Pharmacochaperones post-translationally enhance cell surface expression by increasing conformational stability of wild-type and mutant vasopressin V2 receptors.

Some membrane-permeable antagonists restore cell surface expression of misfolded receptors retained in the endoplasmic reticulum (ER) and are therefore termed pharmacochaperones. Whether pharmacochaperones increase protein stability, thereby preventing rapid degradation, or assist folding via direct receptor interactions or interfere with quality control components remains elusive. We now show that the cell surface expression and function (binding of the agonist) of the mainly ER-retained wild-type murine vasopressin V2 receptor GFP fusion protein (mV2R.GFP) is restored by the vasopressin receptor antagonists SR49059 and SR121463B with EC50 values similar to their KD values. This effect was preserved when protein synthesis was abolished. In addition, SR121463B rescued eight mutant human V2Rs (hV2Rs, three are responsible for nephrogenic diabetes insipidus) characterized by amino acid exchanges at the C-terminal end of transmembrane helix TM I and TM VII. In contrast, mutants with amino acid exchanges at the interface of TM II and IV were not rescued by either antagonist. The mechanisms involved in successful rescue of cell surface delivery are explained in a three-dimensional homology model of the antagonist-bound hV2R.

Some membrane-permeable antagonists restore cell surface expression of misfolded receptors retained in the endoplasmic reticulum (ER) and are therefore termed pharmacochaperones. Whether pharmacochaperones increase protein stability, thereby preventing rapid degradation, or assist folding via direct receptor interactions or interfere with quality control components remains elusive. We now show that the cell surface expression and function (binding of the agonist) of the mainly ER-retained wild-type murine vasopressin V 2 receptor GFP fusion protein (mV 2 R⅐GFP) is restored by the vasopressin receptor antagonists SR49059 and SR121463B with EC 50 values similar to their K D values. This effect was preserved when protein synthesis was abolished. In addition, SR121463B rescued eight mutant human V 2 Rs (hV 2 Rs, three are responsible for nephrogenic diabetes insipidus) characterized by amino acid exchanges at the C-terminal end of transmembrane helix TM I and TM VII. In contrast, mutants with amino acid exchanges at the interface of TM II and IV were not rescued by either antagonist. The mechanisms involved in successful rescue of cell surface delivery are explained in a three-dimensional homology model of the antagonist-bound hV 2 R.
Water homeostasis in mammals is regulated through arginine-vasopressin (AVP), 1 acting through the vasopressin V 2 receptor (V 2 R) expressed in the renal collecting duct (1). In Xlinked nephrogenic diabetes insipidus (NDI), the kidney shows a resistance to the action of AVP, caused by inactivating mu-tations of the human V 2 R (hV 2 R) gene (2). More than 150 different mutations have been described (for review, see Ref. 3), 50% of which are missense mutations resulting in the substitution of a single amino acid. Most of the hV 2 R mutants with a single amino acid exchange are retained within the ER and not transported to the cell surface (for review, see Ref. 3). Most likely, the amino acid exchanges result in improper folding of the mutant hV 2 Rs and subsequently prolonged association with molecular chaperones. For example, for the NDI mutant hR337X, a prolonged association with the ER-chaperone calnexin has been observed (4). Chaperone association prevents the aggregation of misfolded proteins, but also inhibits the exit of improperly folded proteins from the ER until correct folding is established.
Recently, it has been found that membrane-permeable antagonists not only inhibit receptor activation, but also promote cell surface expression of misfolded, ER-retained G proteincoupled receptors (GPCRs). This concept represents an intriguing new approach for the therapy of congenital diseases caused by mutations in genes encoding GPCRs. For the ER-retained rhodopsin mutant P23H (a frequent cause of autosomal-dominant retinitis pigmentosa), it has been shown in vitro that the inverse agonist 9-cis-retinal or the non-hydrolyzable inverse agonist 11-cis-7-ring-retinal promoted cell surface expression (5,6). Restoration of cell surface expression by antagonists or inverse agonists has also been found for various mutants of the hV 2 R (7) and the gonadotropin releasing hormone receptor (8).
