ATP-independent activation of natriuretic peptide receptors.

Natriuretic peptide receptor A (NPR-A) is an essential cardiovascular regulator that is stimulated by atrial natriuretic peptide and B-type natriuretic peptide, whereas natriuretic peptide receptor B (NPR-B) stimulates long bone growth in a C-type natriuretic peptide-dependent manner. Many reports indicate that ATP is essential for NPR-A and NPR-B activation. Current models suggest that natriuretic peptide binding to receptor extracellular domains causes ATP binding to intracellular kinase homology domains, which derepresses adjacent catalytic domains. Here, we report 100-fold activations of natriuretic peptide receptors in the absence of ATP. The addition of a nonhydrolyzable ATP analog had no effect at early time periods (measured in seconds) but increased cGMP production about 2-fold after longer incubations (measured in minutes), consistent with a stabilization, not activation, mechanism. These data indicate that ATP does not activate natriuretic peptide receptors as has been repeatedly reported. Instead, ATP increases activity primarily by maintaining proper receptor phosphorylation status but also serves a previously unappreciated enzyme stabilizing function.

NPR-A, NPR-B, and StaR. A single membrane-spanning region joins the extracellular and intracellular portions of these enzymes. The latter portion consists of kinase homology, coiledcoiled dimerization, and carboxyl-terminal guanylyl cyclase domains (10).
Activation of NPR-A requires natriuretic peptide binding to its extracellular domain at a stoichiometry of 1:2 (11), which results in a "tightening" of its juxtamembrane region (12). ATP was first reported to inhibit ANP binding (13) but shortly thereafter was shown to increase ANP-dependent guanylyl cyclase activity (14 -17). Subsequent reports indicated an absolute requirement for ATP in the activation of NPR-A (18 -20) and NPR-B (21). In analogy to GTP-dependent activation of adenylyl cyclase, these reports led to an ATP-dependent activation model for guanylyl cyclase-linked natriuretic peptide receptors. Because adenine analogs like AMPPNP that presumably cannot serve as substrates for protein kinases partially mimic the effect of ATP, it was suggested that ATP allosterically regulates NPR-A (15). Consistent with this idea, cDNA cloning revealed the presence of a motif within the kinase homology domain of NPR-A that resembles the canonical ATP-binding glycine elbow found in many protein kinases (22,23). Because the mutation of this motif resulted in diminished hormonal responsiveness of NPR-A and NPR-B, it was termed the ATP regulatory module (24,25). Together, these data led to a two-step model in which natriuretic peptidebinding facilitates direct ATP binding to the kinase homology domains of NPR-A and NPR-B, which ultimately causes the formation of two active sites per catalytic dimer (10). This description represents the model for natriuretic peptide receptor activation that is most frequently cited in reviews of the field (10, 26 -29).
More recent studies suggest that the ATP-dependent regulation of the kinase homology domains of NPR-A and NPR-B also involves changes in their phosphorylation state, a process that is required for natriuretic peptide receptor activation (30,31). For instance, mutations that disrupt the putative ATP regulator module in NPR-B reduce the phosphate content of the receptor (30). Additionally, ATP␥S sensitizes NPR-A to subsequent activation by ANP and AMPPNP, indicating that in broken cell assays ATP serves as a substrate for the protein kinase that phosphorylates NPR-A (32). However, even in studies in which ATP was implicated in receptor phosphorylation, an adenine nucleotide was still required for enzyme activation, presumably because ATP binding to the kinase homology domain is essential for activation (32).
Here, we show that when membranes are prepared in the presence of phosphatase inhibitors and assayed for short time periods, NPR-A and NPR-B are activated 50 -200-fold in the absence of ATP. Surprisingly, the addition of ATP had no effect on activity at early time periods. However, upon prolonged exposure, it increased hormone-dependent guanylyl cyclase activity by a mechanism involving enzyme stabilization, not ac-* This work was supported in part by National Institutes of Health Grant RO1HL66397 and Scientist Development Award 0130398 from the National Division of the American Heart Association (to L. R. P.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
§ Supported by National Institutes of Health Training Grant AR07612.
ʈ To whom correspondence should be addressed: 6-155 Jackson, 321 Church St. S. E., Minneapolis, MN 55455. Tel.: 612-624-7251; Fax: 612-624-7282; E-mail: potter@umn.edu. 1 The abbreviations used are: NPR, natriuretic peptide receptor; AMPPNP, 5Ј-adenylyl imidodiphosphate; ANP, atrial natriuretic peptide; CNP, C-type natriuretic peptide; GC, guanylyl cyclase; ATP␥S, guanosine 5Ј-O-(thiotriphosphate). tivation. For the first time, these data demonstrate that ATP is not an allosteric activator of NPR-A and NPR-B. Instead, they indicate that the primary effect of ATP is to keep NPR-A and NPR-B phosphorylated and that a secondary effect is to stabilize the cyclase activity of these receptors. Together, these data support a new single-step model for natriuretic peptide receptor activation.

