Agonistic Induction of a Covalent Dimer in a Mutant of Natriuretic Peptide Receptor-A Documents a Juxtamembrane Interaction That Accompanies Receptor Activation*

The natriuretic peptide receptor-A (NPR-A) is composed of an extracellular domain with a ligand binding site, a transmembrane-spanning domain, a kinase homology domain, and a guanylyl cyclase domain. In response to agonists (atrial natriuretic peptide (ANP) and brain natriuretic peptide), the kinase homology domain-mediated guanylate cyclase repression is removed, which allows the production of cyclic GMP. Previous work from our laboratory strongly indicated that agonists are exerting their effects through the induction of a juxtamembrane dimeric contact. However, a direct demonstration of this mechanism remains to be provided. As a tool, we are now using the properties of a new mutation, D435C. It introduces a cysteine at a position in NPR-A corresponding to a supplementary cysteine found in NPR-C6, another receptor of this family (a disulfide-linked dimer). Although this D435C mutation only leads to trace levels of NPR-A disulfide-linked dimer at basal state, covalent dimerization can be induced by a treatment with rat ANP or with other agonists. The NPR-AD435C mutant has not been subjected to significant structural alterations, since it shares with the wild type receptor a similar dose-response pattern of cellular guanylyl cyclase activation. However, a persistent activation accompanies NPR-AD435C dimer formation after the removal of the inducer agonist. On the other hand, a construction where the intracellular domain of NPR-AD435C has been truncated (ΔKCD435C) displays a spontaneous and complete covalent dimerization. In addition, the elimination of the intracellular domain in wild type ΔKC and ΔKCD435C is associated with an increase of agonist binding affinity, this effect being more pronounced with the weak agonist pBNP. Also, a D435C secreted extracellular domain remains unlinked even after incubation with rat ANP. In summary, these results demonstrate, in a dynamic fashion, the agonistic induction of a dimeric contact in the juxtamembrane domain of NPR-A. In addition, this process seems to require membrane attachment of the receptor. Finally, the intracellular domain represses this contact at the basal state, showing its potent influence on the outer juxtamembrane domain.

The natriuretic peptide receptors (NPRs) 1 are members of a family of single-transmembrane domain receptors that mediate their effects through the production of cyclic GMP (1). Three different NPRs have been identified, and two of these, NPR-A and NPR-B, respond to agonists by the activation of their guanylyl cyclase catalytic domain. The production of intracellular cGMP mediates their effects on diuresis, vasorelaxation, and the inhibition of the renin-angiotensin-aldosterone system (2). A third receptor, called NPR-C or the clearance receptor, displays only 37 amino acids in its intracellular domain and is devoid of guanylyl cyclase activity. NPR-C is a disulfidebridged dimer that internalizes through a fast intracellular cycle process (3) and might be involved in signal transduction (4). NPR-A is stimulated by two peptides, ANP and BNP, whereas CNP is the only known agonist of NPR-B (2,5). NPR-C has nearly equal binding affinity for all of these natriuretic peptides (6,7).
NPR-A is an ϳ130-kDa protein that contains four structural domains: an extracellular domain (ECD) with a ligand binding site, a transmembrane domain (TM), a kinase homology domain (KHD), and a guanylyl cyclase domain (GC) (2). Several studies have demonstrated that this receptor is spontaneously preassociated in noncovalent dimers or oligomers (8 -10). Taken together, these studies have indicated that both extracellular and intracellular domains are involved in NPR-A dimerization.
According to the current model of agonist activation, NPR-A signal transduction includes these five sequential steps (11). 1) The binding of the natriuretic peptide to the ectodomain induces a conformational change. 2) This modification corresponds to a signal that migrates through the TM domain. 3) The KHD responds to this signal by adopting a conformation that allows ATP binding. 4) ATP binding has two major effects in derepressing the guanylyl cyclase activity and increasing the off-rate of ANP from the receptor (12). 5) Subsequent desensitization results from reduction in phosphorylation state of the KHD (13,14).
We have previously brought out the remarkable conservation of spacing between the cysteine residues found in the extracellular domain of nearly all of the guanylyl cyclases (15). * 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. This work was supported by grants MT-13753 and MT-15165 from the Medical Research Council of Canada.
By analogy to the cysteine distribution of NPR-C5, we previously designed the mutation C423S in NPR-A, which eliminates its first juxtamembrane cysteine (15). The expectation was that, in the absence of this cysteine (equivalent to Cys 423 in NPR-A), it would permit interchain linkage of the second cysteine (Cys 432 in NPR-A, equivalent to Cys 469 of NPR-C5). Indeed, this mutation led to a spontaneously disulfide-bridged NPR-A C423S dimer.