The molecular mechanisms by which antagonists or inverse agonists promote cell surface delivery remain elusive. Antagonists may act on misfolded proteins by increasing protein stability, e.g. inhibiting their rapid degradation or by preventing misfolding and aggregation of the nascent proteins. Alternatively, although less likely, hydrophobic pharmacochaperones could interfere with components of the quality control system. To explore the mechanisms by which antagonists rescue intracellularly retained GPCRs, we used the wild-type murine V 2 R (mV 2 R), which is predominantly retained within the ER as an immature protein (9). In contrast, the hV 2 R is mainly located within the plasma membrane as a complex glycosylated protein. These differences are caused by a variant amino acid at the junction of the second transmembrane domain and the first extracellular loop (lysine 100 in hV 2 R, aspartate 100 in mV 2 R, Ref. 9). We show here that antagonists increase the conformational stability of the mV 2 R at a post-translational level via direct interactions. Antagonist-mediated cell surface delivery was also found for a subset of mutant hV 2 Rs, which showed amino acid exchanges at the C-terminal end of transmembrane regions TM I and TM VII. In contrast, mutant hV 2 Rs with amino acid exchanges at the interface of TM II and TM IV did not show antagonist-mediated cell surface delivery, most likely because of more severe folding defects. The mechanisms involved in successful antagonist-mediated restoration of cell surface delivery are explained in a three-dimensional homology model of the antagonist-bound hV 2 R.
Plasmids-The plasmids encoding mV 2 R⅐CFP and mV 2 R⅐YFP fusion proteins are derivatives of the plasmid mV 2 R⅐GFP described previously (9). Plasmids pcDNA3⅐V 2 R and pmV 2 R⅐A3 encoding the wild-type hV 2 R and mV 2 R were also described previously (9,10). The plasmid encoding wild-type and dominant-negative dynamin I were kindly provided by Dr. S. Schmid (La Jolla, CA) and plasmids ER⅐CFP, Endo⅐YFP, and Mem⅐YFP were from BD Biosciences. The plasmids encoding the mutant hV 2 Rs (hL62P, h⌬LAR62-64, hH80R, hW164R, hK100D, hD136A, hS167L, hS167T, hC319Y, hP322S, hW323H, hF328A, hD368K/S371X) were generated by site-directed mutagenesis of the plasmids pcDNA3⅐V 2 R and phV 2 R⅐GFP. The sequence of all plasmids was verified by DNA sequencing using the FS Dye Terminator kit (PerkinElmer Life Sciences). The nucleotide sequences of oligonucleotides used are available upon request.
Cell Culture and Transfection-HEK293 cells were cultured in Dulbecco's modified Eagle's medium supplemented with 10% heat-inactivated fetal calf serum, 100 international units/ml penicillin, and 100 g/ml streptomycin at 37°C in a humidified atmosphere of 95% air and 5% CO 2 . Transient transfection was carried out with LipofectAMINE or with FuGENE 6 essentially as described (9,11). For co-transfections of plasmids encoding the mV 2 R or the hV 2 R mutants with plasmids encoding wild-type or mutant dynamin, the ratio of DNA was 0.7/0.3. In all experiments cells were analyzed 16 -48 h after transfection.
Incubation of Cells with Non-peptide Antagonists-The effects of the non-peptide antagonists SR121463B and SR49059 were studied in LSM, biotinylation, immunoblotting, and [ 3 H]AVP binding experiments. If not described otherwise, cells were incubated 6 h after transient transfection with SR121463B and SR49059 (both at 1 M) for additional 16 h. SR49059-treated cells were washed twice with KRH (125 mM NaCl, 3 mM KCl, 1 mM NaH 2 PO 4 , 1.2 mM MgSO 4 , 2.4 mM CaCl 2 , 22 mM NaHCO 3 , 5.5 mM glucose, 10 mM Hepes) for 2 min at 4°C or 37°C and then further investigated. For the inhibition of protein synthesis HEK293 cells were incubated with 20 g/ml cycloheximide or 10 g/ml puromycin for 30 min or 2 h prior to the treatment with SR49059 or SR121463B (both at a final concentration of 1 M) for up to 6 h.
Immunoblots and Cell Surface Biotinylation Experiments-Preparation of membranes of HEK293 cells transiently expressing the mV 2 R or mutant hV 2 Rs and the procedure for immunoblotting were as described previously (9). Cell surface biotinylation experiments were performed as described previously (12). In brief, HEK293 cells transiently expressing GFP fusion proteins grown on 60-mm Petri dishes were incubated with sulfo-NHS-biotin in PBS (0.5 mg/ml, 30 min, 4°C). The reaction was terminated with 50 mM NH 4 Cl in PBS (10 min, 4°C). Cells were then washed three times with PBS and incubated with lysis buffer (1% Triton X-100, 0.1% SDS, 50 mM Tris-HCl, 150 mM NaCl, 1 mM Na-EDTA, pH 8.0; 1 h, 4°C). Insoluble debris was removed by centrifugation (20 min, 20,000 ϫ g). Biotinylated proteins were precipitated with NeutrAvidin-agarose beads (3 min, 17,000 ϫ g), and finally solubilized with Laemmli buffer. Biotinylated V 2 R⅐GFP was detected in immunoblots with polyclonal GFP antibodies and alkaline phosphatase-conjugated anti-rabbit antibodies as described (12).
