The Role of Activin A in Regulation of Hemopoiesis
Activin A, a cytokine member of the transforming growth factor-β superfamily, is expressed locally by the mesenchymal component of the hemopoietic microenvironment. Its expression is regulated on the mRNA level by different cytokines, and the biological activity of the protein is tightly controlled by several inhibitory molecules. Activin A affects hemopoietic cells of various lineages, as evidenced by in vitro studies of leukemia and lymphoma cell lines, which were used to elucidate the mechanism of its action. In the B-cell lineage, activin A is a cell cycle inhibitor, a mediator of apoptosis, and a cytokine antagonist. Limited information is available on the effects of activin A on normal hemopoietic cells. Recent studies suggest that it might be a negative regulator of normal B lymphopoiesis. Whereas the functions of activin A in vitro are well established, further research tools are needed to elucidate its role within specific hemopoietic microenvironments in vivo.
Structure and Function
Activin A is a pleiotropic cytokine belonging to the transforming growth factor (TGF)-β superfamily . Various functions have been attributed to the activin A molecule since its discovery in the late 1980s as a follicle-stimulating hormone (FSH)-releasing protein, following purification from ovarian follicular fluid. Among these different functions, those commonly studied are its erythroid differentiation factor (EDF) capacity, plasmacytoma growth inhibition, mesoderm stimulation, and promotion of nerve survival. Like other members of the superfamily, activins are dimeric proteins. Five types of activin subunits have been identified, designated βA through βE . Activin A is a homodimer of two βA subunits, linked by a disulfide bond. Other studied members of the family are activin B (βBβB) and activin AB (βAβB). Inhibins, which are antagonists of activins, are formed by the heterodimerization of βA or βB subunits with the inhibin α subunit. The function of the dimers composed of the other subunits, βC, βD, and βE, is still unknown.
Activin A signals intracellularly via a membrane receptor complex consisting of two types of activin receptors that belong to the TGF-β receptor superfamily [3–5]. Several subtypes of these receptors have been identified. Activin type IIA (ActRIIA) and type IIB (ActRIIB) receptors are constitutively phosphorylated proteins containing an extracellular ligand-binding domain and an intracellular serine/threonine kinase domain. Type I receptors for activin, designated activin receptor-like kinases (ALKs), are the ActRIB/ALK-4 and ActRIA/ALK-2, which contain a cytoplasmic protein kinase domain. Different combinations of type I and type II receptors form membranal complexes that can bind several ligands other than activin A .
Activin receptors type IIA or IIB bind the dimeric activin A ligand through their extracellular domain, and this interaction recruits activin type I receptors into this membranal complex. Type II receptors transphosphorylate the type I receptors, which in turn serve as docking sites for the intracellular mediators of activin A signaling . These signaling proteins are known as Smads. Unphosphorylated receptor-regulated Smads (R-Smads: Smad2 and Smad3) bind the type I receptors and are activated by phosphorylation with the aid of Smad anchor for receptor activation . Subsequently, they are released and interact with common-mediator Smads (Co-Smad: Smad4). This hetero-oligomeric complex translocates to the nucleus and induces transcriptional activity of several genes by binding to sequence-specific promoters. Inhibition of the activin-signaling pathway is controlled by inhibitory (anti) Smads (I-Smads: Smad7), which bind to activated type I receptors and interfere with R-Smad activation. The TGF-β-induced Smad protein cascade was initially discovered following the identification of proteins in the Drosophila decapentaplegic (Dpp)-signaling pathway. Recent studies imply that mammalian Schnurri, a homologue of one component of the Dpp cascade, controls positive selection in T lymphocytes, and thus may also be involved in mammalian TGF-β signaling .
Various components of the activin-signaling cascade, such as Co-Smads, are shared by several TGF-β superfamily members. In addition, activin receptors have more than a single ligand . The competition for these molecules is one mode by which the functions of activin A are regulated. In addition, several antagonistic molecules control activin A functions; inhibin antagonizes activin A-induced release of FSH by competing for the binding to the same receptor. Follistatins, on the other hand, bind directly to activin A and inhibit its action.