This NPR-A C423S mutant was also found to be constitutively activated, and it displayed an important increase in the binding affinity of pBNP, a weak agonist (15). Using these observations, we proposed a model where agonists are inducing a dimeric "tightening" in the juxtamembrane region of NPR-A, hence allowing catalytic activation of the guanylyl cyclase. However, we indicated at the time that we could not exclude the contribution of a conformational change induced by the mutation independently of the interchain disulfide linkage (15). For instance, it was not known if the disruption of the Cys 432 -Cys 423 bond might by itself take part in the constitutive activation of NPR-A C423S .
In the current study, our objective is to definitively demonstrate that a juxtamembrane dimerization event is associated with NPR-A activation. To limit eventual structural alterations, we chose to avoid the disruption of the Cys 423 -Cys 432 internal bond. For this, we referred to a minor splicing isoform of NPR-C (NPR-C6) that displays a supplementary juxtamembrane cysteine also forming an accessory interchain disulfide bridge (17). By comparing the juxtamembrane regions of NPR-C6 and NPR-A, this supplementary cysteine in NPR-C6 aligns with the aspartate 435 in NPR-A (Fig. 1). We thought that the addition of a cysteine at position 435 might lead again to a covalently dimerized NPR-A. We thus verified if NPR-A D435C forms a disulfide-linked dimer. We found that although this mutant displays only trace levels of spontaneous covalent dimerization, agonists can induce such a dimeric linkage. This characteristic allowed us to define important constraints involved in receptor activation.

EXPERIMENTAL PROCEDURES
Construction of NPR-A Mutants-rNPR-A mutants were engineered in the expression vector PBK-Neo (Stratagene). The construction of the His-tagged HT-ECD has already been described (15). This construct includes all of the extracellular domain up to Leu 440 followed by Arg-Ser-His 6 . HT-ECD D435C was obtained by mutating the Asp 435 in Cys with the mutagenic primer 5Ј-CCTGCAACCAATGCCACTTTTCGA-C-3Ј using the Transformer mutagenesis kit from CLONTECH. NPR-A D435C was obtained by mutating Asp 435 in Cys using the mutagenic primer 5Ј-CCTGCAACCAATGCCACTTTTCCAC-3Ј. A deletion mutant of the entire intracellular domain of NPR-A (⌬KC) was obtained by PCR using Taq polymerase. The amplification was realized using a sense primer 5Ј-ATGCCTTCAGGAATCTGATGC-3Ј and two antisense primers. The first antisense primer (5-AGAGCCTCTTTCACCCTTCCTGT-ATATGAAGAAAGA-3Ј) was limiting (1 pmol), and the other (5Ј-TTT-TGGTACCTTAACCTCTGGTAGAAGAGCCTCTTTCACCCTT-3Ј) was in excess (100 pmol). The amplified fragment included codons 217-464, followed by the epitope GERGSSTRG, a stop codon, and finally a KpnI site. The polymerase chain reaction product was codigested with EcoRV-KpnI. This fragment (codons 416 to Stop) was inserted in PBK-NPR-A, which had been previously codigested with the same enzymes.
The resulting construction included the whole ectodomain, the transmembrane-spanning domain, and the two first intracellular residues (Arg 463 -Lys 464 ), followed by the epitope and a stop codon. The ⌬KC D435C was obtained by following the same strategy but using NPR-A D435C as an initial polymerase chain reaction template. The constructions and the mutations were confirmed by sequencing on the two strands using the Sequenase kit from Amersham Pharmacia Biotech.
Cell Culture-The human embryonal kidney cell line 293 (American Type Culture Collection) was grown in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum and 100 units of penicillin/streptomycin in a 5% CO 2 incubator at 37°C. For the cyclic GMP stimulation experiments, cells of the NPR-A and NPR-A D435C clones were seeded at 10 5 cells/well onto 24-well cluster plates. Experiments were performed when the cells reached subconfluence.
Transient and Stable Expression in HEK 293 Cells-Transient expression of ⌬KC, ⌬KC D435C , HT-ECD, and HT-ECD D435C was obtained by the transfection of constructs in PBK-Neo using the CaHPO 4 precipitation as described elsewhere (18). For HT-ECD and HT-ECD D435C , 20 g of DNA/10-cm plate was transfected. For the truncated receptors (⌬KCs), the quantity of transfected PBK-⌬KC had to be reduced to obtain a level of expression comparable with the full-length receptor. This level was obtained by transfecting 2.5 g of PBK-⌬KC or PBK-⌬KC D435C mixed with 17.5 g of the PBK-Neo vector. For the stable expression of the rNPR-A and rNPR-A D435C , the cells were transfected with 20 g of DNA/10-cm plate, and clones were selected in 600 g/ml G-418 (Geneticin; Roche Molecular Biochemicals) in culture medium.