[ 3 H]AVP Binding Experiments with Membrane Preparations and with Intact Cells-[ 3 H]AVP binding experiments with intact HEK293 cells were essentially performed as described (9). In brief, cells treated without or with SR49059 were washed twice with ice-cold KRH, pH 7.4 for 1 min followed by an incubation with [ 3 H]AVP in PBS in the absence (total binding) or presence of 10 M unlabeled AVP (unspecific binding) at 4°C for 2 h. After washing and lysis with 0.1 nM NaOH, radioactivity was determined in a liquid scintillation counter. In control experiments, we verified that this procedure was sufficient to remove SR49059 from the mV 2 R and the hV 2 R completely. However, in the case of SR121463B treatment, different washing procedures including high and low salt buffers with different pH values (ranging from pH 6.8 to 2.5) removed only a small and invariable fraction (Ͻ10%) of the antagonist from the mV 2 R and hV 2 R.
The procedure for membrane preparation of transiently transfected HEK293 or primary-cultured inner medullary collecting duct (IMCD) cells and [ 3 H]AVP binding experiments with membrane preparations were carried out as described previously (9,13).
LSM Analysis, LSM-FRET Imaging, and Fluorescence Recovery after Photobleach-Coverslips with HEK293 cells transiently expressing the different GFP fusion proteins were mounted in a temperable insert (Zeiss, Jena Germany) and analyzed with an LSM 510 META system using an Axiovert 135 microscope equipped with Plan-Apochromat 63ϫ/ 1.4 and PlanNeofluar 100ϫ/1.3 objectives (all from Zeiss). GFP and YFP were investigated at exc ϭ 488 nm and em Ͼ515 nm. The plasma membrane was visualized after addition of 20 l of trypan blue (0.05% in PBS) at exc ϭ 543 nm and em Ͼ 590 nm. The thickness of optical sections was between 0.3 and 0.6 m. For quantitative analysis of antagonist-mediated cell surface expression, the plasma membrane identified in the trypan blue image was marked as region of interest (ROI) and subsequently transferred to the GFP image. In the GFP image a second ROI representing the cell interior was set, and the average fluorescence intensities in the plasma membrane and the cell interior were determined. For each single cell, the ratio of fluorescence in the plasma membrane and the cell interior was calculated.
For FRET analyses, HEK293 cells grown on coverslips were transiently co-transfected with mV 2 R.CFP and mV 2 R.YFP. LSM-FRET imaging and fluorescence recovery after photobleaching were performed as described in detail previously (11,14). LSM-FRET imaging was performed with the LSM510 META and for fluorescence recovery after photobleaching we used an inverted microscope (Axiovert 100; Zeiss) equipped with a Plan-Apochromat 63ϫ/1.4 objective, a monochromator (Polychrome II; TILL Photonics, Grä felfing, Germany) and a cooled CCD camera (Imago; TILL Photonics). For excitation a dual reflectivity dichroic mirror (Ͻ460 nm and 490 -530 nm; Chroma Technology, Rockingham, VT) and for emission band pass filters of 460 -500 nm (CFP) or 535-580 nm (YFP) were used.
Molecular Modeling of Antagonist-bound hV 2 R-The initial threedimensional structure of the transmembrane helices of the hV 2 R was established on the basis of the three-dimensional structure of bovine rhodopsin (15). The construction of the complete hV 2 R model has been described previously (9,16). The energetically preferred conformations for SR49059 and SR121463 were selected from searches of their conformational spaces by the random search module in the Sybyl 6.9 package (TRIPOS Inc., St. Louis, MO). The starting orientation for docking of the V 1 R-specific antagonist SR49059 to the hV 2 R model was comparable to the reported SR49059-bound V 1A R model (17). In the case of SR121463, the initial orientation was derived after superposition of the common ring systems of SR121463 and SR49059. After minimization of the starting complex using the AMBER 5.0 force field (18), molecular dynamics simulations were performed at 300 K for 500 ps, where only hydrogen bonds of the TM backbones maintaining the TM helices were restrained. Low energy conformations of the last 50 ps were compared.

Investigated Wild-type and Mutant V 2 Rs and V 2 R⅐GFP Fusion
Proteins-To clarify the mechanisms of antagonist-promoted restoration of cell surface expression, we investigated the predominantly ER-retained mV 2 R (9). In addition, eight NDI-causing hV 2 R mutants ( Fig. 1, gray boxes) (19 -23) and several in vitro mutants ( Fig. 1, white boxes) were included in this study. The mutants hW164R and hC319Y represent novel NDI-causing mutations. The mutant hW164R was identified in a patient with severe NDI, whereas patients with the mutant C319Y suffered from partial NDI, indicating that this mutant has retained residual activity. The in vitro mutant hD368K/ S371X codes for a hV 2 R with an engineered dibasic ER-retrieval motif at the very C terminus of the hV 2 R. For all mutants with the exception of hD368K/S371X, plasmids were generated, which encode C-terminal GFP fusion proteins suitable for LSM of living cells and immunoblotting using GFP antibodies. In the case of the hD368K/S371X mutant, an Nterminally Myc epitope-tagged receptor was used (10).
