Activation of Protein Kinase C Stimulates the Dephosphorylation of Natriuretic Peptide Receptor-B at a Single Serine Residue

The binding of atrial natriuretic peptide and C-type natriuretic peptide (CNP) to the guanylyl cyclase-linked natriuretic peptide receptors A and B (NPR-A and -B), respectively, stimulates increases in intracellular cGMP concentrations. The vasoactive peptides vasopressin, angiotensin II, and endothelin inhibit natriuretic peptide-dependent cGMP elevations by activating protein kinase C (PKC). Recently, we identified sixin vivo phosphorylation sites for NPR-A and five sites for NPR-B and demonstrated that the phosphorylation of these sites is required for ligand-dependent receptor activation. Here, we show that phorbol 12-myristate 13-acetate, a direct activator of PKC, causes the dephosphorylation and desensitization of NPR-B. In contrast to the CNP-dependent desensitization process, which results in coordinate dephosphorylation of all five sites in the receptor, phorbol 12-myristate 13-acetate treatment causes the dephosphorylation of only one site, which we have identified as Ser523. The conversion of this residue to alanine or glutamate did not reduce the amount of mature receptor protein as indicated by detergent-dependent guanylyl cyclase activities or Western blot analysis but completely blocked the ability of PKC to induce the dephosphorylation and desensitization of NPR-B. Thus, in contrast to previous reports suggesting that PKC directly phosphorylates and inhibits guanylyl cyclase-linked natriuretic peptide receptors, we show that PKC-dependent dephosphorylation of NPR-B at Ser523 provides a possible molecular explanation for how pressor hormones inhibit CNP signaling.

The natriuretic peptide family consists of atrial natriuretic peptide (ANP), 1 brain natriuretic peptide (BNP), C-type natriuretic peptide (CNP), and urodilatin, a variant of ANP that contains four additional amino-terminal residues and is found primarily in the kidney (1,2). ANP and BNP activate the natriuretic peptide receptor-A (NPR-A, also known as guanylyl cyclase A (GC-A)), and CNP activates the natriuretic peptide receptor B (NPR-B, also known as guanylyl cyclase B (GC-B)) (3,4). Once in the circulation, ANP and BNP migrate to their target tissues such as the kidney, adrenal gland, and the peripheral vasculature. Natriuretic peptide binding to NPR-A present in these tissues stimulates the elevation of intracellular cGMP concentrations, which result in renal sodium and water excretion, decreased aldosterone secretion, and vascular smooth muscle relaxation (5)(6)(7). The combined effect of these processes is decreased blood pressure. Recent studies have begun to paint a broader picture of cardiac natriuretic peptide function because ANP has been shown to inhibit cell proliferation in culture (8), and male NPR-A "knockout" mice die before 6 months of age because of apparent cardiac hypertrophy (9). Whether the hypertrophy is a direct effect of the loss of the antiproliferative actions of NPR-A, an indirect result of the increased cardiac load in these animals, or both is not known.
The biological role of CNP is less apparent. Although CNP was first isolated from porcine brain, it is now known to be present in extraneural tissues, including uterus, trachea, seminal plasma, endothelial cells, and chondrocytes (10 -13). The physiological function of CNP in many of these tissues is not known, but recent data are providing clues as to its function in blood vessels and bone. With respect to the former, it has been proposed that CNP and NPR-B comprise a vascular natriuretic peptide paracrine system with ligand and receptor being expressed in endothelial and smooth muscle cells, respectively (14). According to this hypothesis, endothelial cells, which are directly exposed to blood, can respond to circulating cytokines such as transforming growth factor-␤ and tumor necrosis factor-␣ by dramatically increasing the synthesis of CNP. The endothelially produced CNP can then communicate in a paracrine manner with the adjacent smooth muscle cells by stimulating NPR-B to exert its vasorelaxant and growth inhibitory actions. This pathway may be an ideal pharmacologic target because CNP has been shown to inhibit intimal thickening in response to certain types of vascular injury, such as coronary angioplasty (15,16).
Another exciting role for CNP is as a regulator of long bone growth. Studies using chondrocytes, osteoclast-containing bone marrow cultures, or osteoblasts have shown that natriuretic peptides can regulate the proliferation and differentiation of these cells (13,17,18). Furthermore, transgenic overexpression of BNP results in skeletal overgrowth in mice (19), and CNP but not ANP increases the height of the proliferative and hypertrophic chondrocyte zones in cultured fetal mouse tibias (20). Consistent with these findings are the observations that mice lacking NPR-C also display skeletal overgrowth (21), and mice lacking type II cGMP-dependent protein kinase, which is one step downstream in the CNP signal transduction pathway, exhibit dwarfism (22). Based on these recent data, it is now clear that the CNP/NPR-B system plays an important regulatory role in bone as well as brain and vascular tissue.