Dose-Effect Study of Cellular NPR-A D435C Covalent Dimerization-Stable clones expressing wild type NPR-A and NPR-A D435C were plated in 10-cm plates and were allowed to grow to subconfluence. After the cells were washed twice with serum-free DMEM, 6 ml of the same medium (37°C) containing 0.5% BSA and varying concentrations (10 Ϫ12 to 10 Ϫ6 M) of rANP were added to individual plates. The induction was allowed to proceed for 30 min in a 5% CO 2 incubator at 37°C. Following incubation, the cells were washed twice with phosphatebuffered saline (37°C), and all liquid was removed. The plates were put directly in a freezer at Ϫ80°C until further used. For membrane preparation, the frozen plates were put on ice, and 4 ml of ice-cold homogenization buffer (5 mM Tris, pH 7.4, 0.2 mM EDTA containing 10 Ϫ6 M aprotinin, 10 Ϫ6 M pepstatin, 10 Ϫ6 M leupeptin, 10 Ϫ5 M pefabloc) was immediately added. The cells were scraped, collected in centrifugation tubes, homogenized twice for 20 s with a Polytron homogenizer, and centrifuged once for 30 min at 35,000 ϫ g. The pellets were immediately resuspended in ice-cold freezing buffer (50 mM Tris, pH 7.4, 0.1 mM EDTA, 250 mM sucrose, 1 mM MgCl 2 , and the protease inhibitors) and stored at Ϫ80°C. The protein concentration was determined using the BCA protein assay kit (Pierce). The induction of receptor dimerization was assessed by Western blotting after the separation of membrane proteins (25 g) on a 5% SDS-PAGE in the presence or absence of ␤-mercaptoethanol in the loading buffer.
Membrane Preparations-Membranes used for the binding studies and the in vitro induction of NPR-A D435C dimerization were prepared as follows. 72 h post-transfection for ⌬KC and ⌬KC D435C , or at subconfluence for the stable clones expressing NPR-A and NPR-A D435C , the cells were rinsed twice with phosphate-buffered saline and lysed in ice-cold homogenization buffer (5 mM Tris, pH 7.4, 0.2 mM EDTA, and the protease inhibitors). The cells were scraped, collected in centrifuge tubes, homogenized twice for 20 s with a Polytron homogenizer, and centrifuged for 30 min at 35,000 ϫ g. The pellets were resuspended and washed twice in the same buffer. Finally, membranes were resuspended in ice-cold freezing buffer (50 mM Tris, pH 7.4, 0.1 mM EDTA, 250 mM sucrose, 1 mM MgCl 2 , and the protease inhibitors), frozen in liquid nitrogen, and stored at Ϫ80°C.
Purification of Secreted Ectodomains-HT-ECD and HT-ECD D435C were purified from cell culture medium collected 72 h post-transfection. Supernatants were dialyzed three times against 90 volumes of 50 mM sodium phosphate buffer, pH 7.4, containing 0.3 M NaCl. After adding 16% glycerol, the dialysate was aliquoted, frozen in liquid nitrogen, and kept at Ϫ80°C. The His-tagged ectodomains were purified on Ni 2ϩnitrilotriacetic acid-agarose gel (Qiagen) as described elsewhere (15). Ectodomains were eluted from the gel with 500 mM imidazole. The eluates were finally dialyzed in sodium phosphate buffer, pH 7.4, containing 0.3 M NaCl using Slide-A-Lyzer cassettes (molecular weight cut-off of 10,000; Pierce).
In Vitro Induction of NPR-A D435C Dimerization-25 g of membrane proteins obtained from stable clones expressing NPR-A or NPR-A D435C were added to 500 l of cold incubation buffer (50 mM Tris, pH 7.4, 0.1 mM EDTA, 0.5% BSA, and the protease inhibitors) containing 1 M agonist. For the specificity study, rANP, pBNP, atriopeptin I (API), CNP, or C-ANF were included in the incubation mixture. The agonist induction of disulfide linkage was allowed to proceed for 22 h at 4°C. Following incubation, the samples were centrifuged in a microcentrifuge at 10,000 ϫ g for 10 min. The pellets were carefully resuspended in ice-cold deionized water, and 2ϫ SDS-PAGE sample buffer (without ␤-mercaptoethanol) was immediately added. The samples were immediately boiled for 5 min. The covalent dimerization was assessed by Western blotting after the separation of membrane proteins on a 5% SDS-PAGE. The induction of HT-ECD D435C was tested at 4°C or at 22°C, with 1 M rANP, for 22 h in 0.1 ml of binding buffer (50 mM sodium phosphate buffer, pH 7.4, 0.3 M NaCl, 1 mM EDTA, 0.1% BSA, 0.05% lysozyme). The presence of covalent dimer was assessed by Western blotting after the separation of proteins (nonreducing conditions) on a 7.5% SDS-PAGE.