Retention of V 2 R Mutants in the ER-HEK293 cells transiently expressing the wild-type mV 2 R or different mutant hV 2 Rs show little or no [ 3 H]AVP binding ( Fig. 2A). For mV 2 R, hK100D and hF328A binding of [ 3 H]AVP ranged between 8 and 16% of that for the wild-type hV 2 R. For all other mutants, including the hD368K/S371X mutant, [ 3 H]AVP binding was less than 3% of that of the wild-type hV 2 R. LSM of transiently transfected HEK293 cells revealed that wild-type mV 2 R⅐GFP and mutant hV 2 R⅐GFPs were predominantly located within the ER (Fig. 2B). The staining patterns were indistinguishable from that of an ER-targeted cyan fluorescent protein derivative (ER⅐CFP).
Membrane-permeable Receptor Antagonists Promote Cell Surface Delivery of the mV 2 R at a Post-translational Level-HEK293 cells transiently expressing the mV 2 R⅐GFP were treated with the non-peptide vasopressin V 1 receptor (V 1 R)antagonist SR49059 or the V 2 R-selective non-peptide antagonist SR121463B. Both antagonists (1 M, 16 h) restored cell surface expression (Fig. 3A). In contrast, ET A and ET B receptor-selective non-peptide antagonists BQ123 and BQ788 (both 10 M for 16 h) had no effect on the subcellular distribution of the mV 2 R (Fig. 3A, panels d and e). Co-transfection of the mV 2 R⅐GFP with dominant-negative K44A⅐dynamin did not increase the cell surface delivery of the mV 2 R⅐GFP, indicating that SR121463-and SR49059-promoted cell surface expression was not caused by an inhibition of internalization. The antagonist effects were preserved when cells were preincubated with cycloheximide (20 g/ml) for 30 min prior to the application of SR121463B (1 M for 6 h; Fig. 3A, panel g). Similar results were obtained when cycloheximide was administered for 2 h prior to the addition of antagonists or when puromycin (20 g/ml) was used instead of cycloheximide (data not shown). These results were further confirmed in [ 3 H]AVP binding experiments. Since only SR49059, but not SR121463B can be removed from the mV 2 R and hV 2 R by washing (see "Experimental Procedures"), the experiments were performed with SR49059. Cells were treated for up to 6 h with buffer, SR49059, cycloheximide or the combination of SR49059 and cycloheximide. Membranes of cells treated with cycloheximide alone or left untreated did not differ in [ 3 H]AVP binding, whereas membranes of cells treated with SR49059 or with cycloheximide and SR49059 revealed a 3-fold increase in [ 3 H]AVP binding (Fig. 3B). These results suggest that the ER-retained wild-type mV 2 R is not rapidly degraded, but remains in the ER for longer periods. Thus, the antagonists do not simply function by preventing degradation. Rather the antagonists promote the proper folding, e.g. increase conformational stability of the already synthesized mV 2 R retained in the ER.
To test whether SR49059 also promotes cell surface delivery in cells that express the V 2 R endogenously, we studied primary cultured rat IMCD (13). The V 2 Rs of rat and mouse are highly homologous and share the aspartate residue at position 100. Both receptors differ only by six amino acids (five in the extreme N terminus, one in the third intracellular loop). Treatment of rat IMCD cells with 1 M SR49059 for 16 h increased specific [ 3 H]AVP binding 2.5-fold (Fig. 3C), indicating that antagonists also promote cell surface expression of the endogenous rat V 2 R.
