B. The Guanylate Cyclase Family
1. Particulate Guanylate Cyclase
Guanylate cyclase is the intracellular enzyme responsible for the synthesis of cGMP from GTP (GTP pyrophosphate-lyase (cyclizing), EC 4.6.1.2). Guanylate cyclase is found in two biochemically different forms. One form of guanylate cyclase is membrane bound (particulate) and the other is cytoplasmic (soluble). The particulate form of guanylate cyclase appears to be evolutionary a very old enzyme which can be found in most animal species and
mono-cellular systems (Tremblay, et al. 1988). The membrane bound family of guanylate cyclase contains at least four different monomeric isozymes, each encoded by a unique gene (Bentley and Beavo, 1992). The main distinguishing feature of this guanylate cyclase family of isozymes, is that they are only activated by direct stimulation of a ligand binding to the enzyme's extracellular domain (Goy, 1991). The particulate form of guanylate cyclase possess a extracellular domain, which binds the enzyme's appropriate signal, a single transmebrane domain, and an intracellular kinase-like or multi-catalytic domain (Bentley and Beavo, 1992). The type A isozyme has a molecular weight of 115 kD, and is activated by the binding of atrial natriuretic peptide (Chinkers, et al. 1989). The second particulate guanylate cyclase isozyme has been designated type B. Similar to type A, type B has a molecular weight of 114 kD, and is activated by the binding of brain natriuretic peptide (Schulz, et al. 1989). The third isozyme is known as type C, and is found almost exclusively in the epithelium of the gut. Interestingly this isozyme is stimulated by enterotoxin which is excreted by E. coli in the intestinal tract (Bentley and Beavo, 1992). Particulate guanylate cyclase type P, is a photoreceptor specific form of the enzyme that plays a major role in visual excitation and recovery. Type P seems to be activated by a polypeptide known as recoverin, in the presence of Ca2+ (Dizhoor, et al. 1991)(Stryer, 1991). Recoverin is a calcium-sensitive stimulatory protein which exhibits three potential calcium-binding sites characteristic of the EF-hand motif.
Particulate guanylate cyclase has been shown to be extensively regulated via post-translational modification. A number of investigators have shown particulate guanylate cyclase to be phosphorylated by PKC (Bently, 1991)(Louis, et al., 1993). In contrast to many other enzymes with cell surface receptors, guanylate cyclase is desensitized by ligand-induced dephosphorylation (Potter and Garbers, 1992). Not only does the phosphorylation state of particulate guanylate cyclase dictate the activity of the enzyme, it has also been shown to regulate the degree of cooperative interaction between catalytic or GTP-binding sites on the enzyme (Ramarao and Garbers, 1988).
Because particulate guanylate cyclase does not play a role in the EDRF initiated cGMP cascade described above, it will not be discussed further, and is only included in this review for the sake of completeness.
The soluble form of guanylate cyclase has been found in most mammalian cells. It is this form of guanylate cyclase which is responsible for the endothelium mediated production of cGMP in the vascular smooth muscle. Although particulate guanylate cyclase is also present in vascular smooth muscle, it is has not been shown to significantly contribute to endothelium mediated vasorelaxation (Papapetropoulos, et al., 1996). Unlike the membrane bound particulate form of guanylate cyclase, the soluble form of guanylate cyclase includes a heme group as a co-factor, and the enzyme itself is located in the cytosol. Therefore, due to the cytosolic positioning of soluble guanylate cyclase, it must be activated indirectly via EDRF, or a number of other compounds (peroxides, unsaturated fatty acids such as arachidonate, or certain lipids such as lysophosphatidyl choline and oleic acid) (Goy, 1991)(Pearce and Harder, 1994). Soluble guanylate cyclase is activated when NO interacts with its heme component (Gerzer, et al., 1982)(Garbers 1989). Although it is conceivable that other agents capable of interacting with the heme group (other free-radicals) could produce changes in guanylate cyclase activity, it has been shown that NO is physiologically the most important and abundant of these agents (Ignarro, et al., 1987). Figure 2 illustrates the functional domains and known isoforms for both particulate and soluble guanylate cyclase
Soluble guanylate cyclase is comprised of two subunits (a and b), with at least one heme prosthetic group loosely held in place with a histidine as the trans-axial ligand. The proposed site involved in the hydrogen bonding of the heme prosthetic group is at the His-105 residue within the b subunit (Wedel, et al., 1994). Mutations at this site, resulted in the apparent loss of the heme group from the soluble guanylate cyclase heterodimer. Although the majority of the literature states that only one heme is bound per soluble guanylate cyclase, recent data suggests that more than one heme may be bound within the enzyme. Stone and Marletta, (1995) have demonstrated that each subunit of the heterodimer binds one moiety of heme at a site conserved between the two subunits. In contrast to Wedel, et al., (1994), Stone and Marletta have shown these sites to be in the central portion of each subunit, corresponding to residues a His-290/b His-220 or a His-407/b His-346 in the bovine a1b1 isoform of soluble guanylate cyclase.
