K. S. Misono, Department of Biochemistry, University of Nevada School of Medicine, Reno, Nevada 89557, USA Fax: +1 775 784 1419 Tel: +1 775 784 4690 E-mail: email@example.com
Atrial natriuretic peptide (ANP) and the homologous B-type natriuretic peptide are cardiac hormones that dilate blood vessels and stimulate natriuresis and diuresis, thereby lowering blood pressure and blood volume. ANP and B-type natriuretic peptide counterbalance the actions of the renin–angiotensin–aldosterone and neurohormonal systems, and play a central role in cardiovascular regulation. These activities are mediated by natriuretic peptide receptor-A (NPRA), a single transmembrane segment, guanylyl cyclase (GC)-linked receptor that occurs as a homodimer. Here, we present an overview of the structure, possible chloride-mediated regulation and signaling mechanism of NPRA and other receptor GCs. Earlier, we determined the crystal structures of the NPRA extracellular domain with and without bound ANP. Their structural comparison has revealed a novel ANP-induced rotation mechanism occurring in the juxtamembrane region that apparently triggers transmembrane signal transduction. More recently, the crystal structures of the dimerized catalytic domain of green algae GC Cyg12 and that of cyanobacterium GC Cya2 have been reported. These structures closely resemble that of the adenylyl cyclase catalytic domain, consisting of a C1 and C2 subdomain heterodimer. Adenylyl cyclase is activated by binding of Gsα to C2 and the ensuing 7° rotation of C1 around an axis parallel to the central cleft, thereby inducing the heterodimer to adopt a catalytically active conformation. We speculate that, in NPRA, the ANP-induced rotation of the juxtamembrane domains, transmitted across the transmembrane helices, may induce a similar rotation in each of the dimerized GC catalytic domains, leading to the stimulation of the GC catalytic activity.
Atrial natriuretic peptide (ANP) (Fig. 1) is secreted by the atrium of the heart in response to blood volume expansion. ANP stimulates salt excretion  and dilates blood vessels [2,3], thereby lowering blood pressure and reducing blood volume. ANP counterbalances the renin–angiotensin–aldosterone (RAA) system and plays a central role in cardiovascular homeostasis. ANP also suppresses cardiac hypertrophy and fibrosis, and is involved in remodeling of the heart and the vascular system [4–8]. B-type natriuretic peptide (BNP) is mainly produced in the ventricle, and has activities similar to those of ANP . C-type natriuretic peptide (CNP) occurs in the brain , vascular endothelium , cartilage , and other peripheral tissues, and plays a variety of local regulatory roles. The physiological and pathophysiological roles of natriuretic peptides and receptor systems are reviewed in this series by Kishimoto et al.  and Pandey . The detailed structure–activity relationship for ANP has been studied with a peptide synthesis approach, and is summarized in .
Natriuretic peptide receptors –molecular topology
The hormonal activities of ANP and BNP are mediated by natriuretic peptide receptor-A (NPRA). NPRA is a single transmembrane segment receptor linked to its intrinsic guanylyl cyclase (GC) (EC 220.127.116.11) activity in the intracellular domain (Fig. 2). Binding of ANP or BNP stimulates GC activity and elevates intracellular levels of cGMP, which in turn elicits physiological responses through cGMP-regulated ion channels, protein kinases, phosphodiesterases, and possibly other effector proteins. CNP activities are mediated by natriuretic peptide receptor-B, which has a molecular topology similar to that of NPRA.
NPRA exists as a homodimer of a single-span transmembrane polypeptide, which contains a ligand-binding extracellular domain (ECD), a transmembrane domain, and an intracellular domain (ICD) consisting of a protein kinase-like domain (PKLD) and a GC catalytic domain (GCD) (Fig. 2) . The ECD contains a highly conserved chloride-binding site near the ECD dimerization interface [17,18]. The ECD also contains in its juxtamembrane region a highly conserved structural motif, referred to as the receptor-GC signaling motif . Single-residue mutations in this motif either render the receptor unresponsive to ligand binding or cause constitutive activation of GC activity , suggesting that this conserved structure plays a critical role in transmembrane signal transduction. ATP is thought to bind to the PKLD and augment GC stimulation by ANP [21,22]. The PKLD is phosphorylated [23,24], and its dephosphorylation leads to receptor desensitization [24,25].