Studies on the regulation of NPR-A and -B have revealed both similarities and differences with other cell surface receptor systems. The basic topology of NPR-A and -B is similar to many growth factor receptors. It consists of an extracellular ligand binding domain, a single membrane spanning domain, a region with significant similarity to known protein kinases called the kinase homology domain (KHD), and a carboxylterminal guanylyl cyclase catalytic domain (23). Unlike most growth factor receptors, NPR-A exists as a higher order homomeric structure in the absence of ligand, and ANP binding does not lead to further aggregation (24 -26). NPR-B is also an oligomer in the absence of ligand (27), and it has been shown to heterodimerize with NPR-A when both receptors are overexpressed in cultured cells (25). In addition to natriuretic peptides, ATP also is required for the activation of NPR-A and -B (28 -31). The effect of ATP has been suggested to be mediated allosterically by the KHD, because nonhydrolyzable adenine nucleotide analogs can substitute for ATP, and deletion mutants lacking the KHD are constitutively active and are not further stimulated by hormone (32). However, recent data indicate that ATP also is required to keep the KHD phosphorylated (see below).
Both NPR-A and -B are constitutively phosphorylated when expressed in tissue culture cells (33)(34)(35)(36), and receptor phosphorylation is essential for hormonal activation (37,38). NPR-A contains four serine and two threonine phosphorylation sites (Ser 497 , Thr 500 , Ser 502 , Ser 506 , Ser 510 , and Thr 513 ) located within the KHD, and NPR-B contains at least five sites (Ser 513 , Thr 516 , Ser 518 , Ser 523 , and Ser 526 ) located in similar positions (37,38). The coordinated dephosphorylation of these sites in response to hormone binding has been shown to correlate with the desensitization of these receptors in whole cells (33,35,36), and the serine/threonine phosphatase inhibitors microcystin and okadaic acid have been shown to increase both the phosphorylation state and the ANP-dependent activity of NPR-A in crude membrane preparations (39). The receptor itself is the target of the phosphatase present in these membrane preparations because a mutant form of NPR-A that cannot be dephosphorylated is resistant to the effects of microcystin and desensitizes more slowly than the wild type receptor (56).
The pressor hormones arginine vasopressin, angiotensin II, and endothelin, which activate PKC through the stimulation of phospholipase C-␤ oppose the actions of natriuretic peptides (40). For example, pressor peptides cause increases in renal sodium and water retention, vasoconstriction, and cell proliferation, whereas natriuretic peptides cause renal excretion of sodium and water and vasorelaxation and inhibit cell proliferation. Furthermore, all three pressor peptides as well as gonadotropin-releasing hormone (41) have been shown to decrease both ANP-and CNP-dependent cGMP elevations in cultured cell lines (42)(43)(44)(45)(46), and these effects are mimicked by the direct PKC activator phorbol 12-myristate 13-acetate (PMA). Although various theories, such as direct receptor phosphorylation (47)(48)(49), receptor dephosphorylation (35), receptor degradation (50), and reduction in ligand-receptor affinities (51), have been put forth as molecular explanations for the PKC-dependent desensitization of NPR-A and -B, conclusive experimental tests of these hypotheses have not been forthcoming. Here, we show that PMA stimulates the dephosphorylation of NPR-B at Ser 523 and that the mutation of this single residue is sufficient to block both the PMA-dependent dephosphorylation and desensitization of this receptor.

EXPERIMENTAL PROCEDURES
Site-directed Mutagenesis and Transient Transfections-The wild type and many of the mutant NPR-B expression constructs have been previously described (38). The glutamate mutations (S523E, S526E, and 5E (S513E, T516E, S518E, S523E, S526E)) were generated on the 304-base pair EcoRV-XbaI fragment of NPR-B, which was subcloned into pBluescript II (Stratagene, San Diego, CA). The mutations were generated using the Quikchange kit from Stratagene according to the manufacturer's protocols. The mutant EcoRV-XbaI fragments were then subcloned back into the corresponding region of the expression plasmid pRK5-NPR-B (38). All indicated mutations and the absence of unwanted mutations were confirmed by automated nucleic acid sequencing. HEK 293 cells were grown to ϳ50% confluence in 10-cm dishes and then transfected by adding 1 ml of a mixture containing 5 g of the various pRK5-NPR-B constructs, 0.125 M CaCl 2 , and 25 mM BES-buffered saline, pH 6.96 -7.00. The cells were incubated overnight in a 3% CO 2 incubator, and the next day the medium was changed. 24 -48 h after the transfection mixture was added to the cells they were either metabolically labeled or harvested for membrane preparation.