SR121463B Promotes Maturation of the Wild-type mV 2 R⅐ GFP-In immunoblots with membrane preparations of HEK293 cells transiently expressing hV 2 R⅐GFP, bands at 55 and 70 -75 kDa were observed (Fig. 4). The broad band at 70 -75 kDa corresponds to the mature, complex glycosylated hV 2 R⅐GFP, whereas the band at 55 kDa represents the immature, core-glycosylated hV 2 R⅐GFP (16). Like the hV 2 R⅐GFP, the mV 2 R⅐GFP yields bands at 75 and 55 kDa, representing the mature and immature receptor, respectively (see Fig. 4). The complex glycosylated mV 2 R⅐GFP (75-kDa band) was less abundant than the corresponding one of the hV 2 R⅐GFP. Preincubation of cells with 1 M SR121463B for 16 h prior to membrane preparation did not qualitatively change the hV 2 R⅐GFP and mV 2 R⅐GFP patterns; however, the intensity of the band representing the mature mV 2 R⅐GFP (75 kDa) was increased. These results were confirmed in cell surface biotinylation assays. HEK293 cells transiently expressing hV 2 R⅐GFP or mV 2 R⅐GFP were treated with 1 M SR121463B or left untreated for 16 h, followed by an incubation with membrane-impermeable sulfo-NHS-biotin. Biotinylated proteins were isolated and analyzed in immunoblot experiments using a polyclonal GFP antibody. Both, the wild-type hV 2 R⅐GFP and the mV 2 R⅐GFP yielded a band at 75 kDa corresponding to the mature, complex glycosylated form. The intensity of the 75 kDa band was slightly enhanced following SR121463B treatment in the case of the hV 2 R⅐GFP. In the case of the mV 2 R; however, the antagonist treatment resulted in a strong increase in the intensity of the 75-kDa band (Fig. 4). The data demonstrate that SR121463B promotes both maturation and cell surface delivery of the wild-type mV 2 R.
Quantitative LSM-In displacement binding analysis with [ 3 H]AVP we found that the affinity of the mV 2 R for SR49059 was about 30-fold lower than for SR121463B (K i values of the mV 2 R for SR49059 and SR121463B were 618 Ϯ 317 nM and 17.2 Ϯ 6.7 nM, respectively). In order to compare the potencies and time courses of both molecules to promote cell surface expression of the mV 2 R, we analyzed transiently transfected HEK293 cells for plasma membrane delivery of the mV 2 R⅐GFP in the presence of SR121463B or SR49059 (1 M each). Every hour, images of GFP and trypan blue (which labels the plasma membrane) were taken. The intensity of GFP fluorescence from the cell interior and the plasma membrane were quantified, and the ratio was calculated. Both antagonists promoted cell surface delivery to an almost identical extent (6 -7-fold), but revealed differences in their half-times (t h ) for maximal increase in cell surface delivery (t h of SR121463-and SR49059-promoted cell surface delivery were 4.95 Ϯ 0.11 h and 6.48 Ϯ 0.14 h; Fig. 5A). In concentration response analyses we found that the half-maximal concentrations (EC 50 ) required for SR121463B-and SR49059-promoted cell surface delivery of the mV 2 R were 22.6 Ϯ 6.4 nM and 382 Ϯ 73 nM, respectively. It is of note that the EC 50 values were very similar to the K i values of both antagonists. To validate the data from quantitative LSM, we also performed [ 3 H]AVP binding experiments. Incubation of HEK293 cells transiently expressing the mV 2 R⅐GFP with SR49059 for up to 16 h resulted in a 10-fold increase in [ 3 H]AVP binding sites (Fig. 5B). Interestingly, the half-time for SR49059-mediated increase in [ 3 H]AVP binding was about 9.5 h, which was significantly slower than that observed in quantitative LSM (compare with Fig. 5A). The reason for this difference is not known. One explanation could be that LSM does not allow us to distinguish between mV 2 R-containing vesicles in close proximity to the plasma membrane (Ͻ200 nm) and mV 2 Rs in the plasma membrane. Alternatively, not all receptors inserted into the plasma membrane are fully functional, but could require other protein or lipid contacts, which are established more slowly.
Oligomerization Is Not Sufficient for the Exit of the mV 2 R from the ER-The hV 2 R and the hV 1A R form dimers/oligomers within the ER. It has been suggested that oligomerization of vasopressin receptors is essential for the transport to the Golgi apparatus (24). Because ER retention of the mV 2 R could be caused by a lack of oligomerization, we analyzed the oligomeric state of mV 2 Rs retained in the ER and expressed at the cell surface after treatment with SR121463B by FRET experiments. In the first series of experiments, we studied HEK293 cells transiently co-expressing mV 2 R⅐CFP and mV 2 R⅐YFP in LSM-FRET analysis. The CFP, YFP, and FRET signals were calculated by linear unmixing as descried previously (14). In the absence or presence of SR121463B, mV 2 R⅐CFP and mV 2 R⅐YFP formed dimers or oligomers within the ER and at the plasma membrane (Fig. 6A). These results were confirmed in quantitative FRET experiments, in which the recovery of the donor fluorescence after photobleaching of the acceptor was determined. Here, the FRET efficiency for the ER-retained mV 2 R (untreated cells; Fig. 6B) was slightly higher than that of the mV 2 R expressed in the plasma membrane (SR121463Btreated cells; Fig. 6C). This difference could be caused by antagonist-mediated changes in the receptor conformation. In fact, when FRET efficiencies were determined for mV 2 R dimers/oligomers after a brief incubation with SR121463B (1 M for 30 min at 37°C; did not restore cell surface delivery, compared with Fig. 5), FRET efficiencies similar to those of the mV 2 R in the plasma membrane were obtained (data not shown). The data suggest that the mV 2 R is retained in the ER, although assembly to dimers/oligomers has already occurred. Thus, the antagonists promote cell surface delivery of the mV 2 R at a level after receptor assembly to dimers/oligomers. SR121463B Restores Cell Surface Expression of Mutant hV 2 R⅐GFPs-In further experiments we tested several mutant hV 2 R⅐GFPs for their ability to undergo antagonist-promoted cell surface delivery. In transiently transfected HEK293 cells both antagonists restored cell surface delivery of the mutants hK100D and hC319Y (Fig. 7). In case of the mutants h⌬LAR62-64, hD136A, hS167T, hP322S, hW323H, and hF328A, cell surface delivery was only restored by SR121463B but not by SR49059. Similar to the mV 2 R, restoration of cell surface delivery was not abolished by cycloheximide (data not shown). In case of the mutants hL62P, hH80R, hW164R, and hS167L neither antagonist was capable of restoring cell surface delivery (Fig. 7). The LSM results were confirmed in immunoblot and cell surface biotinylation experiments (Fig. 8).
