Atrial Natriuretic Peptide Induces Natriuretic Peptide Receptor-cGMP-dependent Protein Kinase Interaction*

Circulating natriuretic peptides such as atrial natriuretic peptide (ANP) counterbalance the effects of hypertension and inhibit cardiac hypertrophy by activating cGMP-dependent protein kinase (PKG). Natriuretic peptide binding to type I receptors (NPRA and NPRB) activates their intrinsic guanylyl cyclase activity, resulting in a rapid increase in cytosolic cGMP that subsequently activates PKG. Phosphorylation of the receptor by an unknown serine/threonine kinase is required before ligand binding can activate the cyclase. While searching for downstream PKG partners using a yeast two-hybrid screen of a human heart cDNA library, we unexpectedly found an upstream association with NPRA. PKG is a serine/threonine kinase capable of phosphorylating NPRA in vitro; however, regulation of NPRA by PKG has not been previously reported. Here we show that PKG is recruited to the plasma membrane following ANP treatment, an effect that can be blocked by pharmacological inhibition of PKG activation. Furthermore, PKG participates in a ligand-dependent gain-of-function loop that significantly increases the intrinsic cyclase activity of the receptor. PKG translocation is ANP-dependent but not nitric oxide-dependent. Our results suggest that anchoring of PKG to NPRA is a key event after ligand binding that determines distal effects. As such, the NPRA-PKG association may represent a novel mechanism for compartmentation of cGMP-mediated signaling and regulation of receptor sensitivity.

Natriuretic peptide binding to type I receptors (NPRA and NPRB) on target cells activates their intrinsic guanylyl cyclase activity, resulting in a rapid increase in cGMP. Diffusible cGMP acts as a second messenger primarily by stimulating PKG (19). PKG is the major mediator of cGMP-induced smooth muscle relaxation (20). Downstream NP effects that have been directly tied to activated PKG include modulation of the L-type calcium channel (21,22) and cross-talk with heterologous receptors, such as G protein-coupled receptors (23,24). Furthermore, there is recent evidence that the membrane-bound guanylyl cyclase, NPRA, but not soluble cyclases that are activated by nitric oxide, has potent effects on plasma membrane control of the calcium ATPase pump (25), suggesting that NO-and NP-mediated effects are compartmentalized in cells.
Although the NPRA cDNA was first cloned more than 13 years ago (26), its regulation remains poorly understood. In its prebound state, the NPRA receptor exists as a homodimer (27), but ligand binding alone is insufficient to induce cyclase activity. Rather, phosphorylation of six serine and threonine residues in the intracellular juxtamembrane-kinase homology domain makes the receptor susceptible to NP activation (28). The protein kinase that mediates receptor phosphorylation is unknown.
We have previously reported that ANP inhibits cardiac hypertrophy through cGMP/PKG-mediated activation of the ERK signaling cascade at the level of MEK (9) but could not demonstrate a direct interaction between PKG and MEK. PKG substrates are membrane-bound (29), cytosolic (30), and intranuclear (31). In an attempt to identify novel proteins that could be candidates for linking PKG to MEK, we used a cytosolic yeast two-hybrid system employing PKG as bait. We found that PKG directly interacts with NPRA. These results were initially quite surprising, because PKG was previously thought to regulate only downstream ANP targets. However, it has been demonstrated previously that PKG is a serine/threonine kinase capable of phosphorylating NPRA in vitro (32). We report for the first time the regulation of NPRA by PKG.

EXPERIMENTAL PROCEDURES
Yeast Two-hybrid Studies-Screening for PKG-interacting proteins was done using a commercially available system (CytoTrap, Stratagene, La Jolla, CA) that identifies protein-protein interactions in the yeast cytoplasm. Rather than relying on transcriptional activation to detect interactions, the RAS signal transduction cascade is initiated by recruitment of hSOS to the plasma membrane in a temperature-sensitive mutant yeast strain, cdc25H, by virtue of the interaction of its bait fusion partner with a myristoylated prey protein, which allows growth at 36°C. A bovine PKG I␣ cDNA was cloned in-frame into the hSOS bait plasmid. A human heart cDNA library in the pMyr plasmid (Stratagene) containing 7.4 ϫ 10 6 independent clones was cotransfected with the pPKG-hSOS expression vector into competent cdc25H cells, which were then grown for 4 days at 25°C on minimal glucose plates. Colonies were isolated and tested for galactose-dependent growth at 36°C. Plasmids were extracted from three initially positive colonies, and the inserts were sequenced before retransformation in cdc25H cells together with pPKG-hSOS. Conventional yeast transformation and manipulation protocols were used. Cells were replica-plated onto either glucose-or galactose-minimal medium containing relevant amino acids, according to the manufacturer's protocols.