Activin A and its receptors are expressed in a wide range of cells and tissues [5, 9]. Indeed, the bioactivities of this factor are observed in adult tissues such as the reproductive organs, brain, heart, and liver. Moreover, activin A functions as a morphogen during embryonal development, inducing the formation of different structures including the hemopoietic system . This review will focus on the expression and function of activin A in cells of the hemopoietic system and in cells of the surrounding stromal microenvironment that control stem cell renewal and differentiation.
Bone marrow-derived stromal cells, including mainly mesenchyme and endothelium, are known to induce and maintain hemopoiesis. This effect is also mediated on different types of leukemic cells that require stromal support for their survival and growth in vitro. By contrast, mouse myeloma (plasmacytoma) cells are growth inhibited by a stromal activity that was first designated restrictin-P [11–13], and was eventually identified as stroma-expressed activin A . Activin A might also influence the growth of stromal cells themselves in an autocrine fashion, since it is found to cause the mitogenic stimulation of activin receptor-expressing stromal cells .
Stromal cells express activin A in a constitutive manner , and this basal expression is enhanced upon stimulation with proinflammatory cytokines and mitogenic molecules such as interleukin-1α (IL-1α) [17, 18], IL-1β [19, 20], lipopolysaccharide (LPS) [17, 18], tumor necrosis factor (TNF)-α [17–20], phorbol myristate acetate [19, 20], basic fibroblast growth factor (bFGF) [15, 21], and platelet-derived growth factor . Activin A expression is inhibited by glucocorticoids such as hydrocortisone and dexamethasone . High levels of activin A have been found in synovial fluid from patients with rheumatoid arthritis and gout , implying a role for activin A in response to inflammation, although the mechanism is unclear.
Experiments with TNF-α show that the mechanism of activin A stimulation is by direct enhancement of activin A mRNA transcription . Varying transcript sizes (6.4, 4.0, 2.8, and 1.6 kilobases) are detected in different types of stromal cells [17, 20, 23], although all are expressed from one 5′ transcription start site . Western blotting and immunochemical staining detect activin A protein in stromal cells [17, 23]. Three promoter elements are identified upstream to the start codon, indicating complex regulation of this gene, which might depend on the type of stromal cell and the cytokines acting on it. Indeed, stromal cell types differ dramatically in the levels of activin A secretion; therefore, some types are inhibitory, while others are permissive to plasmacytoma cells . This matter is more complex since activin A-producing stromal cells also express very low levels of inhibin α subunit  and follistatin mRNAs, both inhibitors of activin A action . bFGF transforms the permissive effect of some stromal cells on plasmacytomas to an inhibitory one by the induced secretion of activin A . Moreover, follistatin-related gene (FLRG), first detected in B-cell leukemia , is also produced in stromal cells and in other hemopoietic cells that produce activin A, physically binding to and inactivating it . TGF-β causes a rise in activin A and FLRG, but not in follistatin, in stromal cells . The biological effects of activin A are thus tightly regulated, and its mere expression by the stroma does not guarantee that a functional molecule is being secreted.
Another level in this regulatory system is the spatial distribution of activin A-expressing cells. Activin A is not uniformly expressed within hemopoietic microenvironments. In mouse spleen, activin A is most abundant in the red pulp, whereas in the lymphoid follicles, it is found at a low titer (Shoham et al., submitted for publication). This seems to correlate with the opposing effects of activin A on erythroid versus B-lymphoid cells (see below).