Whole Cell Cyclic GMP Elevations-24 h after transfections, cells were split into 12-well dishes and incubated overnight. The next day these cells were 50 -75% confluent and were washed once with 1 ml of Dulbecco's modified Eagle's medium (DMEM). They were then incubated with 0.5 ml of DMEM containing 0.5 mM 1-methyl-3-isobutylxanthine (a phosphodiesterase inhibitor used to block cGMP degradation) and either 200 nM or no PMA for 30 min. The plates were then moved to a bench top at ambient temperature, and 55 l of 1 M CNP was added to each well (final concentration, 100 nM). The cells were incubated for 10 min, and then 0.5 ml 6% trichloroacetic acid was added to each plate to terminate the production of cGMP. The amount of cGMP contained in each well (cells and medium) was determined using a cGMP radioimmunoassay kit from DuPont according to the manufacturer's protocol.
Metabolic Labeling, Receptor Purification, and Phosphopeptide Mapping-A detailed description of the phosphopeptide mapping procedure used in this study has recently been published (53). Briefly, transfected HEK 293 cells were washed twice with phosphate-deficient DMEM and then changed to 95% phosphate-deficient DMEM, 5% dialyzed fetal bovine serum, 100 units/ml penicillin, 100 g/ml streptomycin, 0.25 g/ml amphotericin B, and 1 mCi/ml [ 32 P]orthophosphate (NEN Life Science Products) and incubated in an atmosphere of 5% CO 2 and 95% air at 37°C overnight. NPR-B was isolated from metabolically labeled cells by immunoprecipitation with rabbit polyclonal antiserum Z658 (kindly provided by David L. Garbers at the University of Texas Southwestern Medical Center), fractionated by SDS-polyacrylamide gel electrophoresis and transferred to Immobilon-P membrane as described under "Immunoblot Analysis." Labeled NPR-B adsorbed to Immobilon-P membrane was localized by autoradiography, and the band corresponding to NPR-B was then cut out and incubated with 0.5% polyvinylpyrrolidone dissolved in 0.1 M acetic acid for 30 min at 37°C. The membrane was then washed, and 200 l of a solution containing 50 mM (NH 4 ) 2 CO 3 , 5% acetonitrile, and 10 g of tosylphenylalanyl chloromethyl ketone-treated trypsin was added to each sample and incubated overnight at 37°C. The remaining buffer was removed by repeated lyophilization using a SpeedVac rotary evaporator. The phosphopeptides were dissolved in a small volume of distilled water and spotted on 100-m-thick cellulose plates (Merck). The phosphopeptides were first separated in the horizontal dimension by high voltage electrophoresis (1000 V) for 25 min in 1% ammonium carbonate, pH 8.9, and then by ascending chromatography in phosphochromatography buffer. The phosphopeptides were visualized by exposing the plates to Kodak XAR film for approximately 1 week at Ϫ80°C with one intensifying screen.
Chemical Synthesis of Phosphopeptide GSS(P)Y-A phosphopeptide with the sequence GSS(P)Y was synthesized on an Applied Biosystems 432A peptide synthesizer and purified according to the manufacturer's protocol. The mass composition of the phosphopeptide was verified by laser desorption mass spectroscopy.
Secondary Digestions-Phosphopeptide C was scraped from the plates and eluted from the free cellulose by vortexing in 200 l of water followed by spinning the mixture through an ultrafree-MC Millipore 0.22 micron filter unit (Millipore, Bedford, MA) in a microcentrifuge for 30 s. This extraction was repeated for a total of three times. The purified phosphopeptide was then dried down in a SpeedVac, resuspended in 50 l of (NH 4 ) 2 CO 3 , pH 8, and digested with 2 g of purified chymotrypsin at 37°C for 2 h. The digestion products were dried down, resuspended in 10 l of water, spotted on a cellulose plate, and electrophoretically fractionated in the presence or absence of 5 g of the synthetic peptide GSS(P)Y at pH 8.9 for 25 min at 1 kV. The radioactive phosphopeptides were visualized by autoradiography, and the synthetic peptide was visualized by staining with 0.25% ninhydrin in acetone.
Immunoblot Analysis-NPR-B was isolated as described above and electroblotted onto a polyvinylidene difluoride (Immobilon-P) membrane. The membrane was then blocked for 1 h in TBST (20 mM Tris (hydroxymethyl) aminomethane, 500 mM NaCl, and 0.05% polyoxyethylene sorbitan monolaurate, pH 7.5) containing 3% bovine serum albumin, washed three times for 5 min with TBST, and then incubated with rabbit antiserum R1215 (kindly provided by David L. Garbers at the University of Texas Southwestern Medical Center) diluted 1:500 in TBST containing 1% bovine serum albumin for 2 h at 25°C. This antiserum was raised against a synthetic peptide corresponding to the last 15 amino acids of NPR-A (33) but cross-reacts with NPR-B on an immunoblot. The membrane was washed three times for 10 min with TBST and incubated for 45 min at 25°C with protein A conjugated to horseradish peroxidase. The membrane was then washed once for 15 min and twice for 5 min in TBST. The NPR-B antibody complex was detected by chemiluminescence using the ECL Western blot Detection System from Amersham Pharmacia Biotech.