EXPERIMENTAL PROCEDURES
Membrane Preparation-HEK293T or NIH3T3 cells stably expressing NPR-A (293T-NPR-A or 3T3-NPR-A) or NPR-B (293T-NPR-B) were grown and maintained as described previously (33,34). Crude membranes were prepared by washing cells at 4°C with phosphate-buffered saline and scraping them off the plate in the presence of either phosphatase inhibitor buffer (25 mM Hepes, 20% glycerol, 50 mM NaCl, 50 mM NaF, 2 mM EDTA, 0.25 M microcystin, and 8.6 mg/10 ml Sigma protease inhibitor mixture (P8465)) or homogenizing buffer (50 mM Hepes, 10% glycerol, 50 mM NaCl, 10 g/ml aprotinin, 10 g/ml leupeptin, and 1 g/ml pepstatin A). Suspended cells were sonicated for 1-2 s and centrifuged at 20,000 ϫ g for 10 min at 4°C. The supernatant was aspirated, and the pellet was resuspended in either phosphatase inhibitor buffer or homogenizing buffer at a protein concentration of 5-10 mg/ml. For experiments shown in Fig. 2, membranes were repeatedly resuspended in 1 ml of phosphatase inhibitor buffer and pelleted by centrifugation.
Guanylyl Cyclase Assays-All of the guanylyl cyclase reactions were at 37°C. Except for those shown in the top panel of Fig. 4, all of the assays contained cyclase mixture 1 (25 mM Hepes, pH 7.4, 1 mM GTP, 1 mM EDTA, 4 mM MgCl 2 , 0.5 M microcystin, and 10 -30 Ci of [␣-32 P]GTP). Where indicated in Figs. 1-3, reactions also included 1 M rat ANP, 1 M rat CNP, 1 mM ATP, 1 mM AMPPNP, 1% Triton X-100, and 3 mM MnCl 2 . Assays were started by adding 50 -55 l of prewarmed reaction mixture to tubes containing 15-20 l of membranes and 30 l of activators (natriuretic peptides and/or adenine analogs). Assays were stopped with zinc acetate, and cGMP was purified as described previously (35). In the top panel of Fig. 4, membranes were prepared in homogenizing buffer and incubated with water or 10 M concentrations of the adenine analogs shown in the legend for 10 min at 37°C in cyclase mixture 2 (25 mM Hepes, pH 7.4, 50 mM NaCl, 0.1% bovine serum albumin, 0.25 mM 1-methyl-3-isobutylxanthine, 5 mM MgCl 2 , 10 mM NaN 3 , 0.1 mM GTP, and 10 -30 Ci of [␣-32 P]GTP) as described originally (32). Then water (basal), 1 M ANP, or 1 M ANP and 1 mM AMPPNP were added, and the reactions were incubated for an additional 10 min. The experiment shown in the bottom panel of Fig. 4 was conducted exactly as described for the top panel of Fig. 4 except that the membranes were prepared in phosphatase inhibitor buffer and the cyclase reaction contained mixture 1.
Immunoprecipitations-Membranes were prepared and assayed as described above in the presence of either mixture 1 or mixture 2, except cyclase reactions were stopped with the addition of 1 ml of ice-cold immunoprecipitation buffer containing microcystin and protease inhibitors. Samples were then transferred to a microcentrifuge tube and precleared by rotating for 20 min at 4°C with 50 l of immobilized protein A beads followed by low speed centrifugation. 800 l of sample were transferred to a new tube and incubated overnight with 2 l of rabbit polyclonal antiserum 6325 against NPR-A. Immunocomplexes were washed three times with immunoprecipitation buffer and fractionated by SDS-PAGE as described previously (36,37). The phosphorylation state and protein levels of NPR-A were determined by ProQ Diamond and SYPRO Ruby staining of the gel, respectively, as reported for NPR-B (37).

ATP Is Not Required for Maximal Natriuretic Peptide
Receptor Activation-In the process of investigating the calcium-dependent inhibition of NPR-B, we performed guanylyl cyclase assays on membranes prepared from HEK293T cells stably expressing NPR-B (293T-NPR-B). In contrast to previous re- ports where the absence of ATP was associated with a complete failure of CNP to activate NPR-B (21), CNP alone increased cyclase activity over 200-fold in our assay (Fig. 1, top left). The inclusion of AMPPNP did not further increase CNP-dependent activity, and the inclusion of ATP resulted in slight increases that were not statistically significant. Similarly, the addition of ANP to membranes from HEK-293T-NPR-A cells increased guanylyl cyclase activity about 50-fold (Fig. 1, top right). As with NPR-B, the inclusion of AMPPNP did not increase cyclase activity, and the ATP-dependent increases were not statistically significant. These data demonstrate that NPR-A and NPR-B are maximally activated in the absence of ATP, which is not consistent with a model in which ATP binding is required for receptor activation.