In the NPRA genes in human  and in rat , exons 1–6, 8–15 and 17–22 (approximately) encode the ECD, PKLD and GCD, respectively. The intervening sequences, the transmembrane sequence and a linker region between the PKLD and the GCD are encoded by exons 7 and 16, respectively.
ANP receptor ECD – biochemical and biophysical properties
We expressed the ECD of rat NPRA in mammalian cells (COS cells and CHO cells), and purified it by ANP affinity chromatography . The purified ECD bound ANP with an affinity (Kd ∼ 1 nm) comparable to that of the full-length NPRA purified previously from bovine adrenal membranes . The ECD contains three disulfide bonds, Cys60–Cys86, Cys164–Cys213 and Cys423–Cys432, in a 1–2, 3–4 and 5–6 linkage pattern (Fig. 3) . Of these, the disulfide bond Cys60–Cys86 occurs in the chloride-binding site (see below), and the disulfide bond Cys423–Cys432 occurs in the juxtamembrane receptor-GC motif. Both disulfide bonds are conserved among the A-type and B-type natriuretic peptide receptors. The ECD also contains five N-linked oligosaccharides . Correct glycosylation is essential for expression of functional NPRA: deletion of any one of the five glycosylation sites by mutagenesis reduces or abolishes NPRA expression . On the other hand, deglycosylation of the native or expressed ECD with endoglycosidase F2 or H has no effect on ANP binding . Together, these findings suggest that glycosylation is essential for folding or transport of the nascent receptor polypeptide to the cell membrane, but that, once the active receptor is formed, the glycosyl moieties are not involved in ANP binding. This notion is consistent with the crystal structure of the ANP–ECD complex , which shows that glycosyl moieties or the glycosylation sites are located away from the ANP-binding site.
By sedimentation equilibrium analyses, we found that the ECD undergoes dimerization with a dissociation constant Kd of ∼ 500 nm. In the presence of ANP, ECD dimerization was strongly enhanced (Kd ∼ 10 nm) .
Crystal structures of the ECD with and without ANP
We have determined the crystal structures of the apo-ECD dimer and the ANP–ECD complex (Fig. 4A,B) [17,34]. Each ECD monomer has the membrane-distal and membrane-proximal subdomains connected by three stretches of the polypeptide backbone. The apo-ECD occurs as a homodimer associated via the membrane-distal subdomain [35,36]. In the ANP-bound complex, two ECD monomers bind one ANP molecule, forming a 2 : 1 complex (Fig. 4B) . The structure reveals detailed ANP binding interactions (Fig. 4C) that include the following: (a) Arg14 of ANP hydrogen bonds with Glu119 of ECD monomer A (Glu119A) and Asp62 of ECD monomer B (Asp62B). Arg95A and Asp62B are also hydrogen bonded, contributing to the stability of the complex; (b) Phe8 of ANP makes a hydrophobic contact with a hydrophobic pocket formed by Tyr154A, Phe165A, Val168A, and Tyr172A; and (c) the C-terminal peptide backbone of ANP (Asn24-Ser25) forms a short parallel β-sheet with the receptor protein backbone (Glu187B-Phe188B). These binding interactions identified in the ANP–ECD crystal structure are consistent with the structure–activity relationship data reported for ANP .
Chloride-mediated control of NPRA
A protein-bound chloride atom occurs near the dimer interface (Fig. 4D) . This chloride is reversibly bound , consistent with the finding that ANP binding to the receptor requires chloride and is chloride concentration-dependent . We have proposed that chloride may allosterically regulate NPRA in the kidney and control ANP-induced natriuresis.