Guanylyl Cyclase Assays-All guanylyl cyclase assays were at 37°C in the presence of 25 mM PIPES, pH 7.4, 50 mM NaCl, 0.25 mM 1-methyl-3-isobutylxanthine, 0.1% bovine serum albumin, 5 mM creatine phosphate and 5-10 units/assay creatine phosphokinase (as a nucleotide regenerating system), 1 mM GTP and 0.1-0.2 Ci of [␣-32 P]GTP. 5 mM MgCl 2 , 1 mM ATP, and 1 M CNP or 1% Triton X-100 and 3 mM MnCl 2 were also included in the reaction mixtures. Assays were initiated by the addition of a solution of the above reagents to approximately 50 g of crude membrane protein in a total volume of 0.1 ml. Assays were terminated by addition of 0.5 ml of 110 mM zinc acetate. [ 32 P]cGMP was purified as follows: 0.5 ml of 110 mM sodium carbonate was added to precipitate the [ 32 P]GTP. The reactions tubes were vortexed, incubated on ice for 10 min and centrifuged at 2000 ϫ g for 10 min. at 4°C. The supernatant was added to chromatography columns (Bio-Rad model-731-1550) containing 0.5 g of (dry) neutral alumina resin (Sigma-A9003) that had been acidified with 5 ml of 1 N perchloric acid. The columns were then washed sequentially with 10 ml of 1 N perchloric acid, and 10 ml of water, and then the 32 P-cGMP was eluted into scintillation vials with 5 ml of freshly prepared 200 mM ammonium formate. [ 32 P]cGMP recovered was quantitated by the method of Cerenkov in a Beckman 3801 scintillation counter.

PMA Inhibits CNP-dependent Cyclic GMP Elevations-To
investigate whether PKC has a direct effect on NPR-B, we first identified a transient expression cell culture system that mimics the effect that PMA has on NPR-B in untransfected cells. For this purpose, we chose HEK 293 cells because they are highly transfectable and do not express detectable levels of endogenous NPR-B, although they do express low levels of NPR-A (37). The basal cGMP concentrations of cells transiently transfected with NPR-B were low (10 pmol/well), and a 30-min incubation with 200 nM PMA reduced their concentration by 50% to 5 pmol/well (Fig. 1). A 5-min stimulation with 100 nM CNP elevated the cGMP concentrations in resting cells more than 25-fold to 254 pmol/well. However, a 30-min preincubation with 200 nM PMA reduced the CNP-dependent cGMP levels in these cells to less than 10% of the control values.
Because these data are qualitatively similar to those reported for PMA effects on CNP responses in primary astrocytes (54), pituitary tumor cells (41,55), and ciliary epithelial cells (51), we concluded that the HEK 293 cells were an appropriate model for further studies.
PMA Inhibits CNP-dependent but Not Detergent-dependent NPR-B Activity-Because PMA treatment reduced basal cGMP concentrations, it was possible that PKC was decreasing CNPdependent cGMP elevations by stimulating a isobutylxanthineresistant phosphodiesterase. Therefore, we tested whether or not the diminished cGMP response was attributable to a direct effect on NPR-B catalytic activity. To this end, transfected HEK 293 cells were treated as described above except that after the 30-min PMA incubation, crude membranes were prepared and assayed for guanylyl cyclase activities in the presence either of CNP, ATP, and Mg 2ϩ or Triton X-100 and Mn 2ϩ . The former conditions estimate the physiological activity of the receptor, whereas the latter conditions are known to maximally activate NPR-B in a hormone independent manner (36). Hence, activity determined in the presence of Triton X-100 and Mn 2ϩ is an excellent indicator of the total amount of NPR-B present in any given membrane preparation. We used this measurement as an internal control for the amount of activity loss attributable to receptor degradation. As can be seen in Fig. 2, the CNP-dependent activity was reduced by more than 60% in membranes derived from PMA-treated cells, whereas the detergent-dependent activity was unaffected. These data suggested (i) that the reduced CNP-dependent guanylyl cyclase activity contributes to the diminished cGMP elevations and (ii) that the reduction must be due to a process other than receptor degradation, because the PMA treatment did not decrease the detergent-dependent guanylyl cyclase activity.
