin inhibits the mesoderm-inducing activity of activin A and the vegetalizing factor from chicken embyo. Roux's Arch Dev Biol200:4-7, 1991
62. Shimonaka M, Inouye S, Shimasaki S, Ling N: Follistatin binds to both activin and inhibin through the common subunit. Endocrinology 128:3313-3315, 1991
63. Krummen LA, Woodruff TK, DeGuzman G, Cox ED, Baly DL, Mann E, Garg S, Wong WE, Cossum P, Mather JP:Identification and characterization of binding proteins for inhibin and activin in human serum and follicular fluids. Endcrinology132:431-443, 1993
64. Vaughan JM, Vale WW:α2-Macroglobulin is a binding protein of inhibin and activin. Endocrinology132:2038-2050, 1993
65. Yayon A, Klagsbrun M, Esko JD, Leder P, Ornitz DM:Cell surface, heparin-like molecules are required for binding of basic fibroblast growth factor to its high affinity receptor. Cell64:841-848, 1991
66. Nakamura T, SuginoK, Titani K, Sugino H:Follistatin, an activin-binding protein, associates with heparan sulfate chains of proteoglycans on follicular granulosa cells. J Biol Chem266:19432-19437, 1991
67. Inouye S, Guo Y, DePaolo L, Shimonaka M, Ling N, Shimasaki S:Recombinant expression of human follistatin with 315 and 288 amino acids:chemical and biological comparison with native porcine follistatin. Endocrinology129:815-822, 1991
68. Dalkin AC, Haisenleder DJ, Yasin M, Gilrain JT, Marshall JC:Pituitary activin receptor subtypes and follistatin gene expression in female rats:Differential regulation by activin and follistatin. Endocrinology137:548-554, 1996
69. Hashimoto O, Nakamura T, Shoji H, Shimasaki S, Hayashi Y, Sugino H:A novel role of follistatin, an activin-binding protein, in the inhibition of activin action in rat pituitary cells:Endocytotic degradation of activin and its accelaration by follistatin associated with cell-surface heparan sulfate. J Biol Chem272:13835-13842, 1997
70. de Winter JP, ten Dijke P, de Vries CJ, van Achterberg TA, Sugino H, de Waele P, Huylebroeck D, Verschueren K, van den Eijnden-van Raaij AJ: Follistatins neutralize activin bioactivity by inhibition of activin binding to its type II receptors. Mol Cell Endocrinol116:105-114, 1996
71. Saito S, Nakamura T, Titani K, Sugino H:Production of activin-binding protein by rat granulosa cells in vitro. Biochem Biophys Rec Commun176:413-422, 1991
72. Hemmati-Brivanlou A, Melton DA:Inhibition of activin receptor signaling promotes neuralization in Xenopus. Cell77:273-281, 1994
73. Hemmati-Brivanlou A, Kelly OG, Melton DA:Fol-listatin, an antagonist of activin, is expressed in the Spemann organizer and displays direct neutralizing activity. Cell77:283-295, 1994
74. Furuya S, Sera M, Tohno-Oka R, Sugahara K, Shiokawa K, Hirabayashi Y:Elimination of heparan sulfate by heparitinases induces abnormal mesodermal and neural formation in Xenopus embyos. Develop Growth Differ37:337-346, 1995
75. Piccolo S, Sasai Y, Lu B, De Robertis EM:Dorso-ventral patterning in Xenopus:Inhibition of ventral signals by direct binding of chordin to BMP-4. Cell86:589-598, 1996
76. Zimmerman LB, De Jesus-Escobar JM, Harland RM: The Spemann organazer signal noggin binds and inactivates bone morphogenetic protein4. Cell86:599-606, 1996

Received for publication June 16, 1997;accepted July 31, 1997.

1 Address correspondence and reprint requests to Hiromu Sugino, Ph.D., Division of Molecular Cytology, Institute for Enzyme Research, University of Tokushima, Tokushima, Japan.