a. Activation of Soluble Guanylate Cyclase
The binding of NO and subsequent activation of soluble guanylate cyclase is unlike that seen in any other hemeprotein. Nearly all hemeproteins which bind NO, O2 or CO, do so only in an anaerobic environment since oxygen will readily bind to the 6-coordinate position. Soluble guanylate cyclase can bind all of these ligands under both anaerobic and aerobic conditions (Stone and Marletta, 1994). This indicates that the micro-environment of the heme group within soluble guanylate cyclase is very different than in most hemeproteins. Recent investigations into the heme coordination and ligand binding properties of soluble guanylate cyclase have lead to a comprehensive model of soluble guanylate cyclase activation (Figure 3). Although basal soluble guanylate cyclase activity is heme independent heme (Ignarro, 1989), activation of the enzyme via NO is heme dependent. Maximal enzyme activity has been demonstrated in soluble guanylate cyclase preparations where the heme moiety is fully reduced (ferrous) (Ignarro, et al. 1984)(Ignarro, 1989). When the endogenous heme groups within the soluble guanylate cyclase were in a oxidized state (ferric), activation of the enzyme via NO produced a 57% is reduction in Vmax compared to reduced conditions. When heme deficient enzymes were reconstituted with pre-formed NO-heme groups, only NO-heme groups in the ferrous state would be incorporated into the enzyme resulting in activation (Stone, et al., 1996). This indicates a significant conformational change within the heme binding pocket of the enzyme associated with the oxidation of the heme iron.
The heme moiety is loosely bound within each subunit in a 5-coordinate high spin ferrous heme with histidine as the axial ligand (Deinum, et al., 1996) (Stone and Marletta, 1994). When NO first binds to the ferrous iron of the heme group, it forms a 6-coordinate nitrosyl complex. The 6-coordinate nitrosyl complex converts relatively rapidly to the 5-coordinated complex which breaks the trans-axial proximal histidine bond. It is the breakage of the histidine ligand that allows for the complete displacement of the ferrous iron from the plane of the porphyrin ring. The conformational change in the heme group leads to a global conformational change in the holoenzyme leading to activation (Saheki, et al., 1990). Although the binding of NO into a 6-coordinate complex activates soluble guanylate cyclase, complete activation is not achieved until the NO-heme complex is converted to the 5-coordinated nitrosyl complex. The conversion from the 6-coordinate nitrosyl complex to the 5-coordinated complex is carried out in one of two ways. For 28% of the heme, the 6-coordinate nitrosyl complex converts very rapidly to the 5-coordinate complex. For the remaining 72% of the heme moieties, the conversion to the 5-coordinate nitrosyl complex is slow and is dependent upon the binding of NO to an unidentified non-heme site within the heme pocket (Stone and Marletta, 1994). This mechanism of activation helps explain the apparent burst activity seen during the early stages of soluble guanylate cyclase activation. O2 and CO can also bind to the heme group of the enzyme in a similar manner. When O2 or CO binds to the iron of heme, the binding angle is slightly bent and tilted, and the bond between the heme iron and the proximal histidine remains intact. The negative polarity of the distal heme pocket drastically reduces the affinity for O2 and CO binding, and de-stabilizes O2 binding (Deinum, et al., 1996). Activation of soluble guanylate cyclase by agents other than free-radicals, such as arachidonate or lysophosphatidyl choline, have been shown to be heme independent (Ignarro 1992). It is thought that activation by such free fatty acids is due to the direct interaction of arachidonate at the site of normal heme binding.