PMA Exposure Causes the Dephosphorylation of NPR-B at Ser 523 -Because we had recently shown that phosphorylation of NPR-B is required for hormone responsiveness (38) and that receptor dephosphorylation is correlated with CNP-dependent (homologous) desensitization of NPR-B (36), we asked whether dephosphorylation was also instrumental in the PMA-dependent (heterologous) desensitization process. To test this hypothesis, we isolated NPR-B from [ 32 P]orthophosphate labeled cells that had been incubated in the presence or absence of 200 nM PMA for 30 min. NPR-B purified from untreated cells was phosphorylated, and PMA treatment caused more than a 55% reduction in the phosphate content of the receptor (Fig. 3A, 32 P content). As with the reduction in enzymatic activity, the PMAdependent decrease in receptor phosphate content was not explained by reductions in receptor protein because immunoblot analysis on the same membrane used for the phosphate determination revealed no detectable differences (Fig. 3A, immunoblot). We next asked whether the loss of phosphate was a result of a complete dephosphorylation of all the phosphorylation sites on a specific subset of the NPR-B molecules, as is the case for CNP-dependent desensitization (36), or whether PMA causes the selective dephosphorylation of a specific phosphorylation site or sites. To address this question, we performed tryptic phosphopeptide mapping experiments. We found that NPR-B contained three major tryptic phosphopeptides (labeled A, B, and C in Fig. 3B) and several minor phosphopeptides when isolated from control cells. These maps were similar to those previously observed for NPR-B isolated from either the same HEK 293 cells or NIH 3T3 cells (36,38). When we isolated NPR-B from cells that had been treated with PMA, we observed that one of these phosphopeptides (B in Fig. 3B) was missing. These data indicate that PMA treatment, unlike CNP-dependent desensitization (36), results in the selective dephosphorylation of either a single site or a small subset of the total NPR-B phosphorylation sites.
We identified the major phosphorylation sites of NPR-B by the same phosphopeptide mapping technique as described above (38). Hence, we knew that the mutation of either Ser 523 or Ser 526 to alanine also resulted in the loss of phosphopeptide B (Fig. 4A). This suggested that phosphopeptide B is a diphosphorylated form of the peptide 521 GSSYGSLMTAHGK 533 where both Ser 523 and Ser 526 are the phosphorylated residues. Based on its mobility relative to phosphopeptide B, phos-phopeptide C is likely to be the monophosphorylated form of this peptide, comprised of two very closely migrating species that contain the same peptide backbone as B, but where only Ser 523 or Ser 526 is phosphorylated. Thus, the absence of phosphopeptide B in maps of NPR-B that was isolated from PMAtreated cells could be due to dephosphorylation of either Ser 523 or Ser 526 .
To determine which site is dephosphorylated in response to PKC activation, we performed chymotryptic digestions on phosphopeptide C isolated from maps of NPR-B obtained from cells treated in the presence or absence of PMA. Because chymotrypsin would be predicted to cleave only after Tyr 524 in this peptide, the products of this digestion are GSS(P)Y and GS(P)LMTAHGK, which contain Ser 523 and Ser 526 , respectively. Importantly, because the phosphoserine 526-containing peptide has a lysine, the net charge of this peptide at pH 8.9 is Ϫ1.5 compared with Ϫ2.5 for GSS(P)Y, the peptide containing Ser 523 . Thus, the phosphorylated form of the Ser 523 -containing peptide (GSS(P)Y) should be readily distinguished from the phosphoserine 526-containing peptide (GS(P)LMTAHGK), because the former would migrate much further toward the anode when electrophoresed at pH 8.9 because of its greater negative charge and smaller mass. To prove that the fastest migrating The cells were then lysed, and NPR-B was purified by immunoprecipitation, fractionation by SDSpolyacrylamide gel electrophoresis and blotting to Immobilon-P membrane. A, the amount of 32 P associated with NPR-B was visualized by autoradiography ( 32 P-content) and quantitated by Cerenkov counting the NRP-B containing membrane fragments. The amount of NPR-B protein that was present on the membrane was determined by immunoblot analysis with a primary antibody that recognizes the carboxyl terminus of NPR-B (immunoblot). The upper band corresponds to the fully glycosylated and phosphorylated receptor. The lower band corresponds to an incompletely glycosylated and unphosphorylated form of NPR-B. B, PMA treatment causes the loss of a specific NPR-B phosphopeptide. The membrane fragments containing NPR-B were digested with trypsin overnight. The tryptic peptides were dried down, redissolved in 10 l of water and applied to a thin layer phosphocellulose plate. The peptides were separated by electrophoresis in the horizontal dimension at pH 8.9. The plate was dried, and the peptides were separated in the second dimension by ascending chromatography. The phosphopeptides were visualized by exposing the plate to Kodak XAR film for 1 week at Ϫ70°C with an intensifying screen. species was GSS(P)Y, we chemically synthesized this phosphopeptide and mixed it with the chymotrypsin digestion products before electrophoretic separation. As shown in Fig. 4C, the chymotryptic digestion of phosphopeptide C isolated from control cell maps (ϪPMA) resulted in two major radioactive phosphopeptides. The peptide that migrated closest to the anode was the most abundant, and its migration pattern was identical to that of the synthetic phosphopeptide GSS(P)Y as indicated by ninhydrin staining (dashed ellipse in Fig. 4C). The digestion of phosphopeptide C from the PMA treated cells also yielded two phosphopeptides (Fig. 4C, ϩ PMA), but in this case the level of radioactivity in the peptide that migrated the closest to the anode was markedly reduced compared with the corresponding peptide that was isolated from untreated cells. In contrast, the peptide that migrated closest to the cathode (GS(P)LMTAHGK) was unaffected by the PMA treatment. These data indicate that PMA treatment causes the selective dephosphorylation of Ser 523 .