Immunocytochemistry-HEK-NPRA and HEK293 control cells were incubated in 4-well chamber slides at a density of ϳ100,000 cells/cm 2 in Dulbecco's modified Eagle's medium with 10% fetal bovine serum and 1% penicillin/streptomycin solution (Invitrogen). After fixation in 3.7% formaldehyde, cells were permeabilized with 0.3% Triton X-100 in phosphate-buffered saline for 10 min and blocked with 1% horse serum, 0.2% bovine serum albumin in phosphatebuffered saline. Fixed slides were incubated with a primary antibody mixture containing 0.1 g/ml rabbit anti-PKG I (Stressgen) or rabbit anti-PKA (Santa Cruz Biotechnology) and 0.7 g/ml mouse anti-FLAG M2 (Sigma). The immunogen used to generate the anti-PKG I antisera has no homology to PKG II making cross-reactivity extremely unlikely. Cells were blocked for 1 h and then for an additional hour with fluorescein isothiocyanate or Cy5-conjugated donkey secondary antibodies (Jackson Laboratories); cells were then mounted on coverslips using Gel/Mount mounting medium (Biomedia Corp.). Cardiac cells were treated identically, except fixed cells were incubated with a 1:1000 dilution of rhodamine-phalloidin (Molecular Probes, Eugene, OR) for 1 h. Fluorescent imaging was performed with a Leica DMRA upright fluorescent microscope, and images were acquired with a Hamamatsu ORCA2 CCD camera.
Cardiac Cell Culture-The care of all animals used in this research was in accordance with institutional guidelines. Ventricular cardiac cells from 1-2-day-old Harlan Sprague-Dawley rats were prepared as described previously (33). Ventricles were dissected free from atria and quartered. Myocytes were dissociated in trypsin and DNase I and preplated to remove non-myocyte cells. In all experiments, cells were plated on gelatin-coated chamber slides (ϳ400,000 cells/chamber) and maintained overnight at 5% CO 2 in Dulbecco's modified Eagle's medium with 17% Media 199, 10% horse serum, and 5% fetal bovine serum. Cells were incubated in 80% Dulbecco's modified Eagle's medium, 20% Media 199 (Invitrogen), and phenylephrine (1 M) 72 h before treatments.
Preparation of Plasma Membranes-Plasma membranes were made as described previously (26). HEK-NPRA and control cells were plated at equal densities and transfected with 1 g of the PKG expression vector using FuGENE (Roche Applied Science). After a 48-h incubation, cells were rinsed with ice-cold phosphate-buffered saline and scraped into 0.5 ml of buffer (20 mM HEPES, pH 7.4, 50 mM NaCl, 5 mM EDTA, 10 g/ml leupeptin, 10 g/ml aprotinin, and 1 g/ml pepstatin). Cells were then mechanically lysed by flushing repeatedly through a 22gauge needle. The particulate fraction was isolated by centrifugation at 5000 ϫ g for 15 min at 4°C, and the pellet was washed once with scraping buffer prior to experiments.

RESULTS
NPRA and PKG Associate in Yeast-To identify potential PKG partners, we screened a human heart cDNA library using a cytosolic yeast two-hybrid system (see "Experimental Procedures"). Employing PKG as bait, the single positive clone identified corresponded to an ϳ300-bp cDNA fragment of NPRA encoding the entire C-terminal cyclase domain and part of the so-called hinge region (35) (Fig. 1).