Receptor Binding Assay-125 I-rANP was prepared using the lactoperoxidase method as described elsewhere (15). Binding to membranes was performed at 4°C for 22 h in 1 ml of binding buffer (50 mM Tris, pH 7.4, 0.1 mM EDTA, 5 mM MnCl 2 , and 0.5% BSA). Competition experiments were done by incubation of 3-5 g of HEK 293 membrane expressing rNPR-A, rNPR-A D435C , ⌬KC, or ⌬KC D435C with 10 fmol of 125 I-rANP and increasing concentrations of nonradioactive peptides. Bound 125 I-rANP was separated from free ligand by filtration on GF/C filters precoated with 1% polyethyleneimine.
Whole Cell Guanylyl Cyclase Stimulation-Cells stably expressing rNPR-A and rNPR-A D435C were allowed to grow to subconfluence on 24-well cluster plates. The wells were washed twice with serum-free DMEM and were incubated in a final volume of 1 ml of the same medium containing 0.5 mM 3-isobutyl-1-methylxanthine (IBMX), 0.5% BSA, and varying concentrations (10 Ϫ12 to 10 Ϫ7 M) of rANP. After 1 h of incubation, the medium was collected, and extracellular cyclic GMP was determined by radioimmunoassay as described elsewhere (19). After the assay, 1ϫ SDS-PAGE sample buffer (95°C) was added to several wells. The wild type and mutant receptor levels were estimated by Western blotting, which was used to normalize their relative cGMP production.
Induction and Measurement of Persistent Guanylyl Cyclase Activity-Stable clones expressing NPR-A and NPR-A D435C were allowed to grow to subconfluence on 10-cm plates. After having washed the cells with serum-free DMEM, 6 ml of DMEM (37°C) containing 0.5% BSA and 10 Ϫ7 M rANP was added to the plates. The incubation was allowed to proceed for 30 min in a 5% CO 2 incubator at 37°C. The cells were then carefully washed twice with DMEM, 0.5% BSA (37°C) and incubated for another 30 min. After this postincubation, the cells were washed again twice with phosphate-buffered saline (37°C). Membrane preparation was done as described above except that homogenization, washings (three times), and freezing were realized in 50 mM HEPES, pH 7.4, containing 20% glycerol, 50 mM NaCl, 10 mM NaPO 4 , 0.1 M NaF, 1 mM Na 3 VO 4 , and the protease inhibitors. The protein concentration was determined, and these membranes were used for guanylyl cyclase assays as described in other studies (13,20). 5 g of membrane proteins were incubated during 10 min at 37°C in 50 mM Tris-HCl, pH 7.6, with 10 mM theophylline, 2 mM IBMX, 10 mM creatine phosphate, 10 units of creatine kinase, 1 mM GTP, and 4 mM MgCl 2 . Different conditions were tested using GTP alone (basal) or by adding 1 M rANP, 1 mM ATP, rANP and ATP together or adding 1% Triton X-100 with 4 mM MnCl 2 instead of MgCl 2 . Cyclic GMP was separated from GTP by chromatography on alumina and evaluated by radioimmunoassay as previously reported (19).
Western Blotting and Immunodetection-Membrane proteins were separated on SDS-PAGE in the presence or absence of 5% ␤-mercaptoethanol in the loading buffer. The proteins were transferred to a nitrocellulose membrane (Bio-Rad) using the liquid Mini Tans-Blot System (Bio-Rad). Detection of NPR-A and ⌬KC was achieved using a rabbit polyclonal antiserum raised against the sequence YGERGSSTRG and purified by affinity chromatography. This sequence corresponds to human NPR-A carboxyl terminus preceded by a tyrosine for radioiodination purposes. The rat NPR-A differs from this epitope at a single position; however, both receptors are recognized. Specific signal was probed with an HRP-coupled anti-rabbit polyclonal antibody according to the ECL Western Blotting Analysis System (Amersham Pharmacia Biotech). For the His-tagged HT-ECD and HT-ECD D435C , purified aliquots were run on a 7.5% SDS-PAGE without ␤-mercaptoethanol in the sample buffer. In this case, 6 M urea was included in the sample buffer and in the gel, since it was found to improve the immunodetection. Proteins were transferred to a nitrocellulose membrane as described above, and the ectodomains were detected using a commercial mouse anti-tetrahistidine antibody (Qiagen) according to the technique provided by the manufacturer. Specific signal was probed with a horserad-ish peroxidase-coupled anti-mouse polyclonal antibody using the ECL Western Blotting Analysis System.
Data Analysis-Dose-response curves were analyzed with the program AllFit for Windows based on the four-parameter logistic equation (21). Radioligand binding data were analyzed with the same program on a model for the law of mass action (22). For simplicity, the same binding data were also analyzed as dose-response curves, and ED 50 values are mentioned in the discussion instead of K d values.

Cellular Induction of a Disulfide-bridged Dimer of NPR-
A D435C by rANP-We considered the possibility that the introduction of a cysteine at position 435 could cause a covalent dimerization of rNPR-A subunits. This site was chosen because of its linear alignment with an accessory cysteine involved in the linkage of NPRC-C6 dimers (Fig. 1).