Mutation of Ser 523 Blocks the Effect of PMA-We have previously shown that Ser 523 is a major NPR-B phosphorylation site (38) and that the conversion of this residue to alanine markedly reduces hormone-dependent but not detergent-dependent guanylyl cyclase activity of NPR-B (this also is evident in Fig. 5A). Therefore, we tested whether the dephosphorylation of Ser 523 was responsible for the PMA desensitization by examining the ability of PMA to decrease the activity of the S523A mutant. We reasoned that if the desensitization was due to the dephosphorylation of this residue, then if it was already effectively dephosphorylated, as in the S523A mutant, PMA should not reduce its activity further. As predicted, the PMA treatment failed to reduce the CNP-dependent guanylyl cyclase activity of the S523A containing receptor (Fig. 5A, top panel). In fact, the activity was elevated by PMA exposure. The lack of inhibition was not due to increased NPR-B expression in the PMA-treated preparation because guanylyl cyclase activity measured in the presence of Triton X-100 was similar between treatments (Fig. 5A, middle panel). Likewise, when the activation ratio was determined by dividing the hormone-dependent activity by the detergent activity to control for differing transfection efficiencies and expression levels (Fig. 5A, bottom panel), no inhibitory effect of PMA was detected for the S523A mutant. Mutation of Ser 526 to Ala (S526A) also decreased CNPdependent activity (Fig. 5A) as shown previously (38). Surprisingly, this mutation resulted not only in a loss of the inhibitory effect of PMA, but also in a PMA-mediated increase in CNPdependent activity similarly to the S523 mutation (Fig. 5A). The reason for the PMA-dependent increases in the activities of the S523A and S526A receptors is not known, but it may be due to an increase in the phosphorylation of Ser 518 (see "Discussion").
Because the mutation of either Ser 523 or Ser 526 to alanine abolished the PMA effect, we were unable to determine which of these two sites is responsible for the desensitization of the wild type receptor. Therefore, we mutated the same amino acids to glutamate to mimic the negative charge of their corresponding phosphate groups, an approach that proved successful for NPR-A (37). We reasoned that the conversion of these residues to glutamate is more likely to result in a structure that is similar to that of the phosphorylated wild type receptor than the analogous alanine mutations. Consistent with this hypothesis the S523E and S526E mutants were more responsive to hormonal stimulation than their alanine counterparts, although they were less responsive than the wild type receptor (Fig. 5B). We predicted that the S526E mutant might be inhibited by PMA exposure similarly to the wild type receptor but that the S523E mutant should respond to PMA similarly to the S523A mutant, because both the Ser 523 mutations abolish the ability of the receptor to be dephosphorylated in response to PMA. We found that the PMA effect was reversed with the S523E containing receptor similarly to that of the S523A mutant. In contrast, the activity of the S526E mutant was inhibited by PMA exposure like wild type NPR-B. We also observed that PMA treatment had no effect on the activity of the NPR-B mutant containing glutamate residues at all five of its known phosphorylation sites (Ser 513 , Thr 516 , Ser 518 , Ser 523 , and Ser 526 ) (Fig. 5B, mutant 5E). Together, these data strongly suggest that the PMA-dependent desensitization of NPR-B is mediated by the selective dephosphorylation of Ser 523 .
We then asked whether the dephosphorylation of NPR-B at Ser 523 completely explained the dramatic PMA-induced decreases in CNP-dependent cGMP elevations in whole cells or whether other mechanisms were involved. To this end, we expressed the wild type, S523A, S523E, S526A, S526E, or 5E NPR-B constructs in HEK 293 cells, treated cells with or without PMA for 30 min, and then stimulated cGMP synthesis with CNP. We found that the cGMP levels in transfected cells essentially recapitulated the results from the guanylyl cyclase assays shown in Fig. 5. The S526E mutant was desensitized like the wild type receptor, and the mutant containing glutamate residues at all five phosphorylation sites was completely unaffected by the PMA exposure. We also observed the same paradoxical effect with the S523A, S523E, and S526A mutants that were activated by PMA treatment. These data clearly demonstrate that the PMA-dependent reductions in cGMP con- The vertical dashed line indicates where chymotrypsin is expected to cleave this peptide. C, Ser 523 but not Ser 526 is dephosphorylated in response to PMA exposure. Phosphopeptide C was extracted from tryptic phosphopeptide maps of wild type NPR-B that was isolated from labeled cells that had been incubated in the presence (ϩPMA) or absence (ϪPMA) of 200 nM PMA for 30 min. The peptide was purified from the cellulose, digested with chymotrypsin, mixed with 5 g of the synthetic peptide GSS(P)Y, and spotted on a phosphocellulose plate. These peptides were subsequently electrophoretically separated for 25 min at 1 kV at pH 8.9. Approximately 30 and 20 cpm were added to the origins (indicated by the arrows) of the ϪPMA and ϩPMA samples, respectively. The plate was exposed to Kodak XAR film for 1 month at Ϫ70°C with two intensifying screens to visualize the radioactive phosphopeptides originating from labeled cells. The migration of the synthetic peptide was determined by ninhydrin staining and is indicated by the dashed ellipse.
centrations in these cells result solely from the inhibition of NPR-B activity and allow us to rule out any potential involvement of other cGMP synthesizing or degrading pathways.