To determine whether adenine nucleotides affect the sensitivity of NPR-A and NPR-B to natriuretic peptide stimulation, we performed concentration-response experiments in the presence or absence of AMPPNP or ATP (Fig. 1, bottom   panels). We observed no statistically significant differences in the CNP or ANP concentrations required to elicit half of the maximum response in the presence or absence of the adenine nucleotides (Fig. 1, insets, bottom panels). Thus, our data indicate that adenine nucleotides neither increase natriuretic peptide-dependent receptor sensitivity (Fig. 1, bottom panels) nor increase maximal activation levels (Fig. 1,  top panels).
Because it was formally possible that the activation of our crude membranes was due to cellular ATP contamination, we performed guanylyl cyclase assays on membranes that were washed successively three or four times with phosphatase inhibitor buffer. We hypothesized that if activation were a function of ATP contamination, then the more highly washed membranes would be less responsive to natriuretic peptide stimulation. We found that washing had no effect on the hormone responsiveness of either receptor (Fig. 2). In fact, we found that natriuretic peptides stimulated these highly washed membranes to about the same level as did manganese and Triton X-100, conditions thought to activate particulate guanylyl cyclases to their maximum levels. We observed similar results in NIH3T3 cells stably expressing NPR-A, indicating that ATP-independent activation of these receptors is not limited to the 293 model system (Fig. 2, bottom panel). Interestingly, unlike the assays in Fig. 1, in these experiments we observed more activity in the presence of ATP and natriuretic peptide than with natriuretic peptide alone. ATP Stabilizes the Guanylyl Cyclase Activity of NPR-A and NPR-B-One possible explanation for the discrepancy between the effects of ATP on the cyclase activities measured in Figs. 1 and 2 is the reaction time. The membranes described in Fig. 1 were assayed for 15 s, whereas in Fig. 2 they were assayed for 1 min. Therefore, we investigated the ability of AMPPNP to increase the guanylyl cyclase activity of NPR-A and NPR-B as a function of time. We used AMPPNP instead of ATP to rule out the effect of phosphorylation on potential activity changes. As shown in the top panel of Fig. 3, AMPPNP had no statistically significant effect on the activity of NPR-A until the membranes had been incubated for 300 s. However, for every subsequent time point, the amount of activity measured in the presence of AMPPNP was significantly higher than that measured in its absence. Similar results were obtained for NPR-B except that a significant effect of AMPPNP was observed after 180 s (Fig. 3,  bottom panel). These are the first data to demonstrate that the increased natriuretic peptide-dependent cyclase activities result from the ability of AMPPNP to maintain initial reaction rates. In other words, these data indicate that AMPPNP is stabilizing, not activating, NPR-A and NPR-B.
Buffers That Yield Highest Cyclase Activity Yield More Highly Phosphorylated Receptors-Finally, we asked why we observed robust ATP-independent activations of natriuretic peptide receptors when other investigations observed either no activation or severely diminished responses. Clearly one reason is the short duration of our assays (15 s). All other reports describe much longer assay periods (10 min). However, another contributing factor became apparent during our attempts to demonstrate ATP␥S-dependent sensitization of NPR-A to ANP stimulation as originally reported (32). In this assay, membranes from 3T3-NPR-A cells were prepared in a phosphatase inhibitor-deficient buffer and incubated in the presence of 10 M concentrations of AMPPNP, ATP, or ATP␥S at 37°C to sensitize NPR-A to subsequent 10-min activations with ANP in the presence or absence of 1 mM AMPPMP.
As in the original report (32), we found that membranes preincubated with ATP␥S were more active than those preincubated with ATP, AMPPNP, or buffer, presumably because thiophosphorylated NPR-A is resistant to dephosphorylation (Fig. 4, top). We also found that cyclase activities were higher in reaction mixtures containing ANP and AMPPNP compared with ANP alone, although unlike in the original report we observed a statistically significant (p ϭ 0.05) 2.5-fold activation in the absence of AMPPNP. Interestingly, the specific activities obtained under these conditions were dramatically lower than those observed in our previous assays using membranes from the same cells (Fig. 2, middle). When we repeated this assay using our normal cell lysis (phosphatase inhibitor buffer) and cyclase reaction buffers that contain phosphatase inhibitors, EDTA, and higher GTP concentrations, specific activities were restored to the previous levels (Fig. 4, middle; please note y axis scale change). Importantly, the ability of ANP to prominently activate NPR-A in the absence of adenine nucleotides was also restored. Although cyclase levels in this assay were increased about 2-fold by AMPPNP, this is what would be expected because of stabilization in a 10-min assay (Fig. 3). These data indicate that the more than 100-fold disparity between our activities and the activities obtained by other laboratories is due to differences in the composition of the buffers used to prepare the membranes and measure cyclase activity.