Initial studies with rNPR-A D435C expressed in HEK293 showed, however, only low levels of spontaneous dimerization. At this point, we proceeded to further testing and explored the possibility that some dimeric linkage might be induced by rANP, the main agonist of NPR-A. This hypothesis turned out to be right, since dimers of rNPR-A D435C could be detected after a cellular incubation with 1 M rANP (30 min, 37°C). We further detailed this induction through a dose-effect study using a clone stably expressing NPR-A D435C (Fig. 2). A clear signal of ϳ260 kDa corresponding to NPR-A D435C dimers is detected on nonreducing SDS-PAGE. It is noteworthy that such an induction is not seen in cells expressing wild type NPR-A. A densitometric analysis of the dimeric induction gave an approximate ED 50 of ϳ900 pM, which is higher than the ED 50 obtained in cellular guanylyl cyclase stimulation (87 pM). The exact explanation for this difference is not known. However, the reaction between cysteines leading to disulfide formation could be a limiting element of the process and might affect the apparent dose-effect curve. Therefore, considering this supplementary element, an ED 50 of ϳ900 pM constitutes an acceptable value.
These results are showing that ANP induces a particular dimeric contact in the juxtamembrane domain of NPR-A D435C . This conclusion is in accordance with our previous hypothesis based on the observation of constitutively dimerized NPR- A C423S mutant (15). However, the ANP-inducible dimerization properties of NPR-A D435C directly proves this mechanism. Furthermore, the near absence of NPR-A D435C covalent dimerization at basal state reveals the existence of a constraint that represses this contact in absence of ANP.
Cellular Guanylyl Cyclase Activation of NPR-A D435C -To assess if the D435C mutation altered the NPR-A function, we studied cellular guanylyl cyclase activation by rANP. As shown in Fig. 3, the ED 50 obtained for the wild type (71 Ϯ 12.8 pM) and the mutant (87 Ϯ 4.6 pM) are essentially comparable, and their maximal levels of stimulation are similar. It should be noted that we reproducibly observed a very slight increase in the basal activity of NPR-A D435C as compared with that of NPR-A WT (Fig. 3). This almost negligible increase might indicate that the D435C mutation has very slightly modified the interactions in the juxtamembrane region. Alternatively, it may be attributed to a trace level of NPR-A D435C dimer sometimes detectable on Western blot through signal overexposure (not shown).
From this result, it can be reasonably concluded that NPR-A D435C displays a response to rANP that is very close to that for wild type NPR-A. Therefore, the conclusions based on the results obtained with this mutant are likely to be applicable to the wild type receptor.
A Persistent Activation Accompanies NPR-A D435C Dimer Formation-We investigated if the induction of NPR-A D435C covalent dimerization goes along with persistent activation. Guanylate cyclase activity was tested in vitro with membrane preparations obtained from ANP-treated cells. Stable clones expressing NPR-A WT and NPR-A D435C were treated for 30 min (37°C) with 10 Ϫ7 M ANP. Removal of the ligand was realized through extensive washing of cells, followed by postincubation without ligand and washings during membrane preparation (see "Experimental Procedures"). The membranes were submitted to several conditions using GTP alone (basal) or together with ATP, ANP, ATP plus ANP, and Triton/Mn 2ϩ . The results were expressed as a percentage of the activation obtained with Triton/Mn 2ϩ . This condition stimulates NPR-A to its maximal catalytic level and is commonly used as an internal reference of enzyme activity.
Several observations can be made from the results. First, the level of costimulation with ATP plus ANP tends to diminish in membranes obtained from ANP-treated cells (NPR-A: nontreated 38.9 Ϯ 4.5%, treated 29.8 Ϯ 5.9%; NPR-A D435C : nontreated 39.8 Ϯ 6.9%, treated 29.6 Ϯ 2.6%). This may be attributed to some desensitization of the receptor. Also, as shown in Fig. 4, cellular ANP treatment results in an increase of basal activity that is 3-fold higher for NPR-A D435C than for the wild type. Finally, ATP stimulation is appreciably increased for both receptors as a result of cellular ANP pretreatment. This latter phenomenon has been previously observed following receptor desensitization of NPR-A (13). However, this increase in ATP response is significantly higher in the D435C mutant than in the wild type receptor (Fig. 4). These results indicate a more pronounced tendency of NPR-A D435C toward persistent activation as a result of ANP pretreatment.