Hemopoietic Stem Cells
The effects of activin A on committed and mature hemopoietic cells have been studied in great detail, as summarized below. By contrast, detailed effects on the stem cell compartment still await further elucidation. Broxmeyer et al.  were first to show that activin A enhances the formation of colony-forming units-granulocyte-erythroid-macrophage-megakaryocyte (CFU-GEMM), while it does not affect CFU-granulocyte-macrophage (GM) numbers [28, 29]. On the other hand, other studies show that activin A suppresses CFU-GM formation [30, 31]. Activin A expression is induced during hemopoietic differentiation. Expression is found in CFU-GEMM and CFU-GM cells, yet, in earlier precursors (CD34+ populations containing multipotent progenitors and committed precursors), it is not expressed . Injection of inhibin leads to reduced numbers of CFU-GM in bone marrow and spleen, implying by inference that activin A may be involved in induction of early stages of hemopoiesis . Conversely, experiments carried out recently in our laboratory show that injection of recombinant activin A into irradiated-bone marrow-transplanted mice results in reduced bone marrow cellularity at 2 weeks posttreatment (Shoham et al., unpublished observations). It therefore remains to be determined whether activin A directly affects early stem cells.
The effect of stroma-derived activin A/restrictin-P on hemopoiesis was first identified by virtue of its suppressive effect on plasmacytoma cell growth . Indeed, activin A causes apoptotic death in cells of the B lineage as observed by experiments performed on tumor cells including human myeloma cells and mouse hybridoma and plasmacytoma cell lines [14, 33, 34]. In MPC-11 plasmacytoma cells seeded with activin A, microvilli are lost and membrane blebbing begins, followed by chromatin condensation, nuclear fragmentation, and, finally, apoptotic death . This function of activin A is abrogated by follistatin . Despite the fact that TGF-β1 shares with activin A a similar intracellular signaling cascade, it does not affect plasmacytoma cell growth or survival.
Activin A induces cell cycle arrest and a shift towards the G0/G1 cell cycle stages in B-lineage cells, including plasmacytomas and B-cell hybridomas [35, 37–39]. The mechanism by which activin A imposes G1 cell cycle arrest on B-cell hybridomas and hepatoma cells involves the hypophosphorylation of retinoblastoma protein (Rb) [38, 40]. Cyclin D2 binds to CDK4 and induces its kinase activity on Rb, whereas p21Waf1/Cip1 binds this complex and negatively regulates phosphorylation. Upon addition of activin A, the levels of cyclin D2 are downmodulated in conjunction with an increased expression of p21Waf1/Cip1. These occurrences do not permit the hyperphosphorylation of Rb required for the transition from G1 to S phase . Ishisaki et al. [39, 42] show that the genes for I-Smads, Smad7 and Smad6, are induced by activin A. Overexpression of Smad7 suppresses the G1 arrest caused by activin A by abolishing the increased expression of p21Waf1/Cip1, and also has an inhibitory affect on apoptosis. The phosphorylation of Smad2, the intracellular mediator for the activin A-signaling pathway, is also abolished by Smad7 overexpression. Smad6, on the other hand, has no effect on either activin A-induced G1 growth arrest or apoptosis .
Bcl-2 family proteins play an important role in apoptosis and were initially identified in B-cell lymphomas. Bcl-2 and Bcl-xL have antiapoptotic activity, while Bax, Bcl-xS and other members of this family have proapoptotic activity and promote cell death. Mouse B-cell hybridomas have very low levels of Bcl-2 and Bcl-xL. Overexpresssion of Bcl-2 in these cells inhibits the apoptotic effect induced by activin A, while Bax levels remain low , although these cells still become arrested at G1 . On the other hand, activin A was shown to cause an increase in the levels of the proapoptotic protein, Bcl-xS. This effect is reduced when protein kinase C (PKC) is inhibited, hence suppressing apoptosis . During this inhibition of PKC, Bcl-xL levels are increased. It is therefore suggested that the induction of apoptosis in B-lineage cells by activin A is controlled by Bcl-xS, which functions as a dominant negative repressor of Bcl-xL.