Given that our lysis and cyclase reaction buffers included phosphatase inhibitors (mixture 1), whereas the buffers that resulted in diminished activity did not (mixture 2), we asked whether the phosphorylation state of NPR-A isolated in the presence of phosphatase inhibitors was higher than that of NPR-A isolated from buffers lacking these inhibitors. To accomplish this goal, we treated 3T3-NPR-A membranes exactly as described in the top and middle panels of Fig. 4. However, instead of stopping the cyclase reactions with zinc acetate as is normally done for a cyclase assay, we solubilized the receptors in immunoprecipitation buffer containing phosphatase inhibitors and purified it by immunoprecipitation and SDS-PAGE. We then measured NPR-A phosphate content and protein levels by incubating the resulting gel with ProQ Diamond and SYPRO Ruby dyes, respectively (Fig. 4, bottom). We found that the amount of NPR-A isolated using the two procedures was unaffected by the buffer constituents (SYPRO Ruby), whereas the phosphorylation state of NPR-A was markedly higher when isolated with buffers containing phosphatase inhibitors (ProQ Diamond). Data from three separate experiments (n ϭ 6) indicated that NPR-A purified in the presence of phosphatase inhibitors (mixture 1 ϭ 1 arbitrary units Ϯ 0.05) contained FIG. 4. Phosphatase inhibitors increase the guanylyl cyclase activity and phosphorylation state of NPR-A. Crude membranes from 3T3-NPR-A cells were incubated for 10 min at 37°C in the presence of 10 M AMPPNP, ATP, ATP␥S, or water. The effect of the first incubation on the hormone responsiveness of NPR-A was accessed by measuring the guanylyl cyclase activity of the membranes in water (basal), 1 M ANP, or 1 M ANP and 1 mM AMPPNP for 10 min at 37°C as indicated. In the top panel, membranes were prepared, and the cyclase reaction was conducted in buffers lacking phosphatase inhibitors (mixture 2, CT2), whereas in the middle panel phosphatase inhibitors were included in both buffers (mixture 1, CT1). The data are from 3-4 separate experiments where n ϭ 6 -8. In the bottom panel, membranes were treated exactly as described above, but NPR-A was purified by immunoprecipitation and SDS-PAGE. The resulting gel was stained with ProQ Diamond and SYPRO Ruby to determine NPR-A phosphate and protein levels, respectively. Results from two separate experiments are shown that are representative of four individual experiments. more than twice as much phosphate as NPR-A purified in the absence of phosphatase inhibitors (mixture 2 ϭ 0.47 arbitrary units Ϯ 0.09). These differences were significant at a p value of 0.0003 and suggest that the differences in enzymatic activity result, at least in part, from changes in the NPR-A phosphorylation state. Hence, the reason that AMPPNP has a greater effect on ANP-dependent cyclase activity in the absence (Fig. 4, top) as opposed to the presence (Fig. 4, middle) of phosphatase inhibitors (Fig. 4, top) is not due to allosteric activation of NPR-A. Rather the stabilization effect of AMPPNP is simply more apparent under these conditions because of the diminished hormonal response of the dephosphorylated receptors. DISCUSSION In this report, we demonstrate for the first time massive ATP-independent activation of guanylyl cyclase-linked natriuretic peptide receptors. Also for the first time, we find that the AMPPNP-dependent increases in cyclase activities result from enzyme stabilization, not from receptor activation, as has been previously reported. These data call into question the current two-step allosteric activation model that suggests that the physiologic activation of these receptors requires ANP binding to their extracellular domains followed by ATP binding to their kinase homology domains. Instead, it suggests a single-step model in which natriuretic peptide binding in the absence of ATP is sufficient for maximal catalytic activation.
ATP has now been shown to stabilize the activity of all members of the human transmembrane guanylyl cyclase family that have been tested. Our data along with data from other laboratories on the regulation of GC-C (38, 39) and GC-E (40, 41) support a unifying stabilization model for ATP-dependent regulation of transmembrane guanylyl cyclases. Whether ATP stabilizes these guanylyl cyclase receptors by directly binding to them is not known. Mutations in the glycine-rich regions of NPR-A (42) or NPR-B (30) do not block receptor activation, but the contribution of these regions to the stabilization process has not been formally investigated. However, the other ATPregulated receptor guanylyl cyclases, GC-C and GC-E, completely lack this motif, suggesting that another region common to all four receptors either binds ATP directly or interacts with an ATP-binding regulatory protein.