The Covalent Dimerization of NPR-A D435C Is Specifically Induced by Agonists in Vitro-We also tested if the agonistinduced dimerization could occur in vitro in membrane preparations. A clear dose-dependent induction of a ϳ260-kDa dimer can be seen when membranes are incubated with rANP for 22 h at 4°C (Fig. 5). Such induction is not seen in the wild type control. Although dimerization is not complete at the highest ANP concentration used, an ED 50 of approximately 52 nM can be estimated. This value is far from what is obtained in cellular activation (Fig. 2). However, at 10 Ϫ6 M the dimerization is more complete than what is seen in cells. Considering what has already been mentioned for the cellular dimeric induction, it is possible that the lower temperature of incubation (used to minimize protein degradation) might directly or indirectly affect the rate of formation of the disulfide bond, which may be limiting in these conditions. Nevertheless, to assess the specificity of the process, we tested the induction using the main selective ligands of the natriuretic peptide receptors (rANP, API, pBNP, CNP, and C-ANF). A concentration of 1 M of the different peptides was chosen, since, due to the high ED 50 of the ANP-induced dimerization in these conditions, this concentration was likely to discriminate the relative potency of the agonists. As shown in Fig. 6, the level of induction correlated well with agonists specificity on NPR-A (rANP Ͼ pBNP Ͼ API; CNP and C-ANF no induction). ⌬KC-NPR-A D435C Is a Disulfide-bridged Dimer-At this point, we wanted to assess the potential role of the intracellular domain in the source of the constraint preventing NPR-A D435C covalent dimerization at the basal state. We thus realized constructions with truncations of the intracellular domain on wild type NPR-A (⌬KC WT ) and on NPR-A D435C (⌬KC D435C ). These were analyzed for their level of covalent dimerization.
Strikingly, ⌬KC D435C covalent dimerization appears complete on nonreducing SDS-PAGE (ϳ130 kDa), whereas ⌬KC WT remains monomeric (ϳ70 kDa) (Fig. 7). Therefore, the intracellular domain is at the origin of a constraint that represses Cys 435 interaction at basal state. On the other hand, ⌬KC D435C spontaneously reaches the dimeric state, allowing Cys 435 disulfide linkage. These observations indicate that a balance between the antagonistic constraints of the extracellular and the intracellular domains modulates the juxtamembrane dimeric interactions.
Intracellular Constraint Represses High Affinity Binding-We wanted to further define the effects of this constraint originating from the intracellular domain on the functions of the extracellular domain. As a tool, we used the binding properties of rANP and pBNP, which are respectively strong and weak agonists of rNPR-A. As we have shown earlier, the competition binding of pBNP against 125 I-rANP 22 is biphasic and can be modeled into high and low affinity components (15,23). The molecular events associated with this complex binding are not fully understood. However, since the proportion of high affinity component is increased in NPR-A C423S , we have previously associated the high affinity state with a tight juxtamembrane conformation corresponding to the activated state (15). We also formerly hypothesized that the binding of pBNP was more influenced by the intracellular domain than that of rANP.
Binding studies were performed on NPR-A, NPR-A D435C , ⌬KC WT , and ⌬KC D435C . As shown in Fig. 8, the respective binding characteristics of rANP and pBNP on NPR-A WT and NPR-A D435C are similar. Also, the binding of pBNP on NPR-A and NPR-A D435C can be modeled with similar biphasic curves. For simplicity, we will provide here the ED 50  Clearly, these results confirm that pBNP binding is much more affected than rANP by the presence of the intracellular domain.
It is noteworthy that the rANP competition curves on ⌬KC WT and ⌬KC D435C are almost identical. Moreover, the pBNP binding on these truncated receptors is quite similar. This indicates that the Cys 435 disulfide linkage in ⌬KC D435C is probably not interfering with the high affinity conformation spontaneously reached by the receptor in the absence of the intracellular domain.
In summary, these results indicate that the ⌬KC spontaneously reaches a high affinity state. It is also reasonable to think that the presence of the intracellular domain prevents the receptor from spontaneously reaching this state in the absence of agonist.
rANP Is Not Inducing a Disulfide Linkage in ECD D435C -We finally tested if the D435C mutation could lead to the dimerization of a secreted NPR-A ECD devoid of transmembrane domain. A His-tagged extracellular domain of NPR-A D435C (HT-ECD D435C ) was produced in HEK293 cells, purified on Ni 2ϩ -nitrilotriacetic acid-agarose gel, and its level of dimerization was verified on nonreducing SDS-PAGE. As shown in Fig.  9, only a trace amount of spontaneous dimer can be seen. We also tested for induction of dimerization with rANP (10 Ϫ6 M, overnight, 4°C), which proved to be ineffective (Fig. 9). rANP induction was also tested at room temperature with the same result (not shown).
Thus, the cysteine 435 in HT-ECD D435C does not show a significant ability to form an intermolecular disulfide bridge. However, we wondered whether this incapacity was due to a lack of ligand-induced noncovalent dimerization. Therefore, we tested on gel filtration if rANP is inducing noncovalent dimerization and found that the dimerization was complete after an overnight incubation (22°C) with 10 Ϫ6 M rANP (not shown).