Based on the low dissociation rate constants for NO-heme seen in other hemeproteins, it had been assumed that dissociation of NO from the heme group of soluble guanylate cyclase would be too slow to account for apparent physiological deactivation of the enzyme. The dissociation of NO 5-coordinate complex is much faster than from typical high-spin ferrous hemoproteins. The dissociation rate of NO is approximately 1 s-1. This increased rate may be explained by the polarity effects of the distal heme pocket, which may facilitate the dissociation of NO from the heme. Therefore, the more important determinant of physiological deactivation, is the slower rate for the reassociation of the histidine with the heme iron (for the major component of the heme population) at 0.02s-1 (Deinum, et al., 1996)(Stone and Marletta,1996). If this is the case, then the half-life for deactivation would be 35s, consistent with the apparent deactivation of soluble guanylate cyclase observed in studies employing whole tissue (White and Pearce, 1996).
Although Stone and his colleagues have provided strong evidence for the model of soluble guanylate cyclase activation described above, some controversy still remains. Stone and others have proposed that the unstimulated heme moiety is complexed into the enzyme only in the 5-coordinate high spin ferrous heme configuration. Several other investigators have shown a mixed population of 5-coordinate and 6-coordinate nitrosyl complexs within the soluble guanylate cyclase pool (Idriss, et al., 1992)(Burstyn, et al., 1995). When the unstimulated enzyme is found in the 6-coordinate configuration, it has been shown to bind to a second histidine in the distal heme pocket (Burstyn, et al., 1995). Idriss, et al. (1992) has shown that soluble guanylate cyclase isolated from the human placenta is 1/5 as active as the enzyme isolated form the bovine lung. Soluble guanylate cyclase isolated from the human placenta was found to exist primarily in the less active 6-coordinate heme configuration, whereas the more active enzyme isolated from the bovine lung, was found to be in the 5-coordinate heme configuration. These observations suggest a mechanism for soluble guanylate cyclase regulation based primarily upon the coordinational configuration of the heme group.
b. Known Subunits of Soluble Guanylate Cyclase
Both the a and b subunits have been cloned, and restriction maps of the cDNAs have been developed for each (Nakane, et al., 1988) (Nakane, et al., 1990). The maps of the two cDNAs are quite different, which indicates that the two subunits are probably derived from different mRNAs rather than a product of alternative splicing. The a subunit is a 82 kD protein and is encoded for by a 5.5-kb mRNA. The b subunit is a 70 kD protein and is encoded for by a 3.4-kb mRNA. These subunits have been shown to be highly conserved across species (Koesling, et al., 1988). Three isoforms for both the a and b subunit have been identified and designated a1, a2, a3 and b1, b2, b3. To date, cDNA sequences for all six a and b subunits have been reported (Nakane, et al., 1990)(Yuen, et al., 1990)(Harteneck, et al., 1991). The greatest homology between the subunits was found within a 300 amino acid region at the carboxyl terminal. This region shares considerable homology with the known cytoplasmic catalytic domain found in particulate guanylate cyclase, and the catalytic domain of adenylate cyclase, suggesting a common functional motif responsible for soluble guanylate cyclase's catalytic activity. Although both subunits possess one of these putative catalytic domains, studies that have expressed the cDNA for each subunit in tandem or individually, have showed that both subunits are required for catalytic activity (Harteneck, et al., 1990) (Nakane, et al., 1990)(Buechler, et al., 1991).
The a subunit ranges in size from 81.3 kD to 82 kD. The a1 and
a3 subunits display 80% homology at the protein level, whereas, the a2 subunit shares only a 32% homology with the a1 and a3 subunits (Papapetropoulos, et al., 1996). While the a2 subunit is similar to the a1 and a3 subunits throughout the middle and C-terminal, there are an additional 31 amino acid present within the catalytic domain. The additional insert is highly homologous to a region within the putative catalytic domain of adenylate cyclases (Harteneck, et al., 1991)(Behrends, et al., 1995). The b subunit ranges in size from 70 kD to 76.3 kD. The b1 and b3 subunits can not be considered true isoforms since they share a 99% homology at the amino acid level (Papapetropoulos, et al., 1996). The b2 subunit is less than 40% homologous with the b1 and b3 subunits. Additionally, the carboxyl terminus of the b2 subunit extends 86 amino acids beyond that of the b1 and b3 subunits, and contains an isoprenylation / carboxymethylation consensus sequence which may serve to localize soluble guanylate cyclase containing the b2 to the cell membrane of the cell (Yuen, et al., 1990). There is also evidence that the alterations in the b subunits are due to alternative splicing (Chhajlani, et al., 1991). Soluble guanylate cyclase mRNA encoding b1 and b2 subunits found in human lung tissue, possess two very different 5' intron splice sites which precedes two alternatively spliced exons. The resulting b subunits differ by 99 amino acids at the carboxyl terminus. Because these subunit alterations occur in the putative catalytic domains of their carboxy terminus, it is very likely that these modifications effect the activity and function of soluble guanylate cyclase. The actual function of the additional sequences or alternative splicing in a and b subunits remains unclear, but this does lead to the suggestion that various heterodimeric isoforms of soluble guanylate cyclase exist which differ from each other by less than 100 amino acids.
