For about two decades, the enzymes (guanylyl cyclases) that synthesize cyclic GMP have been described as soluble or particulate, depending upon where they are found in a cell homogenate (Kimura and Murad, 1974; Chrisman et al., 1975). The particulate forms have been often further separated into those that are easily solubilized with nonionic detergents and those that are not. Early studies suggested that the enzymes found in the soluble and particulate fractions were different proteins, and the cloning of the mRNA not only confirmed this, but demonstrated that multiple subtypes existed within both compartments. The particulate forms appear to contain a single transmembrane domain, a protein kinase homology domain, and a single putative catalytic site, whereas the soluble forms contain two subunits, each of which exhibits a putative consensus cyclase catalytic domain (Figure 1). Four different cDNA clones (al, a2, f1, (2) encoding homologous proteins have been obtained for subunits of the soluble guanylyl cyclases (Koesling et al., 1991; Garbers, 1992; Yuen and Garbers, 1992), but only the heterodimer, a1fA3, has been purified from tissue extracts and shown to contain associated heme (Figure 1) (Garbers, 1979; Gerzer et al., 1981; Humbert et al., 1990). Therefore, the pairing of the remaining subunits, whether they are normally found as heteromers and whether they are actually present within the cytoplasm, is not known. Coexpression of a2/f1 results in the formation of an active enzyme (Harteneck et al., 1991), but it is not known if a2f1 exists normally. Nitric oxide appears to be a ligand for al1f1 and activates the heterodimer by as much as 170-fold (Humbert et al., 1990). The known particulate guanylyl cyclases contain a single putative transmembrane domain that separates an extracellular ligand binding domain from intracellular protein kinase-like and cyclase catalytic domains (Figure 1). Four unique cDNA clones have been isolated from various mammalian cDNA libraries for members of this part of the family and have been designated as GCA(NPR-A), GC-B(NPR-B), GC-C(STaR), and RetGC (Koesling et al., 1991; Garbers, 1992; Shyjan et al., 1992). Atrial natriuretic peptide (ANP), C-type natriuretic peptide (CNP), and heat-stable enterotoxins/guanylin appear to represent the natural ligands for GC-A, GCB, and GC-C, respectively (Koesling et al., 1991; Garbers, 1992). The ligands are synthesized in many different tissues and likely function both locally and as circulating hormones. The RetGC has the same signature domains as the other particulate forms but has been suggested to function as the photoreceptor guanylyl cyclase, not responding to extracellular ligands (Shyjan et al., 1992); however, a limited number of potential ligands have been examined, and whether the enzyme is localized to rod outer segments has not been confirmed. A guanylyl cyclase purified from rod outer segments (Margulis et al., 1993), in fact, could be encoded by a different mRNA than the one cloned from retina (Shyjan et al., 1992). The ANP-clearance receptor (Fuller et al., 1988) is another member of this family; it binds ANP and other natriuretic peptides but contains neither a protein kinase-like nor a cyclase catalytic domain (Figure 1). There has been controversy about the function of this receptor. Some have suggested that it functions as a clearance receptor, whereas others have argued that it is a signaling molecule. The requirements for basal cyclase activity or for ligand activation of the cyclases are just beginning to be understood, and although principally at a speculative stage, this will be area of focus for the essay.