The natriuretic activity of ANP has been well documented experimentally. However, the physiological role of ANP as a natriuretic hormone continues to be debated, because there are certain physiological and pathological conditions in which salt is retained despite elevated plasma ANP levels [38–40]. For example, in normovolemic animals, infusion of high-dose ANP does not cause a corresponding increase in natriuresis . In edematous diseases such as congestive heart failure, nephrotic syndrome, and hepatic cirrhosis, plasma levels of ANP are markedly elevated, but sodium is retained [42–44]. In ANP-overexpressing transgenic animals, plasma ANP levels are markedly elevated, but salt is retained [45,46]. Insensitivity to the natriuretic effects of ANP is also observed in salt-depleted rats; this occurs independent of the RAA and sympathetic systems, and is not caused by receptor downregulation . Although it is beyond the scope of this review to analyze individual cases, there is apparently a common mechanism that can independently switch off natriuresis irrespective of the presence of ANP, and sodium retention prevails over ANP natriuresis in situations where the RAA system is activated either as a normal physiological response or as a compensatory (often aggravating) response in disease states such as heart failure. We speculate that chloride control of NPRA occurs in the kidney on the luminal surface of the collecting duct, and that this mechanism switches off NPRA (and hence prevents ANP natriuresis) in response to a reduced luminal chloride concentration.
In vitro, ANP binding to NPRA is chloride-sensitive over a chloride concentration range 0.1–10 mm [18,37]. When volume depletion activates the RAA system, for example, the luminal chloride concentration at the collecting duct may decrease to < 1 mm [48–50]. At such low chloride levels, the receptor is unable to bind ANP, blocking natriuresis. Undoubtedly, for this mechanism to operate, both NPRA and the ligand ANP must be localized on the luminal side of the renal tubule (but not on the basolateral side, where the chloride concentration is stable at 90–110 mm). There is a plethora of evidence in support of this view.
Both NPRA and other natriuretic peptide receptors are expressed along the nephron tubule. The subcellular localization (or polarization) of the natriuretic peptide receptors has been studied by immunofluorescence staining with antibodies against receptors. Although the results are not in complete agreement, NPRA is found predominantly on the apical (or luminal) membrane of the medullary collecting duct cells , where it is proposed to regulate sodium transport [52,53]. On the other hand, natriuretic peptide receptor-B is localized on the apical membrane of intercalated cells, where it is thought to interact with CNP and regulate acid–base homeostasis .
The presence of natriuretic peptides in the urine (and hence the luminal fluid) has long been known [55–60]. Urodilatin, originally discovered and isolated from human urine, is a 32-residue natriuretic peptide derived from the common ANP precursor polypeptide being differently processed in the kidney . It is synthesized in the tubular cells, is luminally secreted, and regulates tubular sodium transport by binding to the luminal surface NPRA. It has been proposed that urodilatin, rather than ANP (of cardiac origin), is mainly responsible for natriuretic and diuretic regulation . Similar tubular synthesis and urinary excretion of CNP has also been reported .
In addition to urodilatin, ANP and other natriuretic peptides of cardiac origin may also be present in the luminal fluid and contribute to the regulation. It is well established that small proteins and peptides (with molecular masses of less than ∼ 20 000 Da) efficiently filter through the glomerulus into the tubular lumen [62,63]. Small proteins are reabsorbed mainly by endocytosis, and are intracellularly hydrolyzed. Small peptides are hydrolyzed by proteases on the luminal brush border membrane of the proximal tubule , and the metabolites are rapidly absorbed. Because of these activities, it is often assumed that no peptide reaches the distal site of the nephron. It is necessary to note, however, that certain peptides are resistant to hydrolysis, and enter the urine intact. Studies involving microperfusion of radiolabeled peptides into the nephron (either the surface nephron in vivo or the isolated nephron in vitro) have shown that vasopressin, oxytocin, and insulin, all containing disulfide bridge(s), are not hydrolyzed at the luminal brush border, and are recovered intact in urine or the collecting fluid, whereas small linear peptides, such as angiotensins I and II, bradykinin, glucagon, and luteinizing hormone-releasing hormone, are hydrolyzed and recovered as amino acids or small fragments [65–67]. To our knowledge, no similar study has yet been reported for ANP or other natriuretic peptides. Natriuretic peptides may be similarly resistant to brush border hydrolysis (in vivo) and reach the distal tubule intact, at least fractionally. Consistent with this view, ANP, BNP and CNP are known to be excreted in the urine [55–60], and their levels are higher in heart failure [68,69], apparently corresponding to their elevated plasma levels. Indeed, urinary natriuretic peptides, especially BNP  and N-terminal BNP , have been proposed as noninvasive diagnostic and prognostic biomarkers for heart failure.