These results indicate that, in response to ligand induction, the juxtamembrane dimeric interactions are probably different in the ECD receptor mutant as compared with its membrane counterpart. On the other hand, it is possible that the transmembrane domain of the receptor might influence the structure of the outer juxtamembrane region, and, thus, its presence might be essential for ligand-induced disulfide linkage of Cys 435 .

DISCUSSION
In this work, we have shown that agonists are inducing a particular dimeric contact in the juxtamembrane domain of rNPR-A. Our results also indicate that the intracellular domain sterically hinders the juxtamembrane tightening associated with receptor activation. At basal state, this negative constraint presumably overcomes the ectodomain positive constraint. Upon agonist binding, the balance is switched toward juxtamembrane dimerization and, consequently, receptor activation (Fig. 10). In that respect, the induction of the C435 dimeric linkage in the full-length NPR-A D435C represents a direct tracer of these molecular events. Finally, membrane localization of this mutant has proven to be essential for the covalent dimeric linkage.
The spontaneous dimerization of ⌬KC D435C indicates that the ⌬KC (extracellular together with transmembrane domains) exerts by itself a spontaneous "positive" constraint toward jux- tamembrane dimerization. On the other hand, our results show a striking increase of pBNP binding affinity for both ⌬KC D435C and ⌬KC WT as compared with the full-length receptor, whereas the affinity of ANP is not affected. Taken together, these elements strongly indicate that the ⌬KC spontaneously reaches a "tight" dimeric state, probably closely related to the ectodomain dimeric positioning that occurs in the activated full-length NPR-A. Since the presence of the intracellular domain affects more deeply pBNP binding, one can think of the existence of a "threshold level" of juxtamembrane dimeric tightening beyond which the intracellular negative constraint might be mainly overcome by the positive influence of the ectodomain. Hence, ANP would be much more efficient than pBNP to overcome this threshold level. Therefore, the observed affinity of an agonist might result from a combination of its "pure" affinity for the membrane-anchored ectodomain together with its capacity to overcome the intracellular negative constraint. According to this model, the binding results observed with the ⌬KCs are providing the "pure affinities" of rANP and pBNP for the membrane-anchored ectodomain. Indeed, the difference of affinity between rANP and pBNP is much less for the ⌬KC (ϳ6 -10fold) as compared with the full-length receptor (ϳ500-fold).
One may wonder which part of the cytoplasmic domain of NPR-A is particularly involved in this steric hindrance. One can suspect a potent influence of the KHD. Of note, it has been shown that the removal of the KHD leads to a constitutive activation of the GC (24,25). Therefore, since the KHD seems to exert a potent regulatory role on GC activity, it might as well influence the ectodomain properties.
Agonistic induction of disulfide linkage in a mutated receptor has been described in one other case (26). In this study, Sorokin et al. have shown that EGF was stimulating such linkage in a mutant of the EGF receptor having a supplementary cysteine in the juxtamembrane domain. The induced dimer was found to possess persistent tyrosine kinase activity and also displayed increased high affinity binding. The authors concluded that the disulfide bridge is stabilizing a particular dimeric arrangement of the two protomers corresponding to the activated state. Interestingly also, a recent study by Tanner et al. has shown that the intracellular domain of the EGF receptor sterically hinders its EGF-induced dimerization rate (27). Moreover, they found that when the EGF is completely removed from the receptor, the activation persists for a long period of time (28). They hypothesized that this phenomenon might be due to interactions between subunits of the cytoplasmic domains, which impart significant stabilization of the dimeric state of the activated enzyme. A similar phenomenon might explain the residual persistent activation that we have seen, in the presence of ATP, with the ANP-pretreated wild type NPR-A (Fig. 4). Since this persistent activation was significantly higher in the case of NPR-A D435C , it is possible that the formation of the disulfide bridge might provide further stabilization of the activated dimeric complex.
Interesting studies have been recently made on molecular aspects of the erythropoietin receptor (EpoR) activation (29 -31). Indeed, a relative correlation between agonist potency and juxtamembrane orientation was deduced from the crystal structures of EpoR complexed with erythropoietin or with less potent synthetic agonists. The authors have proposed a model where the juxtamembrane domains of two subunits are brought into a closer proximity in response to agonist induction (32). This mechanism has been supported by in vivo studies using a protein fragment complementation assay (33). It should be noted that these studies are showing the intracellular domain as passively responding to the agonistic stimulation of the ectodomain. In addition, truncation of the intracellular domain of the EpoR is presumably not leading to spontaneous juxtamembrane dimerization (33). Therefore, the balance of constraints present in the EpoR seems to differ form that in NPR-A. Indeed, as we are showing here, the intracellular domain of NPR-A exerts a potent constraint on its ectodomain.