In cell surface binding experiments with SR49059-treated HEK293 cells transiently expressing the hC319Y mutant we observed only a small increase in [ 3 H]AVP binding when compared with untreated controls (Table I). This small increase in [ 3 H]AVP binding was unexpected, since SR49059 promoted the cell surface delivery of hC319Y to a similar extent as for mV 2 R ( Table I). The increase in [ 3 H]AVP binding was comparable to that of the hS167T or hD136A mutants, which showed no significant SR49059-mediated restoration of cell surface delivery in LSM and biotinylation experiments. Thus, it is likely that treatment with SR49059 yields transport-competent, but binding-defective hC319Y receptors.
Transport-competent Folding of hV 2  mutants h⌬LAR62-64, hS167T, hP322S, hW323H, and hF328A was restored by SR121463B only. Thus, we hypothesized that these mutants have a reduced affinity for SR49059. Since the hF328A mutant shows significant residual cell surface delivery in the absence of the antagonists (5-10% of the wild-type hV 2 R; see also Fig. 2), we used this mutant to determine its affinity for AVP, SR121463B, and SR49059. The K d value of the hF328A mutant for AVP was similar to that observed for the wild-type hV 2 R (K d values for the F328A mutant and the wildtype hV 2 R were 1.8 Ϯ 0.8 nM and 2.6 Ϯ 1.2 nM, respectively). The K i values of the hF328A mutant for AVP, SR121463, and SR49059 were 2.7 Ϯ 3.7 nM, 5.9 Ϯ 0.1 nM, and 134 Ϯ 31 nM, respectively, and thus very similar to those of the hV 2 R (5.3 Ϯ 4.1 nM, 7.3 Ϯ 4.5 nM, and 129 Ϯ 24 nM, respectively). Further we observed by immunocytochemistry that SR49059 changed the subcellular distribution of the hF328A mutant although it did not promote its cell surface delivery. In the absence of SR49059, the hF328A⅐GFP mutant revealed an ER-like distribution, which was not altered in the presence of bafilomycin A1, an inhibitor of the vacuolar H ϩ -ATPase. In cells treated with SR49059 for 13 h followed by an additional incubation with bafilomycin A1 for 3 h, the mutant was found mainly in the ER-Golgi intermediate compartment (ERGIC, Fig. 9). The results demonstrate that both antagonists can bind to the hF328A mutant, but only the SR121463B is able to restore the cell surface expression.
Homology Model of the Antagonist-bound hV 2 R-To gain insight into the interactions of the hV 2 R with the two antag-onists and to understand the potential mechanisms underlying antagonist-promoted folding, we established a computerassisted homology model of the SR121463B-and SR49059bound hV 2 R. The three-dimensional structure of rhodopsin was used as a template (15), which was refined by molecular dynamics simulations. Since SR49059 and SR121463B share a common structural core, consisting of an aryl-sulfonylindoline ring-system (Fig. 10A), we started with an orientation of SR121463B that was similar to that of SR49059 bound to the human vasopressin V 1A receptor (hV 1A R, Ref. 17). The aromatic interactions of the core ring system with Phe 225 (TM V), Tyr 300 , Trp 304 , Phe 307 , Phe 308 (all TM VI) found for SR49059 in the V 1A R also apply for SR121463B and hV 2 R. Here, the aryl-sulfonyl-indoline core of SR121463B interacts with the corresponding residues Phe 214 (TM V) and Tyr 280 , Trp 284 , Phe 287 , and Phe 288 (all TM VI) in the hV 2 R. The additional morpholinoethoxy-cyclohexane group of SR121463B most likely projects between TM II and TM VII and interacts mainly with the side chains of phenylalanine 307 (cyclohexane group) and valine 308 (morpholinoethoxy group) of TM VII ( Fig. 10B and Supplement 1). These interactions could also contribute to the V 2 R-selectivity of SR121463B since the corresponding amino acids isoleucine 330 and threonine 331 in the hV 1A R can only weakly interact, if at all with the cyclohexane and morpholinoethoxy groups.