NPRA and PKG Associate in Mammalian Cells-Experiments were carried out in HEK293 cells stably transfected with an expression vector encoding FLAG epitope-tagged NPRA (27) (HEK-NPRA). HEK-NPRA cells exhibit a hormone-stimulated generation of cGMP, whereas untransfected HEK293 cells ex-press minimal endogenous NPRA (28). 2 Using FLAG antibody, it was not possible to co-immunoprecipitate levels of PKG from HEK-NPRA membranes that were detectable by Western immunoblotting, presumably because detergent solubilization disrupted the NPRA-PKG association. Attempts to release NPRA from membrane lysates while maintaining PKG association using a variety of concentrations of both ionic and nonionic detergents were unsuccessful. However, these same immunoprecipitates clearly contained low levels of PKG based upon highly sensitive PKG activity assays ( Fig. 2A). Because NPRA is a transmembrane protein and membrane localization of PKG had not been previously reported, we prepared detergent-free lysates from crude membranes of HEK-NPRA and control cells that had both been transiently transfected with the PKG expression vector. PKG protein (Fig. 2B, upper panel) and kinase activity (Fig. 3C) were detectable in the HEK-NPRA membranes but not the control membrane fraction. Treatment 2 M. Chinkers, personal communication. with ANP for 15 and 60 min, but not 5 min, demonstrated an apparent increase in membrane-associated PKG protein in the HEK-NPRA membranes only (Fig. 2B). PKG protein was present equally in cytosolic fractions of HEK-NPRA and control cells (Fig. 2B, middle panel), and NPRA protein was observed only in the HEK-NPRA membranes (Fig. 2B, lower panel). In additional experiments, ANP treatment for either 15 or 30 min resulted in a significant increase in PKG activity in crude membrane preparations of HEK-NPRA cells compared with untreated cells (Fig. 3B). Taken together, these data suggest that PKG is prebound to unliganded receptor and that ANP binding to NPRA recruits additional PKG to the cell membrane.
We next used immunofluorescence microscopy to verify the ANP-dependent co-localization of NPRA and endogenous PKG. After ANP treatment for 15, 30, and 60 min, we observed markedly increased staining of PKG at the membrane in HEK-NPRA cells (Fig. 3A) but not in control cells (data not shown). Furthermore, the PKG inhibitor KT5823 completely blocked the ANP-induced translocation of PKG to the plasma membrane (Fig. 3C), suggesting that activation of PKG is necessary for ANP-induced translocation. In parallel control experiments, cytosolic PKA immunofluorescence was readily detectable, but PKA did not migrate to the plasma membrane after ANP treatment indicating that the effect is specific for PKG (data not shown). Thus, even after prolonged exposure to the ligand for up to 60 min PKG membrane staining persisted. To determine whether the presence of NP was required to maintain membrane localization of PKG, cells were treated for 30 min, washed, and incubated in NP-free medium for different lengths of time. Within 5 min of incubation in NP-free medium, membrane staining was lost (Fig. 4). Because we initially identified the NPRA-PKG interaction by screening a heart cDNA library and to determine whether ANP-induced translocation of PKG also occurred in non-transfected primary tissue-derived cells, we examined cultured neonatal rat cardiomyocytes. In these primary cells, endogenous PKG migrated to the sarcolemma after ANP treatment (Fig. 5).
PKG Increases Receptor Cyclase Activity-To determine whether ANP-dependent recruitment of PKG to the plasma membrane regulates NPRA function, we measured the effect of PKG on ANP-induced guanylyl cyclase activity (Fig. 6). To eliminate the possibility that receptor function was down-regulated as a result of receptor internalization, increased cGMP phosphodiesterase activity, or phosphatase action, studies were performed using isolated membranes prepared from HEK-NPRA cells using an array of phosphatase inhibitors. Furthermore, to increase the specificity of the reaction, the PKA inhibitor PKI was added to the lysates. In these in vitro experiments, PKG treatment alone had no effect on guanylyl cyclase activity whereas ANP increased cyclase activity by ϳ70%. The addition of recombinant PKG to the ANP treatments, however, increased cyclase activity by ϳ150%. Thus, the addition of PKG increased the modest effect of ANP alone by 2-fold.