Experiments in hepatoma cells show that activin A causes attenuation of the cell cycle but does not direct the cell towards death by apoptosis . Also in the case of plasmacytoma cells, cell death mediated by activin A is not solely related to cell cycle arrest. It is found that activin A interferes with the IL-6-signaling pathway required for plasmacytoma and hybridoma cell growth . The antagonism of the IL-6 signal is caused by activated Smad2 or Smad3 proteins in cooperation with Smad4, which interferes with the transcriptional activation of IL-6-induced promoters . The coactivator p300 is shared by activin A and IL-6-signaling cascades and participates in the formation of transcription complexes. Apparently, activin A activates Smads that compete on the available cellular p300, thus causing disintegration of IL-6 transcription complexes. In addition, Smad4 may have a separate function since its overexpression reduces the binding of the IL-6 transcription factor, CCAAT/enhancer-binding protein β, to specific promoter sites .
Activin A protein is stored abundantly in bone matrix  and was purified both from bone marrow stromal cell-conditioned medium  and a macrophage cell line . These findings suggest that activin A might be produced locally within the bone marrow microenvironment and thus serves as a B-cell regulator. Interestingly, in B cells, neither activin A, follistatin, nor FLRG are expressed , and therefore, the effect of activin A on these cells is the outcome of the relative concentrations of these molecules in the surrounding tissue. Some experimental evidence points to a possible in vivo role for activin A as a regulator of the B-lymphocyte lineage; activin A inhibits the formation of plasmacytoma tumors in an in vivo mouse model . In human nasal polyps that contain infiltrating lymphocytes, the spatial distribution of B cells correlates negatively with the presence of activin A in the tissue .
Suppression of activin A action by addition of follistatin to long-term bone marrow lymphopoietic cultures (LTBMLCs) causes the early onset of B lymphopoiesis , suggesting a negative effect of activin A on normal B-lineage cells. Our recent studies show that there is an inverse relationship between functional activin A titer and the abundance of B cells in LTBMLC. Furthermore, isolated B-cell precursors seeded in vitro in the presence of activin A are halted at an earlier differentiation stage compared with cells seeded in the absence of this cytokine (Shoham et al., submitted for publication).
The identification of activin A as a stromal factor that negatively regulates hemopoiesis in a lineage-specific manner, led to the formulation of the model of “restrictive mode of cell organization within tissues” ; the existence of lineage-and stage-specific inhibitors expressed by stromal cells is proposed. These inhibit/kill specific target hemopoietic cells at specific loci within the bone marrow environment [50, 51]. This theory was updated  and was recently substantiated by the identification of molecules that act specifically to suppress specific hemopoietic lineage only. Limitin is found to preferentially inhibit B lymphopoiesis , while WECHE (for WEird CHEmokine) is a specific erythroid-differentiation inhibitor  and adiponectin suppresses myeloid progenitors and spares erythroid progenitors or lymphoid cell lines .
Activin A does not affect the viability of T cells [13, 33], whereas TGF-β is known to be a potent mediator of T-cell death. However, several studies utilizing [3H]thymidine uptake assays show that activin A does have an antiproliferative activity on mitogenically stimulated mature thymocytes in vitro [56–58]. This inhibition of proliferation also occurred in the presence of IL-1β, a stimulator of T cells, and entails the inhibition of IL-6 production by these cells. IL-6 is a stimulator of T-cell proliferation and exerts its action in an autocrine manner. Yet, when both activin A and IL-6 are added to thymocyte cultures, proliferation is enhanced . In addition, activin A was found to reverse the increased interferon-γ (IFN-γ) production induced by inhibin . It therefore seems that activin A has both positive and negative effects on T-cell growth and cytokine production.
An activity purified from monocytic cells, which causes the differentiation of erythroleukemia cells, therefore termed EDF , is identified as activin A . Activin A is shown to be expressed by tissue plasminogen activator-induced K562 erythroleukemia cells .
Activin A induces the differentiation of mouse Friend erythroleukemia cells (MEL) at low concentrations (1-10 ng/ml) but suppresses their growth in soft agar . This is also shown in human K562 cells together with an increase in hemoglobin synthesis  and in the transcription of globin genes . The effect of activin A on erythropoiesis is also shown in vitro [28, 29] and in vivo by the administration of the molecule to rodents. The in vivo studies show an increase in erythroid precursors , circulating red blood cells , and reticulocyte release  upon treatment, whereas follistatin causes a reduction in progenitor numbers .