ANP inhibits sodium reabsorption (thus stimulating natriuresis) via the second messenger cGMP by inhibiting the cGMP-sensitive and amiloride-sensitive cation channels on the luminal membrane of collecting duct cells (by direct inhibition of the channels by cGMP and by suppression via cGMP-dependent protein kinase and Gi)  and by inhibiting Na+/K+-ATPase on the basolateral membrane . The former is believed to function in rapid and direct control of sodium transport, whereas the latter functions in slower and longer-term regulation. It has been pointed out that the affinity of cGMP-regulated channels for cGMP is weak, with Kd values of 20 μm or greater, whereas the cGMP levels in most cells are below 100 nm . It is likely, then, that the channels on the luminal membrane are inhibited by local elevation of cGMP by activation of NPRA by luminal natriuretic peptides also on the luminal membrane.
Taken together, these findings show that both NPRA and natriuretic peptides (ANP, BNP, CNP, and urodilatin) are present on the luminal side of the collecting duct, where the final and rate-limiting regulation of sodium reabsorption occurs. The ANP–NPRA regulatory mechanism may then be governed by the change in the chloride concentration in the lumen.
The kidney filters some 60 times the plasma volume or more than 10 times the total extracellular fluid volume per day and, consequently, almost all of the filtered salts and water must be returned to the circulation . Preventing excessive salt loss in the process is essential for survival. It is conceivable, then, that there is a mechanism enabling sodium reabsorption to override natriuretic stimuli when necessary. Deactivation of NPRA at low luminal chloride concentrations (which change in parallel with sodium concentrations) allows for sodium reabsorption even in the presence of high natriuretic peptide levels.
It needs to be acknowledged that, at present, the direct evidence for the proposed chloride control of the ANP–NPRA system is limited to the observation of the chloride effects on ANP binding and cGMP production in vitro [18,37] and the conserved chloride-binding site identified in the X-ray structures [17,18]. However, it is worth noting that the data in the literature are consistent with and strongly suggest the proposed mechanism operating in the kidney. This control mechanism may account for the renal insensitivity to ANP that has long been recognized but has been unexplained to date. Additional focused studies are necessary to determine how this control mechanism may operate in vivo and ultimately to allow the utilization of such knowledge for improved cardiovascular disease therapy.
ANP-induced structural change in the ECD
Binding of ANP to the ECD does not cause appreciable intramolecular structural changes (rmsd of the assigned 426 Cα atoms in the ECD between the ANP-bound and unbound structures, 0.64 Å) . ANP binding, however, causes a large change in the ECD dimer quaternary structure. The ECD monomers undergo a twisting motion (Fig. 5A) [17,73], which causes the two juxtamembrane domains in the dimer to translate in opposite directions. This movement alters the relative angular relationship between the two juxtamembrane domains, equivalent to rotating each domain by 24° counterclockwise (looking towards the cell membrane; Fig. 5B). We have proposed that this ligand-induced rotation mechanism in the juxtamembrane region triggers transmembrane signal transduction [17,36].
The structures of the apo-ECD dimer and the ANP–ECD complex were also observed by single-particle electron microscopy (Fig. 5C) . This method allows determination of the native structures in the absence of crystal contacts and in solution conditions closer to those of the native environment. The three-dimensional reconstructions of the apo-ECD dimer and the ANP–ECD complex revealed an ANP-induced conformational change similar to that identified from the X-ray structures. These electron microscopy data confirm that the ECD occurs as a preformed homodimer in the head-to-head configuration and undergoes a large and distinct quaternary structural change, as seen in X-ray structures, in response to ANP binding.