To determine whether PKG translocation is a specific effect of ANP-induced NPRA activation or a general consequence of raising cGMP levels in the cell, we compared the effect of the nitric oxide donor S-nitroso-N-acetylpenicillamine (SNP) to ANP treatment. Using fluorescence immunohistochemistry, we observed robust PKG translocation after ANP treatment. However, SNP treatment had no effect on the translocation of endogenous cytosolic PKG (Fig. 7A). PKG translocation was not observed in control HEK cells with either SNP or ANP treat- ment (data not shown). To confirm that HEK-NPRA cells, in fact, contained activable soluble guanylyl cyclase, we measured cyclase activity in ANP-or SNP-treated cells. Soluble guanylyl cyclase activity was more than 2-fold higher in the SNP-treated cells compared with ANP-treated cells (Fig. 7B). DISCUSSION We have demonstrated, both in yeast and in mammalian cells, a novel association of NPRA and PKG. Although the original finding in yeast was unexpected, the notion that PKG interacts with NPRA was compelling because PKG was already well established as a downstream component of the ANP-NPRA signal. The initial co-immunoprecipitation experiments did not demonstrate this interaction because the solubilization of membrane-bound NPRA with either ionic or non-ionic detergents presumably also reduced PKG binding. Nevertheless, highly sensitive PKG in vitro kinase activity assays of these same NPRA immunoprecipitates demon- Images are from a single experiment and are representative of three independent identical experiments. B, SNP-induced guanylyl cyclase activity is greater than ANP-induced activity in HEK-NPRA cells. Cells were treated as above, and guanylyl cyclase activity was measured in the cytosolic fractions (as described in the legend to Fig. 6, except that cGMP was omitted from the reactions). Data are expressed as the percent increase from the counts obtained from the control (mean Ϯ S.D. of two independent experiments, each performed in triplicate; *, p Ͻ 0.001). strated that they contained PKG activity. Additionally, we have shown that PKG is found in membrane extracts from cells that had been stably transfected with an NPRA expression vector but not from non-expressing control cells. Finally, ANP-dependent co-localization of NPRA and PKG at the plasma membrane was subsequently corroborated by immunofluorescence microscopy.
We have also demonstrated ligand-dependent "feed-forward" regulation of NPRA by PKG. The requirement for an NPRA kinase was first suggested by Potter and Hunter (28), who demonstrated that phosphorylation of serine/threonine residues in the NPRA kinase-like domain was essential for receptor function (18). Our studies suggest that prebound PKG is the NPRA kinase and that ANP binding is necessary and sufficient for both recruitment to and maintenance of membrane-bound PKG. Our findings may also be of relevance to brain natriuretic peptide because it is nearly identical in structure to ANP, and both ANP and brain natriuretic peptide preferentially bind with high affinity to NPRA (36). Finally, this ligand-dependent process appears to regulate the intrinsic guanylyl cyclase activity of NPRA. Thus, NPRA-PKG interaction may play a role in determining receptor sensitivity.
We have demonstrated that ANP, but not the NO donor SNP, induces PKG translocation. These data suggest that PKG membrane recruitment is an effect of NPRA activation and not a general consequence of raising cGMP levels. Thus, soluble and particulate guanylyl cyclase appear to be compartmentalized with respect to PKG migration. How this observation relates to the regulation of diverse downstream PKG actions will require additional study. It is well established that both nitric oxide and NPs signal through cGMP generated by soluble and membrane-bound guanylyl cyclases, respectively. Most of the downstream actions of GMP (37) occur through its binding and subsequent activation of PKG, but cGMP also directly regulates ion channels (38 -41) and phosphodiesterases (42). There is recent evidence that particulate, but not soluble, cyclases have potent effects on plasma membrane control of calcium homeostasis (25). Indeed, whereas it is generally believed that such spatially and temporally distinct cGMP signals can coexist within cells, a precise mechanism for differentiating NP-versus NO-specified cellular events has never been proposed. Our data support the speculation that NP binding recruits PKG to a membrane pool that may segregate it from soluble guanylyl cyclase-generated cGMP and may simultaneously promote PKG activation via receptor cyclase-generated cGMP through NP binding. The existence of locally increased cGMP at the plasma membrane or elsewhere may explain early observations that despite extremely low cGMP concentrations in whole-cell lysates, cGMP-targeted ligands have potent and important biological effects.
Manipulation of the NPRA-PKG interaction may favor enhanced ANP signaling in disease states. Although ANP is a potent vasodilator and diuretic in normal subjects, its beneficial effects are blunted in heart failure (15), despite increased circulating ANP levels (43). Therefore, ANP-dependent receptor desensitization (16,17,44) may limit the therapeutic utility of natriuretic peptide-based therapies. In fact, homologous desensitization of NPRA may be triggered specifically by a reduction in NPRA kinase activity (45). Although our results indicate that the recruitment of PKG to the plasma membrane amplifies cyclase activity, further studies will be required to define the critical sequence or motif of the NPRA receptor responsible for PKG translocation and to determine whether exclusion of PKG from its NPRA-binding sites results in desensitization.