DISCUSSION
For several years now there has been a controversy in the natriuretic peptide receptor field regarding how PKC inhibits natriuretic peptide-dependent cGMP elevations. Early in vitro experiments conducted with impure preparations suggested that PKC directly phosphorylates and inhibits NPR-A (47-49). Thus, direct receptor phosphorylation was proposed as a mechanism for the heterologous desensitization of this receptor. But with the availability of immunoprecipitating antibodies, it became clear that NPR-A was highly phosphorylated in resting cells and that ANP or PMA exposure results in decreases in both receptor phosphate and hormone-dependent guanylyl cyclase activity levels (33)(34)(35). Additional experiments revealed that the in vitro dephosphorylation of NPR-A and -B with protein phosphatase 2A catalytic subunit also reduces cyclase activity (33,36). Based on these data, it was suggested that the homologous desensitization of these receptors was mediated not by phosphorylation but by dephosphorylation (33,36). However, because these experiments were carried out in membrane and not purified receptor preparations, the possibility existed that another associated protein, which was also dephosphorylated by the phosphatase treatment, was responsible for the inhibition. In an effort to answer this question more definitively, we identified the phosphorylation sites of the receptor (37,38). Consistent with the desensitization by dephosphorylation hypothesis, we found that the mutation of the phosphorylated residues to alanines results in hormonally unresponsive enzymes. However, because the alanine-substituted receptors cannot be stimulated by ligand, we could not use them to directly test the homologous desensitization process (37,38). Fortunately, when we mutated all six phosphorylation sites in NPR-A to glutamate, we found that this receptor is responsive to ANP stimulation but resistant to homologous desensitization, suggesting that dephosphorylation is necessary for homologous desensitization (56).
Heterologous desensitization in response to PMA treatment was also correlated with NPR-A dephosphorylation (32), but definitive proof that dephosphorylation of this receptor is required for desensitization is lacking. Here, we tested whether receptor dephosphorylation is involved in the heterologous desensitization of NPR-B. We show that PMA exposure causes its dephosphorylation and desensitization, but unlike the homologous process, which results in a coordinated dephosphoryl- ation of all the NPR-B phosphorylation sites, the heterologous (PKC-dependent) process results in the selective dephosphorylation of a single site. We identified this dephosphorylated residue as Ser 523 . Importantly, we tested whether the dephosphorylation of Ser 523 could explain the PMA-dependent decreases in NPR-B activity. We found that the mutation of Ser 523 to alanine or glutamate, but not the mutation of Ser 526 to glutamate, completely blocked the ability of PMA to inhibit CNPdependent guanylyl cyclase activities as well as cGMP elevations in whole cells. Based on these data, we conclude that dephosphorylation of NPR-B at Ser 523 is the mechanism for reductions in CNP-dependent cGMP elevations observed in cells treated with activators of PKC. To our knowledge this is the first and only study to identify a residue within either NPR-A or B that is either phosphorylated or dephosphorylated in a PKC-dependent manner.
One observation that we consistently made during the course of these studies was that PMA exposure results in the activation of the S523A, S523E, and S526A receptors. We detected the increase in both guanylyl cyclase assays and whole cell stimulations (Figs. 5 and 6). In additional experiments, we found that PMA treatment increased the phosphate content of the S523A and S526A mutant receptors. However, it only resulted in the loss of phosphopeptide C from maps of the S526A but not the S523A mutant receptor (data not shown). Again, these data indicate that PMA treatment causes the dephosphorylation of Ser 523 but not Ser 526 . We could not detect any new spots in the resulting tryptic phosphopeptide maps, but we did notice that the peptide containing Ser 518 (phosphopeptide A in Figs. 3 and 4) was markedly increased (data not shown). It is difficult to interpret these results because these mutants do not exist in nature. Clearly, the wild type receptor is dephosphorylated on Ser 523 in response to PMA (Figs. 3 and 4). Thus, it is logical to expect that the mutation of this residue would block both the dephosphorylation and desensitization, if in fact dephosphorylation of this residue is the mechanism for the desensitization. This is what we observed. The reversal of the PMA response in the Ser 523 mutant is unexpected and at the moment without explanation, although it could be due to an increase in the phosphorylation of Ser 518 , which lies in a reasonable PKC phosphorylation consensus sequence, i.e. Ser-hydrophobic-Arg. Regardless of the exact mechanism, we do not believe that the loss of cyclase activity and PMA-dependent inhibition is a nonspecific effect of the alanine mutation, because we observed similar effects when we mutated the same serine to glutamate to mimic a phosphorylated residue.