Rotation mechanism for transmembrane signaling by NPRA
We speculate that the ANP binding-induced rotation of the juxtamembrane domains in the dimerized receptor is transmitted across the transmembane helices and reorients the two intracellular domains into the active conformation, thereby enabling GC catalysis [17,35,36] (Fig. 6 ). This is the first example of the rotation mechanism for receptor signaling that has been structurally demonstrated.
NPRA belongs to the family of membrane-bound receptor GCs. Receptor GCs and receptor protein kinases represent two major families in the superfamily of single transmembrane segment, enzyme-linked receptors. Signaling by receptor protein kinases is thought to be mediated by agonist-induced ‘association’ mechanisms, whereas signaling by receptor GCs may be mediated by agonist-induced ‘rotation’ mechanisms.
The intracellular domain consists of the PKLD and the GCD (Fig. 2). The PKLD is thought to be the site for ATP binding. ATP is a positive allosteric effector of NPRA, which augments GC activation by ANP [21,22]. In contrast to this model, others have reported the absence of such stimulatory effects by ATP . The PKLD is phosphorylated at multiple sites [23,24]. Desensitization of NPRA in cultured cells upon extended exposure to ANP is accompanied by dephosphorylation [24,25]. The PKLD structure has been modeled on the basis of sequence homology with protein tyrosine kinases . This model has found some support from site-directed mutagenesis studies. However, the actual structure of the PKLD has not been reported. Thus, the structure and the regulatory role of this domain remain largely unsolved.
The PKLD is connected to the C-terminal GCD by a ∼50-residue linker region. Deletion mutagenesis studies have suggested that this region is necessary for receptor dimerization and GC activity . From its amino acid sequence, this region has been predicted to form an amphipathic helix in the monomer and a coiled-coil structure in the receptor dimer. On the other hand, more recent studies involving systematic site-directed mutagenesis of the guanylin receptor (or GC-C) have suggested that this region does not contain a coiled-coil structure and is not necessary for dimerization . Thus, the structure and role of this linker region remain uncertain.
Recently, two independent groups have reported the crystal structures of GCDs: a 188-residue GC catalytic core of Cyg12, a GC from a eukaryotic unicellular green alga , and a 202-residue catalytic core of Cya2, a GC from a prokaryotic cyanobacterium . These are the first structures for any GCs that have been reported, more than 10 years after the first reports of the adenylyl cyclase (AC) structures [81–83]. GCs and all known ACs belong to the class III nucleotide cyclase family, and share high sequence similarity . By amino acid sequence comparison, Cyg12 is homologous to mammalian soluble GCs, whereas Cya2 appears to be related to membrane-bound GCs.
Both Cyg12 GCD and Cya2 GCD were expressed in Escherichia coli, and without the putative linker region discussed above. Yet, both formed and crystallized as dimers. The Cyg12 GCD had a specific activity of 2.8 μmol·min−1·mg−1 in the presence of Mn2+, but much lower activity (less than 1%) in the presence of Mg2+ , as is generally observed for mammalian GCs  and ACs . The activity of the Gya2 GCD was reported to be significantly lower, at 1.5 nmol·min−1·mg−1 . However, the GC activity showed a similar metal ion dependence, exhibiting a significantly higher specific activity in the presence of Mn2+ than in the presence of Mg2+, and an even higher specific activity when both metal ions were present. In eukaryotes, manganese is a trace element, and magnesium ions are assumed to be the physiological active site ions. The enhancement of the catalytic activity by manganese ions is considered to be unlikely to have any physiological meaning . Nevertheless, the observed homodimerization of the GCD and the metal ion dependence of the catalytic activity support the integrity of the expressed proteins.