The G1 growth arrest caused by activin A in K562 cells is associated with the hypophosphorylation of Rb , as described above for B cells. Some erythroleukemia cells need both erythropoietin (EPO) and activin A for induction of differentiation, whereas mouse Friend cells have a constitutively active EPO receptor, and activation with EPO is unnecessary . The latter study shows that activin A acts prior to EPO, thereby proposing that activin A can also act as a commitment factor for erythroid differentiation. Interestingly, MEL cells overexpressing the inhibitory Smad7, which is normally not expressed by erythroleukemia cells, do not synthesize hemoglobin upon activin A stimulation. Although both types of activin receptors are found in these cells, inhibition of ActRIA signaling by Smad7 is more potent than its inhibition of ActRIB signaling, indicating that a more complex mechanism for activin A signaling exists .
It is suggested that activin A affects erythroid differentiation by directly interacting with receptors on erythroid cells, whereas proliferation of these cells is mediated indirectly via effector molecules produced by activin A-stimulated monocytes [28, 73]. The function of activin A is also observed at earlier stages of erythropoiesis. In these cases, activin A enhances the formation of BFU-E [66, 74]. This is also seen in conjunction with EPO [28, 29]. As for erythroid colony-forming units, there is some discrepancy between studies. Activin A is shown to have either inducing effects, both in vivo and in vitro [64, 66], inhibitory effects [31, 74], or no effect at all .
Myelomonocytic cells produce and secrete activin A [17, 61, 75–78], although stromal cells are approximately 10-fold more potent producers . As mentioned above, EDF was first found in the culture medium of activated THP-1 monocytic cells . The production of activin A by monocytes is stimulated by GM-CSF, IFN-γ, and LPS, and is downregulated by hydrocortisone, dexamethasone, and retinoic acid [17, 78]. In contrast to bone marrow-derived stromal cells, activin A production by monocytes is not affected by treatment with IL-1α and TNF-α .
The outcome of activin A treatment on monocytic differentiation and cytokine expression is dependent on the cell line tested. The IL-6-dependent differentiation of monocytic M1 cells into macrophages is abolished upon addition of activin A. Activin A alone has no effect on phagocytosis and acts only as an antagonist of IL-6 . On the other hand, activin A is shown to induce the generation of IL-6 by peripheral blood monocytes, which probably leads to the augmentation of IgE release by B cells . In addition, activin A inhibits the growth of HL-60 promyelocytic leukemia cells and induces differentiation towards the monocyte/macrophage lineage [80, 81]. Apoptosis in this cell line is inhibited by activin A . In activated monocytic cells, activin A inhibited IL-1β secretion and enhanced IL-1 receptor antagonist secretion, hence causing a decrease in overall IL-1 biological activity . Finally, one study shows that activin A has a positive effect on monocyte chemotaxis .
A recent study shows that activin A caused growth arrest and apoptosis in human chronic myelogenous leukemia cells (KU812). Bax protein is induced and caspases are activated in these cells. When EPO is added together with activin A, these cells become hemoglobin-synthesizing cells .
Whether activin A might have a role in megakaryopoiesis is still unclear. Activin A can induce the differentiation of a megakaryocytic cell line (L8057), as observed by determination of acetylcholinesterase activity in these cells [85, 86]. This work further shows that a MEL cell line has the potential to differentiate not only to the erythroid lineage but also to the megakaryocytic lineage via activin A induction .
Like most other cytokines, activin A is pleiotropic and affects different cell types within the hemopoietic system. Probably the most well-studied and most prominent activities are those related to the erythroid and B-lymphocyte lineages. In vivo studies suggest that activin A enhances erythropoiesis, while it is apparently inhibitory to the generation of B-lineage cells. Such discriminating effects and the confinement of activin A to specific sites within the hemopoietic microenvironment imply a role for this factor in the organization of the hemopoietic system into functional domains that differ in their cell type composition.
Dov Zipori is an incumbent of the Joe and Celia Weinstein professorial chair at the Weizmann Institute of Science.