We also do not know why the Ser 523 or Ser 526 to glutamate NPR-B mutants are not more responsive to ligand. We have directly compared the activities of the glutamate and the alanine mutants in at least three separate experiments, and in every case, the glutamate mutants are more responsive to hormonal stimulation than the alanine mutants, but we had hoped the glutamate mutants would have had activities closer to those of the wild type receptor based upon our successful single (37) and multiple (56) glutamate substitutions for NPR-A. However, the inability of glutamate or aspartate to substitute fully for phosphorylated serine or threonine residues in a functional way is commonly observed. Indeed, in many instances these acidic acid substitutions fail to mimic the effect of the phosphorylated residue at all, presumably because the stereochemisty is sufficiently different or the single negative charge of an aspartate or glutamate does not mimic the double negative charge of phosphate. For instance, the S218E/S222E mutant of the mitogen-activated protein kinase kinase, MEK1, has only 1% of the activity of MEK1 doubly phosphorylated at Ser 218 and Ser 222 (57). Nonetheless, the S218/222E MEK1 mutant is significantly more active than unphosphorylated MEK1 and has proved extremely useful in elucidating the function of MEK1 in activation the ERK/mitogen-activated protein kinase pathway. By analogy, even though the S523E and S526E NPR-B mutants are not as active as wild type NPR-B, we believe that the activity of these mutants is sufficient to allow us to draw conclusions about the function of phosphate at Ser 523 and Ser 526 , respectively. In addition, because these mutant receptors bind CNP normally and have detergent stimulated guanylyl cyclase activities equivalent to that of wild type, it is likely that they are correctly folded, processed, and expressed on the cell surface. Why the S526E but not S526A mutant can be desensitized by PMA treatment is unknown, but one explanation is that Ser 526 needs to be phosphorylated for a phosphatase to have access to Ser 523 . Finally, the reduced activity of the S526E mutant in guanylyl cyclase assays may be a function of its instability in broken cell preparations because it is markedly more responsive in whole cells (compare Figs. 5 and 6).
Protein kinase C is involved in the desensitization of other receptor systems as well. For instance, PKC has been shown to phosphorylate Thr 654 of the human epidermal growth factor receptor, which results in decreases in its epidermal growth factor-dependent tyrosine kinase activity (58,59). Likewise, PKC has been shown to phosphorylate and inhibit the ability of the ␤-adrenergic receptor to activate adenylyl cyclase (60). Interestingly, in these situations the desensitization results from receptor phosphorylation, not dephosphorylation, as is the case for NPR-B. We are not aware of any other cell surface receptors that are specifically dephosphorylated in response to PKC activation. However, activation of PKC has been shown to decrease the phosphorylation of c-Jun at sites that negatively regulate its ability to bind DNA (61).
With this report, both ends of the heterologous desensitization pathway have now been identified. It is initiated by the binding of a pressor hormone (angiotensin II, vasopressin, or endothelin) to its cognate heptahelical receptor, and it ends with the dephosphorylation of NPR-B at Ser 523 . However, FIG. 6. Mutation of Ser 523 but not Ser 526 to glutamate blocks the ability of PMA to inhibit CNP-dependent cGMP elevations. HEK 293 cells were transfected with the wild type (W.T.), S523A, S523E, S526A, S526E, or 5E (S513E/T516E/S518E/S523E/S526E) NPR-B expression constructs and then split into 12-well dishes 24 h later. The next day the medium from these cells was aspirated and replaced with 0.5 ml of DMEM containing 0.5 mM 1-methyl-3-isobutylxanthine and either 200 nM PMA or no PMA for 30 min. The plates were then moved to a bench top at ambient temperature. 50 l of 1 M CNP was added to individual wells. The cells were incubated for 10 min, and then 0.5 ml of 6% trichloroacetic acid was added to each plate to terminate the cGMP production. The amount of cGMP contained in each well (cells and medium) was estimated by radioimmunoassay. The error bars represent the S.E. for four separate wells. No bars indicate that the S.E. was less than 3%. much work needs to be done to identify the steps in the middle. The most pressing question is how Ser 523 is dephosphorylated. One scenario is that PKC phosphorylates and activates a specific phosphatase that then dephosphorylates Ser 523 . However, it is also possible that the dephosphorylation could result from the phosphorylation and inactivation of the protein kinase that phosphorylates Ser 523 . Clearly, the identification of the enzymes that phosphorylate and dephosphorylate these guanylyl cyclase receptors will be the next major step in the dissection of this signal transduction pathway. Finally, we note that Chrisman and Garbers (62) have recently reported that CNP-dependent cGMP elevations can be inhibited by prior exposure to platelet-derived growth factor. Because platelet-derived growth factor is known to activate PKC via phospholipase C-␥, it is possible that the dephosphorylation of Ser 523 will be instrumental in this process as well.