One of the main advantages of the current NPR-A mutant is that ligand induction of covalent dimerization occurs without the addition of any cross-linking reagent. Notably, the distance range between the ␣-carbons participating in a disulfide bond is Յ7 Å (34,35). The disulfide linkage of cysteines separated by more than 7 Å necessitates motion of the protein backbone (35). The rate of disulfide formation might be influenced by the inherent reactivity of cysteines, including accessibility of the sulfhydryl pair and the frequency of structural fluctuations that cause collisions between the reactive residues (35). This latter parameter might have influenced the rate of dimer formation in our in vitro assay at 4°C. It is noteworthy that studies with the EGF receptor have shown that a reduction of temperature diminishes the rate of EGF-induced receptor dimerization, which might be related to reduced structural transition rate to the active state (27). Nevertheless, the cellular induction of NPR-A D435C dimerization is detectable after a 30-min incubation at 37°C, indicating a good reactivity of Cys 435 under these conditions.
In our previous work, we mutated the Cys 423 of NPR-A into serine, which left the other cysteine 432 free to form an interchain disulfide bridge. This NPR-A C423S dimer displayed an elevated constitutive activity (15). We then hypothesized that agonists, during the activation process, are inducing a dimeric tightening of the juxtamembrane domain of NPR-A. However, we mentioned then that we could not exclude the possibility that the mutation had induced a conformational change that activates the receptor independently of the disulfide bridge. Following our study, Huo et al. (36) realized the double mutant NPR-A C423S,C432S , which eliminated both juxtamembrane cysteines. Since this mutant also displayed constitutive activity, they indicated that the disulfide bridge in NPR-A C423S was not responsible for the constitutive activation. Unfortunately, the reciprocal NPR-A C432S mutant was not provided in their study, which would have definitively completed this structure-function analysis and enabled them to unambiguously support their conclusion. They hypothesized that the disruption of the Cys 432 -Cys 423 linkage had altered the structure of the receptor in this region, which is essential for receptor signaling. In view of the results that we are presenting here, it is possible that this structural modification resulted in the increase of the spontaneous activating "positive" constraint of the ectodomain.
Recently, van den Akker et al. (37) have provided the crystal structure of the NPR-A extracellular domain. This structure is documented up to Asp 435 and shows a ligand-free dimer. Indeed, spontaneous noncovalent dimerization of NPR-A ECD has been shown to occur at very high protein concentrations typical of crystallization conditions (38). The C-terminal region (residues 423-435) forms a protruding irregular structure, which shows residue 435 as nonburied, as expected from our results. To complete the dimeric structure, the authors extrapolated the unresolved region 426 -435 of one of the monomers from the known structure of the other. From this reconstitution, they calculated a distance of ϳ14 Å between the C-␣ of the two chains at position 435 and concluded that their structure corresponded to an active dimer.
Our results complement these structural data. As already mentioned, in response to agonist binding, the C-␣ of residue 435 of the full-length receptor reaches a distance of Յ7 Å. On the other hand, the 14 Å provided by the structural data is not close enough to mediate the disulfide linkage at this position. This, taken together with the absence of ligand-induced covalent dimerization in the ECD D435C , suggests that the conformation of the liganded ECD does not exactly correspond to the agonist-activated state of the full-length receptor. Furthermore, according to our results, it is more likely that the ⌬KC spontaneously approaches the "activated" conformation. This suggests a significant influence of the transmembrane helix and/or membrane proximity on the tertiary and quaternary structure of the juxtamembrane region.
In conclusion, this study with NPR-A D435C allowed us to define several fundamental constraints that occur in this receptor. Also, it has provided us with the opportunity to define an activation model supported by the detection of a precise dimeric molecular interaction. Furthermore, this kind of information constitutes a significant asset to interpret the crystallographic data of NPR-A. In summary, these results definitively demonstrate that agonists induce a tight dimeric interaction in the juxtamembrane domain of NPR-A, an event that is closely related to its activation process.  (11). According to this model, the signal transduction occurs through five sequential steps as follows. 1) The binding of the natriuretic peptide to ECD induces a conformational change. 2) This modification corresponds to a signal that migrates through the TM domain.
3) The KHD responds to this signal by adopting a conformation that allows ATP binding. 4) ATP binding has two major effects in derepressing the guanylyl cyclase activity and increasing the off-rate of ANP from the receptor. 5) Subsequent desensitization results from reduction in the phosphorylation state of the KHD (13,14). The results obtained with ⌬KC WT and ⌬KC D435C indicate that the anchored ectodomain exerts by itself a positive constraint on juxtamembrane dimerization. However, the intracellular domain counteracts this tendency at basal state. Agonists are inducing a displacement of the balance toward activation through juxtamembrane dimerization. Following the intracellular response to juxtamembrane dimerization, ATP presumably binds to the KHD, which induces guanylyl cyclase activation and also leads to an increase of ANP dissociation rate.