The hV 2 R mutants, which can be rescued by SR121463 only harbor amino acid replacements which cluster at the C-terminal part of both TM I and TM VII, whereas in the mutants lacking antagonist-mediated rescue, amino acid replacements are found mainly at the interface of TM II and TM IV (Fig. 10B and Supplement 1).

DISCUSSION
In this study we show that the wild-type mV 2 R, which is predominately retained in the ER, reveals an almost complete restoration of cell surface delivery in the presence of the antagonists SR121463B or SR49059. However, antagonist treatment fails to increase cell surface delivery of the hD368/S371X mutant, which is most likely properly folded but retained in the ER via a dibasic retention signal at the very C terminus (25). In a previous study we have shown that aspartate 100 in the mV 2 R confers a low conformational stability, whereas the exchange of aspartate by lysine, found at the corresponding position in the mainly cell surface expressed hV 2 R increases conformational stability. The role of lysine 100 for proper folding may be explained by its participation in a hydrogen bond network formed between the side chains of Lys 100 (TM II), Glu 40 (TM I), and Asp 191 (2nd extracellular loop, Ref. 9). With an acidic aspartate at position 100 in the mV 2 R, this hydrogen bond network cannot be established. Instead a repulsion of Asp 100 , Asp 191 , and Glu 40 is likely. As a result the free energy between the properly folded, native, and the improperly folded, non-native states could be at a similar level, so that the mV 2 R switches between the different states. Both, substitution of aspartate 100 by lysine (mD100K, Ref. 9) or binding of an antagonist (this work) may enhance conformational stability of the mV 2 R, resulting in the exit from the ER. The fact that the EC 50 values for antagonist-promoted cell surface delivery of the mV 2 R were similar to their K i values, is strong evidence for the notion that the antagonists interact directly with the ERretained receptor via the binding cleft. In addition, the data obtained with cycloheximide-treated cells show that antagonists act on a post-translational level, suggesting that antagonists are unlikely to serve as folding templates for nascent proteins preventing misfolding.
For hV 2 R mutants, variable effects of antagonists on cell surface delivery were found. In the hV 2 R mutants studied, cell surface delivery was restored with the antagonists SR49059 and SR121463B (hK100D, C319Y) or only with SR121463B (h⌬LAR62-64, hS167T, hP322S, hW323H, hF328A). In addition, some hV 2 R mutants did not respond to either antagonist. In the hV 2 R mutants, which do not undergo antagonist-mediated cell surface expression (hH80R, hW164R, hS167L), the amino acid replacements affect residues, which are of major structural importance for receptor folding. For example, tryptophan 164 in TM IV is the most conserved residue among class A GPCR, and at position 167 in TM IV, only small amino acids, such as alanine, serine or cysteine are found (26). In agreement with these structural considerations, the introduction of a methyl group at position 167 (hS167T) causes only a slight distortion of the interaction with the neighboring valine 121 (TM III) and allow rescue, whereas the insertion of an isopropyl group (hS167L) results in a severe disturbance of receptor folding, and a rescue is not possible. Similarly, replacement of highly conserved tryptophan 164 may lead to a severe folding defect.
