D. GRAHAME HARDIE CELL SIGNALLING Reactin’ with activin The receptor for activin, a peptide messenger possibly involved in early differentiation steps in the embryo, may be ‘a ligand-activated serine kinase. Transduction of eflects of extracellular messengers into the cell is achieved by changes in the concentra- tion of intracellular second messengers (for example, CAMP, cGMP, diacylglycerol, Ca2+), which then acti- vate serine/threonine-specific protein kinases (known for simplicity as setie kinases), or by binding of extracellu- lar messengers to transmembrane receptors, which acti- vates the protein kinase functions carried on the recep- tors’ intracellular domains, Until recently the protein ki- nase activities of the latter (for example, the epidermal growth factor, plateilet-derived growth factor and insulin receptors) were thought to be exclusively tyrosine-spe- cific. However, cDNAs encoding a couple of putative re- ceptors were recently cloned in which the sequence of the intracellular domain resembles serine/threonine-spe- cilic protein kinases [ 11. These are the &#l gene prod- uct of the nematodie Caenorlxzbditk elegans, and the ZmPKl gene product of the plant Zea muys. As yet, the enzymic activity of these putative receptors has not been studied biochemically, nor has their extracellular ligand, if any, been identified. However, the latter importznt step has now been achieved by Mathews and Vale [2] in the case of another system, the activin .receptor. Activins are peptide messengers in the transforming growth factor (TGF)-P superfamily. They consist of homodimers or heterodimers of two closely related polypeptide chains, that is, activin A (&,(JA), activin AB(&&), and act&in B (&&). The inhibins, which an- tagonize some actions of the activins, consist of PA or & chains complexed with an unrelated (a) polypeptide. The activins and inhibins were first characterized by their ability to modulate the release of follicle-stimulating hor- mone from cells of the anterior pituitary gland. It has sub- sequently been found that the activins have many actions. Particular excitement was generated by findings that fac- tors from various Xenopus and mamma&an cell lines that induce mesoderm formation in Xenopus embryos are similar or identical to the activins [3-7] (Fig. 1). Classic cell biology experiments as early as 1939 181 had indi- cated the existence in Xenopus embryo of mesoderm- inducing factors that stimulate a switch in development of ectodermal tissue into mesoderm, which subsequently gives rise to neural and muscular tissue. In truth, it is not clear that the ac%s :are the mesoderm-inducing factors in the embryo itself, because activin mRNAs do not seem to be expressed early enough [7]. It remains possible, however, that release of stored maternal activin protein is the key event. Mathews and Vale have taken an important step in un- derstanding the mechanism of action of these intriguing messengers by cloning an activin receptor 121. Compli- mentary DNA from a pituitary cell line (AtT20) was used to make an expression vector library. Potions of the library were expressed in COS cells, and activin bihd- ing was detected autoradiqgraphically by binding of 12% activin to individual cells. Repeated subdivision and re- screening of positive portions of the library resulted in cloning of cDNA, and eventually a full-length cDNA was obtained and sequenced. Fig. 1. Induction of axial structures in a Xenopus embryo by injec- tion of synthetic activin mRNA. The mRNA was injected into the embryo (bottom) at the 32-cell stage, and a second dorsal axis was induced as a result 171. An uninjected control embryo of the same age is shown (top). (Photograph courtesy DA Melton.) Cells transfected with the cloned cDNA, but not untrans- fected controls, expressed a single class of high-affinity binding site for W-activin A (I&J = 150pM), and bind- ing was abolished by unlabelled actitiri A or B. Binding was also abolished with lower potency by inhibin A, but because the biological potency of inhibins on susceptible cells is normally several times higher than that of activin, the authors suggest that a distinct, high-a&&y receptor for inhibin must exist. The activin receptor cDNA encodes an open reading frame of 513 amino acids, and the authors’ model for the protein structure includes a hydrophobic leader se- quence, an extracellular domain of 116 amino acids, a single transmembrane sequence, and an intracellular do- main of 452 amino acids (Fig. 2). The putative extracel- lular domain is not related to any other protein in the database, but the putative intracellular domain is clearly related to the protein kinase family, and -eight out of Volume 1 Number !5 1991 321