As expected from the high sequence homology with ACs, both Cyg12 GCD and Gya2 GCD monomers have the same protein fold as the mammalian AC catalytic domain. Each GCD monomer contains a seven-stranded β-sheet surrounded by several α-helices. In the dimer, two GCD monomers are related by a two-fold symmetry axis that runs through the central dimer cleft and form a wreath-like structure (Fig. 7A, Cyg12 GCD dimer). The central cavity between the two monomers contains two symmetric active sites. The catalytic residues in each active site are supplied jointly by both monomers. The active site residues in each monomer are located at positions homologous to their counterparts in ACs. Such conserved active site residues include two metal-binding aspartic acids, a ribose-orienting asparagine, a transition state-stabilizing arginine, and triphosphate-binding arginine and lysines . The guanine base-recognition residues glutamic acid and cysteine in Cyg12  and glutamic acid and glycine in Cya2  are similarly conserved at the positions close to the locations of the adenine base-recognizing lysine and aspartic acid in ACs .
The AC catalytic core consists of a C1 and C2 subdomain heterodimer. In the AC catalytic core, C1 and C2 domains, related by a pseudo-two-fold symmetry, form a heterodimeric wreath-like structure. By structural comparison, the dimer structure of Cyg12 GCD (Fig. 7A) is similar to the open, inactive conformation of the AC catalytic domain, which must close to be catalytically active . On the other hand, the Cya2 GCD dimer is in a closed conformation that must open in order to bind the substrate GTP for catalysis . This closed structure of the Cya2 GCD dimer may explain its low specific GC activity.
Interestingly, the specific activity of the Cyg12 GCD, at 2.8 μmol·min−1·mg−1, is roughly comparable to those observed for the full-length receptor GCs purified from various tissues and species, which range from 1.8 μmol·min−1·mg−1 to 23 μmol·min−1·mg−1 [29,88–90]. Together, these data seem to suggest that the structure of the Cyg12 GCD dimer may reflect the structure of the GCD in the dimerized full-length NPRA in its basal state.
Possible mechanisms for GC activation and NPRA signaling
Signaling by G-protein-coupled receptors may involve stimulation of AC by Gsα, which is released from the heterotrimeric G-protein upon receptor activation by a ligand. A possible mechanism for this AC activation by Gsα has been proposed, based on the crystal structures of the AC catalytic domains [81–83]. In the proposed mechanism, Gsα binds to the C2 domain of the AC C1–C2 heterodimer. This binding causes a 7° rotation of the C1 domain around an axis that runs through the C1 domain and roughly parallel to the central cleft axis. This movement brings the catalytic residues on the C1 domain closer to the catalytic residues on the C2 domain, thereby forming the catalytically competent active site .
In the Cyg12 GCD crystal structure, two GCD monomers are reported to be in an inactive, open conformation (Fig. 7A) . It has been suggested that activation of the Cyg12 GCD may be mediated by a domain rotation similar to the AC C1 domain rotation induced by Gsα binding to C2. The Cyg12 GCD monomer contains a surface groove similar to the groove on the AC C2 domain to which Gsα binds. In the dimerized Cyg12 GCD, binding of certain regulatory elements, similar to the H-NOX sensor domain in soluble GC, to this groove may cause a domain rearrangement or rotation in the GCD monomers, leading to stimulation of GC activity (Fig. 7B) [79,80].
NPRA and other receptor GCs exist as homodimers. Their GCDs are similarly expected to form homodimer structures. We speculate that the ANP-induced rotation of the two juxtamembrane domains in the ECD [17,36] may be transduced across the transmembrane helices and through the PKLD, causing a rotation of each of the two GCDs [17,36] (Figs 5A and 6). This rotation may bring the GCD dimer into a closed and active conformation (modeled in Fig. 7C), thereby enabling GC catalysis. In this signaling process, the PKLD may play a regulatory role by binding to the allosteric effector ATP or by its phosphorylation state. The actual and detailed mechanism of GC activation by ANP, namely the signal transduction mechanism, must await determination of NPRA’s GCD structure and ultimately the structure of full-length NPRA with and without bound ANP.
This work was supported by grants HL54329 from the National Institutes of Health and 09GRNT2250064 from the American Heart Association to K. S. Misono, and grants from the Canadian Institutes for Health Research, the Canada Foundation for Innovation and the Alberta Science and Research Investments Program to H. S. Young. H. S. Young is a Senior Scholar of the Alberta Heritage Foundation for Medical Research. We thank X. Zhang for able technical assistance.