The mutant hV 2 Rs which show cell surface delivery only in response to SR121463B (h⌬LAR62-64, hS167T, hP322S, hW323H, hF328A), point to structural properties of the antagonists, which are of crucial importance for the rescue activity. These mutants form a class characterized by amino acid exchanges at the C-terminal ends of TM I and TM VII (Fig. 10B). The amino acid replacements most likely cause a slightly altered arrangement of the transmembrane helices, which results in ER retention. Although the substitutions are distant from the ligand binding site, SR121463B could promote folding by its extended morpholinoethoxy group. The latter most likely binds between TM II and TM VII of the hV 2 R and via side chains interactions with TM II and TM VII forces a proper rearrangement of the transmembrane helices either pushing or attracting TM I, TM II, and TM VII on the extracellular side. As a consequence, the transmembrane helices are forced together on the intracellular side and slightly disturbed helixhelix interactions are corrected. Consequently, SR49059, which lacks the extended morpholinoethoxy group, does not restore cell surface delivery of these mutants. It is unlikely, that the mutants lack affinity for SR49059, since the hF328A mutant, a small fraction of which is expressed at the cell surface in untreated cells, exhibits the same affinity to SR49059 as the wild-type hV 2 R. Evidence that SR49059 interacts with the hF328A mutant even in its ER-retained form is derived from the fact that SR49059 promotes transport of the hF328A mu-tant from the ER to the ERGIC in bafilomycin A1-treated cells. Thus, the SR49059-bound hF328A mutant adopts a conformation, which can exit the ER, but underlies ER retrieval. The results also show that the ERGIC functions as a further site of FIG. 8. SR121463B promotes maturation and cell surface expression of ER-retained hV 2 R⅐GFP mutants. HEK293 cells transiently expressing several NDI-causing and in vitro V 2 R⅐GFP mutants were incubated with SR121463B (1 M) for 16 h. A, cells were subjected to membrane preparation as described under "Experimental Procedures." Immunoblot analysis was performed as described in legend to Fig. 4. B, biotinylated membrane proteins following cell surface biotinylation were detected as described in Fig. 4 (for details see "Experimental Procedures"). The immunoblots are representative of at least three independently performed experiments. #, mature receptor; *, immature receptor.  quality control for proteins, which exit the ER, but have not established the native conformation. 2 Analysis of functional activity following treatment with antagonists could only be preformed for the mV 2 R and the hC319Y mutant, since only these V 2 Rs showed cell surface expression with SR49059, which in contrast to SR121463B can be removed by washing. While SR49059-promoted cell surface delivery of the mV 2 R was paralleled by a strong increase in [ 3 H]AVP binding, this was not the case for the hC319Y mutant. Here, SR49059 yielded a mainly binding-defective receptor population. Thus, it is likely that some hV 2 Rs mutants not only display a reduced conformational stability, but also a functional impairment. Similar results have been described for the hV 2 R mutant hR202C. This mutant is delivered to the cell surface to the same extent as the wild-type hV 2 R, but only a small fraction of these receptors (Ͻ5%) revealed functional activity (27). In the case of the ER-retained hV 2 R mutant h⌬V278, SR121463B restored cell surface transport. However, the SR121463B-bound h⌬V278 does not show complex glycosylation and is targeted to the apical instead to the basolateral plasma membrane (28). Similar observations have also been made for other mutant proteins, for which small moleculeassisted folding was demonstrated. For example, small, membrane-permeable peptide and non-peptide ligands increased the conformational stability of several p53 mutants, finally resulting in a restoration of DNA binding activity (29,30). However, the stabilized mutant p53 proteins did not inhibit growth (as does the wild-type p53) and restored apoptosis only partially (30).
While the investigated antagonists can enhance conformational stability of mutant hV 2 Rs, thereby enabling their exit from the ER, they fail to restore complete functional activity. Thus, further studies have to focus on the identification of pharmacochaperones, which not only restore cell surface delivery, but also functional activity. FIG. 9. Subcellular localization of the hF328A⅐GFP mutant in the absence and presence of SR49059 and bafilomycin A1. HEK293 cells transiently expressing the hF328A⅐GFP mutant were incubated without or with 1 M SR49059 for 16 h and, if indicated, additionally treated in the last 3 h with bafilomycin A1. The cells were then fixed, permeabilized, and incubated with a monoclonal ERGIC-53 antibody and a Cy3-conjugated goat anti-mouse secondary antibody. The mutant hF328A⅐GFP (green) shows a typical ER pattern in the presence or absence of SR49059 and did not reveal significant colocalization with ERGIC-53 (red). When cells incubated with SR49059 were additionally treated with bafilomycin A1 an extensive co-localization of hF328A⅐GFP with ERGIC-53 close to the nucleus was found. The figures are representative of three independent experiments, in which at least 30 different cells were analyzed, respectively. Bars, 5 m.
FIG. 10. Homology model of the SR121463B-and SR49059bound hV 2 R. A, comparison of the three-dimensional structure of the V 1 R-selective and the V 2 R-selective antagonists SR49059 (yellow) and SR121463B (magenta). B, homology model of the antagonist-bound hV 2 R based on the three-dimensional structure of bovine rhodopsin. Shown are structural frames as a transmembrane view of the superimposed SR121463B-and SR49059-bound hV 2 R conformation of molecular dynamics-guided docking. Seven transmembrane helices are labeled TM I to TM VII. For clarity, TM V-VII are cyan, TM I-IV are gray. The short helix (H8) in the C terminus following TM VII is also depicted in gray. Green, amino acid exchanges sensitive to SR121463B/SR49059mediated rescue; orange, amino acid exchanges sensitive to SR121463B-mediated rescue; red, amino acid exchanges insensitive to antagonist-promoted rescue.