Although soluble guanylate cyclase has been well characterized in terms of amino acid sequence and heme-dependent activation, dramatic evidence of direct regulation has not been widely demonstrated for the soluble form. It has been speculated, that as with other enzymes which are present as different isoforms, soluble guanylate cyclase may potentially be subject to different modes of regulation for its catalytic activity as well as different tissue localization and expression (Ahmad and Barnstable, 1993)(Papapetropoulos, et al., 1994).
From the limited data available there is evidence of tissue specific expression. Soluble guanylate cyclase comprised of the a1 and b1 subunits has been shown to be expressed in many different tissue types (Nakane, et al., 1988)(Nakane, et al., 1990). Soluble guanylate cyclase isoforms comprised of a2 and b1 subunits, a1 and b2 subunits, and a3 and b3 subunits have also been found expressed primarily in the vasculature of the rat kidney, liver, and human vascular tissue respectively (Yuen, et al., 1990)(Ujiie, et al., 1993) (Papapetropoulos, et al., 1996). The genes encoding a3 and b3 subunits have been shown to be colocalized to the same region of the human chromosome. This represents a possible mechanism for the coordinated regulation of expression for both of the soluble guanylate cyclase subunits in a given tissue (Giuili, et al., 1993). The responsiveness of these varying soluble guanylate cyclase isoforms to NO does seem to be altered, supporting the speculation that soluble guanylate cyclase activity may be tissue specifically regulated by the presence of different soluble guanylate cyclase isoforms.
Some form of molecular regulation my also be seen in the substrate specificity of soluble guanylate cyclase. Both cyclic nucleotide enzymes, soluble guanylate cyclase and adenylate cyclase, have been linked to a common ancestor (Krupinski, et al., 1989), yet both enzymes are capable of precise substrate discrimination from structurally similar substrates. Point mutations of the highly conserved amino acids Gly-107 and/or Val-124 within the consensus sequence of the substrate binding region of soluble guanylate cyclase and adenylate cyclase (GTP binding region and ATP binding region, respectively), results in a major shift in substrate preference and synthetic rates for each enzyme (Beuve and Danchin 1992). This suggests that any modification of the consensus substrate binding amino acids may cause soluble guanylate cyclase activity to be altered.
The catalytic activity of most enzymes can be regulated by covalent modification of the enzyme. Although the phosphorylation of particulate guanylate cyclase has been clearly demonstrated by many investigators, the phosphorylation of soluble guanylate cyclase has not been widely reported. However, Zwiller has demonstrated the phosphorylation of soluble guanylate cyclase by PKC in response to various phorbol esters, in intact cells and purified preparations (Zwiller, et al., 1981) (Zwiller, et al., 1985)(Louis, et al., 1993). Phosphorylation produces a stable increase of 70% in crude soluble guanylate cyclase basal activity. The stimulation of PKC via phorbol esters rather than a cAMP-dependent system, indicates a possible long term modulation of soluble guanylate cyclase. To what extent PKC mediated phosphorylation may occur in vivo is still uncertain.
Covalent modification to soluble guanylate cyclase are not necessarily required to alter the activity of soluble guanylate cyclase. Changes in the level of stimulation to the vascular smooth muscle by NO and other free-radicals (either under basal or activated conditions) can also significantly alter the rates of cGMP synthesis without effecting soluble guanylate cyclase itself. Alteration these diffusible activators under basal conditions may increase the vascular smooth muscle's response to subsequent stimulation of soluble guanylate cyclase. Related to this form of regulation would be the formation of free-radical scavengers such as superoxide anion and/or any oxidizing compound in the tissue. Formation of these scavengers would effectively inhibit the same soluble guanylate cyclase activity by lowering the concentration of these diffusible activators (Lincoln, 1989).